Spallation Neutron Source Drift Tube Linac Water Cooling and Resonance Control System Final Design Report

Transcription

1 (SNS DE0001-R01) Spallation Neutron Source Drift Tube Linac Water Cooling and Resonance Control System Final Design Report (SNS DE0001-R01) by: J. D. Bernardin, R. Brown, S. Brown, G. Bustos, M. Crow, J. Gioia, W. Gregory, M. Hood, J. Jurney, D. Katonak, Z. Konecni, P. Marroquin, I. Medalen, A. Owen, L. Parietti, and R. Weiss Mechanical Engineering Group Spallation Neutron Source Division Los Alamos National Laboratory April 4, 2001

2 TABLE OF CONTENTS 1. Introduction Project Scope, Deliverables, and Design Criteria Drift Tube Linac (DTL) Water Cooled Environment DTL Resonance Control Basic Philosophy Water Cooling Resonance Control Technique DTL Cooling Requirements Mechanical and Electrical Interfaces Comments and Action Items from Preliminary Design Review DTL Water Cooling and Resonance Control System Design Summary Water System Layout Manifolding on the RF Structure Water Skid Transfer Lines Facility Chilled Water Source Instrumentation and Controls Water Cooling Analyses DTL Water Cooling Loops - Lumped-Parameter Flow Network Modeling DTL RF Structure Cooling Loop Design Goals Design Specifications Tank 3 Global Model Description Drift Tube Sub-model Description Drift Tube Sub-model Results Slug Tuner Sub-model Description Slug Tuner Sub-model Results Post Coupler Sub-model Description Post Coupler Sub-model Results Dipole Magnet Sub-model Description Dipole Magnet Sub-model Results Tank Wall Sub-model Description Tank Wall Sub-model Results End Wall Sub-model Description End Wall Sub-model Results Drive Iris Sub-model Description Drive Iris Sub-model Results Tank 3 Global Model Design Studies/Results Summary DTL Water Skid Design Goals Design Specifications Model Description Design Studies/Results Summary ii

3 3.1.3 SINDA/FLUINT Uncertainty Analysis DTL Water Cooling Loops Stability and Response Modeling Design Goals Model Description Design Studies/Results Summary Mechanical Design Introduction Engineering Codes and Drawing Standards Plumbing Materials Radiation Damage Assessment Material Selection for Design General Manuracturing and Assembly Techniques DTL RF Structure Water Manifolds and Lines Piping and Instrumentation Diagrams Major Components Assemblies Water Skid Piping and Instrumentation Diagrams Performance Specifications Vibration Isolation Noise Level Requirements Major Components and Specifications Structure Plumbing Pump Heat Exchanger Control Valves Heater Water Purification System System Performance Parts Database and Naming Convention Water System Purity Introduction Water Purification Techniques Particle Accelerator Specific Issues Operating Parameter Specifications Water Purification System Design Prototype Design and Testing Facility-related Issues Instrumentation and Controls Local Controls Introduction and Design Requirements Instrumentation and Control System Architecture Control Methodology and Logic Safety Interlocks and Equipment Protection iii

4 6.1.5 Signal List Global Controls Interfaces Configuration Interlocks Operator Interface Archiving Alarm Management SNS Facility Interfaces Klystron Gallery Linac Tunnel Chases Safety Hazard Analyses and Protective Measures Personnel Safety Procurement Water Skid Procurement Water Manifold Procurement Hardware Costs Delivery and Inspection Quality Assurance Assembly, Installation, and Certification Plans Operation, Reliability and Maintenance Operation Reliability Maintenance Decommissioning Project Summary and Schedule Project Summary and Ongoing Work Cost Summary Schedule Appendix A ASME B31.3 Code Tables Appendix B Engineering Drawings Appendix C Water Skid Specifications Appendix D Hardware Costs Appendix E Parts Database/Device Name List for DTL Tank Appendix F DTL Drift Tube Heat Load and Cooling Requirements Appendix G Orifice Plate Spreadsheet Calculations for DTL Drift Tubes Appendix H Flexible Tubing Data Appendix I Procurement Specification for the Water Purification System Appendix J Resin Handling and Disposal Plan Appendix K Preliminary SystemView Calculations References iv

5 1.0 Introduction The Spallation Neutron Source (SNS) is an accelerator-based neutron research facility being designed for scientific and industrial research and development. Specifically, SNS will generate and use neutrons as a diagnostic tool, much like X-rays, for medical purposes as well as physical, chemical, biological, and material sciences. The SNS will produce neutrons by bombarding a heavy metal target with a high-energy beam of protons, generated and accelerated with a linear particle accelerator, or linac. To effectively accelerate the protons, the linac uses high electrical fields, established in copper resonance cavities with Radio Frequency (RF) energy. The low energy end of the SNS linac consists of a room temperature copper structure that dissipates roughly 60-80% of the RF energy in the form of heat. To deal with this waste heat, a water cooling system has been designed as an integral par of the room-temperature linac. The water cooling system is responsible for removing the undesired RF waste heat and maintaining the electrical resonance of the copper RF structures by manipulating their operating temperature. One of the two room-temperature accelerating structures in the SNS Linac, is the Drift Tube Linac (DTL). The DTL accelerates the SNS proton beam from 2.5 MeV to 87 MeV, before injecting it into the second copper structure, the Coupled Cavity Linac (CCL). The basic design criteria and features of the DTL can be found in [1.1, 1.2] and are summarized briefly in Section 1.4 of this report. A preliminary design for the Drift Tube Linac (DTL) water cooling and resonance control system was completed in August of 2000, and documented in [1.3]. This report summarizes the final design of that DTL water cooling and resonance control system. 1.1 Project Scope, Deliverables, and Design Criteria The complete project scope associated with the DTL Water Cooling and Resonance Control System includes the design, analyses, fabrication, assembly, installation, testing, and certification of the cooling system components. The efforts associated with this project scope include performing final design engineering calculations and developing corresponding engineering drawings, preparation of 5

6 procurement packages, liaison with vendors and participation in assembly, installation, and testing at Oak Ridge National Laboratory (ORNL). This report covers the final design efforts, based on a preliminary design outlined in [1.3]. To develop a functional, reliable, and affordable water cooling and resonance control system, the following final design deliverables were identified [1.2]: 1. Revision of preliminary design aspects as directed by LANL SNS-PO following the DTL Water Cooling and Resonance Control System PDR [1.4]. 2. Completion of all engineering calculations and supporting R&D experimentation. 3. Completion of all water purification studies. 4. Completion of P&IDs as well as assembly and detail drawings for the water skids, distribution manifolds, support fixtures, etc. 5. Complete design of instrumentation, controllers, and software for the local control system and global control system integration plans. 6. Specifications and procedures for water cooling system material preparation, cleaning, handling, and shipping. 7. Completion of detailed mechanical drawings and procurement plans with bill of materials for procurement of off-the-shelf items and fabrication plans for specialized components. 8. Completion of assembly, installation, testing, and certification/quality assurance plans. Table 1.1 lists the general design criteria that were applied to the SNS DTL water cooling and resonance control system design. Each criterion has a brief description and a weighting factor associated with it. The weighting factor is intended to give a measure of the criterion s importance in the overall DTL water cooling and resonance control system design, and consequently, assist the engineering design team in selecting between various design alternatives. An example of the use of the design criteria and weighting factors in assessing two different design alternatives can be found in [1.2]. 6

7 Table 1.1. SNS DTL water cooling and resonance control system design criteria. Design Criteria Weighting Description Factor Functionality 5 Resonance control criteria must be met (i.e. heat removal, temperature stability, etc) Water cooling system hardware must integrate with support structure Water system must be resilient to react to design and operational beam line changes Safety 5 Proper controls and safety features, following appropriate DOE guidelines, should be employed to protect personnel Procurement, Fabrication, Assembly Durability/ Reliability and the beam line (equipment and operation) 3 Design with standard, off-the-shelf parts Avoid using exotic materials Assembly and maintenance issues should be incorporated in the design to ensure consistency with other subsystems (i.e, support structure, vacuum system) 4 Best engineering practices should be employed in the design of the water cooling system to maximize its availability and reliability. Pumps should be selected for 30 year lifetime and have a 5 year maintenance period. Pumping redundancy should be considered in order to meet the reliability and duty factor requirements of the accelerator. Cost 4 Optimize functionality to minimize procurement, fabrication and assembly costs to fit within the allocated budget (based on the conceptual design). Maintainability 3 Pumps and hardware should be accessible for maintenance/replacement with minimal impact on beam down-time Consistency 2 Every effort should be made to use the same type of water system components throughout the Linac. In addition, these components should be consistent with those used elsewhere in the SNS facility (i.e., RF systems, RFQ, storage ring, target, etc) 5 = very important, 1 = least important 1.2 Drift Tube Linac (DTL) Water Cooled Environment The SNS linear particle accelerator, or linac, is comprised of three main structures including the Drift Tube Linac (DTL), the Coupled Cavity Linac (CCL), and the Super Conducting Linac (SCL), as displayed in Figure 1.1. The first proton accelerating structure following the ion injector and RFQ, is the DTL. The MHz Alverez DTL [1.5], is used to accelerate the H- beam from 2.5 MeV to 86.8 MeV. The SNS DTL is comprised of six tanks, the first of which is roughly 4 m in length, and the remainders are approximately 6 m in length. Tank 1, as shown in Fig. 1.2{a), is made up of 2 seamless copper-plated, carbon-steel cylinders that are bolted together with RF and vacuum seals 7

8 DTL CCL SCL1 SCL2 Proton Beam DTL CCL Tank 1 Module 1 Tank 6 Module m 6 R F cooling skids, 1 per tank for drift tubes, for slug tuners, post couplers, drive loops, and tank walls 56.5 m 4 modules, 12 segments/module, 8 cavities/segment 4 R F cooling skids (1/module) 1 EMQ cooling skid (1/ four modules) Figure 1.1. General layout of the SNS Linac and basic descriptions of the water cooling systems. 8

9 (a) Figure 1.2. The Drift Tube Linac R F structure, support structure, and main vacuum pumps for (a) tank #1 and (b) tank #2.. (b) 9

10 at each joint, and tanks 2 through 6 are made up of 3 sections each, as shown in Fig 1.2(b). The RF structure provides a stable platform for an array of drift tubes, post couplers, and slug tuners, all used to shape and tune the structure to maintain precise resonance and optimal acceleration of the proton beam. These components, and other design details, are shown in the cut-away view of tank #1 in Fig A more detailed description of these components and their functionality can be found in [1.1] and [1.5]. Table 1.2 summarizes the number of tank sections, cells, drift tubes, post couplers, slug tuners, and drive irises within each of the DTL tanks. Table 1.2. Summary of DTL tank component distributions. DTL Tank # # of cells # of post couplers Tank section # # tank wall cooling channels # of drift tubes # of endwall noses (half of drift tube) # of slug tuners # of drive irises 1 1 & , 2, & , 2, & , 2, & , 2, & , 2, &

11 Figure 1.3. Cut-away details of DTL tank #1. 11

12 Under normal operation (beam on), approximately 60 to 80% of the RF power is dissipated in the DTL copper structural components. The dissipated power causes thermal distortions (i. e., shape change) which result in a frequency shift of the RF energy. To maintain the desired resonant frequency, the thermal distortions of the various DTL components are controlled by water forced-convection cooling. The water cooled components include the tank wall sections, end walls, post couplers, slug tuners, drive irises, drift tubes, RF windows, faraday cups, and dipole electro-magnets. Detailed engineering frequency shift and thermal/fluid analyses have been conducted for each of these DTL components and documented in [1.5]. Consequently, only brief descriptions of these components are provided in this report. Each tank section is cooled via 12 rectangular stainless-steel cooling channels (1 in wide by 0.5 inches deep) that are bonded and clamped in machined groves on the tank walls, as shown in Figure 1.4. The tank endwalls are cooled by water flowing through a series of machined cooling channels, as depicted in Figure 1.5. The post coupler is used to tilt and adjust the shape of the standing RF field inside the DTL tank. Figure 1.6 displays the construction of one such post coupler and its internal water passages. The concentric water passages allow the cooling water to enter the post coupler stem and pass along the outermost passage, turn around at the post coupler tip, and return within the innermost water passage. To further shape the RF field within the DTL tanks, several slug tuners are spaced along the bottom of the DTL. A slug tuner is solid copper cylinder which extends several inches into the tank body. RF energy creates waste-heat electrical currents in the slug tuners. To remove this waste heat, cooling water is circulated through a circular cooling channel machined in one end of the slug tuner, as displayed in Figure 1.7. The drift tubes serve to form the RF cells which accelerate the packets of protons and also shield the accelerating proton packets as they pass from one RF cell to another. Quadrupole magnets, housed within each drift tube, provide the required focusing of the proton beam. Each drift tube has cooling channels machined in its body, as shown in Figure 1.8. The flow of water is fed through the outer tube of the stem, splits in half, circulates around the drift tube body, and exits through the stem inner tube. 12

13 Cooling Channel Figure 1.4. Stainless steel water cooling channels mounted in grooves on a DTL tank wall. 13

14 Steel alignment Bushings press-fit in place Endwall base OFE copper Beam SST coolant fittings Copper tubing Magnet housing with coolant plenum OFE copper PMQ alignment pin press-fit into magnet housing Endwall coolant passages Cover plate OFE copper Exploded view Figure 1.5. DTL tank end-wall cooling passage design. 14

15 Copper tip Copper Stem SST Coolant fittings Rf seal groove Copper Body Brazed Assembly (a) Flow Diverter Water Outlet Water Inlet Figure 1.6. (a) Solid model and (b) cross-section of a SNS DTL post coupler. (b) 15

16 Copper Body and flange SST backing flange For Rf seal SST Coolant fittings Brazed - Welded Assembly (a) 4.5 in 2.25 in (b) Cooling channel Figure 1.7. (a) Solid model and (b) cross-section of a SNS DTL slug tuner. 16

17 Cooling/vacuum manifold All metal vacuum & rf seal mount assembly Stem assembly Bore tube Samarium Cobalt PMQ with Aluminum yoke Body Coolant channels Figure 1.8. Assembly of a typical DTL drift tube. 17

18 RF power is transmitted to the DTL via rectangular gas-filled waveguides. Separating the atmospheric pressure waveguide and the vacuum environment of the DTL tank, is a ceramic RF window. On the DTL tank-side of the RF window is a narrow slit, termed an iris, which allows the RF energy to pass into the DTL tank. The transition waveguide which connects the RF window to the iris, is displayed in Figure 1.9. The RF losses in the transition waveguide and iris must be removed with internal water cooling channels. The RF window has a separate water cooling jacket. Dynamically adjusting the cooling water temperature in the drift tubes and other RF structural components will maintain resonance of the DTL. A uniform frequency shift for all DTL cells can be obtained by balancing the water flow rate, tailoring the cooling channels for each individual drift tube, and adjusting the water inlet temperature. The heat loads and cooling water flow requirements for each of the DTL components are contained in [1.2] and summarized in Section 1.4 of this report. Rf Vacuum Pumping Grills Turbo Pump Turbo-V70 Cooling Tubes Gate Valve Getter Pump Capacitor-B Figure 1.9. DTL RF window waveguide transition piece. 18

19 1.3 DTL Resonance Control Basic Philosophy The electromagnetic field resonant frequency of a particle accelerator is a function of its internal geometry [1.11]. In the case of the DTL, the resonant frequency is mainly a function of the geometry of the tank walls and drift tubes. In the CCL, the resonant frequency is primarily a function of the geometry of the cavities, side coupling cells, and bridge couplers. By manipulating the dimensions of these RF structures, the resonant frequency of the particle accelerator can be finely adjusted and tuned. In practice, resonance control of a room temperature linac is maintained by both the Low Level RF (LLRF) control system and a water cooling system (RCCS). The goal of the SNS resonance control systems is to resonate the DTL at MHz and the CCL cavities at MHz under nominal loaded conditions. In practice, the DTL will be designed and pre-tuned (with post couplers and slug tuners) to a frequency which is offset from the desired resonant value of MHz, with coolant flowing through the structure at a temperature of 20.0 C and no RF heating applied. This frequency offset, which is yet to be defined, will account for the expansion of drift tubes, tank walls, and other tank components, as they heat up under RF power. In practice, the CCL cavities will be designed, manufactured, and pre-tuned to a frequency of MHz, (140 khz above the target frequency) with coolant flowing through the structure at a temperature of 20.0 C and no RF heating applied. As RF power is introduced into the cavity, the copper cavity will heat up, expand, and its resonant frequency will decrease. Engineering analyses have been performed and estimate that the decrease in the cavity resonant frequency (due to RF heating) will be approximately 140 khz under full RF power and a coolant inlet temperature of 20.0 C. The CCL s steady-state operational resonant frequency, MHz, is achieved and maintained by manipulation of the CCL cavity dimensions (expansion/contraction) by adjusting their wall temperatures with the Water Cooling System. The LLRF Control and the RCCS share the responsibility of the resonance control of the DTL and CCL. Consider the DTL as an example. From system start-up, when RF power is gradually introduced to the DTL tank, to full-on steady-state accelerator operation, there are many complicated thermal, fluidic, structural, and electrical 19

20 interactions occurring which influence the resonance of the DTL structure. To deal with these effects, and achieve and maintain resonance of the DTL structure, the LLRF Control and Water Cooling Systems have individual, as well as shared responsibilities. Figure 1.10 displays the responsibilities of the LLRF Control and RCCS as a function of the DTL resonant frequency. Frequency Agile only RCCS / Agile combo. Dead Band outer Frequency Agile only inner Fag- Fhi- Flo- F0 Flo+ Fhi+ Fag+ Fno RF F0-33kHz F0-10 khz MHz F khz F0 + 33kHz F khz Frequency Agile only: Water RCCS is inactive, holding at a saturation position of the valves, while the Resonance Control Module brings the drive frequency into the RCCS / Agile band. RCCS / Agile Band: Dead Band: RCM and the water RCCS act to control the cavity resonant frequency and bring it into the deadband. LLRF control system locks to the fundamental frequency (master oscillator) and the water RCCS takes over to control the cavity resonance within the deadband limits (as determined by operator through the RCM). Figure 1.10 Resonance control responsibility diagram for the SNS DTL and CCL. 20

21 During the early stages of introducing RF power into the DTL RF structure, the RF control system will monitor the structure s resonant frequency and adjust the LLRF Control system output drive frequency to the klystron to match it. The RF control system will thus continuously change the RF frequency as the cavities warm up, and follow the cavity resonant frequency to the desired operational resonant frequency ( MHz). This chase the cavity s resonance activity is referred to as a frequency agile mode of operation. The signal that determines the output RF drive frequency is also used to send an error signal to the water system which indicates how far off the cavity resonant frequency is from the desired operational resonant frequency, and in which direction. For the DTL, a 0V to 10V analog signal, sent from the LLRF to the RCCS, will be used to represent this frequency error. In particular, the analog signal ranges and resulting RCCS actions are as follows: 0V to 0.5V negative frequency saturation. RCCS: Cool the water and structure by forcing all circulating water through heat exchanger. 0.5V to 5.0V error signal is proportional to the -50 khz to 0 khz frequency error (the lower frequency error limit is software selectable). RCCS: use PID algorithm to gradually cool the structure and push the frequency error signal towards 5V, or zero frequency error. 5.0V to 9.5V error signal is proportional to the 0 khz to 50 khz frequency error (the higher frequency error limit is software selectable). RCCS: use PID algorithm to gradually warm the structure and push the frequency error signal towards 5V, or zero frequency error. 9.5V to 10.0V positive saturation. RCCS: Warm the water and structure by forcing all circulating water through the heat exchanger by-pass line. When the resonant frequency of the cavities gets within ±33 khz of the operational resonant frequency, F o, the Water Cooling System begins to perform active resonance control by adjusting a water mixing proportional valve in an attempt to bring the cavity resonant frequency to F o. This ±33 khz frequency band is termed the RCCS/Agile Band. During this mode of operation, the LLRF Control System continues to monitor the resonant frequency of the DTL and attempts to match the output RF drive frequency to it. In addition, the Water Cooling System reads the operational resonant 21

22 frequency error from the LLRF Control System and attempts to adjust the DTL resonant frequency by manipulating the water inlet temperature. The DTL resonant frequency shift induced by a mean temperature change of the DTL drift tube copper is approximately 6.5 khz/ C. Thus by adjusting the cooling water temperature, the DTL resonant frequency is brought closer to F o, and the operational resonant frequency error is reduced. This control logic, similar to that used for the Accelerator Production of Tritium/Low Energy Demonstration Accelerator RFQ and CCDTL Hot Model resonance control systems, is depicted in Figure Note that this resonance control methodology is much different from that used on the LANSCE accelerator, where a particular cooling water temperature is sought, but no feedback is provided by the RF system. When the resonant frequency of the cavities gets within ±10 khz of the operational resonant frequency, F o, the LLRF Control System locks to the operational resonant frequency and the Water Cooling System takes over active cavity resonance control. This narrow frequency range is referred to as the Dead Band. Note that the limits on the Dead Band will be software selectable Water Cooling Resonance Control Technique In the case of the SNS DTL and CCL, a closed loop water cooling system extracts heat from the RF structure and transfers it to a facility chilled water supply via a liquidliquid heat exchanger, as depicted in the P&ID diagram of Figure 1.12(a). In this closedloop circuit, water temperature control is achieved by manipulating the hot-side (Linac side) heat exchanger water flow rate while holding the cold-side water inlet temperature and flow rate constant. This is achieved by using a proportional control valve that divides the circulating water between the heat exchanger and by-pass line, as shown in Figure 1.12(a). By changing the hot-side water flow rate, the overall heat transfer coefficient of the heat exchanger is varied. Since the heat removal rate must effectively remain constant for quasi-steady-state conditions (heat rate into system equals heat rate out of the system), the hot-side water temperature must change inversely to the overall heat transfer coefficient to achieve a new operating condition. Consequently, increasing the water flow through the heat exchanger results in an increase in the overall heat transfer coefficient, and an associated decrease in the mean water temperature. And 22

23 conversely, decreasing the water flow through the heat exchanger results in a decrease in the overall heat transfer coefficient, and an associated increase in the mean water temperature. This water temperature dependence on heat exchanger hot-side flow rate is depicted in Figure 1.12(b). Choose frequency gain or water temperature gain e f e T PID Valve (position) Cavity (temperature and frequency) - Water Temperature Water Temp. Set Point + Low Level R F System Figure Resonance control system logic proposed for the SNS Linac RCCS. 23

24 While being robust and versatile, the water cooling systems for the DTL and CCL will possess limited working ranges and stabilities in water temperature, flow rate, and pressure drop. The nominal operating conditions that the water cooling systems are being designed to are listed in the following section. R F Structure By-Pass Control Valve Reservoir/ Expansion Tank Hot Side Water Inlet Manifold Pump Water Inlet Temperature Cold Side Heat Exchanger Facility Chilled Water Source Heat Exchanger Hot Side Water Flow Rate (a) (b) Figure (a) R F structure water cooling loop schematic and (b) Water temperature control through heat exchanger hot side water flow rate manipulation. 24

25 1.4 DTL Cooling Requirements As mentioned previously, the DTL water cooling system removes the waste heat from the copper RF structure and maintains resonance through active temperature control. In designing the water cooling system, the copper waste heat loads, RF structure mean operating temperature, temperature range and sensitivity required for resonance control, and control methodology (variable flow or variable temperature) needed to be defined a priori. First of all, RF cavity physics computer codes were used to determine the distribution of the RF waste heat in the DTL structural components. Next, a mean DTL copper structure temperature was chosen along with the desired range and resolution of the RF resonance control provided by the water cooling system. Finally, finite element and computational fluid dynamics codes were employed to optimize the design of the water cooling passages and determine the required water flow rates and temperatures. Much of this is discussed in more detail in [1.2, 1.3]. The following tables give the heat loads, required cooling water flow rates and supply temperatures, as well as flow pressure drop specifications for the DTL drift tubes, post couplers, slug tuners, end and side walls, dipole electro-magnets, RF windows, and Faraday cups.. All the heat loads reported in the tables assume a 7.02% RF duty factor. A representative set of data for the heat load, water flow rates, water temperature, and resonance control parameters for the DTL drift tubes is given in Table 1.3. A complete listing of individual drift tube heat load and cooling flow requirements is given in Appendix F. The frequency shift characteristics of the DTL in response to drift tube water inlet temperature changes, in KHz/ C, are also given in Table 1.3. The maximum range of frequency control, ±33 KHz, corresponds to an inlet water temperature range of ±5.1 C about the mean value of 20.0 C. The drift tube pressure drop corresponds to the flow path between the inlet and outlet connectors on the drift tube stem. Note that the heat dissipated on the drift tube outside wall increases as the energy level increases. To avoid field errors and frequency mismatch, the frequency shift needs to be the same for all the DTL cells. This will be achieved by properly tailoring the water flow rate to each drift tube by using an orifice plate upstream of each individual drift tube. 25

26 Table 1.3. DTL drift tubes nominal water cooling system design and operation parameters. Parameter Value Comments/References Nominal heat load per drift tube kw Heat load is different for each drift tube (see Tables 4.5 to 4.10 for individual heat loads) Nominal average operating temperature of drift tubes Nominal water flow rate per drift tube Water flow rate accuracy and stability per drift tube Flow resistance across drift tubes Drift tube water inlet temperature prior to RF power and during steadystate, full RF power Temperature range of water Nominal temperature rise of cooling water through 1 drift tube Temperature accuracy Temperature resolution Temperature stability Frequency shift per change in water inlet temperature Range of frequency control Stability in frequency control Chilled water supply temperature Chilled water supply temperature stability 26.6 C (79.9 o F) x10-4 m 3 /s ( gpm) ± 5% 1.06x x10 14 Pa/m 6 s ( psi/gpm 2 ) 20.0 C (68.0 o F) 5.1 o C (9.2 o F) about mean inlet temperature o C ( o F) ± 0.5 o C ± 0.1 o C ± 0.1 o C 6-7 khz/1 o C ± 33 khz ± 0.7 khz 7.2 o C (45.0 o F) ±0.5 C (± 0.9 o F) Average drift tube operating temperature is identical to average CCL copper temperature. Flow rate is adjusted for each individual drift tube. A uniform frequency shift for all cells within a tank is obtained by balancing the flow rate and tailoring the cooling channels for each individual drift tube. Flow resistances for each drift tube are estimated from standard pipe flow correlations. Tables 4.5 to 4.10 give flow rate requirements and flow resistances for each drift tube. Initial drift tube water inlet temperature chosen to equal mean desired operating temperature. Drift tube water inlet temperature must remain constant as RF power is introduced. An approximate ±8.3 o C band about this mean temperature will be required for full resonance control (i.e. to get ±50 khz frequency adjustment). Tables 4.5 to 4.10 give water temperature rise for each drift tube. A 0.1 o C change in water temperature corresponds to a change in RF frequency of 0.6 to 0.7 KHz. Resonance is maintained by dynamically adjusting the water temperature in the drift tube circuit and the tank circuit simultaneously. 26

27 The heat load and water cooling requirements for the tank walls are given in Table 1.4. Note that the heat loads and water flow rate requirements vary for each tank. The flow resistances for the cooling channels were estimated from standard pipe flow correlations. Table 1.4. DTL tank wall nominal water cooling system design and operation parameters (parameters given per DTL tank). Tank # Heat Load (kw) Water flow rate (gpm) Flow rate accuracy/ stability (gpm) Flow resistance per unit length of channel (psi/gpm 2 / m) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) ± 5% / ± 5% / ± 5% / ± 5% / ± 5% / ± 5% /0.1 The heat load and water cooling requirements for the DTL slug tuners are given in Table 1.5. Note that the heat load on the slug tuner depends on it penetration length inside the tank. The heat load given in Table 1.5 assumes a maximum slug tuner penetration of 2.25 inches. The flow resistance for the slug tuners was estimated from a standard pipe flow correlation. Table 1.5. DTL slug tuner nominal water cooling system design and operation parameters (parameters given per slug tuner). Heat Load (kw) Water flow rate (gpm) Flow rate accuracy/ stability (gpm) Flow resistance (psi/gpm 2 ) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ Stability ( C) ± 5% /0.5 The heat load and water cooling requirements for the DTL post couplers are given in Table 1.6. Note that the heat load on the post coupler depends upon its penetration length inside the tank. The heat load given in Table 1.6 assumes a maximum post 27

28 coupler penetration of 6.2 inches. The flow resistance for the post couplers was estimated from a standard pipe flow correlation. Table 1.6. DTL post coupler nominal water cooling system design and operation parameters (parameters given per post coupler). Heat Load (kw) Water flow rate (gpm) Flow rate accuracy/ stability (gpm) Flow resistance (psi/gpm 2 ) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) ± 5% /0.5 Tables 1.7, 1.8, and 1.9, give the heat loads and water cooling requirements of the end walls, drive irises, and electromagnets, respectively. Table 1.7. DTL end walls nominal water cooling system design and operation parameters. Tank # Heat Load (kw) Flow resistance (psi/gpm 2 ) Water flow rate (gpm) Flow rate accuracy/ Stability (gpm) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) 1 front ± 5% /0.1 1 end ± 5% /0.1 2 front ± 5% /0.1 2 end ± 5% /0.1 3 front ± 5% /0.1 3 end ± 5% /0.1 4 front ± 5% /0.1 4 end ± 5% /0.1 5 front ± 5% /0.1 5 end ± 5% /0.1 6 front ± 5% /0.1 6 end ± 5% /0.1 Table 1.8. DTL drive iris nominal water cooling system design and operation parameters. Heat Load (kw) Water flow rate (gpm) Flow rate accuracy/ Stability (gpm) Flow resistance (psi/gpm 2 ) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) ± 5% /0.5 28

29 Table 1.9. DTL dipole electromagnet nominal water cooling system design and operation parameters (two water passages in parallel per magnet). Parameter Value Comments/References Nominal heat load per magnet Number of magnets 0.35 kw 24 magnets Dipole magnet design excel summary sheet provided by Ted Hunter on 7/3/00. Nominal water flow rate per magnet (for 3.4 C rise in water temp through magnet at normal heat load). Water passes through 2 coils in series. Pressure drop across magnet at nominal water flow rate Water flow rate accuracy and stability per magnet Magnet water inlet temperature Nominal temperature rise of cooling water through 1 magnet Temperature accuracy Temperature resolution Temperature stability Chilled water supply temperature Chilled water supply temperature stability m 3 /s (0.38 gpm) per coil m 3 /s (0.38 gpm) per magnet 11.8 psi ± 5% 20.0 C 3.4 C ±0.5 C ±0.1 C ±0.5 C 7.2 C (45 F) ±0.5 C (±0.9 F) Dipole magnet design excel summary sheet provided by Ted Hunter on 7/3/00 Dipole magnet design excel summary sheet provided by Ted Hunter on 7/3/00 Tables 1.10 and 1.11 give the heat load and cooling requirements for the DTL RF window and Faraday Cup, respectively. Table DTL RF window nominal water cooling system design and operation parameters. Heat Load (kw) Water flow rate (gpm) Flow rate accuracy/ Stability (gpm) Flow resistance (psi/gpm 2 ) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) ± 5% > /0.5 29

30 Table DTL Faraday Cup nominal water cooling system design and operation parameters. Beam Energy (Far. Cup) (MeV) Heat Load (kw) Water flow rate * (gpm) Flow rate accuracy/ Stability (gpm) Flow resistance (psi/gpm 2 ) Steadystate water inlet temp. ( C) Nominal water temp. rise ( C) Water temp. accuracy/ stability ( C) ± 5% TBD / ± 5% TBD / ± 5% TBD / ± 5% TBD / ± 5% TBD /0.5 * A nominal 0.5 gpm flow rate will be used to cool each faraday cup. From the water skid partitioning scheme along with the heat load and flow rate data from Tables 1.3 through 1.11, the individual water skid performance specifications were derived (water flow rate, water inlet temperature, total waste heat dissipation rate, etc.). These water skid performance specifications for the DTL RF structures are summarized in Tables 1.12 through Table Summary of heat loads for the DTL water pumping stations. DTL Tank # Total Drift Tube and Endwall Nose heat load (kw) Tank side and endwall heat load (kw) Total Slug Tuner Heat Load (kw) Total Post Coupler Heat Load (kw) Total Drive Iris Heat Load (kw) Dipole Magnet Heat Load (kw) Total RF Module Waste Heat Load (kw)

31 Table Summary of water flow rates and water inlet temperatures for the DTL water pumping stations. DTL Tank # Total Drift Tube and Endwall Nose flow rate (gpm) Tank side and endwall flow rate (gpm) Total Slug Tuner flow rate (gpm) Total Post Coupler flow rate (gpm) Total Drive Iris flow rate (gpm) Dipole Magnet flow rate (gpm) Water Inlet Temp. ( o C) Total Tank Water Flow Rate (gpm) Based upon the final design of the DTL water cooling systems, the water capacities for each flow loop have been estimated and are summarized in Table Table Water capacities of the DTL water cooling systems. Water Cooling System Flow Loop Water Capacity (gallons) DTL Tank DTL Tank DTL Tank DTL Tank DTL Tank DTL Tank

32 1.5 Mechanical and Electrical Interfaces The key mechanical interfaces between the DTL Water Cooling and Resonance Control System hardware and the DTL RF structure are summarized in Table All mechanical connections on the DTL components, will not be the responsibility of the DTL water cooling system design team. Table Mechanical Interfaces between the DTL Water Cooling and Resonance Control System and DTL RF structure, magnets, and diagnostics. Interface Description Mechanical Connection Supplied on DTL Component Water Cooling System Impact Drift tube water line Stainless steel tube, ½ Water cooling team will provide a 90 Swagelok ports Tanks 1 and 2 fitting and beaded for flexible hose attachment. Drift tube water line Stainless steel tube, ¾ Water cooling team will provide a 90 Swagelok ports Tanks 3 thru 6 fitting and beaded for flexible hose attachment. Post coupler water line Swagelok compression Supply proper sized beaded tube for the interface ports fitting 3/8 between the flex line and DTL compression Slug tuner water line ports Tank end wall water line ports Tank wall water line ports Dipole electro-magnet water line ports RF window water line port Drive iris water line port Swagelok compression fitting 3/8 Swagelok compression fitting 3/8 Swagelok compression fitting 1/2 Swagelok compression fitting 3/16 Male NPT threaded port - ½ Swagelok compression fitting ½ fitting. Supply proper sized beaded tube for the interface between the flex line and DTL compression fitting. Supply proper sized beaded tube for the interface between the flex line and DTL compression fitting. Supply proper sized beaded tube for the interface between the flex line and DTL compression fitting. Supply proper sized beaded tube for the interface between the flex line and DTL compression fitting. Provide proper connectors that results in a beaded hose connector for the supply and return flex lines. Supply proper sized beaded tube for the interface between the flex line and DTL compression fitting. TBD Faraday cup water line Presently not defined. port Assume ½ beaded tube DTL tank body None There is currently no plan to support any of the water cooling lines or manifolds off of the tank body. RF structure support stand/manifolds Pipe supports fastened to the support stand Main supply and return manifolds will be mounted, one above the other, to the main support structure running parallel to the beam line. The manifolds will be attached with pipe supports on the non-aisle side of the linac tunnel. All sub-manifolds will be attached to pipe supports that are connected to the RF stupport structure. 32

33 All water cooling system equipment (pumps, instrumentation, valves, etc) shall operate from the klystron gallery utilities. The SNS conventional facility requirements for the Linac are specified in [1.10]. Table 1.16 summarizes the SNS facility chilled water and electrical requirements for the DTL water cooling and resonance control systems. Note that the electrical requirements listed in Table 1.16 do not include any surpluses required by electrical codes and do not call for any clean electrical power. Finally, In the event of an electrical power failure, uninterruptible electrical power service (UPS) will not be required for the water cooling system diagnostics or PLC on the DTL. Table Summary of utilities required for a single DTL tank water cooling system. Linac Structure Typical DTL Tank Chilled Water Supply Temp. ( C) Chilled Water Supply Mean Load Removal Capacity (kw) Heat Chilled Water Supply Mean Flow Rate Water Skid (GPM) to Chilled Water Supply Pressure Designed Pressure Loss (psig) & Electrical (Qty/KVA/V/Phase) Note that KVA is per unit 7.2 ± /15 6/25/480/3 (pump) 6/30/480/3 (heater) 6/1.8/120/1 (water skid) 6/1.8/120/1 (elec. rack) 6/1.8/120/1 (elec. rack) The communication interfaces between the DTL Vacuum Control System and the SNS Global Control System are described in detail in Section 6 of this report. All other facility-type interfaces are covered in Section 7 of this report. 1.6 Comments and Action Items from Preliminary Design Review The DTL and CCL Water Cooling and Resonance Control System Preliminary Design Review (PDR) committee s comments and the corresponding design team responses and/or action items are given in Table Each item of concern that was raised by the PDR committee has been addressed and documented in this final design report. 33

34 Table Preliminary design review committee comments and the corresponding responses and/or actions taken during the final design of the DTL and CCL water cooling systems. Comment Review Committee Comment # 1 It is recommended that the make up water system for the DI cooling loops be improved. An option is to have a recirculating DI loop to provide the make up for all loops. This will help avoid loop contamination due to maintenance errors. In addition, since recycling of loop water is currently planned (to reduce waste), reinjecting this water back into the system should be accommodated by the design. 2 How the loop heater will be used during start up and during events like RF trips should be investigated in more detail. The heater operational requirements resulting from this investigation should define detail parameters like heater power Design Team Response or Action Two options for providing for make-up water, as well as filling and draining the water systems were considered. The first was a permanent facility-based system, with water lines and fixtures incorporated in the facility design. This option was rejected in favor of a portable water service cart, which would be available throughout the facility for filling, treating, and draining any and all water cooling systems. This service cart was beyond the scope and budget of the DTL and CCL water cooling system designs. However, a proposal (LANL-SNS Memo SNS-00-76) for the design and procurement of such a cart was drafted and submitted to ORNL-SNS for review. The loop heater may be used in system start up. However, the heat supplied to the water to heat up the DTL or CCL is relatively very small when compared to the RF waste heat dissipation rate of a DTL tank or CCL module. Q = mc p T where m = ρv, c p =4179 J/Kg K, T=(26-20) K, ρ=1000kg/m 3 Q = 1000Kg/m m J/Kg K (26-20) K Q = 37,109,520 J 3 How much RF power is available as a function of structure temperature should be determined. This will influence the control system design and response. 4 The appropriate RF duty factors that the cooling system should be able to accommodate and how it should respond, should be more clearly defined. Requirements in this area should be developed in conjunction with the SNS division in OR. 5 The requirements for the control systems of each cooling system should be defined in more detail. The following For the Heater [Q] = w = J/s, [Q] t = Q [Q] = Q/ t Plot [Q] as a function of length of time. Plot Cost ($) as a function of [Q]. Plot Cost (&) as a function of the length of time. An evaluation of the heater hardware cost versus time to heat the volume of water 6 C was performed to size the water skid s inline electrical water heater. RF power is not a direct function of structure temperature, but rather of the transfer matching between the RF drive frequency and the resonance frequency of the cavities (the latter of which is a function of structure temperature). The RF power and duty factors for steady-state operation are currently well known and documented. These parameters are what the current water cooling system is designed around. Lower RF power and duty factors during early commissioning are expected and proper flexibility was incorporated in the design of the water cooling and resonance control systems to handle these off-normal conditions. The RF power and duty factors for steady-state operation are currently well known and documented. These parameters are what the current water cooling system is designed around. Lower RF power and duty factors during early commissioning are expected, but have not been specifically defined or designed to. The water cooling skid has been designed with a variable speed pump, and control valves on the hot and cold sides of the heat exchanger, so as to have enough flexibility in handling off-normal heat loads or operational conditions. No additional water system requirements were given to the design team from the operations or RF structures design teams. Consequently, the water system design was based around the criteria and requirements given in the latest SNS DTL and CCL Water Cooling and Resonance Control System 34

35 operational requirement should be defined: permissible time period to attain design temperature following a cold start permissible time period to attain design temperature following an RF trip (for purposes of design, a design basis RF trip should be established with assistance from the responsible physicist and/or project office to define a credible trip scenario - if one does not already exist and any requirements for thermal transients) response time for minor temperature transients in the linac cooling loop required turndown for temperature control system (the minimum level of power input expressed as a percentage of the design value at which the linac cooling system must be able to deliver water within the specified temperature tolerances on a continuous basis) 6 Air eliminators are needed at system high points. The float type air eliminator has been used successful at LANSCE. 7 The need for redundant pumps (and other components) should be considered in detail and if not included in the design, justification should be included in the design documentation. The RAMI allocation for the cooling system should be part of this assessment. It is noted that all of the other cooling systems for SNS (ring component cooling systems, target component cooling systems, klystron cooling systems, etc) have redundant pumps. It is recommended that the designers of the cooling systems mentioned above, be Description Document. Each of the operational issues presented by the PDR committee at left, have been addressed and are contained in the DTL and CCL Water Cooling and Resonance Control System Final Design Reports. The transient conditions (start-up, RF trip, and minor temperature transients) were studied with a Systemview model on both the DTL and CCL. Design features, including a variable speed pump, water heater, and electrically actuated control valves on the hot and cold sides of the heat exchanger, were incorporated in the water skid design to handle off-normal operating conditions. Air eliminators have been incorporated at the high point on each water skid. We reviewed the LANSCE air eliminator devices which consist of a float type device. We are also contacted vendors (Spirax Sarco and Nalco) for product information. One brand in consideration is the Spriotherm Air Eliminator, removes all air, including entrained microbubbles from the cooling system with a patented screening process. This air eliminator also has the float for air release and does not need the air separator component. The air eliminators will be branched off the transfer main cooling lines at the high point in parallel with a vent valve. Vent and drain valves, with quickconnect, non-leakage fittings, have been incorporated on each water skid as well as each DTL tank and CCL module manifold system in the linac tunnel. No requirement exists for a redundant system. However, in reviewing systems that do have redundancy on LEDA (Power Supply Specifications B5473 and E2289) and in talking with those engineers involved, redundancy was designed into the system for either of two primary reasons. The failure of the system could cause injury to personnel or cause significant damage to the accelerator or the system itself. The failure of the water skid does not meet this criterion. There are plenty of safety interlocks to shut down the beam and RF power if the pump operation and cooling flow should cease. Inclusion of a redundant system would result in a significant cost and schedule impact. Additionally, a redundant system will significantly increase the size of the skid that will have a detrimental affect on the layout within the Klystron Gallery. Finally, the SNS RAMI program has been replaced by Best Engineering Practices, and hence the need for pump redundancy can not be incorporated into a full reliability and availability analysis. 35

36 consulted regarding the decision they made to include redundant pumps. 8 It is recommended that the error bands on thermal and fluid analyses be quantified to help determine if design and safety margins are adequate. 9 Additional thought should be given to how and if individual circuit flow blockages might be detected 10 The use of orifices to balance the flow to individual circuits should work well. However, since changing one or more orifices to correct for unexpected hardware differences will be time consuming, verifying the flow characteristics of individual components will be important. It is important that full scale tests are performed (not just component testing) to verify the accuracy of the design calculations. 11 It appears that the temperature control limit previously specified for the primary side An uncertainty analysis on the SINDA/FLUINT modeling was accomplished by comparing flow and pressure drop predictions (from a S/F model), to empirical data obtained from the CCL hot model prototype water cooling system. The comparison and a thorough discussion of agreements and discrepancies are presented in the DTL and CCL Water Cooling and Resonance Control Systems Final Design Reports. The flow blockage problem is quite complicated and an easy and costeffective fix was not available. It is not cost effective, nor practical to supply enough flow sensors for every piece of DTL hardware or CCL cooling passage. Design alternatives included placing flow switches on each DTL drift tube, and flow meters on every outlet line on the CCL. This was found to be unrealistic due to cost implications. The design compromise, which satisfies operations, safety, maintainability, and cost, was to use flow meters on the DTL submanifold return lines, as well as the return lines on the CCL cavity lines. These flow meters will allow operators to detect significant flow blockages. Other design features include the use of screens at the entrance to the main supply manifolds, and well as all drift tube sub manifolds. RTDs will also be placed on each CCL segment and bridge coupler, which will serve as a safety indicator of flow blockage on those systems. In addition, several design features and detailed quality assurance measures were defined such as filters, proper cleaning procedures, exclusion of teflon tape sealant, high levels of water cleanliness, etc., to minimize the chances of flow blockage. Flow checks of the individual pieces of the DTL hardware and CCL segments will occur prior to assembly. Portable, nonintrusive ultrasonic flow meters, which attach to the outside of the water lines, will be available to trouble shoot flow blockage problems, should they arise. The pressure drop across six prototype drift tubes will be measured using an existing flow loop facility. The geometry of these prototypes is representative of the different drift tubes. Based on the results of these tests, analytical correlations to evaluate the pressure drop will be derived for all the DTL drift tubes. Before final installation, the pressure drop across each drift tube and orifice plate assembly will be checked at nominal flow rate. Orifice plate tests have already been performed to insure that the empirical correlation for orifice plate performance agrees with measured performance. Flow tests have been performed on the CCL cavity cooling passages to bench mark the numerically predicted pressure drop. Each of these tests shows good agreement between experiments and analytical/numerical models. In regards to the CCL, full scale flow tests will be made on prototype full scale segments (cavities and short coupling cells) and bridge coupler flow lines. From this data, empirical flow resistance factors will be calculated and compared against analytical valves used in the Sinda/Fluint flow models. This will develop confidence in the accuracy of the Sinda/Fluint models to correctly size orifice plates to correctly distribute the flow in the CCL. Finally, a flow test will be performed on the CCL during assembly in the RATS building. Flow meters, incorporated on the outlet lines of the CCL cavities, SCCs, and BCs, will be used to determine the adequacy of the flow distribution being generated by the orifice plates. These tests will determine if the orifice plates need to be revised for better flow distribution. Flow tests on the DTL will be performed to set the flow control valves to obtain desired water flows (measured with flow meters) in the various DTL cooling passages. Transient thermal modeling was performed to study the impact of disturbances to the temperature of the cooling water on the facility-side of the heat exchanger. Analyses indicate that temperature swings of +/- 2.0 F 36

37 of the heat exchanger (i.e.: +/- 0.5 F) can be loosened. Since this will simplify the chiller design, it is recommended that the new limit be quantified and conveyed to CF ASAP. 12 It is recommended that the requirements for vibration isolation be investigated in more detail to determine appropriate design features (e.g.: should the entire cooling skid be vibrationally isolated or is it sufficient to only isolate the pump). In addition, the vibration issues associated with the variable speed pump (wider frequency range of operation than fixed speed pumps) should be investigated in more detail. 13 It is recommended that the expected cooling system noise level and sound attenuation requirements be investigated in more detail. 14 The Design Pressures for the cooling loops (as defined by the ASME code) should be defined and specified so structural calculations on related systems can be performed using the appropriate pressure. In addition, consider making the design pressure of the process water loop greater than the shut-in (i.e. deadhead) pressure of the pump plus any static liquid head on the system plus a comfortable safety margin. This will help prevent nuisance discharges of safety relief valves. 15 Based on experience at LANSCE, globe valves should be avoided (they sometimes lose their setting). A better choice is a solid stem type valve. 16 The design of the cooling skids and how they are positioned in the facility should facilitate required maintenance activities for both the pump and the DI bottles. This may require additional clear space on the sides of the skid where these items are located. A (about the mean of supply temperature of 45 F) can be tolerated by the DTL and CCL water cooling and resonance control systems. No dynamic vibration requirement currently exists in the SNS Systems Requirements Document. However, LANL will require that the pump be mounted on dynamic isolators. The skid may be mounted on isolators and will be a recommendation to the skid builder. No thorough analysis was performed by LANL, however, a water skid vibration reduction memo was drafted (LANL-SNS-00-80) which describes the vibration concerns and design features/requirements to minimize the potential for undesired vibrations. No noise level or sound attenuation requirement exists in the SNS Systems Requirements Document. However, LANL will require a common sense requirement to the skid builder that follows OSHA regulations. A LANL- SNS memo (SNS-00-83) describes in detail, the noise concerns, lack of SNS requirements, and acceptable engineering practices and codes related to noise levels and attenuation. The design pressure for the cooling loops has been specified in the SNS DTL and CCL Water Cooling and Resonance Control System Description Document 150 psig. This is approximately 5 times greater than the largest anticipated pressure at the RF structure water manifolds in the Linac tunnel, and 2 times higher than anticipated maximum pressures in the water skids. Valve closure times will be sufficiently large to prevent water hammer. Pressure relief valves will be set at 100 psig (50 psi below the maximum design pressure). Solid Stem type globe valves should prevent the vibration loosening observed on the LANSCE Linac water cooling system. Several vendors (Warren Valves, Conval, and Flow serve) were contacted and supplied us with design information on solid stem valves which have been designed to eliminate valve position changes from vibrations. This requirement has been included in a general globe valve specification document that will be used for hardware procurement. Valve specification requirements also included valve locking devices to prevent accidental movement of valve setting positions. A great deal of thought has been put into the layout of the skid with reference to maintenance needs. The 2 items most likely to need maintenance that are somewhat personnel intensive are the purification system tanks and the pump motor. These have been located at the most accessible side of the skid and will be a requirement for the supplier/builder of the skid. Skids locations within the klystron gallery have been optimized and incorporated in the facility layout drawings to allow for required maintenance access. Maintenance of these components can usually occur during scheduled maintenance periods, and should not cause a shut-down of 37

38 maintenance plan to replace a skid, or components on the skid, should also be developed. This may influence the decision to include redundant components as part of the design or not. It will also give a check of the RAM analysis, which can yield misleading (fairy tale) results. 17 An updated list of spare parts should be developed, considering the current cooling system design. 18 To stream line the procurement process for the cooling system, pre-qualifying vendors is recommended. 19 Since 3-way valves are complex components and can be troublesome over time due to dead heading and erosion, alternate design options should be considered. An alternative design that could be considered is to put a valve on the by-pass line around the pump. Since the flow will follow the least resistant path, most will go to the pump suction. Granted, not all the flow will by-pass the cooling circuit, but the reliability issue may dictate not using a 3-way valve. 20 To avoid contamination problems, the oxygen scavenger bottle should be upstream of the mixed bed resin bottles. 21 Cooling the magnets in the SC part of the tunnel with the same loop as is used to cool the magnets in the warm part of the tunnel may reduce cost and should be considered. 22 The issue of radionuclides should be addressed in more the accelerator operation. A maintenance plan for component replacement will be developed during the procurement phase. Replacement of the entire skid should not be required and will not be part of any standard maintenance procedure. Component redundancy usually driven by operational and safety requirements. No specific requirements were given to warrant the inclusion of redundant subsystems on the water skid. The RAMI analysis, originally proposed by the ORNL SNS-PO, has been replaced by Best Engineering Practices, due to man-power and budget limitations. Consequently, the RAMI analysis will not be performed and will not influence the water skid design. A spare parts recommendation list was created, based on the final design, and submitted to ORNL-SNS. These spare parts are only a recommendation. No criteria, requests, or funding was made or allocated in the DTL and CCL water cooling system work packages to acquire spare parts. A selected list of potential skid suppliers/builders was developed and vendor survey forms were sent out to each of these companies via a LANL UC buyer. Presently, this list for the water skid fabrication includes 12 very interested and capable suppliers who work with very reputable component suppliers. We are waiting on the responses of these surveys and will plan to take facility inspection tours of down-selected vendors to review and assess their fabrication/inspection/certification capabilities, before selecting the final vendor(s). Several valve suppliers have been contacted to supply design and operational information on 3-way valves. These valves have been used in various services for well over 50 years and show no significant problems that would not be encountered with 2-way valves. In addition, the incorporation of the 3-way valve offers a wider range of flow control through or around the heat exchanger for standard and off-normal operating conditions, than was available with a previous design which incorporated a standard 2-way valve in the heat exchanger by-pass line. Several vendors produce 3-way valves with design features that minimize erosion. In addition, the normal operation of the 3-way valve will be to divide water flow between the heat exchanger and by-pass line, and thus dead heading will not be a regular occurance. Consequently, the design team has chosen to use the 3-way valve for controlling flow through the heat exchanger. The pump by-pass line has been removed since it is no longer required for operation of the water cooling system. The 3-way valve can by-pass cooling water around the heat exchanger if the need to prevent cooling to the DTL or CCL arises. The pump is a variable speed type and thus will not need to have a by-pass to control flow rate to the RF structures. Removal of the pump by-pass also eliminates a dead leg line which could have been a source for bacteria growth and dissolved oxygen. We agree, and this change has been incorporated in the final design of the water purification system. We have decided to move ahead with our previous plan of keeping the CCL and SCR magnets on separate cooling lines. Factors such as installation schedule, system pressure drop, water manifold placements for the two systems (on the support structure for the CCL and on the wall or floor for the SRF), and differing magnet configurations on the CCL and SCL, have led to two separate magnet cooling systems. We have decided to consolidate the 2 SRF magnet cooling systems into a single cooling system. We believe this will save both space and costs without a degradation in performance. The collection and disposal plan for the ion exchange resins will follow that used by LANSCE (unless a better technique is found). The ion 38

39 detail. Joe DeVore (SNS OR) should be consulted on this topic. Specific issues which should be dealt with are: When will radionuclides be collected for disposal? How long are the resin tanks estimated to last before they are depleted? Will the resin tanks require shielding? How will resin bottles be handled? Are there other sources of radionuclides and what is their impact? 23 The plan to integrate and test the control system with the cooling skid only after it has been installed at the SNS site will delay the uncovering of problems and leave little time for required modifications. Ways to integrate the cooling skid and controls earlier should be considered. 24 Waveguide cooling requirements (particularly in the chases) is still an open issue. This should be investigated by LANL and ORNL- SNS so that the required capacity can be included in the cooling design as appropriate. exchange bottles will be blown down and dried, sealed off, and be allowed for burial. This process must occur each time the resins are depleted. Cost of regenerating radioactive resins would be extremely high and handling procedures would be difficult. We will not recommend regeneration of the resins. The 9 resin bottles on LANSCE are changed out every 8 to 12 months. The SNS linac water cooling system will probably follow the same or longer exchange period. The exact maintenance period will depend on water/system cleanliness and will come from operational experience. Radiological Control Technician measurements on the resin bottles from the LANSCE water cooling system do not show significant amounts of activation. Shielding is currently not required around the LANSCE resin bottles and it is anticipated that shielding will not be required for those on the SNS water cooling systems. ORNL operations engineers have been contacted about the possibility of activation of the resin bottles and water skid components. US DOE regulations for radiation area designations have been discussed, but no radiation level requirements have been specified by the ORNL SNS- PO. We will follow the same design and operational procedure as used on the LANSCE water cooling system water purification hardware and recommend that a radiological scan be performed during operation of SNS to ensure that radiation levels are sufficiently low. If the levels of radiation are high on any point on the water skid, then shielding can be added. Resin handling and disposal plans have been adopted from LANSCE and are included in the DTL and CCL Water Cooling and Resonance Control System Final Design Reports. There are no other sources of radionuclides that we are aware of. Budget and schedule limitations will prevent full-scale testing of the water control systems prior to installation in the klystron gallery. A compromise was reached to perform the following tasks: 1. Test the resonance control system logic on the CCL hot model water cooling and resonance control system (using the Labview-based control system), and gain valuable operations experience. 2. Develop a prototype water cooling and resonance control system electronics rack, complete with PLC and touchscreen interface. Incorporate this prototype system on the CCL hot model water cooling system to test a minimal number of functional capabilities. Also use these tests to interface with an SNS global control system IOC, running EPICS, to test interfaces, communication drivers, etc. 3. Using the prototype control system developed in (2), test each water skid s basic functioning characteristics (pump and valve control, instrumentation output, etc.) before shipment from the skid vendor to ORNL. 4. Install a complete DTL water skid and corresponding electronics rack in the RATS building to perform flow checks on each DTL tank and CCL half-module, following the assembly process and prior to installation. These tests will not allow for resonance control testing. Resonance control tesing will occur following installation of the systems in the Linac tunnel and klystron gallery. This is the current agreement between LANL and the SNS Project Office at ORNL. The waveguide cooling is a design requirement of the RF engineering team. That team is responsible for cooling the klystrons, and other RF hardware (up to the RF structures), with the exception of the RF windows, and has not requested the RF structure water cooling system design team to deal with waveguide cooling. To reduce the influence of undesired waveguide heat on the water cooling lines in the waveguide chases, the water lines in those sealed chases will be insulated. 39

40 2.0 DTL Water Cooling and Resonance Control System Design Summary The SNS DTL water cooling and resonance control system is comprised of multiple closed-loop water systems, one per DTL tank. Each loop is a modular system, comprised of a water skid (pump, expansion tank, valves, heat exchanger, etc.), water transfer lines, and manifolding/cooling passages at the DTL RF structure. Each loop removes waste heat from a single DTL tank and transfers it to the SNS facility chilled water source via a liquid-to-liquid heat exchanger. Since each modular water system is close-looped, the water simply circulates between the DTL RF structure and the water skid, and hence does not require continual make-up feed water. The closed-loop modular water cooling system, similar to that used in the Advanced Photon Source design [1.6], was chosen for the SNS linac over a fully integrated, open-loop design [1.7, 1.8, 1.9] for the following reasons: Modular, closed-loop design allows for enhanced temperature control and stability during start-up and steady-state operation. Modular water system is consistent with the modular design approach used on the DTL and CCL RF structures. This modularity allows each water system to be installed and commissioned with its corresponding RF structure tank or module. Closed-loop systems mitigate spreading of contamination (radioactive, water purity). Modular system provides consistency in design and ease of manufacturing and installation. Modular system lends itself to reduced manufacturing and assembly costs. Modular design lends itself to easy maintenance (fixing leaks, performing scheduled maintenance, maintaining spare parts, etc.). 2.1 Water System Layout Each DTL tank water cooling system is responsible for removing the RF waste heat from the tank s copper structure and providing resonance control of the RF in the DTL cells. The water cooling system configuration for a single DTL tank can be divided into four main sections including the manifolding on the RF structure, water skid, 40

41 transition lines, and facility chilled water source. Summary details of each of these main sections are provided below Manifolding on the RF Structure The DTL RF structure, discussed previously, contains all of the water-cooled components and the associated internal water passages, as well as the external plumbing manifolds and water lines. Flow is distributed to the various components by way of a water manifold and jumper line distribution system, and metered by valves/flow meters and orifice plates. Figure 2.1 displays the flow diagram for the water cooling system on DTL tank #1, and Figure 2.2 is a top level assembly drawing of the water cooling lines attached to DTL tank #1. A pumping skid delivers water to a main supply manifold on the DTL tank. From the main supply manifold, the water is diverted to a number of submanifolds, which in turn feed the drift tubes, post couplers, tank walls, slug tuners, dipole electro-magnets, RF window, Faraday cup, and drive iris. Proportional valves in combination with flow meters are used to accurately meter the correct amount of water to each sub-manifold. With the exception of the drift tubes, each DTL component in a particular group (i.e., post couplers), has the same heat load and thus will require the same cooling water flow rate. Consequently, all components in that group are ganged together on a common supply sub-manifold and plumbed in parallel. This eliminates the need of flow metering equipment for the majority of the RF structural components. However, each drift tube has a slightly different heat load, and thus each requires a unique cooling water flow rate. This is accomplished by placing an orifice plate upstream of each drift tube. The orifice plates contain a hole of a specified diameter to meter the desired amount of flow to a particular drift tube from a common supply submanifold. To guard against flow blockage problems in the narrow drift tube channels or orifice throats, a fine mesh screen filter is provided at the inlet to the drift tube supply sub-manifold. Other design features include pressure and temperature transducers as well as drain, vent, and pressure relief valves on the main supply and return manifolds. Flow meters on the return lines of the various sub-manifolds are used to correctly distribute coolant flow to the various components, and serve as safety interlocks should leaks or blockages occur in the water lines. More details concerning the component sizing, 41

42 DTL Tank 1, 59 drift tubes, 2 endwall tubes, 2 tank sections, 20 post couplers, 8 slug tuners 4 dipole steering magnets, 1 R F window and 1 drive iris DTL PID - Tank 1 John Bernardin Date Last Modified: DTL Tank 1 Cooling Loop for Drift Tubes and Tank Walls Vent T P FM FM FM Drain FM FM FM FS P T Orifice Plate Globe Valve Ball Valve Flow Meter Flow Switch Pressure Trans. Temperature Trans. Drift Tube Return Lines Drift Tube Supply Lines Vent P T Drain Faciltiy CW Inlet Water Skid #1 Vent Drain Facility CW Outlet DTL Tank 1 Cooling Loop for Post Couplers, Slug Tuners, and Drive Iris T P FM FM FM FM FM FM Drive Iris Slug Tuners Post Couplers Dipole Electromagnets R F Window Vent P T Drain Vent Water Skid #1 Drain Faciltiy CW Inlet Facility CW Outlet Figure 2.1. Flow diagram for the water cooling system on DTL tank #1. 42

43 Figure 2.2. Water manifolds and lines on DTL tank #1. plumbing materials, joining techniques, etc., will be covered in later sections of this report Water Skid The second major component of the water cooling system is the water skid, shown in the flow diagram and model of Figure 2.3. The water skid is a self-contained unit with all of the necessary plumbing, water treatment hardware, instrumentation, and pumping/heat transfer equipment required for delivering water at a desired flow rate and temperature to the DTL RF structural components. A small capacity tank serves as a water reservoir and allows for expansion or contraction of the water associated with temperature changes. The tank is equipped with 43

44 a Nitrogen gas source for controlling pressure and reducing the presence of dissolved oxygen in the water. A pressure relief valve, vent valve, and a liquid low-level indicator were added for safety purposes. The water reservoir feeds the main water line on the suction side of the pump through a manual valve. The reservoir tank volume will be kept to a small capacity (10-20 gallons) to minimize the effect of its large thermal mass on the time response of the water loop s temperature control system. A high capacity, variable speed centrifugal pump and a flow meter connected to the programmable logic controller (PLC), will be used to supply a constant water flow rate to the RF structure. Consequently, flow loop pressure fluctuations induced by the by-pass control valve will not upset the constant supply of water to the DTL. To provide for heating of the water loop (preheating of the copper structure), an inline electrical water heater was placed downstream of the pump. A solenoid valve, plumbed in parallel with the heater, will be used to direct all of the water flow through the heater when it is in use. To remove the waste heat from the cooling loop and maintain the desired water temperature, a stainless steel flat plate counter-flowing heat exchanger was incorporated. This type of heat exchanger is relatively cheap to manufacture, compact, corrosion resistant and extremely efficient. The cold side of the heat exchanger is fed with chilled 7.2 C (45 F) water from the SNS conventional facilities. To maintain steady flow on the conventional facility side of the heat exchanger, a 2-way control valve, connected to a flow meter and the PLC, was incorporated. A PLC will monitor the flow rate through the cold side of the heat exchanger and adjust the control valve to maintain the desired flow rate. To minimize contamination of the heat exchanger from the facility chilled water supply, a 100 mesh filter was added to the upstream cold side of the heat exchanger. Flushing ports will be incorporated on the cold side of the heat exchanger to allow acid cleaning to remove potential scale build-up. The water temperature in the flow loop is manipulated by adjusting the distribution of water flow between the heat exchanger and the heat exchanger by-pass line. This is achieved using a proportional 3-way valve on the return leg from the DTL tank. The 3-way valve directs a portion of the water flow to the heat exchanger, and directs the remainder of the flow through the heat exchanger by-pass line (see Fig. 2.3). 44

45 Steady-state operation requires that all of the waste heat from the DTL be transferred to the facility s chilled water. By raising or lowering the velocity of the hot water through the heat exchanger (via the proportional by-pass valve), the overall heat transfer coefficient of the heat exchanger is raised or lowered, respectively. The net effect is that the effective thermal resistance between the DTL water and the facility s chilled water is inversely proportional to the hot water flow rate through the heat exchanger. If the chilled water source temperature and flow rate, and the thermal load of the DTL are constant, the DTL cooling water temperature (water temperature leaving the pump) will increase with a decrease in hot side heat exchanger water flow rate, and decrease with an increase in hot side heat exchanger water flow rate. In the event of RF power failure or trip, and hence a loss of heat load to the water cooling system, it is desirable not to continue to cool the DTL structure. The motivation is to keep the RF structure as close to its resonance dimensions as possible during the RF trip so that when RF power is restored, little time is lost trying to get the structure back to its desired resonance frequency. To minimize cooling of the DTL during an RF trip, the 3-way valve upstream of the heat exchanger, will adjust its position and force all cooling water to by-pass the heat exchanger and thus minimize the amount of heat loss from the system. At the same instant, the 2-way control valve on the facility-side of the heat exchanger, will close and prevent further cooling of the heat exchanger volume. Once RF power is restored, the 2-way valve will open to its previous setting and the 3-way valve will redirect cooling water through the heat exchanger. 45

46 Ball Valve FM Heat Exchanger By-Pass Control Valve Reservoir/ Expansion Tank Fluid Low-Level Indicator N 2 Pressure Relief Valve Drain Vent Valve Carbon Bed FM UV 5 µ msource Filter Mixed Ion Bed Resin P T Cation Resin Variable-Speed Pump Deoxygen. P 5 µ m Filter Filter 60 mesh S FM T In-Line Heater FM Heat Exchanger Filter Reservoir/ Expansion Tank Heater WP FM Water Purity Transducer (Ph, elect. Cond., Diss. O) 2 Flow Meter Pump FM P FM T T P Valve for Heat Exchanger acid flush P T T P Flow Control Valve Filter 100 mesh T P Temperature Transducer (RTD) Pressure Transducer Facility Chilled Water Outlet Facility Chilled Water Inlet Water Purification Equipment (a) (b) Figure 2.3. (a) Flow diagram and (b) solid model representation of the DTL water skid. 46

47 A water purification system was included in the design of the water skid to minimize the formation of deposits, scale buildup, biological growth, corrosion and activation, all of which can be of significant threat to the performance of the SNS linac water cooling system. This system consists of several filters for removal of debris, a carbon bed for extraction of hydrocarbons, several ion exchange resins for the removal of salts, minerals, dissolved oxygen, and radionuclides, and an ultraviolet lamp to kill bacteria. The water treatment hardware was placed in a small side loop in which approximately 3% of the total flow will be circulated. Electrical resistivity, ph, and dissolved oxygen sensors will monitor the water purification system performance. Additional information concerning water purification and related particle accelerator issues is provided in a later section of this report Transfer Lines Connecting the water skid to the RF structure manifolds, are water supply and return lines. The transfer lines, shown in Figure 2.4 for a particular DTL system, are routed from the klystron gallery to the linac tunnel, through circular chases. In the klystron gallery, the transfer lines will need to be routed overhead, around other plumbing, cable trays, waveguides, etc. In the Linac tunnel, the transfer lines will need to be routed along the floor between the chase exit and the RF structure manifold junctions. Cover plates can be used to avoid the potential tripping hazard caused by these lines on the non-isle side of the accelerator. The transfer lines will contain isolation valves on either end for maintenance purposes. In addition, they will contain short flexible sections to aid in their installation and minimize the transmission of mechanical vibrations. Figure 2.5 shows the routing of the water transfer lines between the water skids and RF structures for all six DTL tanks. 47

48 Klystron Gallery DTL Tank 3 Waveguide Chase Water Transfer Lines Water Skid Linac Tunnel Figure 2.4. Water transfer line routing between a water skid and a DTL tank. 48

49 Figure 2.5. Water transfer line routing for the six DTL water cooling systems. 49

50 2.1.4 Facility Chilled Water Source Chilled water from refrigerated source within the klystron gallery, will be used to remove the waste heat from each DTL closed loop water system. The chilled water will be drawn from a facility supply main to the water skid, pass through the counter-flow heat exchanger, and exit to a facility return main. The current SNS System Requirement s Document [2.1], specifies that chilled water will be supplied at a temperature of 7.2 C (45 F), with a maximum deviation of ±0.56 C (±1.0 F). The total mean heat removal of the six DTL water cooling systems is 472 kw, requiring a total chilled water supply flow rate of approximately 250 gpm, and a maximum heat exchanger pressure drop of 15 psi [1.2]. 2.2 Instrumentation and Controls A variety of transducers are strategically placed at various points in the water skid to monitor pressure, temperature, and flow rate. Several of these transducers will be used for control purposes during operation, while the remainder will be employed for system monitoring during commissioning and trouble shooting situations. A programmable logic controller (PLC), located in an electronics rack in the klystron gallery, will be responsible for overseeing the operation of the water skid and logging necessary data. Some of the PLC functions will include controlling the water temperature and resonance of the DTL, maintaining desired water flow rates on the hot and cold sides of the heat exchanger, monitoring and recording the water purification system parameters, monitoring the flows, temperatures, and pressures at various locations throughout the skid, and providing alarms for off-normal operating conditions. The PLC will possess the ability to operate in a stand-alone mode for commissioning and maintenance purposes, and will also have a direct interface to the SNS global control system for steady-state operation. More detailed information concerning the instrumentation and control system is provided in Section 6 of this report. 50

51 3.0 Water Cooling Analyses 3.1 DTL Water Cooling Loops Lumped Parameter Flow Network Modeling Numerical calculations from a lumped-parameter computer code were used to compute all the pressures, temperatures, and flow rates for the SNS DTL water cooling system models. The computer code used is called SINDA/FLUINT (Systems Improved Numerical Differencing Analyzer with Fluid Integrator) [3.1]. This computer code is ideally suited for piping networks that will be used to cool the RF structure. In a piping network one has a length of pipe called a path and at each end of a path are points called junctions. The path lines usually calculate flow, while at the junctions, values of pressure and/or temperature are calculated. Figure 3.1 illustrates the correspondence between a simple physical model and a SINDA/FLUINT representation. The work discussed below, explains the SINDA/FlUINT modeling plan and the results for the DTL water cooling system. The end goal was to have a master model of an entire DTL water cooling system comprised of a series of sub-models (drift tube cooling circuit, water skid, etc.). Subdividing the simulation of the DTL water cooling system into separate, but coupled, models allowed the simulation to be efficient, tractable, and convenient for debugging DTL RF Structure Cooling Loop Design Goals The design goal of this work was to specify the piping configuration of the SNS Drift Tube Linac (DTL) and optimize their designs by performing engineering analyses to determine the flow and pressure drops, as well as temperature distributions throughout the systems. This task involved designing a system that provided the necessary water flow to support cooling of the RF structure. In particular, a cooling loop had to be designed to provide the required water cooling flow rate to each of the components in the DTL structure. These components include the drift tubes, tank walls, slug tuners, post couplers, dipole electric magnets, and drive iris. A more detailed discussion of the cooling passages and cooling requirements for each DTL component, can be found in Section 1 of this report. A cut-away view of DTL tank 1 is shown in Figure

52 Tank Heat Exchanger Pump Flow System Pump SINDA/FLUINT Representation Figure 3.1. Simple flow system and the corresponding SINDA/FLUINT representation. 52

53 Figure 3.2. Assembly drawing of DTL tank 1 (note that the water manifolds and lines are not included in the schematic). 53

54 Although there are 6 different tank sections that comprise the DTL, tank 3 was selected as the representative structure and was modeled in detail. The results from tank 3, along with additional analyses necessary to model any significant differences, were used to design the piping systems for the remaining tanks. The focus of this work was to analyze each cooling loop with a computer model to study fluid flow and pressure throughout the system. Pipe line sizes, orifice sizes for flow control, and overall pressure drops throughout the fluid circuits were determined. This information was required to size the pump, heat exchanger, flow control valves, etc. The design goals for DTL RF structure model are summarized in Table 1. Note that all line diameters listed in this section of the report correspond to internal diameters. All water velocities correspond to mean or average flow rates through a pipe of a given cross-section. 54

55 Table 3.1. SINDA/FLUINT modeling goals for the DTL RF structure model. Design Goal Outcome Determine orifice plate sizes required to properly distribute the water flow to the drift tubes Optimize supply and return main manifold diameters to minimize pressure variations along length and minimize overall cost Optimize transfer line diameters to minimize pressure drop, erosion, and cost From main manifolds to water skid From main manifolds to end walls From main manifolds to drive iris From main manifolds to RF window From main manifold to drift tube submanifold From main manifold to all other submanifolds Optimize sub-manifold diameters for the drift tubes post couplers slug tuners tank walls DTL magnets to minimize pressure variations, pressure drops, and overall cost Determine pressure distribution around the flow loop and the maximum pressure drop across the system (help size pump for flow rate and pressure drop) Repeat necessary analyses (from DTL tank 3) to meet the above mentioned design goals for DTL tanks 1, 2, 4, 5, and 6 The orifice plates were sized to provide each drift tube with its design flow rate. Values of β ranged from 0.43 to " ID minimum 4" ID minimum 0.5 " ID minimum 0.313" ID minimum 0.5" ID minimum 2.5" ID minimum Same size as submanifold diameter 1.75" ID minimum 1.00" ID minimum 1.25" ID minimum 1.5" ID minimum 0.5" ID minimum Pressure drop = approximately 24 psi Required that post coupler sub-manifold be 1.25" ID minimum Design Specifications Reference [1.2] contains the design specifications for the DTL cooling loops. Each module of the DTL is to have its own separate cooling loop complete with heat exchanger, pump, and instrumentation, etc. A separate facility cooling supply of chilled water is available with inlet temperature of 7.2 C, ±0.28 C. The temperature of cooling water delivered to the DTL tanks is specified to be 20.0 C. The design flows specified for the DTL tanks 1 through 6 include flows of 0.2 to 5.1 gpm to the drift tubes, flows of 19.2 to 79.2 gpm to the tank walls, a flow of 1.0 gpm to each slug tuner, a flow of 0.65 gpm for each post coupler, and flows ranging from 0.2 to 1.0 gpm for the end walls. Section 1 of this report contains detailed descriptions of the heat loads and cooling requirements for all of the DTL components. Separate supply and return sub-manifolds 55

56 are used to provide distribution of the cooling water to each of these subsystems for a given tank. A single supply and return manifold combination will feed all sub-manifolds. This large manifold will be connected to the water skid by transfer lines Tank 3 Global Model Description A flow diagram in Figure 3.2 displays the water-cooled components and the water distribution lines on DTL tank 3. A pumping skid delivers water to a main supply manifold on the DTL tank. From the main supply manifold, the water is diverted to a number of sub-manifolds, which in turn feed the drift tubes, post couplers, tank walls, slug tuners, dipole electro-magnets, and drive iris. Proportional valves in combination with flow meters are used to accurately meter the correct amount of water to each submanifold. With the exception of the drift tubes, each DTL component in a particular group (i.e., post couplers), has the same heat load and thus will require the same cooling water flow rate. Consequently, all components in that group are ganged together on a common supply sub-manifold and plumbed in parallel. This eliminates the need of flow metering equipment for the majority of the RF structural components. However, each drift tube has a slightly different heat load, and thus each requires a unique cooling water 56

57 DTL Tank 3 Cooling Loop for Drift Tubes and Tank Walls Vent T P FM FM FM Drain FM FM Orifice Plate FS FS FS FS Drift Tube Return Lines FS FS FS Globe Valve Ball Valve FM Flow Meter FS Flow Switch P Pressure Trans. T Temperature Trans Drift Tube Supply Lines Vent P T Drain Faciltiy CW Inlet Water Skid #3 Vent Drain Facility CW Outlet DTL Tank 3 Cooling Loop for Post Couplers, Slug Tuners, and Drive Iris T P FM FM FM FM FM Drive Iris Slug Tuners Post Couplers Dipole Electromagnets Vent P T Drain Faciltiy CW Inlet Water Skid #3 Vent Drain Facility CW Outlet Figure 3.3. Flow diagram for DTL Tank 3. 57

58 flow rate. This is accomplished by placing an orifice plate upstream of each drift tube. The orifice plates contain a hole of a specified diameter to meter the desired amount of flow to a particular drift tube from a common supply sub-manifold. The approach to modeling the maze of water lines on DTL tank 3, was to generate a global model, comprised of a series of detailed sub-models for each subsystem of components. Figure 3.4 displays a representation of the global model and its sub-model construction for the DTL tank 3 water cooling system. The SINDA/FLUINT global model for DTL tank 3 is displayed in Figure 3.5. The sub-models were developed to optimize the water line geometries of the submanifolds, size orifice plates, and to determine subsystem pressure drops. Subsystem models, once completed, were represented as a corresponding flow branch in the global model. The pressure losses that occur in series through each branch of the global model, are summed as follows: P branch = P subsystem + P flow meter + P other losses (4 tees and line friction) + P valve where P subsystem is the pressure drop of a particular subsystem, determined from its corresponding SINDA/FLUINT submodel, P flowmeter is the pressure drop of flow meter in that branch, P other losses is the pressure drop of the supply and return plumbing not included in the submodel, and P valve is the pressure drop across the globe valve used to meter the flow to that particular subsystem. Note that the flow resistance coefficient of the branch valves were numerically adjusted to obtain the desired flow rates through each subsystem. 58

59 Main Return Manifold Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter End Wall Drift Tube Sub-Model Drive Iris Sub-Model Slug Tuner Sub-Model Post Coupler Sub-Model Tank Walls Sub-Model Magnet Sub-Model Side Wall Main Supply Manifold Water Cart Junction Junction Junction Junction Junction Junction Junction Junction Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter Flow Meter End Wall Losses Drift Tube Losses Drive Iris Losses Slug Tuner Losses Post Coupler Losses Tank Wall Losses Magnet Losses Junction Junction Junction Junction Junction Junction Junction Junction Water Skid Losses VF Figure 3.4. Water-cooling system model representation for a single DTL tank. 59

60 Figure 3.5. SINDA/FLUINT global model for DTL Tank 3. The next several sections will present the model descriptions and numerical studies/results for each subsystem model. These sections will be followed by a discussion of the global model results Drift Tube Sub-model Description As mentioned previously, each drift tube requires a different cooling water flow rate. The first step in modeling the drift tube subsystem, therefore, was to determine the orifice plate sizes needed to deliver the required water flow rate to each drift tube. To do this, all K factors and flow resistances in the drift tube subsystem were calculated or taken from reference [3.2]. Figure 3.6, a cross section of a DTL tank and drift tube, displays the pressure loss components for a drift tube flow circuit. 60

61 Adapter Contraction Elbows Bends Adapter Expansion Flow Switch Tube Expansion Branch Out Tube Contraction Orifice Plate Branch In Drift Tube Figure 3.6 Cross section of a DTL tank and drift tube flow circuit. The following table displays the loss factor values and resistances used for the drift tube circuits of DTL Tank 3. 61

62 Table 3.2. K factor values for the drift tube circuits of DTL Tank 3. Loss Description Dimensions (inches) Loss Ref. Factor Value (dimensionless) K1 Inlet Tee 1.75 Straight to.5 1 Parietti, L [3.3] Branch K2 Tee to Union.5 to White, F [3.4] K3 Union to Orifice Fitting.62 to White, F [3.4] K4 Orifice Plate See Table 4 Idelchik [3.2] K5 Orifice Fitting to Union.65 to White, F [3.4] K6 Union to Adapter.62 to White, F [3.4] K7 Adapter.59 to White, F [3.4] K8 Adapter to Tubing.63 to White, F [3.4] K9 Tubing to Adapter.75 to White, F [3.4] K10 Adapter.63 to White, F [3.4] K11 Adapter to Elbow.59 to White, F [3.4] K12 Elbow White, F [3.4] K13 Elbow to D.T..62 to.5.15 White, F [3.4] K14* Drift Tube Resistance Varies See Table White, F [3.4] K15 D.T. to Elbow.5 to White, F [3.4] K16 Elbow White, F [3.4] K17 Elbow to Adapter.62 to White, F [3.4] K18 Adapter.59 to White, F [3.4] K19 Adapter to Tubing.63 to White, F [3.4] K20 Tubing to Adapter.75 to White, F [3.4] K21 Adapter.63 to White, F [3.4] K22 Adapter to Union.59 to White, F [3.4] K23 Union to Flow Switch.62 to.5.15 White, F [3.4] K24 Flow Switch.5" NPT 1.5 psi AutoFlow cat. K25 Flow Switch to Union.5 to White, F [3.4] K26 Union to Tee.62 to.5.15 White, F [3.4] K27 Outlet Tee.5 Branch to White, F [3.4] Straight K28 Straight Passage Tee On sub-manifold.2 White, F [3.4] *Drift tube losses given in [1.2] as a resistance with units of psi/gpm 2. Once all losses were characterized, the pressure drop associated with each loss was calculated. The following equation was employed: where, P = pressure drop for each loss factor (psi) K = loss factor (dimensionless) ρ = density P Ki = KρV (3.1) 62

63 V = water velocity i = loss number (1 through 28) The resistance values given in Reference [1.2] for each drift tube were converted into pressure losses using the following relation: where, P = pressure drop across a single drift tube (psi) Fr = Flow rate through drift tube (gpm) R = Resistance of flow through drift tube (psi/gpm 2 ) 2 P dt = R * Fr (3.2) As mentioned previously, the drift tubes are plumbed in parallel branches. The pressure losses that occurred in series through each branch were summed as follows: P branch = P dt + P k1 + + P k28 (3.3) Next, equivalent overall loss or K factors were calculated to represent each branch of the drift tube sub-manifold. This was accomplished by rearranging Equation (3.1), substituting P branch for P k, and solving for K. The equivalent overall loss factor calculated for losses through a single branch are presented in the Results section. Once the equivalent loss factors were determined, a SINDA/FLUINT model was developed for the drift tube subsystem. A graphical representation of this model is shown in Figure 3.7. The equivalent loss or K factors for a drift tube branch, were incorporated into the loss, or L components, of the model shown in Figure

64 FS FS FS FS Drift Tube #1 Drift Tube #34 Sinda/Fluint Model J L J L Drift Tube Return Manifold J L J L VF Drift Tube Supply Manifold J J Drift Tube Supply Manifold J J Figure 3.7. Generic Sinda/Fluint model representation of the drift tube circuit for DTL tank 3. Figure 3.8 is the actual drift tube model created using the SINDA/FLUINT computer code. SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). The magnitude of the flow in a path is also described with a line thickness thicker lines denote larger volumetric flow. Notice that the lines are described with a T meaning Tube, a L meaning pressure loss to account for fittings, bends, reducers, etc, and a VF meaning a constant volume pump. Although difficult to see from this image, the two rows of 35 junctions are connected together with tubes to form the supply and return manifolds. Figure 3.8. SINDA/FLUINT model for the drift tube circuit of DTL tank 3. 64

65 Drift Tube Sub-model Results Orifice Plates To fulfill the first modeling goal of determining the individual drift tube orifice geometries, spreadsheet calculations were used. A portion of the spreadsheet calculation is shown in Table 3.3. The flow calculations performed in the spreadsheet, predicted the pressure drop for each of the drift tubes and their corresponding orifice plates and inlet/outlet water lines, using the flow rates, passage geometries, flow resistance coefficients, and Eqn. (3.3). The main objective of the spreadsheet calculations, was to determine the orifice plate sizes required to get the correct flow rate through each drift tube. The drift tubes in Table 3.3 are numbered from 1 to 33. The end noses are numbered 0 and 34. The flow rates required to achieve the desired hardware temperature are shown in column 2. Column 4 represents the pressure loss across the flow switch. Assuming an adjustable flow switch will be employed, a pressure drop across each flow switch of 1.75 psi was assumed. The analytically determined flow resistance of each drift tube is shown in column 5 in units of psi/gpm 2. These resistances are multiplied by the respective drift tube flow rate to determine the drift tube pressure drop. Using Eqn. (3.3), and the flow resistance data presented previously, the orifice plate pressure drop and hence the orifice plate geometry, required to give the desired drift tube flow rate could be determined. The sharp-edged orifice correlation from Reference [3.2] was used to calculate the pressure drop for a given orifice-to-tube diameter ratio, β. This empirically-based correlation was found to have the best accuracy for β ranging from 0.2 to 0.8 [3.5]. Since the last drift tube in a particular tank requires the greatest flow rate of the drift tubes in that tank, it will have the largest corresponding β value for its orifice plate. Using β=0.72 for the 33 rd drift tube of tank 3, the pressure drop calculated for the orifice plate was 0.58 psi, resulting in a total pressure drop for the 33 rd drift tube of psi. Orifice plates for the remaining drift tubes and end noses were subsequently sized to match the overall pressure drop of the 33 rd drift tube. Based on these total pressure drop numbers, the corresponding loss factors were then calculated based on a ½ ID tube size (last 65

66 column of Table 3.3). The results for the orifice plate sizes for tank 3 are given in Table 3.3. See Appendix G for drift tube orifice plate geometries for the additional five DTL tanks. Table 3.3. Orifice plate sizing spreadsheet for the tank 3 drift tube cooling system. Drift Tube Flow Rate DP branch total DP flow switch R drift tube DP drift tube A orifice/a pipe b ID orifice DP orifice, required # (gpm) (psi) (psi) (psi/gpm 2 ) (psi) (in) (psi) (psi) DP total K total.5id

67 Inlet/Outlet locations In an effort to create a uniform pressure profile along the drift tube supply and return manifolds, while minimizing potential interference between flow ports and the supply/return liens, it was decided that the manifold connections be placed where the flow naturally splits or converges. The following table identifies the location for the supply and return water line connections of each drift tube sub-manifold. Table 3.4. Supply and return water line connection locations on the drift tube submanifolds of the six DTL tanks. Tank Number Inlet and Outlet Locations (from front of tank) 1 Between Drift Tubes 44 and 45 2 Between Drift Tubes 29 and 30 3 Between Drift Tubes 19 and 20 4 Between Drift Tubes 15 and 16 5 Between Drift Tubes 13 and 14 6 Between Drift Tubes 13 and 14 Sub-Manifold Sizing and Configuration After the orifice plates and the inlet/outlet locations were characterized, the submanifolds were sized. Figure 3.9 shows the SINDA/FLUINT model predictions for the flow rates and pressure drops in the drift tube cooling circuit of DTL Tank 3 with submanifold diameters of 1.25 inches. 67

68 Figure 3.9. SINDA/FLUINT model results for the drift tube cooling circuit of DTL tank 3 (Units: Pressure = Pascals, Flow Rate = kg/s). The color red denotes the highest value for both pressure and flow with blue denoting the lowest value. The total flow rate within the model is gpm. The supply and return manifolds have been sized to an inside diameter of 1.25 inches. Using SINDA/FLUINT, a series of runs were made with a varying manifold internal diameter from 1 inch to 3 inches in ¼ in. increments. The results of this diameter study are shown in Figure

69 Line Size Effect on Flow Rate Difference from Required (%) " 1.5" 1.75" 2" 2.25" Drift Tube Number Figure Drift tube flow rate variance (from required value) versus drift tube number for five different sub-manifold inner diameters for DTL tank 3. As the sub-manifold diameter increases, the percentage difference between the actual and required drift tube flow rates decreases. This trend occurs because the pressure in the sub-manifold becomes increasingly more uniform along its axis as the manifold diameter increases. As the manifold pressure becomes more uniform, the flow distribution within a given drift tube segment converges on the designed flow rate. For a manifold diameter of 1.25 inch, the maximum flow deviation in the drift tube assembly was 14.3 %. For a manifold diameter of 1.75-inch (ID), however, the maximum flow deviation decreases to an acceptable level of 3.75 %. Therefore, a 1.75-in. inner diameter is recommended for the drift tube supply and return sub-manifolds. A pressure drop of 12.9 psi exists when this diameter is employed for the drift tube system. Summary From the drift tube flow modeling studies discussed above, the following key results were obtained: 69

70 Orifice plates hole diameters range from 0.12 to 0.47 inches for the drift tubes of DTL tank 3. Orifice geometries of the remaining 5 DTL tanks are given in Appendix G. Locations for supply and return water line connection should follow that outlined in Table 3.4. The drift tube sub-manifold diameter must be a minimum of 1.75 inches ID to produce a maximum flow deviation of approximately 3.75%. This will apply for all 6 DTL tanks. The supply and return transfer lines for the drift tube submanifolds should have an inner diameter of 2.5 to minimize pressure drop and erosion. A pressure drop of 12.9 psi is produced through the system Slug Tuner Sub-model Description The slug tuner cooling system for DTL tank 3 is represented in Figure Slug Tuners Return Manifold Supply Manifold Figure Generic representation of the slug tuner cooling system for tank 3. 70

71 The first step in modeling the slug tuner cooling system was to determine the locations where pressure drop was expected.. Figure 3.12 labels the pressure loss locations for a typical DTL post coupler and Table 3.5 assigns a loss factor to each location. Slug Tuner Outlet Manifold Branch Adapter Adapter Adapter Adapter Inlet Manifold Branch Outlet Manifold Inlet Manifold Outlet Union Inlet Union Figure Locations for pressure drops in the slug tuner cooling circuit. Table 3.5. K factor values for the slug tuner cooling circuit of DTL Tank 3. Loss Description Size (inches) Quantity Value K1 Inlet Tee 1.25 Straight to Branch K2 Tee to Union.25 to K3 Union to Adapter.28 to K4 Adapter.28 to K5 Adapter to Tubing.19 to K6 Tubing to Adapter.25 to K7 Adapter.19 to K8 Adapter to Union.28 to K9 Union to S.T..28 to Rst Resistance psi/gpm 2 K10 S.T. to Union.25 to K11 Union to Adapter.28 to K12 Adapter.28 to K13 Adapter to Tubing.19 to K14 Tubing to Adapter.25 to K15 Adapter.19 to K16 Adapter to Union.28 to K17 Union to Outlet Tee.28 to K18 Outlet Tee 1.25 Branch to Straight K19 Elbow May not be used 2 K20 Additional losses

72 Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed. where, P = pressure drop K = loss factor ρ = density V = velocity i = loss factor number P Ki = KρV (3.4) The post coupler resistance value given in Table 3.5 was converted into a pressure loss using the following relation: where, P = pressure drop across slug tuner (psi) Fr = Flow rate through slug tuner (gpm) R = Resistance of flow through slug tuner (psi/gpm 2 ) 2 P pc = R * Fr (3.5) The pressure losses that occurred in series through each sub-manifold branch were summed as follows: P branch = P st + P K1 + + P K20 (3.6) The total pressure drop across a single branch was calculated to be approximately 12.5 psi. Next, equivalent overall loss or K factors were calculated to represent each branch of the slug tuner sub-manifold. This was accomplished by rearranging Equation (3.4), substituting P branch for P k, and solving for K. 72

73 Several SINDA/FLUINT models were created which employed losses (L) and tubes (T) to characterize the slug tuner system. From Figure 3.13, notice that the SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). Path lines are described with a T meaning Tube, a L meaning loss to account for fittings, bends, reducers, etc, or a VF meaning a constant volume pump. The two rows of 12 junctions are connected together using a loss connector. Since flow must be distributed evenly to each slug tuner, the ideal location for the supply inlet exists at the center of the sub-manifold where flow evenly divides. Similarly, the outlet may be ideally located at the center where flow converges evenly from both sides of the return sub-manifold. Unfortunately, due to interference with other objects on the DTL, it may not be possible for the supply inlet and return outlet to be positioned at their ideal locations. Subsequently, several different SINDA/FLUINT models were created to study the effect of inlet and outlet placement on flow distribution. Figure 3.13 displays some of the models created for studying the slug tuner cooling system. Figure 3.13a allows the inlet and exit to be located at extreme opposites. This configuration forces flow, through each post coupler, to travel the same distance. In Figure 3.13b the inlet and outlet are located between slug tuner three and four. In Figure 3.13c the inlet is shifted over and located at the center of the sub-manifolds. In all cases, the equivalent loss or K factors for a slug tuner branch as determined above, were incorporated into the loss, or L components, of the model shown in Figure

74 (A) (B) Figure Various SINDA/FLUINT models of the slug tuner cooling system. 74

75 Slug Tuner Sub-model Results Figure 3.14 compares the average flow rate error (difference between desired and predicted flow rate) through each slug tuner branch as a function of sub-manifold diameter for various supply and return water line connection location. Results from this study confirm that the best location for the placement of the inlet and outlet are at the center of the sub-manifolds. If deviation from the central location is necessary to accommodate other devices on DTL Tank 3, then, as Figure 3.14 indicates, an increase in error for the average flow rate through each slug tuner will occur. This is especially apparent when smaller diameter sub-manifolds are employed. This implies that the slug tuner sub-manifold should have a minimum diameter of 1.25 inches so that inlet and outlet placement does not have a major effect on flow distribution Side Entrance And Exit Entrance And Exit Between Slug Tuners 2 & 3 Percent Error from Required Flow Rate Entrance and Exit Between Slug Tuners 3 & 4 Entrance and Exit Between Slug Tuners 4 & 5 Center Entrance and Exit Manifold Diameter (in) Figure Average flow error versus supply and return sub-manifold diameter for different water line connection locations for the slug tuner cooling system. 75

76 Inlet and outlet placement has little effect on the overall error if a sub-manifold diameter of at least 1.25 inches is employed. The Sinda/Fluint model with the centralized inlet and outlet was chosen as a representative system to study the slug tuner system in closer detail. This model investigated sub-manifold diameters ranging from 1.0 to 1.5 inches. Figure 3.15 displays the Sinda/Fluint model pressure and flow predictions for a sub-manifold diameter of 1.25 inches. Figure SINDA/FLUINT predictions of slug tuner flow and pressure with 1.25 inch sub-manifold diameters. 76

77 In Figure 3.15, the pressure drop as well as the flow rate are nearly equal for each slug tuner. Figure 3.16 shows the results for all sub-manifold diameters studied. Figure 3.16 shows that a sub-manifold with a diameter of 1 inch may allow over 3 % deviation in required flow rate whereas a sub-manifold with a diameter of 1.5 inches will allow a maximum deviation of less than.3%. A sub-manifold with a diameter with a minimum diameter of 1.25 inches allows less than 1 % error and produces a pressure drop of 3.6 psi exists across the model. Line Size effect on Flow Rate Flow Rate Difference From Required (%) " 1.25" 1.5" Slug Tuner Number Figure Slug tuner flow rate variance (from required value) versus slug tuner number for three different sub-manifold inner diameters for DTL tank 3. Summary From the slug tuner flow modeling studies discussed above, the following key results were obtained: Water transfer line connection placement has little effect on the overall error if a submanifold diameter of at least 1.25 inches is employed. A sub-manifold with a 1.25 inches or greater diameter will be required to achieve the minimum amount of error in the system. 77

78 A pressure drop of 3.6 psi exists across the slug tuner cooling system. Identical slug tuner sub-manifold diameters can be used for all six DTL tanks Post Coupler Sub-model Description The post coupler cooling system for DTL tank 3 is represented in Figure Return Supply DTL Tank 3 Supply Return Figure Generic representation of the post coupler cooling system for DTL tank 3. The first step in modeling the slug tuner cooling system was to determine the locations where pressure drop was expected. Figure 3.18 labels the pressure loss locations for a typical DTL post coupler and Table 3.6 assigns a loss factor to each loss location. 78

79 Post Coupler Elbow Unions & Hose Adapters Inlet Branch Tee Outlet Branch Tee Figure 3.18 Representation of a DTL post coupler and its flow path. 79

80 Table 3.6. Losses for the post coupler cooling circuit. Loss Description Sizes (in) Quantity Value K1 Inlet Tee.75 Straight to.25 Branch 1 1 K2 Tee to Union.25 to K3 Union to Adapter.28 to K4 Adapter.28 to K5 Adapter to Tubing.19 to K6 Tubing to Adapter.25 to K7 Adapter.19 to K8 Adapter to Union.28 to K9 Union to S.T..28 to Rpc Resistance psi/gpm 2 K10 S.T. to Union.25 to K11 Union to Adapter.28 to K12 Adapter.28 to K13 Adapter to Tubing.19 to K14 Tubing to Adapter.25 to K15 Adapter.19 to K16 Adapter to Union.28 to K17 Union to Outlet Tee.28 to K18 Outlet Tee.25 Branch to.75 Straight 1 1 Kelbow Elbow May not be needed 1 Kadd Additional 1.3 Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed. where, P = pressure drop K = loss factor ρ = density V = velocity i = loss factor number P Ki = KρV (3.7) The post coupler resistance value given in Table 3.6 was converted into a pressure loss using the following relation: 2 P pc = R * Fr (3.8) 80

81 where, P = pressure drop across post coupler (psi) Fr = Flow rate through post coupler (gpm) R = Resistance of flow through post coupler (psi/gpm 2 ) The pressure losses that occurred in series through each sub-manifold branch were summed as follows: P branch = P pc + P K1 + + P K18 (3.9) The total pressure drop for a post coupler branch was calculated to be 3.7 psi. Next, equivalent overall loss or K factors were calculated to represent each branch of the post coupler sub-manifold. This was accomplished by rearranging Equation (3.7), substituting P branch for P k, and solving for K. Next, several SINDA/FLUINT models were created which employed losses (L) and tubes (T) to characterize the post coupler system. Since flow must be distributed evenly to each post coupler, the ideal location for the supply inlet exists at the center of the sub-manifold where flow evenly divides. Similarly, the outlet may be ideally located at the center where flow converges evenly from both sides of the return sub-manifold. Unfortunately, due to interference with other objects on the DTL, it may not be possible for the supply inlet and return outlet to be positioned at their ideal locations. Subsequently, several different SINDA/FLUINT models were created to study the effect of inlet and outlet placement on flow distribution. In Figure 3.19a the inlet and outlet are located between post coupler two and three. In Figure 3.19b the inlet is shifted over and located between post coupler three and four. Figure 3.19c allows the inlet and exit to be located at extreme opposites. This configuration forces flow, through each post coupler, to travel the same distance. Finally, in Figure 3.19d, the inlet and outlet are located at the center of the submanifold. In all cases, the equivalent loss or K factors for a post coupler branch determined above, were incorporated into the loss, or L components, of the model shown in Figure

82 (A) (B) (C) (D) Figure Various SINDA/FLUINT models created for the post coupler study. 82

83 Post Coupler Sub-model Results Figure 3.20 compares the average flow rate error (difference between desired and predicted flow rate) through each post coupler branch as a function of sub-manifold diameter for various supply and return water line connection location. Results from this study confirm that the best location for the placement of the inlet and outlet are at the center of the submanifolds. If deviation from the central location is necessary to accommodate other devices on DTL Tank 3, then as Figure 3.20 indicates, an increase in error for the average flow rate through the post couplers will occur. This is especially apparent when smaller diameter submanifolds are employed. This implies that the post coupler submanifold should have a minimum diameter of 0.75 inches so that inlet and outlet placement does not have a major effect on flow distribution. Inlet and Outlet Location Comparison Between PC 2 and 3 Between PC 3 and 4 Extreme Oposites Center Average Error (%) Sub-Manifold Diameter (inches) Figure Average flow error versus supply and return sub-manifold diameter for different water line connection locations for the post coupler cooling system. 83

84 Inlet and outlet placement has little effect on the overall error if a sub-manifold diameter of at least 0.75 inches is employed. The SINDA/FLUINT model with the centralized inlet and outlet was chosen as a representative system to study the post coupler system in closer detail. This model will look at sub-manifold diameters ranging from to 1.25 inches. Figure 3.21 is the SINDA/FLUINT model results for a submanifold diameter of 0.75 inches. Figure SINDA/FLUINT flow and pressure predictions for the post coupler cooling circuit with a 0.75 inch sub-manifold diameter (Units: Pressure = Pascals, Flow Rate = kg/s). 84

85 In Figure 3.21, the pressure drop as well as the flow rate are nearly equal for each branch. Figure 3.22 flow rate variance as a function of post coupler location for five different sub-manifold diameters. Percent Error from required (%) Post Coupler Number.375".5".75" 1" 1.25 Figure Post coupler flow rate variance (from required value) versus post coupler number for five different sub-manifold inner diameters for DTL tank 3. Figure 3.22 shows that a sub-manifold with a diameter of inches may allow over 12 % deviation in required flow rate whereas a sub-manifold with a diameter of 1.25 inches will allow a maximum deviation of less than 0.04%. A sub-manifold with a minimum diameter of 0.75 inches allows less than 0.5 % error. 85

86 In an attempt to standardize all post coupler sub-manifold diameters, additional studies were performed to determine whether the 0.75 inch diameter sub-manifold is suitable for use on all DTL tanks. DTL tank 4, because it requires the largest number of post couplers, recognizably presents the worst case for flow distribution. Therefore, an additional model was created to study DTL tank 4. The new model resembled that used to study tank 3, except the number of post couplers increased from 16 to 30. The submanifold inlet was placed between post couplers 8 and 9. Figure 3.23 shows the results for the additional study of tank 4. Percent Error from Required (%) ".5".75" 1" 1.25" Post Coupler Number Figure Post coupler flow rate variance (from required value) versus post coupler number for five different sub-manifold inner diameters for DTL tank 4. 86

87 Figure 3.23 indicates that a sub-manifold diameter of 1.0 inch will introduce less than 1 % error in the system. The results from this additional analysis indicate all post coupler sub-manifolds require an internal diameter of 1.0 inch. Summary From the post coupler flow modeling studies discussed above, the following key results were obtained Water transfer line connection placement has little effect on the overall error if a submanifold diameter of at least 1.0 inches is employed. A post coupler sub-manifold with an internal diameter of 1.0 inch or greater will be required to achieve proper flow distribution in the DTL. A post coupler transfer water line with an internal diameter of 1.0 inch or greater will be required to provide acceptable pressure drops and minimize erosion. The pressure drop across the post coupler branch was calculated to be 3.7 psi. Identical post coupler sub-manifold diameters can be used for all six DTL tanks Dipole Magnet Sub-model Description The first step in modeling the magnet cooling system was to determine the locations where pressure drop was expected. Figure 3.24 labels the pressure loss locations for a typical DTL magnet and Table 3.7 assigns a value to each loss location. 87

88 Outlet Branch Tee Inlet Branch Tee Unions & Hose Adapters Magnet Figure 3.24 Representation of a DTL magnet s water flow path. Table 3.7. Loss values for the magnet cooling circuit on DTL tank 3. Loss Description Description Quantity Value K1 Inlet Tee.75 straight to Branch K2 Branch to Union.25 to K2 Union to Adapter.19 to.19 0 K3 Adapter to Tubing.19 to K4 Tubing to Adapter.25 to K5 Adapter to Reducing Union.19 to K6 Reducing Union to Magnet.12 to Dpmag Magnet Pressure Drop psi K7 Magnet to reducing Union.125 to K8 Reducing Union to Adapter.12 to K9 Adapter to Tubing.19 to K10 Tubing to Adapter.25 to K11 Adapter to Union.19 to K12 Union to Branch.19 to K13 Outlet Tee.25 Branch to.75 1 Straight Kadd Additional losses 1 88

89 Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed. where, P = pressure drop K = loss factor ρ = density V = velocity i = loss factor number P Ki = KρV (3.10) The magnet flow resistance value given in Table 3.7 was converted into a pressure loss using the following relation: where, P = pressure drop across magnet (psi) Fr = Flow rate through magnet (gpm) R = Resistance of flow through magnet (psi/gpm 2 ) 2 P mag = R * Fr (3.11) Finally, the pressure losses that occurred in series through each sub-manifold branch were summed as follows: P branch = P mag + P K1 + + P K13 (3.12) The total pressure drop across a single dipole magnet branch was calculated to be approximately 12.5 psi. Next, equivalent overall loss or K factors were calculated to represent each branch of the dipole magnet sub-manifold. This was accomplished by rearranging Equation (3.10), substituting P branch for P k, and solving for K. 89

90 Once the equivalent loss factors were determined, a SINDA/FLUINT model was developed for the dipole magnet subsystem. A graphical representation of this model is shown in Figure Notice in Figure 3.25, that the SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). Path lines are described with a T meaning Tube, a L meaning loss to account for fittings, bends, reducers, etc, or a VF meaning a constant volume pump. The equivalent loss or K factors for a dipole magnet branch, were incorporated into the loss, or L components, of the model shown in Figure Figure SINDA/FLUINT model of the magnet cooling circuit for DTL tank 3. Since flow must be distributed evenly to each magnet, the ideal location for the supply inlet exists at the center of the sub-manifold where flow evenly divides. Similarly, the outlet may be ideally located at the center where flow converges evenly from both sides of the return sub-manifold. 90

91 Dipole Magnet Sub-model Results Figure 3.26 is the SINDA/FLUINT model results for a sub-manifold diameter of 0.5 inches. As the figure indicates, the pressure drop as well as the flow rate, is nearly equal for each magnet. Figure SINDA/FLUINT flow and pressure predictions for the DTL magnet water circuit with a 0.5 inch sub-manifold diameter (Units: Pressure = Pascals, Flow Rate = kg/s). 91

92 Figure 3.27 shows that a sub-manifold with a diameter of 0.25 inches may allow over 0.2 % deviation in required flow rate whereas a sub-manifold with a diameter of 0.75 inches will allow a maximum deviation of less than 0.05 %. A sub-manifold with a minimum diameter of 0.5 inches allows less than.05 % error and will be suitable for use in the system. Percent Error from Required (%) ".5".75" Magnet Number Figure Magnet flow rate variance (from required value) versus magnet number for three different sub-manifold inner diameters for DTL tank 3. Summary From the dipole magnet flow modeling studies discussed above, the following key results were obtained Water transfer line connection placement has little effect on the overall error if a submanifold diameter of at least 0.5 inches is employed. 92

93 A sub-manifold with a diameter of 0.5 inches or greater diameter will provide uniform flow to the magnets. A magnet transfer water line with an internal diameter of 0.25 inch or greater will be provide acceptable pressure drops and minimize erosion. The pressure drop across a magnet water line branch was calculated to be 12.5 psi. The identical sub-manifold diameter can be used for all magnet cooling circuits on all six DTL tanks. 93

94 Tank Wall Sub-model Description The first step in modeling the tank wall cooling system was to determine the locations where pressure drop was expected. A value of.018 psi/gpm 2 /m was given for flow resistance of each cooling channel in Reference [1.2]. Figure 3.28 displays the locations for pressure loss components and Table 3.8 lists the particular loss coefficient values. Kbend Kadptcont Kadptexp Kexpansion Kcontraction Kbranch Figure Representation of a DTL tank wall s supply/return water line. 94

95 Table 3.8. Loss factors for the DTL tank wall cooling circuit. Symbol Description Size (inches) Quantity Individual K factor K1 Inlet Tee 1.5 Straight to Total K factor Branch K2 Tee to Union.5 to K3 Union to Adapter.41 to K4* Adapter.41 to K5* Adapter to Tubing.402 to K6* Tubing to Adapter.5 to K7* Adapter.402 to K8* Adapter to Elbow.41 to K9* Elbow K10* Elbow to T.W..62 to K11* Tank Wall Resistance 3 See Table K16* T.W. to Elbow.5 to K17* Elbow K18* Elbow to Adapter.5 to K19* Adapter.41 to K20* Adapter to Tubing.402 to K21** Bend in Tubing K22** Tubing to Adapter.5 to K23 Adapter.402 to K24 Adapter to Union..41 to K25 Union to Tee.41 to K26 Outlet Tee.5 Branch to 1.5 Straight ** Occur twice in DTL Tank 3. *Occur for each of the 3 sections that comprise DTL Tank Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed. where, P = pressure drop K = loss factor ρ = density P Ki = KρV (3.13) 95

96 V = velocity i = loss factor number The tank wall resistance values given in Table 3.8 was converted into a pressure loss using the following relation: where, P = pressure drop through tank wall (psi) Fr = Flow rate through tank wall (gpm) R = Resistance of flow through tank wall (psi/gpm 2 ) L = Length of cooling channel 2 P tw = R *Fr * L (3.14) Finally, the pressure losses that occurred in series through each sub-manifold branch were summed as follows: P branch = P tw + P K1 + + P K13 (3.15) The total pressure drop across a single branch was calculated to be approximately 8.5 psi. Next, equivalent overall loss or K factors were calculated to represent each branch of the tank wall sub-manifold. This was accomplished by rearranging Equation (3.13), substituting P branch for P k, and solving for K. Once the equivalent loss factors were determined, a SINDA/FLUINT model was developed for the tank wall cooling circuit, as shown in Figure From Figure 3.29, notice that the SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). Path lines are described with a T meaning Tube, a L meaning loss to account for fittings, bends, reducers, etc, or a VF meaning a constant volume pump. The equivalent loss or K factors for a tank wall branch, were incorporated into the loss, or L components, of the model shown in Figure

97 Figure SINDA/FLUINT model of the tank wall cooling circuit Tank Wall Sub-model Results Figure 3.30 displays the SINDA/FLUINT model flow and pressure predictions for a DTL tank wall cooling circuit with a sub-manifold diameter of 1.5 inches. As the figure indicates, the pressure drops and flow rates across each tank wall cooling passage branch, are identical. 97

98 Figure SINDA/FLUINT model flow and pressure predictions of the DTL tank wall cooling circuit with a 1.5-inch diameter sub-manifold (Units: Pressure = Pascals, Flow Rate = kg/s). Figure 3.31 is a plot of flow variance (% difference of predicted vs. desired flow) in the tank wall cooling lines as a function of the line number over a range of submanifold line diameters. Figure 3.31 shows that a sub-manifold with a diameter of 0.75 inches will produce over 15 % deviation in required flow rate, while a sub-manifold with a minimum diameter of 1.5 inches provides less than 1% error and will be suitable for use in the cooling circuit. 98

99 Tube Diameter vs. Percent Error 20 Percent Error from Required (%) " 1" 1.25" 1.5" Tank Line Number Figure Tank wall flow rate variance (from required value) versus tank line number for four different sub-manifold inner diameters for DTL tank 3. Summary A sub-manifold with a diameter of 1.5 inches or greater diameter will provide uniform flow to the tank wall cooling passages. A tank wall transfer water line with an internal diameter of 1.5 inch or greater will provide acceptable pressure drops and minimize erosion. The pressure drop across a tank wall water line branch was calculated to be 8.5 psi. The identical sub-manifold diameter can be used for all tank wall cooling circuits on all six DTL tanks End Wall Sub-model Description The first step in modeling the end wall cooling system was to determine the locations where pressure drop was expected. A value of psi/gpm 2 for flow resistance in each end wall was taken from Reference [1.2]. Figure 3.32 is a 99

100 representation which shows the main components comprising the end wall cooling system. Table 3.9 gives the loss values associated with each component. Tees Inlet Adapter Union Outlet End Wall Figure Representation of a DTL end wall cooling water flow path. 100

101 Table 3.9. Losses for flow through the end wall cooling system. Loss Description Size (inches) Quantity Value K1 Tee to Union.25 to K2 Union to Adapter.28 to K3 Adapter.28 to K4 Adapter to Tubing.19 to K5 Tubing to Adapter.25 to K6 Adapter.19 to K7 Adapter to Union.28 to K8* Union to S.T..28 to Rend wall Resistance psi/gpm 2 K9* S.T. to Union.25 to K10 Union to Adapter.28 to K11 Adapter.28 to K12 Adapter to Tubing.19 to K13 Tubing to Adapter.25 to K14 Adapter.19 to K15 Adapter to Union.28 to K16 Union to Outlet Tee.28 to * Values change as transfer line size increases to account for contraction and expansion Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed P Ki = KρV (3.16) where, P = pressure drop K = loss factor ρ = density V = velocity i = loss factor number The end wall resistance value given in Table 2 was converted into a pressure loss using the following relation: 101

102 where, P = pressure drop across end wall (psi) Fr = Flow rate through end wall (gpm) R = Resistance of flow through end wall (psi/gpm 2 ) 2 P tw = R *Fr * L (3.17) The pressure losses that occurred in series through the end wall system were summed as follows: P branch = P ew + P K1 + + P K16 (3.18) A pressure drop of approximately 16.4 psi was calculated for each end wall. Next, equivalent overall loss or K factors were calculated to represent each branch of the tank endwall circuit. This was accomplished by rearranging Equation (3.16), substituting P branch for P k, and solving for K. Once the equivalent loss factors were determined, a SINDA/FLUINT model was developed for the tank endwall cooling circuit, as shown in Figure From Figure 3.33, notice that the SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). Path lines are described with a T meaning Tube, a L meaning loss to account for fittings, bends, reducers, etc, or a VF meaning a constant volume pump. The equivalent loss or K factors for a tank wall branch, were incorporated into the loss, or L component, of the model shown in Figure

103 Figure SINDA/FLUINT model of the DTL end wall cooling system End Wall Sub-model Results The results for this study are as expected. Since there is only one path for the fluid to flow through, there is no concern that the required quantity of cooling water will reach the end walls. Pressure loss was the criterion used in selecting the diameter of the transfer lines to the end walls. Figure 3.34 plots the relationship between pressure drop through the end wall system and transfer line diameter. 103

104 Pressure Drop (psi) Transfer line Diameter (inches) Figure DTL end wall cooling system pressure drop versus line diameter. From Figure 3.34, it is shown that a large system pressure drop occurs when the transfer line diameter is less than inches. A pressure drop of just over 18.3 psi occurs when a line diameter of 0.25 inches is employed whereas a pressure drop of approximately 16.5 psi occurs at diameters of inches and above. Therefore, a minimum internal diameter of inches is required to produce a reasonable pressure drop in the end wall system. 104

105 Drive Iris Sub-model Description The first step in modeling the drive iris cooling system was to determine the locations where pressure drop was expected. A value of psi/gpm 2 for flow resistance in the drive iris was taken from Reference [1.2]. Figure 3.35 is a representation of the main components comprising the drive iris cooling circuit. Table 3.10 gives the loss values associated with each component. Outlet Branch Tee Inlet Branch Tee Unions & Hose Adapters Figure Representation of a DTL drive iris s water flow path. 105

106 Table Losses for flow through the drive iris cooling system. Loss Description Size (inches) Quantity Value K1 Tee to Union.25 to K2 Union to Adapter.28 to K3 Adapter.28 to K4 Adapter to Tubing.19 to K5 Tubing to Adapter.25 to K6 Adapter.19 to K7 Adapter to Union.28 to K8* Union to S.T..28 to Rend wall Resistance.535 psi/gpm 2 K9* S.T. to Union.25 to K10 Union to Adapter.28 to K11 Adapter.28 to K12 Adapter to Tubing.19 to K13 Tubing to Adapter.25 to K14 Adapter.19 to K15 Adapter to Union.28 to K16 Union to Outlet Tee.28 to * Values change as transfer line size increases to account for contraction and expansion Once all losses were characterized, the pressure drop associated with each loss factor was calculated. The following equation was employed. where, P = pressure drop K = loss factor ρ = density V = velocity i = loss factor number P Ki = KρV (3.19) The drive iris resistance value given in Table 2 was converted into a pressure loss using the following relation: 2 P tw = R *Fr * L (3.20) 106

107 where, P = pressure drop across drive iris (psi) Fr = Flow rate through drive iris (gpm) R = Resistance of flow through drive iris (psi/gpm 2 ) The pressure losses that occurred in series through the drive iris system were summed as follows: P branch = P di + P K1 + + P K16 (3.21) A pressure drop of approximately 1.54 psi was calculated for the drive iris. Next, equivalent overall loss or K factors were calculated to represent each branch of the drive iris cooling circuit. This was accomplished by rearranging Equation (3.19), substituting P branch for P k, and solving for K. Once the equivalent loss factors were determined, a SINDA/FLUINT model was developed for the drive iris cooling circuit, as shown in Figure From Figure 3.36, notice that the SINDA/FLUINT models contain junctions (where pressure is calculated) that are connected by path lines (where flow is calculated). Path lines are described with a T meaning Tube, a L meaning loss to account for fittings, bends, reducers, etc, or a VF meaning a constant volume pump. The equivalent loss or K factors for a tank wall branch, were incorporated into the loss, or L component, of the model shown in Figure

108 Figure SINDA/FLUINT model of the DTL drive iris cooling circuit Drive Iris Sub-model Results The results for this study are as expected. Since there is only one path for the fluid to flow through, there is no concern that the required quantity of cooling water will reach the drive iris. Pressure loss was the criterion used in selecting the diameter of the drive iris supply and return lines. Figure 3.37 plots the relationship between pressure drop through the drive iris circuit and transfer line diameter. 108

109 Pressure Loss Vs. Tube Diameter 7.3 Pressure Loss (psi) Series Tube Diameter (in.) Figure Drive iris cooling system pressure drop versus line diameter. From Figure 3.37, it is shown that a large system pressure drop occurs transfer line diameters less than 0.5 inches. It may be apparent that a minimum diameter of 0.5 inches is required to fulfill this study s goals (minimize pressure drop), however, since the pressure drop induced by the (5/16") inch diameter line is still less than that required for the DTL tank 3 system, the inch diameter line will be adequate for use in cooling the drive iris. To be more specific, when all subsystems are combined in parallel to form the entire cooling system for DTL tank 3, it is required that a uniform pressure drop occurs across all systems. To achieve this, globe valves in place near the entrances of each subsystem are opened or closed as needed. For the drive iris system, some of the needed pressure drop is introduced by using a inch diameter line instead of a 0.5 inch diameter. This means that less pressure drop occurs across the globe valve. 109

110 Tank 3 Global Model Design Studies/Results After the pressure loss and flow distribution through each DTL subsystem (drift tubes, post couplers, etc.) was determined, the results were incorporated in the DTL tank 3 global model, described previously in Figures 3.4 and 3.5. Also required for the DTL tank 3 global model, were the flow resistances of the flow meters, valves, and plumbing components not included in the individual submodels. Table 3.11 summarizes pressure drops and loss factors for each of the DTL subsystems, as well as the additional components needed for the tank 3 global model. Note that since each DTL subsystem is connected to common supply and return manifolds, the total pressure drop across each DTL subsystem is required to be equivalent. This is achieved by using a proportional globe valve on the supply line to each DTL subsystem. By adjusting the globe valve loss factors, the flow and pressure drop across each subsystem could be adjusted to its correct value. Table 3.11 displays the required globe valve loss factors (column 7) and the total pressure drop of psi across each subsystem. From the system pressure drop, transfer line diameters, and flow velocities of the DTL subsystems listed in Table 3.11, equivalent loss (K) factors were determined for each subsystem branch. These K factors, listed in the last column of Table 3.11, were input into the subsystem branch losses, or L s, of the DTL tank 3 global SINDA/FLUINT model, displayed previously in Figure 3.5. Using the global SINDA/FLUINT model, a series of runs were made with varying supply and return manifold diameters from 2 inch to 4 inches in 1/2 in. increments. Figure 3.38 displays the results of this trade study. Figure 3.38 shows that as the main manifold diameters increase, the percent difference in actual flow rate to the required flow rate, summarized across all DTL subsystems, decreases. This results because the pressure in the manifold becomes increasingly uniform along its axis as its diameter increases. Or in other words, as the manifold pressure becomes more uniform, the flow distribution within a given drift tube segment converges on the design flow rate. For a manifold diameter of 3 inches (ID) or greater, the maximum flow deviation decreases to an acceptable level of less than 1%. Unfortunately, the use of a 3 inch main manifold creates flow velocities in excess of 3 m/s, which may cause undesired erosion of the manifold walls. Therefore, it is recommended that at least a 3.5 inch (ID) main manifold diameter is used to drop the 110

111 mean water velocity below 3 m/s in the manifolds. Figure 3.39 shows the results of a SINDA/FLUINT calculation with a main manifold diameter of 3.5 inches. In Figure 3.39, the color red denotes the highest value for both pressure and flow with blue denoting the lowest pressure and lowest flow rate. The total flow rate within the model is gpm. The supply and return manifolds have been set to an inside diameter of 3.5 inches. As seen in Figure 3.38, there is a system pressure loss of approximately 21 psi. 111

112 Table Summary of the optimized line sizes and pressure losses in the various subsystems of the DTL tank 3 cooling system. System Information System ID Transfer Line Size (in) Flow Rate (gpm) Velocity (m/s) S/F models Tees & FrictionGlobe Valves Flow Meters Subsystem Other k Globe Globe Valve k Flow Flow Meter Pressure Pressure Valve Pressure meter Pressure Loss (psi) Losses (psi) Loss (psi) Loss (psi) Branch Pressure Loss (psi) Globe Valve Pressure Drop (psi) Total Pressure Loss (psi) End Wall Tank Wall Post Coupler Post Coupler Drive Iris Slug Tuner Drift Tubes Magnets Tank Wall End Wall RF Window k Total 112

113 6 Percent Error in Required Flow Rate (%) End Wall 1 Tank Wall 1 Post Coupler 1 Post Coupler 2 Drive Iris RF Window Slug Tuner Drift Tubes Magnets Tank Wall 2 2.5" 3" 3.5" 2" End Wall 2-3 DTL Subsystem Figure Percent difference between actual and required water flow rates for each DTL sub-system in DTL tank 3 for various main supply and return manifold diameters. Figure SINDA/FLUINT model predictions of the flow rate and pressure distribution in DTL tank 3 with 3.5 inch (ID) main manifolds (Units: Pressure = Pascals, Flow rate = kg/s). 113

114 Summary From the SINDA/FLUINT modeling studies discussed above, the following key results were obtained Sub-manifold and transfer line diameters as well as pressure drops for all the DTL subsystems have been optimized and are listed in Table Connection locations for transfer lines on sub-manifolds for the DTL subsystems have been optimized and are listed in Sections through Locations for main supply and return manifold water transfer line connections should be made near the midpoints of the manifolds where flow evenly splits/converges. The DTL main supply and return manifold diameters must be a minimum of 3.5 inches to produce a maximum deviation of less than 1% between actual and required flow rates of the subsystem components. A pressure drop of 21 psi is produced across the DTL tank 3 water cooling circuit (from the main supply manifold to the main return manifold). 114

115 3.1.2 DTL Water Skid Design Goals The water skid is responsible for delivering cooling water to the DTL structure. It must actively adjust the temperature of the water sent to the DTL by manipulating a control valve and bypassing an appropriate quantity of water through a heat exchanger to be cooled. The design goal of this work is to size heat exchangers, pumps, and line sizes for operation of the SNS DTL water skid system. This task involves designing a system that provides the necessary water flow and water temperature to support cooling of each DTL tank while minimizing pressure losses and material costs. Table 3.12 summarizes the goals for this study. Table Water skid cooling system goals. Water Skid Model Design Goal Outcome Optimize line diameters in skid to minimize pressure drop, erosion, costs, and ease the manufacturing of the plumbing. Size heat exchanger for heat load and flow rates. For the heat exchanger, develop a relationship between the hot side flow rate and overall heat transfer coefficient. Determine the pressure drop through the skid for mean flow conditions (combine with flow loop model to determine pressure drop across the pump). Determine the pressure drop versus flow rate required for the proportional control valve to give needed temperature control (20 C +/- approximately 5 C). Determine the water skid pressure drop variance as a function of control valve position (and hence flow rate variance from the pump) and determine if action is required to maintain constant water in the loop. Size pumps (based on flow rate and pressure drop for DTL tank 3). Line diameters within the skid to connect the main components will be 3.0 inch (ID) tubing constructed of stainless steel. A 10 inch x 20 inch FlatePlate heat exchanger with 70 plates. From the data supplied by the manufacturer, relationship was developed using a fifth degree polynomial curve fit. The DTL Tank 3 cooling system produces a total pressure loss of approximately 45 psi. When operating at worst case (Tmix =14 o C) pressure drop is 55 psi. See Figure 7 for pressure loss across the pump with respect to heat exchanger flow rate. See Figure 10. See Table 5 for pump specifications. 115

116 Design Specifications The design specifications for the DTL water skids were taken from the SNS Drift Tube Linac and Coupled Cavity Linac Water Cooling and Resonance Control System Description Document [1.2]. During steady state, full RF power, the target operating temperature of cooling water delivered to the each DTL tank is specified to be / C, with an operational range between 14.9 C and 25.1 C required for resonance control. The heat loads as well as the cooling water flow rates and temperatures for each of the six DTL water cooling skids are summarized in Table Table Nominal heat loads, total water flow rates, and water supply temperature ranges for the DTL water skids. Water Skid Mean Heat Total Cooling Mean Water Load (kw) Water Flow Rate (gpm) Supply Temperature ( C) Water Supply Temperature Range ( C) DTL Tank to 25.1 DTL Tank to 25.1 DTL Tank to 25.1 DTL Tank to 25.1 DTL Tank to 25.1 DTL Tank to Model Description The water skid serves as the water supply for the RF structures and thus acts as a key element in the closed-loop water cooling system. As discussed previously, the primary water skid components consist of a heat exchanger, variable speed pump, expansion tank, water purification system, and a control valve. A simplified schematic of the basic water skid components are shown in Figure 3.40(a). A variable speed pump was incorporated in the design to maintain a constant desired flow rate. As discussed previously, water temperature control is maintained by adjusting the proportion of the total system water flow between the heat exchanger and the heat exchanger by-pass line. This is achieved by use of an electrically actuated control valve located on the heat exchanger by-pass line. Since the focus of this study deals with simulation of the water skid s pressure drop and temperature control, the bypass water purification system was neglected in the current model. Additional water skid 116

117 features include control valves, temperature transducers, pressure transducers, and flow meters, which are strategically placed in the system to provide a way of controlling and monitoring the temperature, flow, and pressure through the system. The objective of this analysis was to calculate the system parameters (flow, pressure, and temperature) needed to size the plumbing and hardware components on the water cooling skid. As described earlier, the SINDA/FLUINT computer code was used to develop a network model of the water skid components. Figure 3.40(b) is the SINDA/FLUINT model representation of the water skid flow diagram shown in Figure 3.40(a). The water skid model in Fig. 3.40(b) is a numerical description of the system shown in Fig. 3.40(a). Recall from the earlier description that the SINDA/FLUINT code models a system as a combination of lumped-parameters. The fluid network is comprised of flow lengths called paths that are joined at ends by points called junctions. In a similar manner, the thermal part of the code uses conductors to describe thermal flow paths and is joined at the ends at points called nodes. Values of mass flow rate and energy flow rate are obtained from the path lines while the junctions and nodes give values of pressure and temperature. The heat conductor lines are labeled as HUS and HN and describe certain properties of the heat flow conductor. These connections are especially important for the heat exchanger portion of the system. The diamond shaped symbols represent connection points between the heat exchanger and the cold loop. The triangle system is a plenum reference point for the hot side and the cold side of the system. The simulation introduces the heat at only one point in the return manifold. This approach was taken since the focus of the analysis was the water temperature control capabilities associated with the water skid and not on the details of the heat transfer in the RF structure. 117

118 R F Structure Heat Input R F Structure FM Filter Proportional Control Valve Pump Filter Hot side of Heat Exchanger FM P P Heat Exchanger By-Pass Proportional Control Valve T T Variable-Speed Pump WP FM T P T P WP FM Water Purity Transducer Flow Meter Heat Exchanger T P Temperature Transducer (RTD) Pressure Transducer Facility Chilled Water Outlet Facility Chilled Water Inlet Cold side of Heat Exchanger (a) (b) Figure (a) Flow diagram and (b) corresponding SINDA/FLUINT model of the water skid. 118

119 Loss Factors An important step in developing the SINDA/FLUINT modeling, was to properly account for all of the pressure loss components within the water skid. All of the fittings, valves, filters and instrument parts and their associated loss factors are identified in Figure 3.41 and Table These plumbing components were accounted for by placing their loss factor, K, in the pipe or path lines of the SINDA/FLUINT water skid model. A modeling simplification was made to account for the flow resistance of the entire system of DTL tank cooling lines as a total pressure loss, L. The value of the pressure loss was taken from Section where the pressure drop across the RF structure was calculated. This pressure loss was used to determine an equivalent K factor, which was used as input to the SINDA/FLUINT model. It was also assumed that for the analysis, a constant volume pump could be used to simulate the performance of a variable speed pump to supply a constant flow rate. To accurately represent the system pressure drop effects of the 70 plate heat exchanger in the SINDA/FLUINT model, pressure drop values at different flow rates across the hot side of the heat exchanger were obtained from the manufacturer. The individual pressure drops were transformed into loss factors using the Equation A 1 inch diameter line was assumed to calculate the fluid velocities. where, P = pressure drop K = loss factor ρ = density V = velocity P HE = KρV (3.22) 119

120 Water Skid K Factors Kskid2 R F Structure Kskid27 Kskid26 FM Kskid24 Kskid3 Kskid5 Filter 60 mesh Kskid23 Kskid22 Kskid20 Kskid4 FM Kskid6 Kskid7 Kskid15 Kskid16 P Kskid17 P Kskid18 Kskid21 By-Pass Proportional Control Valve Kskid14 Drain Kskid19 Variable-Speed Pump T T WP Kskid25 Kskid8 Kskid9 FM Kskid10 T P Heat Exchanger T Kskid11 Kskid12 P Kskid13 WP FM T P Water Purity Transducer (Ph, elect. Cond., Diss. O) 2 Flow Meter Temperature Transducer (RTD) Pressure Transducer Facility Chilled Water Outlet Facility Chilled Water Inlet Figure Simplified flow diagram of a typical DTL water skid. 120

121 Table Summary of loss factors for a typical DTL water skid (refer to Fig. 3.41). Symbol Description Diameter (in) Length (m) K factor Kskid1 DTL tank 3 total Loss Kskid2 Supply Transfer Line S/F friction Kskid3 Ball Valve (100% open) 3 NA.15 Kskid4 Flow Meter 3 NA 6.4 Kskid5 Skid Water Line S/F friction Kskid6 Proportional Control Valve Variable Kskid7 Ball Valve (100% open) 3 NA 0.15 Kskid8 2 two diameter bends Kskid9 Flow Meter 3 NA 4.3 Kskid10 Temp/Press Trans Port 3 NA 0.1 Kskid11 2 two diameter bends 3 NA 1.8 Kskid12 2 two diameter bends 3 NA 1.8 Kskid13 Temp/Press Trans Port 3 NA 0.1 Kskid14 Globe Valve (50% open) 3 NA 6.1 Kskid15 1 two diameter bend Kskid16 Temp/Press Trans Port 3 NA 0.1 Kskid17 Size Transition (2.5 to 3 ) Kskid18 1 two diameter bend 3 NA 0.9 Kskid19 Temp/Press Trans Port Kskid20 Ball Valve (100% Open) 3 NA 0.15 Kskid21 1 two diameter bend Kskid22 Strainer/filter Kskid23 Ball Valve (100% open) 3 NA 0.15 Kskid24 Flow Meter 3 NA 6.4 Kskid25 PH/O2 Measurement Port Kskid26 Ball Valve (100% open) 3 NA 0.15 Kskid27 Return Transfer Line S/F friction 121

122 The loss factors associated with each flow rate were plotted against heat exchanger flow rate to produce a relationship as given in Figure Note that y represents the loss factor and x represents the heat exchanger hot side flow rate y = -2E-05x x x x x Loss Factor (k) Flow Rate (kg/s) Figure Loss factor vs. mass flow rate for the hot side water flow in the 70 plate heat exchanger. The relationship in Figure 3.42 was put into the SINDA/FLUINT model to represent the pressure drop across the heat exchanger. The proportional valve to control the flow was modeled as a globe valve that could be adjusted as needed to obtain the desired flow through the heat exchanger (and hence obtain the desired water mix temperature). This operation will be discussed in further detail below. Heat Exchanger As discussed previously, a closed loop water cooling system extracts heat from the RF structure and transfers it to a facility chilled water supply via a liquid-liquid heat exchanger, as depicted in the flow diagram of Figure 3.40(a). In this closed-loop circuit, water temperature control is achieved by manipulating the hot-side (Linac side) heat 122

123 exchanger water flow rate while holding the cold-side water inlet temperature and flow rate constant. This is achieved by using a proportional control valve that divides the circulating water between the heat exchanger and by-pass line. By changing the hot-side water flow rate, the overall heat transfer coefficient of the heat exchanger is varied. Since the heat removal rate must effectively remain constant for quasi-steady-state conditions (heat rate into system equals heat rate out of the system), the hot-side water temperature must change inversely to the overall heat transfer coefficient to achieve a new operating condition. Consequently, increasing the water flow through the heat exchanger results in an increase in the overall heat transfer coefficient, and an associated decrease in the mean water temperature. And conversely, decreasing the water flow through the heat exchanger results in a decrease in the overall heat transfer coefficient, and an associated increase in the mean water temperature. The cooling performance of the water skid, including water temperature range, accuracy, resolution, and stability, will be highly dependent on the design choice made for the liquid-liquid heat exchanger. The heat load and flow requirements that the DTL water cooling systems are being designed to were summarized previously in Table Note that the most significant variable in the resonance control is the water temperature being delivered to the RF structure. In the case of the six tanks, the mean water delivery or mixture temperature, T mix, was specified to be 20.0 C, with a required range of ±6 C about this mean value. In addition to those parameters listed in Table 3.13, the pressure drop across the heat exchanger needed to be kept below 10 psi for maximum flow rates for the cold side and 5 psi for the hot side. The water inlet temperature on the cold side of the heat exchanger was specified as 7.2 C ± 0.5 C. The next step, prior to initiating the numerical studies, was to size a commercially available heat exchanger so that its performance could be included in the SINDA/FLUINT model of the water skid. The steps used to size the heat exchanger were as follows (Refer to Figure 3.43 to aid in the discussion): 1. For a known heat load, cold side inlet temperature, and hot side inlet temperature, determine acceptable cold side and hot side flow rates that give the desired mixture temperature of 20.0 C. The relationships used in these calculations included the 123

124 enthalpy balance on the hot side flow rates (Eqn. 3.23), and the energy balance for the flow on either side of the heat exchanger (Eqn. 3.24) m T c p T mix = m h c p T ho + (m T - m h ) c p T hi (3.23) m c = q/(c p (T co T ci )), (3.24) where the variables are m = mass flow rate, q = DTL Tank 3 heat dissipation: 95000W, c p = heat capacity of water: 4,179 J/kgK, T = water temperature, and the subscripts correspond as follows: c = cold side h = hot side mix = mixture of water to RF structure T = total (to RF structure) o = outlet condition i = inlet condition 2. With the inlet and outlet temperatures known on each side of the heat exchanger for a given set of flow rates and a known heat load, determine the heat exchanger s overall heat transfer coefficient, UA, from Eqn. (3.25) [3.6]. UA = q/((t ho T ci ) (T hi T co ))/ln((t ho T ci )/(T hi T co )) (3.25) 3. Choose a commercial heat exchanger that satisfies the heat transfer coefficient and temperature conditions from steps (1) and (2), while satisfying the pressure drop limitations listed previously. 4. For the heat exchanger selected in step (3), vary the hot side flow rate and repeat steps (1) and (2) for six more operating conditions. The vendor heat exchanger performance criteria are needed to perform this step. From this step, a plot can be generated of the heat exchanger s overall heat transfer coefficient versus hot side flow rate. 124

125 5. Repeat steps (1) through (4) for three other cold side flow rates to generate a family of curves of overall heat transfer coefficient versus hot side flow rate. 6. Size heat exchangers for other DTL tanks. Step (1) m T By-Pass Proportional Control Valve m T m h T hi q T ho m c T co Heat Exchanger T ci Step (4) Overall Heat Transfer Coefficient, UA Heat Exchanger Hot Side Mass Flow Rate Figure Pictorial representation of the heat exchanger sizing for the SINDA/FLUINT modeling of the DTL water skids. 125

126 Using the heat load, flow rate, and water temperature ranges for DTL tank 3, the heat exchanger was sized using the five steps listed above. The particulars of this exercise were as follows: 1. The given variables included a mix water temperature of 20.0 C (delivered to the structure), a hot side water inlet temperature of 21.5 C (this is the temperature of the water leaving the RF structure after heating up from 20.0 C), a cold side inlet temperature of 7.2 C, a heat load of 95 kw, and a total flow rate of gpm (15.17 kg/s). Next, flow rates of 63.4 (4.00 kg/s)and gpm (4.73 kg/s) were chosen for the hot side and cold side flow rates, respectively. This resulted in a hot side outlet temperature of 15.8 C and a cold side outlet temperature of 12.0 C. 2. With the inlet and outlet temperature as well as the heat load specified, the overall heat transfer coefficient of the heat exchanger, UA, was calculated to be 10,506 W/ C. 3. A 10"x20"- 40 plate, counter-flow, compact, multi-pass heat exchanger from FlatPlate Inc was identified by the manufacturer to satisfy these heat transfer conditions. Unfortunately, the pressure loss limits were exceeded. The size of heat exchanger was increased to accommodate the pressure limits. In doing so, an oversized heat exchanger was selected which had approximately 262.6% additional surface area than what was needed to provide for the overall heat transfer coefficient identified in step 2 above. In particular, a flat plate heat exchanger with seventy 10 by20 (FP 10x20-70) stainless steel plates was selected. See the mechanical design section of this report for the "new" sizing procedure. 4. After the heat exchanger was selected, several additional cases with different hot side water flow rates were studied to obtain multiple values of UA. This information is especially important since the variable flow on the hot side is used to control the water temperature returned to the RF structure. Knowing the product of the heat transfer coefficient and the surface area, as well as the flow rate information, an 126

127 EXCEL spreadsheet was used to plot these data, as shown in Fig A polynomial least squares fit of the data produced the following relationship. UA = m h m h m h m h m h ,(3.26) where UA is the product of heat transfer coefficient times the area, (W/ C) and m h is the heat exchanger hot side water mass flow rate (Kg/s). At this point the heat exchanger sub-model was ready for input into the SINDA-FLUINT model of the water skid. HUS heat transfer ties in the heat exchanger need a heat transfer versus flow rate relationship and Eqn. (3.26) provides the HUS heat transfer tie to determine the heat transfer as a function of hot side flow rate. Here the heat transfer tie (HUS) UA value was adjusted to dissipate the heat between the eight nodes of the SINDA/FLUINT model s heat exchanger. 5. Steps (1) through (4) were repeated for cold side flow rates of gpm (5.97 kg/s), gpm (3.91 kg/s), and gpm (3.34 kg/s) as given in Table Sized heat exchangers for other 5 DTL tanks. See section of this report on mechanical design. Table Cases studied for heat exchanger sizing on DTL tank 3. Hot Side Flow Rate (kg/s) Hot Side Flow Rate (gpm) Tco (deg C.) Cold Side Flow Rate(kg/s) Cold Side Flow Rate(gpm) Tco (deg C.) Cold Side Flow Rate(kg/s) Cold Side Flow Rate(gpm) Tco (deg C.) Cold Side Flow Rate(kg/s) Cold Side Flow Rate(gpm) Tco (deg C.) Cold Side Flow Rate(kg/s) Cold Side Flow Rate(gpm)

128 Figure 3.44 displays the heat exchanger overall heat transfer coefficient (W/ C) versus hot side mass flow rate (kg/s) for all cases listed in Table Overall Heat Transfer Coefficient y = x x x x x y = x x x x x y = x x x x x y = x x x x x Flow Rate (kg/s) Figure Overall heat transfer coefficient (W/ C) versus heat exchanger hot side mass flow rate (kg/s) for four heat exchanger cold side outlet temperatures for DTL tank Design Studies/Results Temperatures Figure 3.45 and show the temperature predictions for the hot and cold side of the heat exchanger at normal operating conditions (Tco =12 o C, Tho = 20 o C) for the water skid associated with DTL tank

129 Figure SINDA/FLUINT temperature predictions on the hot side of the heat exchanger for the water skid on DTL tank 3. In Figure 3.45, notice the heat distribution through the system. A sudden rise in temperature occurs at junction 1. This is where waste heat from DTL tank 3 is introduced into the system (95kW). The heated water is carried from junction 1 to junction 3 where approximately 27.3 gpm (1.72 kg/s) of the flow is diverted to the heat exchanger loop. This is done using control valves labeled as CT. To achieve the correct flow to the heat exchanger and a Tmix (temperature at junction 10) temperature of 20 o C, the K factor employed at the CT following junction 3 was set to In addition, the CT following junction 9 was set to 100. At junction 10, flow from the heat exchanger loop, after being cooled, is recombined with the bypassed flow. Figure 3.46 represents the cold side of the heat exchanger. Notice that the temperatures range from 7.2 o C at the inlet to 12 o C at the outlet. A cold side flow rate of approximately 75 gpm (4.72 kg/s) was employed to achieve the 12 o C outlet temperature. 129

130 Figure SINDA/FLUINT temperature predictions for the cold side of the heat exchanger on the DTL tank 3 water skid. The solver option in Sinda/Fluint was employed to adjust the K factor value applied to the control valve, located immediately after junction 3, to achieve several different flow rates, from17.9 to 95.2 gpm (1.13 to 6 kg/s), through the heat exchanger. Doing this allows T mix to satisfy the required temperature range of 20.0 o C +/- 5.0 o C. Figure 3.47 shows the relationship between T mix and the hot side flow rate through the heat exchanger. Notice that valve control is more sensitive when the flow rate through the heat exchanger is low. The curve is allowed to flatten out by changing the cold side flow rate. Figure 3.47 also shows the additional temperatrue curves for the various cold side flow rates. 130

131 70 Plate Heat Exchanger for DTL Tank 3 Tmix (deg. C) Flow Rate (kg/s) Figure T mix versus heat exchanger hot side flow rate for DTL tank 3. Pressure Drop A relationship of induced pressure versus flow rate through the hot side of the heat exchanger is given in Figure Note that a 3 inch diameter line was employed for all water skid lines, excluding the 4 inch transfer line from the water skid to the DTL RF structure. 131

132 70 60 Pressure Drop (psi) Pump Heat Exchanger Flow Rate through Heat Exchanger (kg/s) Figure Induced pressure drop versus flow rate through the hot side of the heat exchanger for Tco = 12 o C. With exception to the transfer lines between the water skid and the main manifolds (4 ich), all lines within the water skid should be 3 inches internal diameter. Figure 3.49 shows the pressure across the pump versus water skid line diameter. 132

133 Pressure Drop Across Pump (psi) Internal Line Diameter (in) Figure Water skid line size versus pressure drop across the pump at normal operating conditions (Tmix = 20 o C, Tco =12 o C). Once the water skid lines were sized and set at 3" internal diameter, the water skid transfer lines were studied. The transfer lines between the DTL structure and the water skid should be at least 4 inches to keep pressure losses as low as possible. Figure 3.50 shows the relationship between the pressure drop and transfer line diameter. 133

134 58 56 Pressure Drop Across Pump (psi) Transfer Line Internal Diameter (in) Figure Water skid transfer line diameter versus pressure drop across the pump at normal operating condtions (Tmix = 20 o C, Tco = 12 o C) Pump Sizing The water skid pump size was based upon the worst case pressure drop scenario. For the tank 3 water skid, this situation is expected to occur when 4.5 kg/s ( at Tco = 13 deg. C) of cooling water is passed through the heat exchanger and a resulting pressure loss of approximately 55.6 psi is produced across the pump. The water skid pump was sized using the following relations for a total flow rate of 241 gpm (15.2 kg/s). Equation 1 calculates the overall head loss to size the pump. Power = P * Q (3.27) where, DP = Pressure loss in the water skid (psf) Q = Volumetric flow rate (ft 3 /s) 134

135 skid: Equation (3.27) yields the following for the pump power on the DTL tank 3 water Power lbf 144in 15.2kg m ft ft * lbf = * * * * = 4238 = 7.6 horsepower in ft s 998kg m s The total power required to deliver the required water flow is 7.6 horsepower. Please note, however, that pumps are usually only around 75 % efficient and must be sized according to a manufacturer's pump curve specifications. Pumps for all tanks should be sized according to the selected manufacturer's curves for the following specifications in Table Table Pressure drop and flow rate specifications for all six DTL tanks. Tank # Flow rate (gpm) Velocity P across Tank (psi) Add P through system (psi) System k P though water skid w/hx valve at k = 100 and Hx Flow Rate = 4 kg/s Approximate Total DP (psi) Control Valve The control valve is responsible for sending the required amount of water to be sent to the heat exchanger to the system water temperature. Figure 3.51 is a plot which compares the control valve K factor to the flow rate sent to the heat exchanger for DTL tank 3. This curve is produced based upon a cold side outlet temperature of 12 deg. C. 135

136 Control Valve K Factor Flow Rate through Heat Exchanger (kg/s) Figure Control valve K factor vs. flow rate to the heat exchanger for DTL tank Summary Normal operating conditions for tank 3 are achieved when a flow of gpm (1.69 kg/s ) of water is bypassed to the heat exchanger and the cold side flow rate and outlet temperature are 75 gpm (4.72 kg/s) and 12 o C, respectively. A FlatPlate Inc. heat exchanger, model FP 10x20-70, is smallest heat exchanger that may be used on the water skid. All water skid line sizes shall be 3" id min. with exception to the transfer lines to the DTL, which shall be 4" id min. Pressure drop across tank 3 will be a maximum of 60 psi. The total power required to deliver the required water flow is 7.7 horsepower. Please note, however, that pumps are usually only around 75 % efficient and must be sized according to a manufacturer's pump curve specifications. 136

137 3.1.3 SINDA/FLUINT Uncertainty Analysis The DTL water cooling system describe previously, is quite complex. In order to construct manageable SINDA/FLUINT models which yield meaningful results, several modeling simplifications and assumptions were required. Of greatest uncertainty in the SINDA/FLUINT modeling, is the incorporation of empirical flow resistance coefficients to describe the pressure drops across various pieces of plumbing hardware. Omission, or incorrect application of these resistance values, can lead to significant modeling errors. In order to determine the uncertainty in the SINDA/FLUINT modeling of the DTL water cooling system, a comparison was made between flow and pressure drop predictions of a SINDA/FLUINT model and empirical measurements from a prototype water cooling system. Such a comparison will provide an assessment of the accuracy in the modeling technique and the SINDA/FLUINT code. The experimental results were taken from a prototype SNS Linac water cooling system that was fabricated for an R&D effort on a CCL hot model comprised of two RF segments of the CCL. The prototype water skid is shown in Figure 5.4. The flow diagram of the prototype cooling system is shown in Figure The pressure measurement locations are identified by the letter P, and the flow rate measurement locations are designated by FM. The pressure and flow measurement devices that were used in the experiment, had the following specifications: Pressure - pressure transducers (OmegaPX63-100G5V) psig full scale, accuracy is 0.4% full scale, or plus/minus 0.4 psi. Flow in large water lines (1.5 and 2 diameter) - Paddle wheel flow meter: accuracy of +/- 0.2 feet per second (this will have to be converted to gallons per minute by multiplying it by the cross-sectional area of the pipe that the flow meter is inserted). Flow in small lines (1 and diameter) - Turbine flow meter: Accuracy = 1% of reading. 137

138 P4 Fm9 Fm8 Fm7 Fm6 Fm5 Fm4 Fm3 Fm2 FM1 P3 FmT Filter 60 mesh P1 P2 By-Pass Proportional Control Valve Drain Variable-Speed Pump Fmhx FM Flow Meter Heat Exchanger T P Temperature Transducer (RTD) Pressure Transducer Facility Chilled Water Outlet Facility Chilled Water Inlet Figure Flow diagram of the prototype water cooling system for the CCL hot model. 138

139 The SINDA-FLUINT model representation of the prototype water cooling system is shown in Figure A modeling procedure, similar to that described previously in this report, was used in developing and running the model depicted in Figure Figure SINDA/FLUINT model representation of the prototype CCL hot model water cooling system. 139

140 Table 3.17 shows a comparison between the experimental and numerically predicted values of pressure drop and flow rate. The numerically predicted pressure drops across the pump and water manifold system are within 4% of the empirical measurements. The difference can be attributed primarily to the uncertainty in the pressure measurement. The numerically predicted flow rates through the pump and heat exchanger (FmT and Fmhx) were also in good agreement (within 1.6%) with the experimental measurements. The agreement of the numerically predicted and experimentally measured flows in the manifold line distribution system ranged from good to fair. The predicted flows in the large lines (diameter = 1 ) were within 4 to 10 % of the measurements, while the predicted flows in the small lines (diameter = ) differed by 4 to 22 % from the experimental values. It should be noted that the flow split for the SINDA-FLUINT values are exactly equal as they should be for the 1 diameter lines and are all equal for the small diameter lines in the cavity area. The fact that the experimental values do not measure evenly indicates some variance in the experimental apparatus that are not brought out in Table For example FM1 and FM2 should have equal values, however they differ by 3 percent. The small lines (0.375 diameter) should also divide evenly, but they vary by as much as 10 percent. These differences can be attributed to instrumentation accuracy limitations and slight variations in flow-control globe valve settings. In general, the SINDA-FLUINT results compared well with the experimental values (in most cases, better than 10%). No effort was required in tweaking the SINDA-FLUINT model to improve the comparison. Further, the K-factors were taken from textbooks or estimated in the same procedure that was used in the modeling procedure for the DTL and CCL water cooling systems. Consequently, the SINDA- FLUINT modeling for the SNS DTL and CCL water cooling systems should yield acceptable results for accurately sizing pumps, heat exchangers, and plumbing hardware. 140

141 Table Comparison of the SINDA/FLUINT model predictions and experimental measurements of flow and pressure for the prototype CCL hot model water cooling system. Parameter Empirical Value SINDA/FLUINT % Difference Value P2-P psi +/ P3-P psi +/ FMhx 34.5 gpm +/ gpm 0.3 FMT 94.3 gpm +/ gpm 1.6 FM gpm +/ gpm 7.4 FM gpm +/ gpm 10 FM gpm +/ gpm 17 FM4 3.3 gpm +/ gpm 18 FM gpm +/ gpm 4 FM gpm +/ gpm 17 FM gpm +/ gpm 22 FM gpm +/ gpm 4 FM gpm +/ gpm 8 141

142 3.2 DTL Water Cooling Loops Stability and Response Modeling The successful design of the DTL water cooling and resonance control system relies on a number of key operating and performance conditions. These conditions include the temperature and availability of the coolant, how fast the cooling system can react to heat load changes, what happens if the facility water does not have a stable temperature, etc. A delivery system for the coolant was designed and provisions made for gathering the data required to understand whether the design would provide what is required by the cavities under the desired rf loads. To keep the cavity tuned, it is important that both the drift tubes and the tank itself be kept at stable temperatures as both affect the cavity tune. This is complicated by the fact that the drive iris, the slug tuners, the post couplers, and the dipole magnets are all cooled in the same cooling loop. To aid in this design process, a dynamic control model was written that attempts to simulate the process of maintenance of coolant temperature and availability, given the diameters and lengths of the pipes in the cooling system, the characteristics of the control valve, and the characteristics of the heat exchanger. This numerical model was developed in SystemView by Elanix Design Goals The SystemView model allows various dynamic or transients conditions to be investigated, such as the time required to reach a steady-state temperature condition for the RF structure and water cooling system from start-up or a trip in RF power. This is quite different from the steady-state analyses performed with the SINDA/FLUINT models, which were used, in part, for sizing plumbing and hardware. The modeling goals and outcomes for this work are summarized in Table

143 Table SystemView modeling goals for the DTL Tanks. Design Goal Outcome Determine the transit time of water through the cooling loop and components. Estimate the thermal response time of the copper drift tubes and tanks walls to changes in either cooling water supply temperature or RF heat load (i.e., response time per C change in cooling water temperature or per Watt in heat load). Determine the optimization techniques (i.e., shorten pipe run lengths) needed to minimize the response time, or suggest how this can be achieved with the proper PID tuning. Estimate the required start-up time of the RF power and water cooling system to reach a thermally steady-state operating condition. Determine stabilizing capability of cooling system and cavities to a trip in RF power. Create corrective measures to keep cavities from over-cooling (i.e., add heater, open bypass valve fully, close solenoid valve on cold side inlet to heat exchanger, etc.). Estimate acceptable fluctuations (magnitude and frequency) in the chilled water supply temperature and flow rate, that do not have a detrimental impact on the DTL resonance control. Transit times determined and listed on the model figure Model has been developed and run for several case studies. Indications are that the system will perform satisfactorily in its present configuration. Model completed and run for a full RF power-on condition. Model was also run for various off-normal heat loads as would be encountered during cavity conditioning or linac tuning and commissioning resulting in proper resonance control Two conditions were studied, an RF trip of 10 seconds and an RF trip of several hours. The resonance point will probably be at some complicated combination of drift tube and tank wall temperature. These simulations show temperature control but can not show the areas of resonance. The model has been run for several different test cases. Indications are that fluctuations much larger than have been specified will not have a detrimental impact on resonance control Model Description The nodal network SystemView model for the DTL tank 3 water cooling and resonance control system is shown in Figure The model includes nodes to account for the drift tubes, tank walls, post couplers, slug tuners, iris, and dipole magnets. Each of these nodes accounts for the thermal mass of its copper component, as well as the heat 143

144 Figure Schematic representation of the SystemView model of DTL tank 3. transfer coefficient between that component and the cooling water. The numerical model also includes a pump, heat exchanger, variable control valve (controlled by an internal PID algorithm), and all of the necessary plumbing. Further specific details regarding the SystemView model are provided below. Structure temperatures: The temperature of the DTL structures can be calculated from the following differential equation:?c p V = Q + ha(t f T) dt dt Converting the equation to a difference equation it becomes T i+1 = T i + [Q + ha (T f T i )]?t k 144

145 = Q?t + hat f?t +(k-ha?t)t i k where? is the density of Cu, C p is the heat capacity of Cu, V is the volume of the structure, Q is the heat load input to the structure, h is the heat transfer coefficient of Cu, A is the surface area in contact with the coolant, T i is the temperature at the ith step, T f is the coolant temperature, and?t is the time step. Simulation: The above equation is implemented as a time simulation in SystemView by Elanix. The simulation follows the diagram above. The temperature of coolant at each node in the diagram is time-dependent on the flow within the pipe and the diameter of the pipe. These time dependencies are implemented as delays in the simulation. Each node in the diagram is initialized at the holding temperature of the cavities, currently thought to be 24 C. As coolant reaches the node from the previous node, the temperature is released to whatever that previous node is supplying. A variable delay is implemented in the leg that sends heated coolant through the heat exchanger. This variability is due to the variable opening of the control valve that is controlled by a PID algorithm that is controlling the temperature of the low energy full segment. A variable delay is also implemented in the cold side of the heat exchanger to provide for different flows from the facility side to the heat exchanger. Heat Exchanger: The heat exchanger is a flat-plate counter-flow heat exchanger. The energy balance equation describing this device follows: Q = U? A(LMTD) Where Q is the heat exchanged U? is the overall heat transfer coefficient for the heat exchanger A is the heat exchange area LMTD is the log mean-temperature-difference. 145

146 U? A is estimated by the following fit to the vendor-supplied heat exchanger data U? A = *m h *m h *m h *m h *m h + where m h is the water mass flow rate through the hot side of the heat exchanger and is in kg/sec and U? A is in W/ C. This data was obtained with input temperature of the hot side at 22.7 C. LMTD = (T hi T co ) (T ho T ci ) ln[(t hi T co )/(T ho -T ci )] where T hi is the input temperature on the hot side, T ci is the input temperature on the cold side, T ho is the output temperature on the hot side, T co is the output temperature on the cold side. Using Q = U? A(LMTD), Q = m h C ph (T hi T ho ), and Q = m c C pc (T co T ci ) One can solve for Q and obtain Q = m c m h C p (T hi T ci )[e U?A (mc mh)/(cp mh mc) 1] m c e U?A (mc mh)/(cp mh mc) - m h U?A (mc mh)/(cp mh In the simulation, if the exponent on e becomes greater than 12, e mc) can be considered large with respect to 1 and m c * e U?A (mc mh)/(cp mh mc) can be considered large with respect to m h. In this case the equation becomes 146

147 Q = m h C p (T hi T ci) This avoids the problem of attempting to compute a number larger than the computer can handle. One might ask how ave(t h ) can be calculated since T ho is going to be calculated given T hi and the flow rates. Ave(T h ) is calculated using T hi in the current time step and T ho from the previous time step. In this simulation, the cold side mass flow rate through the heat exchanger, m c is held constant at 2.82 kg/sec or 44.7 gal/min. This complies with the constant flow regulating valve in the circuit leading to the cold side of the heat exchanger. Variable Valve Control The variable valve that controls coolant flow through the heat exchanger is controlled by a PID algorithm whose input is taken from the average temperature of the drift tubes. The value of the gain for the differential term is set to zero since this is a first order system. Values for the other two terms were set to give good response. However, they are not necessarily the best values. Use of the Ziegler-Nichols algorithm would give optimum values. Since the PID algorithm is controlling a physical valve, the characteristics of the valve need to be included. The valve that was chosen for temperature control is actuated by a stepping motor. The valve requires 500 steps from fully opened to fully closed and vice versa. Depending on valve design, fully opened to fully closed can require from 15 to 27 seconds. 25 seconds was arbitrarily chosen because the actual valve has not yet been identified. Using the 25 second actuation time, the motor is able to take 1 step (1/500 th of full actuation) every 0.05 seconds. The values that the PID algorithm generates are moderated to provide no faster response than 1 step every 0.05 seconds. Thus, the physical characteristics of valve motion are included in the simulation. The flow versus valve opening is assumed to be linear. This makes the computation a bit easier but as long as the system is stable there is no requirement for linearity. The PID algorithm is able to provide the necessary temperature control in either case. A series of preliminary thermal engineering calculations were required to support the SystemView model. These calculations are included in Appendix L. 147

148 3.2.3 Design Studies/Results Profile of Startup and Stable Operation These simulations were run using the temperature of the center drift tube (#17) as input to the control valve. The initial temperature of the tank components was 20.0 C, the desired setpoint temperature was 26.6 C, and the full rf power was introduced at time = 0.0. In reality, the RF power will be gradually ramped up, however, for the purpose of this simulation, the application of full RF power will be fairly representative of a normal startup condition. The PID controller has gains that are optimized to provide as fast a response as possible while maintaining stable operation. Temperatures of the lowest energy and the highest energy drift tubes in the tank along with the center drift tube were monitored. In addition, the temperature of the tank wall was monitored and is shown. The temperatures of the drift tubes and the tank wall all contribute to the resonant frequency of the tank. The results of the transient simulation are displayed in Figure With T ci set to 7.2 C, the setpoint set to 26.6 C, the following steady-state conditions were found: Temperature of the drift tube is 26.6 C, as expected. The temperature of the low energy tube is slightly higher and the temperature of the high energy tube is essentially identical. Bypass valve operates at 88.7% open. T mix is 19.9 C. T hi is 21.4 C. 148

149 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a normal RF startup condition of DTL tank

150 As the startup simulation begins, the bypass valve starts to open and the applied RF power heats the structures. Tube #17 approaches the setpoint in about 140 seconds. Next, the bypass valve begins to close, anticipating the need for cooling. From this point, another approximately 120 seconds are required to stabilize the temperature at the setpoint of 26.6 C. Due to the tremendous difference in mass, the tank wall heats at a much slower rate not quite coming to the target temperature (it reaches 25.7 C ) even at the end of the simulation (1200 seconds). Note also the plateau that the drift tube temperatures pass through. The drift tubes can react so quickly because of their small thermal mass that the proportional gain of the PID algorithm must be kept very small. Because of the small proportional gain the valve reacts very slowly. The drift tubes react very quickly to the heat being applied by the rf. The tank wall is a large enough thermal mass that it acts like a heat exchanger keeping the coolant temperature from rising. The valve is still open enough to allow coolant from the actual heat exchanger into the drift tubes. As the valve continues to close, the drift tubes are allowed to heat and finally reach the setpoint. Effects of Variation in Coolant Temperature To explore the effects of variations of the SNS facility chilled water temperature on the CCL cavity temperatures, a sine wave with an amplitude of 2.0ºC and a period of 8 seconds was imposed on the chilled water temperature. The results of this simulation are displayed in Figure Note that this variation in coolant temperature shows up as a small variation (about ±0.02 C) in the temperature of the center drift tube. The variation also shows up as a variation in the temperature of the other two monitored drift tubes leading one to the conclusion that it will cause a variation in all the drift tubes in the tank. Note also that this variation doesn t affect the tank wall at all. This is due to the large mass of the tank wall. This small variation in the drift tube temperature is so small as not to change the tune of the cavity. However, it is obvious from the bypass valve action that the bypass valve would be under continuous motion (±0.22% about 88.9%). 150

151 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a sinusoidal disturbance of the facility chilled water temperature (amplitude of 2.0 C and a period of 8 sec.) condition of DTL tank

152 The simulation was repeated with a coolant temperature variation with the same amplitude but a cycle time of 0.1 second. The results of this simulation are displayed in Figure The shorter cycle time introduce an alias into the signal which shows up as an additional variation with a different cycle time. Even a cycle time this short shows a small variation in the temperature of the drift tube. This is most likely because the drift tube is small and will react to almost any temperature change. It is interesting to note that the variation is damped. Since the tank as a whole will respond to changes in temperature and those changes will cause changes to the resonant frequency of the tank and that is the quantity of interest, it is likely that the stability of the tank will offset the lack of stability of the drift tubes. At 8 seconds per cycle the variation of temperature of the drift tubes pretty much disappears at or below a coolant temperature variation of ±0.5 C. 152

153 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a sinusoidal disturbance of the facility chilled water temperature (amplitude of 2.0 C and a period of 0.1 sec.) condition of DTL tank

154 Momentary Loss of Rf Power A system such as the one described will lose rf power on occasion due to a spark in a waveguide or in a klystron. A question arises as to how much of an upset (temperature and deviation) will such a loss cause. A 10 second loss of rf power was input in the next simulation. The results of this simulation are displayed in Figure The power loss (spark) was started at 800 seconds and lasted for 10 seconds. The temperature of the middle drift tube drops to about 23.9 C. The rf power comes back on and the temperature increases, overshooting to about 27.8 C before stabilizing back to the setpoint. Note that a 10 second rf power loss translates to approximately 200 seconds of temperature upset to the controlled structure. The tank wall also reacts to the loss of rf with a small drop in temperature. However, the loss is much less and the original trajectory is regained much more quickly. 154

155 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a momentary loss of RF power condition in DTL tank

156 Total Loss of Rf Power Should there be a total loss of RF power, management of the temperature of the structure becomes an important task. The lower the cavity temperature falls, the more thermal stress will be applied to the structure and the further the system will deviate from the steady-state resonance condition. For a brief period of time, the water cooling system can minimize the temperature drop in the RF structure, however, at some point, some other method of maintaining structure temperature must be applied. In this simulation, the RF power was lost at time = 800 seconds. The results of this simulation are displayed in Figure The question here is, will something need to be done to preserve the tank and drift tube temperature? When the rf power is lost, the temperature of the controlling drift tube drops almost immediately and within about 77 seconds has fallen to 20.7 C. The temperatures in the other two drift tubes also drop copying the profile of the central drift tube. The tank wall drops as well but not nearly so quickly. Because the three-way valve can completely bypass the heat exchanger, the temperature in the drift tube tank can be maintained. One could assume that after reaching a minimum, the temperatures will slowly rise due to heat being added by the circulating pump. If that is indeed the case, it might be prudent during such a loss of power, to change the setpoint to follow temperature rather than the low-level rf error signal and to set the temperature to follow to something less than the 26.6 C operating temperature so re-initiation of RF power will again follow a normal pattern. 156

157 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a total loss of RF power condition in DTL tank

158 Change of Copper Temperature Setpoint In the next simulation, the copper setpoint temperature was increased by 1.0 C at 800 seconds. This simulates an operator s decision to change the tuning of the device. The question is how long will it take for the tank to stabilize after such a change? The results of this simulation are displayed in Figure The plots indicate that when the setpoint is changed, the bypass valve reacts by opening to its full extent. The drift tubes take about 150 seconds to settle to the new setpoint. The central drift tube lags the change in the bypass valve by about 9 seconds. The bypass valve closes down by about 0.1%. The tank wall shows a change in temperature trajectory and again shows an asymptotic approach to the new temperature Summary The results of these simulations indicate that the temperatures of the accelerating structures for the DTL in the accelerator can be controlled to the specifications required using the currently designed water cooling system. This system uses one control valve that is manipulated by a PID algorithm. The coolant is supplied from the facility chilledwater system and is fed into a counter-flow, thin-plate heat exchanger. Start-up times, as well as responses to thermal disturbances appear to be reasonable. It is well to note that the structure seems to be riding on the edge of stability. Slight changes in either the proportional or integral gains will send the unit into oscillation. In addition, if either gain is reduced there seems to be a need to reduce the other to prevent the instability. However, this may be an artifact of the numerical model, and may not be representative of the actual structure. It is speculated that initial testing and commissioning of the actual DTL water cooling systems will allow proper tuning of the control systems to occur and provide stable and accurate operation. 158

159 Figure Tank wall temperature, drift tube 1, 17, and 33 temperatures, and by-pass valve positioning versus time for a 1 C increase in the copper setpoint temperature condition in DTL tank

160 4.0 Mechanical Design 4.1 Introduction This section of the report discusses many facets associated with the mechanical design of the DTL water cooling system. These topics include the types and quantities of engineering drawings being developed, engineering codes and drawing standards being followed, and general mechanical design processes that were followed for the design of the water cooling system hardware. 4.2 Engineering Codes and Drawing Standards To ensure that the DTL water cooling system design meets reasonable reliability and safety standards, design guidelines and specifications provided by the ASME Boiler and Pressure Vessel [4.1] and ASME Piping Process (B31.3) [4.2] codes are being followed. The ASME B31.3 codes are sponsored, published, and maintained by the American Society of Mechanical Engineers (ASME). The scope of the codes as used in this report, is to provide guidance for the design, fabrication, assembly, installation, inspection, and testing of piping and piping components for the SNS Linac water cooling system. In Appendix A, B31.3 topics related to the design of the SNS Linac water cooling system are listed, and where applicable, background information is presented. Failure theories are discussed and their relationship to pipe stress equations and limits presented in B31.3 are established. Welding practices and inspection techniques are some of the major topics of concern in the water cooling system design. The ASME B31.3 is very specific in discussing the various welding procedures including inspection and testing techniques used in pipe fabrication. It must be emphasized that the ASME B31.3 code does not serve as an instruction list for design, rather it is provided to assist engineers and designers in their efforts to produce a safe piping system. It is the responsibility of the piping designer, the manufacturer, the fabricator, and the installer, as applicable; to follow the guidelines set forth by the B31.3 code and provide sufficient documentation of its implementation. The main organizational 160

161 responsibilities, as they apply to the SNS Linac water cooling system, are summarized below: Vendor (Manufacturer and Fabricator): Responsibilities of the vendors include examinations and certification of all contracted work. The vendor shall provide certified examiners and document all dates and results of examinations. Piping and piping elements shall be visually examined to the extent necessary to satisfy the examiner that components, materials, and workmanship conform to the requirements of the B31.3 code and of engineering design. All records shall be made available to the LANL design team. Los Alamos National Laboratory (LANL): The responsibilities of LANL s design team are to verify that all required examinations and testing have been completed and to inspect the system to the extent necessary to be satisfied that is conforms to all applicable examination requirements of the B31.3 code and the engineering design. LANL will also conduct visual inspections during and after all stages of manufacturing, fabrication, assembly, installation and testing. LANL will be responsible for final certification of all equipment prior to facility operation. Oak Ridge National Laboratory (ORNL): ORNL is the owner of the SNS project and is responsible for all final inspections. ORNL s Inspectors shall have access to any place where work associated with piping installation is being performed. This includes manufacture, fabrication, assembly, installtion, examination, and testing of the system. ORNL inspectors shall have the right to audit any examination, to inspect the system using any examination method specified by the engineering design, and review all certifications and records necessary to satisfy the stated owners responsibility. All engineering drawings generated for the DTL water cooling system will adhere to the ESA-DE Drafting and Design Standards and Guidelines [4.3], which closely follow the standards listed in the Global Engineering Drawing Requirements Manual [4.4]. The formating standards used for the Piping and Instrumentation Diagrams (P&IDs), were 161

162 taken from Sherwood and Whistance [4.5], and the ISA S5.5 [4.6] and ANSI Y32.11M [4.7] national standards. 4.3 Plumbing Materials Material selection for the SNS linac water cooling system plumbing is driven by a number of design criteria such as functionality, strength, durability, radiation hardness, cleanliness, manufacturing capability, maintainability, availability, and cost. Copper, stainless steel, brass, carbon steel, and PVC have been evaluated for use as the primary tubing material. Copper tubing is desirable because of its mechanical properties of high strength and corrosion resistance, as well as being economical and easy to form and join. For more specialized water transfer systems such as those found in chemical processing facilities, microchip processing facilities, power plants, nuclear reactors, synchrotrons, and particle accelerators, stainless steel is commonly used. Brass and mild carbon steels are not recommended for water systems where high water purity is an issue as both of these materials are susceptible to relatively high levels of corrosion or erosion. Plastic and PVC materials are being avoided for several reasons including their lack of strength, high diffusion coefficients for oxygen (oxygen promotes bacteria growth and enhances corrosion of copper), and high susceptibility to radiation damage. Selection of flexible hoses was also evaluated. Flexible hoses will serve both as vibration and electrical isolators and allow for greater flexibility in the system design. The engineering design tolerances for the assembly drawings may be relaxed and potential plumbing misalignment can be absorbed during installation. These hoses will greatly simplify assembly and installation of the water skids, manifolds, and submanifolds. Selected flexible polymers have been chosen for short jumper and connection lines. In identifying the correct material of hose, radiation affects, compatibility with deionized water, oxygen permeability, and flexibility have been considered Radiation Damage Assessment The radiation emanating from a particle accelerator can degrade mechanical properties of materials in close proximity to the beam line. The extent of this degradation 162

163 will depend on the dose rate and cumulative radiation dose, as well as other factors such as operating temperature, mechanical stress, and exposure to air [4.8]. Scientists and Engineers at CERN have compiled a fairly extensive database, which relates radiation damage to cumulative dose rate for a variety of materials [4.9]. Table 4.1 lists the radiation damage (cumulative radiation dose) limits for various materials used around high-energy particle accelerators [4.9]. Table 4.1. Radiation damage limits for materials used around high-energy particle accelerators [4.9]. Material Cumulative Dose Limit (Rad) Metals Polyvinyl Chloride (PVC) Ethylene-Propylene Rubber (EPR) Polyurethane Rubber (PUR) Styrene-Butadiene Rubber (SBR) Polychloroprene Rubber (Neoprene) Chlorosulfonated Polyethylene (Hypalon) Acrylonitrile Rubber (Buna-N) Viton Nylon Plexiglass Silicone Rubber (SIR) Fluoro Rubber Acrylic Rubber Butyl Rubber Phenolic Resin Tefflon (PTFE) Assuming a particle beam loss of 1 Watt/meter along the entire SNS linac, the prompt radiation dose rate, at 1 foot from the beam line, will be approximately 1 Rad/hour at the low energy end (10 MeV) of the DTL, 8 Rad/hour at the high end energy (80 MeV) of the DTL, and 18.5 Rad/hour at the high energy end (185 MeV) of the CCL [4.10]. If the SNS accelerator were to run for 300 days/year, the maximum cumulative dose for a year would be approximately Rads in the DTL and Rads in the CCL. To determine which materials will be acceptable for the water cooling system tubing (from a radiation performance perspective), the material cumulative dose limits need to be compared to the annual dose present during accelerator operation. Assuming a 163

164 thirty year desired lifetime for materials in the Linac water cooling system, the total cumulative dose would be Rads in the DTL and Rads in the CCL. Thus all water cooling system materials for the DTL and CCL linac beam line should be able to withstand a radiation dose of at least Rads. Referring to Table 4.1, the metals, such as copper and stainless steel, as well as many of the nonmetallic materials such as Buna-N, Hypalon, Nylon, Neoprene, meet the cumulative dose criteria. However, among the nonmetallic materials, it is only Buna-N and Neoprene that have an historical usage base to refer to. Both have been used on the LANSCE 800 MeV particle accelerator at Los Alamos National Laboratory with good success. The flexible Buna-N lines on the LANSCE CCL have been observed to harden over time by a combination of radiation and atmospheric damage, however they have maintained working lifetimes of well over ten years [4.11]. In addition, Buna-N/Neoprene hoses have been used as flexible jumper lines for the majority of the focusing and steering magnets on the LANSCE accelerator for the last twenty years [4.11]. Note that the annual cumulative dose rate estimated above was based on the high-energy end of the SNS linac and is thus very conservative for the majority of the room temperature linac structure. Nonmetallic materials will also be needed as flange seals and thermal insulation (within the chases) of water system components as well as electrically insulating the power and signal lines. Components such as valves or flow meters are likely to require some type of gasket material for sealing at the connections. Choices are often available from a vendor and the selections will need to meet the radiation dose criteria of Rads. Viton and Teflon are among the most common options however only Viton is acceptable. Teflon should be avoided whenever possible. For insulation, selections will need to be made by consulting Table 4.1. Due to the threat of leakage, threaded joints will be avoided whenever possible but not in all situations. Teflon is a common thread sealant but, as was stated earlier, is unacceptable due to material failure in the SNS radiation environment. LANL recommends the usage of RectorSeal NO. 5 as the pipe thread sealant. 164

165 4.3.2 Material Selection for Design The design of the water cooling system for the room temperature linac will require consistency between components on the DTL and CCL, and will necessitate a clear definition of acceptable materials for the plumbing components. Various aspects of the accelerator and water cooling system designs dictate a need for certain characteristics in the plumbing material. For example, the water purification system requires clean, corrosion resistant, and impermeable (to oxygen) tubing. The RF structure alignment criterion makes the use of flexible jumper lines desirable. Flexible hoses also reduce high tolerance requirements in positioning of the water skids relative to the transfer and facility water lines. To satisfy these wide ranges of needs, it was necessary to define acceptable water cooling system plumbing materials. To determine the correct material for the water lines in this closed loop system, water quality and purity has been identified as a significant factor. A comparison between copper and stainless steel was diligently researched. Stainless steel was identified as the only acceptable material for several significant reasons. Although the DTL and CCL tanks are made of copper, the percentage of surface area relative the remaining portion of the cooling loop is small. Therefore, this factor was not as significant as several others were. The ultrapure deionized water is very aggressive and will attack materials such as copper, brass, and bronze. The water will begin to remove iron oxide particulates and allow them to reattach on other surfaces. This is called rouging and can easily be recognized by a reddish tint in the water. This will directly affect the efficiency and lifetime of critical components such as the pump, 3-way control valve, heat exchanger, and inline heater. Additionally, these particulates are likely to become activated and increase the handling risk of the water. Cost was identified as another criterion. Copper tube and pipe is slightly less expensive then stainless steel. To maintain material compatibility, fittings and valves would be constructed from either bronze or brass, which is significantly less expensive then stainless steel. However, brass is significantly more susceptible to the aggressive nature of the ultrapure water. The use of these materials would require a periodic flushing of the entire closed loop system to remove the iron oxide. This would increase 165

166 the system complexity, the system cost, and the system down time. The material cost savings of copper is countered by the added requirements in maintaining a clean system. System components such as the heat exchanger and the pump would be constructed from stainless steel whether the piping system was copper or stainless steel. This would require a galvanic insulating material to prevent corrosion at the joint if the plumbing material were copper. Even copper to copper brazed or soldered joints may create potential problems. Many of the standard flux materials used in solder joints are susceptible to the aggressive nature of deionized water. Such joints may potentially cause soldered particulates to break off and cause damage to the pump impeller or inhibit flow through an orifice plate. Having weighed the benefits and risks of both copper and stainless steel, the need for a dependable and safe system requires LANL to employ only the stainless steel as the piping and tubing material. From the complete criteria, which includes functionality, durability, radiation hardness, cleanliness, manufacturability, maintainability, and cost, the following materials have been deemed acceptable: 300 Series Stainless Steel Stainless steel is extremely durable, strong, clean, and corrosion/erosion resistant. While Stainless steel tubing is more difficult to form and join than copper tubing, its cost per unit length and availability are similar. In addition, stainless steel will provide a cleaner environment for the water purification system and be less susceptible to erosion and radionuclide-induced corrosion than copper tubing [3.12]. For deionized water systems, stainless steel joints that are welded will provide a more reliable leak-free system compared to copper soldered joints. Most of the compression fittings, valves, orifice plates and housings, and instrumentation probes will be fabricated from 300 series stainless steels because of its strength and corrosion resistance. Nonmetallic Hose Several polymers are being considered for use as flexible jumper and connection lines. The stringent criterion in selecting the correct material quickly narrows the possible choices. Selected hose materials will be the same for both the DTL and CCL 166

167 and must therefore meet the higher of the two dose rates (CCL portion of the linac). The minimum survivable radiation dose is Rads and the ultrapure water electrical resistivity 10 to 15 MΩ. In addition, the hose must be reasonably flexible and its oxygen permeability must be very low. These last two criteria are more subjective and result in a list or ranking of potential materials (see Appendix H). Additionally, any history of use for a given material in a similar environment will strongly be considered. From these criteria, only Buna-N, Hypalon, Nylon, and Neoprene will be considered for flexible hose material General Manufacturing and Assembly Techniques Manufacturing techniques used to fabricate and join the water manifolds and transfer lines may include the use of the T-Ball extrusion technique, welding, flanges, threaded, and compression fittings. Qualified personnel will perform all manufacturing processes and will follow procedures outlined under ASME B31.1 code of practice. Many of the fluid lines require bending of the stainless steel tubing. All bends will follow the SAE-AS33611 standard. The detailed plans for fabrication, assembly, and installation for all water cooling system hardware can be found in Sections 9 and 10 of this report. The general manufacturing and assembly techniques for the water manifolds are defined below. The T-Ball technique is used to produce multiple extruded flow ports in manifolds or headers using a die to control the radii of the extrusion. The process involves drilling a small hole in the side of a pipe and inserting a ball-shaped puller. The ball-shaped puller is then extracted outward through a round die that is placed on the outside of the manifold. As the ball is pulled through the die, metal is drawn with it to form a cylindrical extruded outlet port. The extruded outlet port extends above the surface of the header a distance at least equal to the external radius of the outlet. This manufacturing technique is especially useful for thin-walled manifolds in which it is not possible to produce a tapped hole for a threaded fitting. This technique also eliminates the need for welding or soldering a T fitting in place for a branch line. 167

168 Finally, the T-Ball technique also provides a rounded entrance to the extruded port, thus creating a lower flow resistance than a standard T junction. Welding procedures and welding operators will be in conformance with the rules specified in AWS and ASME code standards. All stainless steel and steel tubing joints will be welded with gas tungsten arc welding (GTAW), better know as TIG welding. In TIG welding, an electrical arc is established between the tungsten electrode and the work-piece, resulting in heating of the base metal. If required, a filler material is used. The weld area is shielded with an inert gas, usually argon or helium. GTAW is ideally suited to weld nonferrous materials such as stainless steel, and is very effective for joining thin-walled sections. Welds used to fabricate manifold support brackets made of mild steel, will be done with Gas Metal Arc Welding (GMAW), better known as MIG welding. GMAW uses a solid metal cored electrode and leaves no residual slag. The shielding gas may be carbon dioxide or a blend of argon with carbon dioxide and/or oxygen. GMAW is ideal for welding thingage materials. Compression Fittings provide a leak-proof, torque-free seal at all tubing connections and reduce the possibility of costly, hazardous leaks in instrumentation and process tubing. The joining action in the fitting moves along the tube axially instead of with a rotary motion. Consequently, no torque is transmitted from the fitting to the tubing, which eliminates any strain that may weaken the tubing. Compression fittings are advantageous for joints that must be taken apart for inspection, replacement, or maintenance purposes (i.e., instrumentation ports, water treatment hardware, etc.) Flanged Fittings provide a very dependable and consistent joining method. For components that may need to be inspected, removed or replaced, flanges are optimal. ANSI flanges are the only acceptable types of flange because they call for specific internal diameters, external diameters, hole pattern, quantity of fasteners, and material type for each tube size. These requirements allow a smooth transition from piping to tubing. Leakage is not a significant concern. Flange seals are available in acceptable 168

169 nonmetallic materials. Rotatable flanges will be used when component or pipe alignment is critical. Threaded Fittings are a very common joining method but will only be used sparingly. This joining method, versus the others discussed above, is the most likely to leak. For insertion RTDs, pressure transducers, pressure relief valves, and potentially a few other components, threaded fittings can not easily be avoided. The goal is to install them in a leak proof way. To do this, a sealant must be applied to the threads to prevent galling. It is recommended that RectorSeal NO. 5 (MSDS0011) be used because it is a soft setting sealant, its provides excellent sealing and antigalling characteristics, and has demonstrated very good performance on previous projects. Beaded Tube Ends will be used to attach all flexible hose not identified as requiring quick disconnects. Document SAE AS5131 specifies bead requirements for tube ends. The bead may be rolled at the end of a tube or a tube stub with a bead may be welded into place. When connecting to a Swagelok fitting, a tube stub may be inserted into one end and locked into place. The flexible hose is installed over the tube end with the bead and then band clamped to lock into place. 4.4 DTL RF Structure Water Manifolds and Lines As mentioned previously, there are a total of six independent DTL RF Structure cooling loops. This section describes the engineering design associated with those cooling loops Piping and Instrumentation Diagrams The DTL RF structure Process and Instrumentation Diagrams (P&IDs) are a practicable way of showing the process flow as well as the instrumentation and controls, the hardware, and show plumbing identification. P&IDs are tools that not only illustrate the process in detail, but also provide information on equipment, valves, lines, and instruments, what the industry commonly refers to as Intelligent P&ID'S. These 169

170 diagrams can also be used for process safety management operational training and maintenance. The information from P&IDs allowed the design team to generate a detailed list of components and instrumentation. Flow meters, valves, pressure transducers, RTDs, etc. are identified in a table (see Section 4.6). The table provides information showing the relationship between the actual components and the P&ID where it is identified. Such information is the drawing number, page, and location where the component can be found. Other essential information related to the component such as flow rate, pressure rating or electrical requirements are provided. Figure 4.1 displays the 4 sheets of the P&ID for the DTL RF structure cooling loop on DTL Tank 1. One of the main features on the P&ID is the supply and return process lines. These lines are called single line drawings. The single line drawings are more compact and can represent more diagrams and instruments on less sheets verses the double line P&ID drawings. DTL P&IDs show the supply coming from the water skid in the center of the module process lines to ensure an even flow rates, and a mirror image for the return process lines. The arrows indicate the flow direction. The DTL P&IDs give each process line an ID naming convention. (e.g., 4 -WS- DTL1-101-SS3). The first number in the ID label is the pipe diameter then after the dash is the Water Supply, DTL RF Module number, process line number and the material specification. Also labeled on the P&ID s are the naming convention for the valves and instrumentation. DTL P&IDs contain diagrams of all supply and return process lines including drain and vent valves. The vent valves are to be located at the high point of the process system and the drain valves will be located at low points. This will insure proper draining and purging of the water cooling system. Valves indicated on the P&IDs are positioned so that the system can be segmented if a section is to be removed or replaced. All valves are manually operated. The DTL supply submanifold for the drift tubes is equipped with orifice plates to distribute the desired amount of flow to each drift tube. These orifices are removable for maintenance purposes. 170

171 Figure 4.1. Piping and instrumentation diagram for DTL tank

172 Figure 4.1. Continued. 172

173 Figure 4.1. Continued. 173

174 Figure 4.1. Continued. 174

175 Instrumentation labeled on the P&ID s are shown as balloons and also with a numbering convention. These instruments will be electronic in nature and will send signals back to the PLC for an accurate reading and a constant hazard control of the entire DTL system. There will be temperature, pressure, and flow sensors on each DTL RF tank Major Components Orifice Plates Swagelok VCR fittings will be used for retaining the orifice plates on the CCL. The line sizing for tanks 1 and 2 is ½ tube and for tanks 3 through 6 is ¾ tube. A detailed effort has been made to minimize the number of unique orifice diameters. A maximum deviation from the analytical orifice diameter to that of the actual drill hole diameter is 5%. Tanks 1 and 2 will require 21 drill hole sizes and tanks 3 through 6 will require 12 drill hole sizes. Orifice plates have unique installation requirements that must be adhered to. Failure to follow these guidelines will result in poor performance or improper operation. Orifice plates will primarily be used in the DTL portion of the linac but will also find usage in the CCL. Each drift tube requires a specific volumetric flow rate to correctly control the RF power. Fluid exiting the orifice must not create fluid cavitation. Straight lengths of tube without any obstruction or disturbance to the flow will create smooth fluid flow, both prior to the orifice as well as after the orifice. Specifically, a straight tube length of 3 diameters upstream and 8 diameters downstream is the minimum installation requirement. Valves Manual globe valves will allow for fine-tuning of the flow to the sub-manifolds. The stem may be adjusted using a standard hand-wheel. At the point where the valve is properly adjusted, a locking mechanism will ensure that no inadvertent contact would alter the valve setting. All wetted metallic valve components will be 300 series stainless steel and non-metallic O-rings will be of Viton. Due to potential sticking or turning of a screwed bonnet, a bolted bonnet is preferred. The desired body end is flanged. 175

176 Manual ball valves are required to isolate sections of the water system for the removal or replacement of system components and system drainage in this closed loop system. The rotary-ball valve will function as a manual on/off flow isolator. The stem rotation will be 90 open to close. The single-seat eccentric version of this ball valve will be used to insure that no leakage occurs. The ball is slightly offset so that it presses into the seat on closure. The type will be a ball valve constructed from 316 or 316L stainless steel. All valves must have a lockout method to prevent unauthorized adjustments. Pressure relief valves will be used to protect the system from over-pressurization. These valves will be set to release at a pressure below 100 psig, so that the system pressure will not at any time exceed the maximum operating pressure of 150 psig. Flow Meters All flowmeters used in the Linac Tunnel will be of the turbine type. Turbine flowmeters produce either a pulsed wave signal or a sine wave signal, the frequency of which is proportional to flow rate. This type of flow meter possesses a reasonable flow resistance and is reasonably priced. It is quite possible that electrical noise in the linac tunnel and waveguide chases will disrupt the standard, low amplitude pulsed wave signal eminating from the turbine flow meter head. Consequently, amplification of the pulsed output signal or conversion of the pulsed signal to 4-20 ma output at the flowmeter is required. The electronics components required for the wave signal amplification or conversion will be at risk due to the radiation environment in the linac tunel. The radiation produced by the accelerator is comprised of Neutrons and Gamma rays. Shielding of the flow meter electronics against the neutron radiation is not feasible. Other types of flowmeters, which may not require radiation shilding, have too great a pressure drop or are extremely expensive and beyond the allowable budget. The only resolution is to plan for scheduled replacement of the electronic amplification or conversion unit. The anticipated prompt radiation dose along the SNS linac was discussed in Section 4.3. At 100 MeV, the beam will produce 10 Rad/hr at a distance of 1 foot and at 176

177 200 MeV, the beam will produce 20 Rad/hr at a distance of 1 foot. Radiation levels drop off proportionally at the rate of 1/r where r is defined as the distance from the beam centerline in feet. For the DTL, the flow meter electronic components are 3 feet away from the beam line and will see a dose rate of 3.33 Rad/hr. The accelerator will operate 300 days per year for 30 years. Consequently, the total radiation dose to the DTL flow meter electronic components would be: 3.33 Rad/hr 24 hr/day 300 days/yr 30 years = Rads over the accelerator s lifetime. Based from historical experience, electronics will survive and remain operable up to a cumulative dose rate of Rad [4.8]. Thus, the flowmeter electronics in the DTL will have a lifetime range of 4.9 to 20 years. The quantity of flowmeters in the CCL portion of the accelerator is less then is required for the DTL, which is of some relief. Scheduled maintenance should correlate with these component lifetime expectations. Turbine flowmeters have unique installation requirements that must be adhered to. Failure to follow these guidelines will result in poor performance or improper operation. A minimum straight tube length both upstream and downstream will produce accurate and consistent flow measurements. The installation requirement is 10 diameters upstream and 5 diameters downstream as a minimum. If a partially closed globe valve, a tight radius bend, or a tee intersection is part of the tube run, these minimum installation requirements need to be increased Assemblies Figure 4.2 displays a solid model of the water lines and manifolds mounted on the DTL tank 1 support structure. The supply and return manifolds have been placed on the back of the RF structure running in horizontal position and in-line vertically. The supply will be the upper manifold and the return will be the lower manifold clocked. The return manifold will be rotated 30 inwards to allow the process lines to be free of potential damage from the work area. The process lines then will be routed over the RF structure support to proper connectors. The support of the manifolds will necessitate standard readily available support brackets with a means for adjustment and the ability to be welded if required. The support structure will be fastened to the RF structure with standard bolt fasteners. This will allow for slight adjustments longitudinally, laterally, 177

178 Figure 4.2. Water manifolds and lines on DTL tank 1 as seen from the front (aisle) and back side of the accelerator. 178

179 and vertically. The key advantage of this support system is the elimination of prestressed joints on the manifolds and sub-manifolds. A secondary benefit is the angling of the main manifold to allow for more efficient draining and venting of the cooling system as required. Additional examples of the subassembly and detailed drawings of the main manifolds are presented in Appendix B. Similar drawings have been generated for all submanifolds on DTL tank 1. Tabularized drawings, where appropriate, are being used to dimension the remaining main and sub-manifolds for DTL tanks 2 through 6. A representative tabularized drawing, which uses a table of dimensions to describe similar manifolds from different DTL tanks, is shown in Appendix B. 4.5 Water Skid The water skid is a modular unit containing various plumbing components (pump, heat exchanger, etc.), a water treatment system, and instrumentation and controls. The water skids for the DTL RF Structures, the CCL RF Structures, the CCL Quadrupole Magnets, and the SCL Quadrupole Magnets will be very similar in function, layout, and operation. The general features and processes are shown on the P&ID of Figure 4.3. The DTL will require a total of 6 RF Structure water. The flow requirement for the RF Structure is approximately 120 to 235. While the pumping requirements all of the skids will be significantly different and may require different size components, it will be desirable to maintain consistency in the design and component selection. The goal of the water system team is to produce the greatest range in performance for each closed loop system. Unfortunately, to produce the greatest amount of thermal range for the cooling loops, twelve identical skids would not produce such result. The significantly diverse range of water volume, flow rate, and heat transfer capabilities drive the need for several unique water skids. Variables such as facility chilled water flow rate and temperature, heat exchanger warm side flow rate and pressure loss, mixed temperature from the heat exchanger and heat exchanger bypass, variable speed pump capabilities are some of the complexities that are to be evaluated. 179

180 4.5.1 Piping and Instrumentation Diagrams The Process and Instrumentation Diagram (P&ID) is a tool used to define the fluid lines, direction of flow, instrumentation/control leads, and facility interfaces. Shown on these diagrams are the components, drains, filters, etc. and it s associated reference or naming designation. The P&ID is created as a simplified layout form for ease of understanding and is not to scale. The water skid P&ID is shown in Figure 4.3. Each water skid contains an expansion tank that will serve as a water reservoir for initially charging the flow loop and pressurizing the system with Nitrogen. The expansion tank is equipped with a water level indicator, a pressure relief valve, and a Nitrogen gas purge. The reservoir will be pressurized with Nitrogen to purge the deionized water of potentially system damaging Oxygen. Additionally, the pressurizing of the system insures a positive priming of the pump. The reservoir will include a pressure relief valve, a flow limiting orifice and a vent valve to insure safety within the system. The water skid will contain the entire water purity system, which is required to remove potential radionuclides (see Section 5 for details regarding water purification). The water remains in continuous circulation driven by one magnetic drive sealess centrifugal pump. The function of the water skid is to produce a correctly tuned RF frequency in the linear accelerator. Each water skid will provide a metered amount of cooling water to the RF Structure, which is monitored by several flow meters along the water s pathway. The RF Structure return line is warm water and passes the first flow meter, which then tees towards either the heat exchanger or to its bypass. The flow to the heat exchanger passes through another flow meter prior to entering the heat exchanger. This second flow meter assists in the temperature control of the system. An electronically actuated 3-way control valve controls the amount of flow that will bypass the heat exchanger. Both flow paths reconnect prior to entering the pump. Upon exiting the pump, the flow will tee into two potential paths. The primary path bypasses an in-line circulation heater. The secondary path is through this heater. The secondary path is likely to be used only when the desire is to slowly raise the temperature of the particle accelerator. These three flow meters within the water skid will help to control the DI water temperature and flow rate. These 180

181 FM Reservoir/ Expansion Tank Fluid Low-Level Indicator N 2 Pressure Relief Valve Vent Valve FM Carbon Bed UV 5 µ m Source Filter Mixed Ion Bed Cation Resin Resin P Deoxygen. P 5 µ m Filter Filter 60 mesh S FM T In-Line Heater FM Heat Exchanger By-Pass Control Valve Drain T Variable-Speed Pump WP Water Purity Transducer (Ph, elect. Cond., Diss. O) 2 FM FM P T T P Valve for acid flush Heat Exchanger P T T P Flow Control Valve Filter 100 mesh FM T P Flow Meter Temperature Transducer (RTD) Pressure Transducer Facility Chilled Water Outlet Facility Chilled Water Inlet Figure 4.3. Water skid piping and instrumentation diagram. 181

182 two variables should allow for reasonable control the RF Structure temperature. In theory, the flow rate will remain constant and the temperature to the RF Structure will be controlled by the amount of water flowing to the heat exchanger. The end result is a properly tuned RF Structure Performance Specifications One procurement specification will be required for all of the DTL RF Structure, CCL RF Structure, CCL Magnet, and SCL Magnet water skids. It will define the system performance requirements, engineering codes that the system must be designed to, fabrication and assembly requirements, and the required acceptability testing prior to delivery. LANL will provide the P&ID to the manufacturer/assembler of the skid. All potential vendors will meet applicable ASME Codes Vibration Isolation No specific vibration isolation requirements have been specified in the SNS requirement documents SNS SR0001-R01 and SNS SR0001-R00 that would guide the design of the water pumping skids. Consequently, it was decided to use best engineering practices to vibrationaly isolate the water skids from other subsystems. On the water skid, the vibration generating components include the 3-way control valve, the flow of water within the plumbing system, and the pump. The control valve does not have an impact on dynamics but only impacts the system with respect to noise levels and not vibration. The system flow will not create a significant dynamic impact based on experience from previous accelerator designs and the engineers who worked on those designs. The only significant source of vibration is the pump motor and rotor. Section identifies pump isolation requirements that will be imposed on the supplier of the water skid procurement. Additionally, LANL Memorandum SNS goes into greater detail and discussion of the process and development of the vibration isolation requirements for the water skid. 182

183 Noise Level Requirements No specific noise level restrictions were specified in the SNS requirement documents SNS SR0001-R01 and SNS SR0001-R00 that would guide the design of the water pumping skids. Consequently, it was decided to use best engineering practices to design a water pumping system that generated reasonable noise levels. With this in mind, the water skid was designed such that no personal protective equipment will be required during system conditioning. The water skid supplier is required to review document OSHA Regulations (Standards 29 CFR) Occupational noise exposure and meet the requirements therein. Another commonly used resource is the American Industrial Hygiene Association (AIHA). These two associations have provided the foundation for what are acceptable noise levels of various equipment for varying lengths of time. An A-weighted response, denoted dba, is often used because it simulates the sensitivity of the human ear at moderate sound levels. To understand typical A-weighted sound levels, freeway traffic at 50 meters produces 70 dba and a loud lawnmower at the operator s ear produces 90 dba. Two other components contribute significantly to the noise level in a defined area. They are the exposure time and the cumulative noise level affects of all equipment in the defined area. The OSHA and AIHA permissible noise exposure for 8 hours is 90 dba. To compute the cumulative affect of various noise sources, an equation was developed. The front end of the Klystron Gallery has 6 noise sources: water skid (qty 2), Klystron cooling skid (qty 2), Klystron transmitter tank (qty 2), SCR high voltage rectifier (qty 1), modulator equipment (qty 1), and equipment racks (qty 3). The equation yielded an allowable noise level of 73 dba for each water skid. Since exposure time may be greater then 8 hours during the accelerator commissioning, the permissible noise exposure will be reduced to 70 dba. Measurements may be taken at any location outside the structural frame of the skid. For a more detailed review of the noise requirements, see LANL Memorandum SNS

184 4.5.3 Major Components and Specifications The general layout configuration of the water skid is currently being developed. All water skids should be similar in construction however it is likely that differences in pump sizes, heat exchanger sizes, and pipe diameters between skids may be required due to differences in required water flow rates and heat loads per module. Considerable effort will be made to maintain consistency in water skid design features and components. The following components have been defined as to the system performance and will be specified in terms of company and model number to the water skid manufacturer/supplier Structure The supporting structure of the water skid will have a base constructed from a flat plate material with supporting cage type structure for vertical attachments. The structure will not corrode due to a moderately humid environment. Best engineering practices will be used in all construction that includes ANSI B1.1 and ANSI B The support structure will be painted using durable enamel as a protective measure to eliminate potential corrosion and rust. Ideally, carbon steel will be used as the construction material. The supplier will follow ASTM guidelines including ASTM A276, ASTM A240, and ASTM A480. A structural envelope will be no greater then 5 feet in width by 8 feet in length by 8½ feet in height. The goal of the supplier is to minimize the water skid envelope to as small a package as possible. In reducing this envelope, the supplier will focus on the reduction of the length and width. A size reduction of the skid will allow greater access to components on the skid as well as more access to various systems in the Klystron Gallery. The orientation of certain components within the skid envelope is critical. LANL anticipates the need for scheduled maintenance on the water purification/filtration unit. Specifically, the carbon bed and mixed bed containers will need to be replaced to ensure the purity of the deionized water. Easy and direct access is required. Due to the location within the building that each skid is located, these containers will need to be accessible on the short side (envelope width). 184

185 The pump is a critical component that will require proper orientation on the water skid. Accessibility to the pump motor on the same short side (envelope width) as the carbon bed and mixed bed containers is required. Although no scheduled maintenance for the pump motor is anticipated, failure of this component has the highest probability of all components within the water skid assembly. The supplier shall develop and implement into the water skid design a method of efficiently draining the system. The drainage system must prevent water from dripping or draining onto the Klystron Gallery floor. The draining/venting scheme must be efficient and simple for maintenance personnel. All low points of the skid must have a method of draining that is valved. The highest point on the skid must have a valved port to vent the system and increase systemdraining efficiency. The structure of the skid must provide a collection tray for any inadvertent leaks due to system water overfill, spillage occurring due to system draining, or an improperly functioning valve. The base of the water skid may be designed to function as a spillage tray Plumbing All tubing design and construction will meet the requirements of ASTM A268, A269, A511, and A554 documents. The primary material used by the supplier to design/fabricate the system will be stainless steel 316 or 316L. Viton is an acceptable nonmetallic seal material. Other materials may be used provided they are commonly used materials, are acceptable for use with deionized water, do not create galvanic corrosion problems, and are acceptable in writing by LANL. The method of joining tube-to-tube used by the supplier to design/fabricate the system will be by compression fittings, welded joints whenever possible, and flanged connections. The method of joining tube-to-components is the same as for tube-to-tube methods however certain components may require threaded NPT connections. All flanged joints will meet the requirements set forth in ANSI B Whenever a threaded fitting is required, a soft setting sealant shall be used. The recommended sealant is RectorSeal NO. 5 (MSDS0011). Teflon does not perform well in the any radiation 185

186 environment and should be avoided. All dissimilar metals require nonconducting dielectric connections and the written approval by LANL. Tubing support shall be in accordance with Manufactures Standardization Society (MSS) for the Valve and Fittings Industry, MSS SP-69. Supports shall be arranged to insure that no structural load is transmitted to the equipment. Based on extensive analysis by LANL, tube sizes have been identified and can seen in the Process and Instrumentation Drawing (P&ID). All tubing will be installed parallel and perpendicular to the skid base frame. Tube cutting will be with tube cutters only. All cut edges will be reamed to remove all burrs. All defects caused by machining, chipping, or grinding will be removed. All stainless steel components/sub-systems shall follow the guideline set forth in document ASTM A380 for precleaning, descaling, and cleaning. The water skid shall be cleaned per PFI ES-5. Flow direction identification is very important to the proper installation of the completed water skid. Each major tubing section shall have directional arrows indicating the water flow path. A major section is defined as any tube length preceding and following a tube intersection. The cold side of the heat exchanger will have typical facility water flowing through its plates. It is likely to cause water scaling and leave mineral deposits on the inside of the heat exchanger. The scaling would have a detrimental affect on the heat exchanger performance and will eventually lower the performance of the entire closed loop system. Therefore, two connection ports are required to do periodic acid wash cleaning on the cold side of the heat exchanger Pump The pump will be a magnetic drive pump (MDP) and be of the horizontal sealless type to maintain constant water flow to the RF Structure and the magnets. It will utilize an outer ring of permanent magnets or electromagnets to drive an internal rotating assembly consisting of an impeller, shaft, and inner drive member (torque ring or magnet ring) through a corrosion resistant containment shell. A flow meter will be located just downstream of the pump to monitor the flow rate. 186

187 The material of construction will be 316 or 316L stainless steel. The selection of a pump will meet all of the requirements of document ASME B73.3M-1997 Specification For Sealless Horizontal End Suction Centrifugal Pumps For Chemical Process. All electric motors must be manufactured and operate per NEMA-MG-1. This document specifies appropriate maximum vibration levels for electric motor assemblies. Each pump assembly (including motor) will be installed on a conventional machinery vibration isolation mount. The mount system must be sized to provide 95% vibration isolation with respect to the pump s fundamental rotational excitation frequency. Isolation must be provided along two perpendicular axes that are in turn perpendicular to the pump axis. Thus, the isolation mount for a horizontally mounted pump could provide isolation vertically and laterally with respect to the pump axis. Isolators may be mounted with their axes angled with respect to each other. Conventional wire rope, helical spring, or isolator styles may be utilized. Correct pump sizing is critical to the performance of the entire water cooling system. Given that the pump manufacturer is not known at this time, an estimation of the required pump sizes has been made. A standard pump performance curve (3 related curves), as seen in Figure 4.4, was used. Along the abscissa is the flow in gallons per minute and down the ordinate is the pressure loss in terms of Head (feet of water). The intersection is found and a curve is selected, usually the curve directly above the point. This curve is the required impeller size. Drawing a vertical line down to the next graph will define the required motor in terms of horsepower. The correct curve is the one that corresponds with the placement of the impeller curve relative to the other curves i.e. if it is the 3 rd curve on the pressure loss graph, use the 3 rd curve on the horsepower graph. The drawn vertical line continues down to the next graph, which identifies the pump efficiency. Selection of the correct curve follows the same steps as that for the horsepower curve. From knowing the pressure loss across the pump in terms of psi and converting it to Head by multiplying by the conversion factor 2.31 as well as knowing the flow rate in terms of gpm, the impeller size, motor requirements, and efficiency can easily be determined. Actual pump sizing will vary from this estimate because each pump supplier has slightly different pump performance curves. 187

188 Figure 4.4. Pump performance curves for a typical centrifugal pump. 188

189 Upon review of standard pump curves, a total of three pump sizes will be required for the twelve water skids. From the SINDA/FLUINT modeling, the flow rate and pressure drop requirements for each closed loop system was determined. This information is available in Section 3 of this report. To allow for any assumption errors in modeling and to allow for future system growth, the pump sizing used 125% of the calculated pressure drop across the pump. The estimated horsepower requirements for the pump motor are 6 hp (for DTL-1, CCL-MAG, and SCL-MAG), 12 hp (for CCL-1, CCL-2, CCL-3, and CCL-4), and 20 hp (for DTL-1, DTL-2, DTL-3, DTL-4, DTL-5, and DTL-6). The pump efficiency ranges from 53% to 74% for these cases. The pump sizes required for the DTL water cooling systems are summarized in Section of this report Heat Exchanger The water skid performance depends greatly on an efficient heat exchanger. A stainless steel brazed plate heat exchanger will be selected. Plate-type heat exchangers outperform traditional shell-and-tube heat exchangers and do so while reducing size and weight. FlatePlate Inc. has been selected as the company to supply heat exchangers for the SNS Linac water cooling systems based on their efficient design, heat transfer characteristics, and history of outstanding performance. Each skid must actively adjust the temperature of the water sent to its respective tank, module, or magnets by remotely adjusting a control valve and bypassing an appropriate quantity of water through a heat exchanger. This task involves the sizing of heat exchangers to support active cooling/heating of the Linac while adhering to pressure loss and temperature requirements. The design specifications needed to size the heat exchangers were taken from the SNS DTL and CCL Water Cooling and Resonance Control System Description Document [1.2]. During steady state, full RF power, the temperature of cooling water delivered to the each DTL tank is specified to be /-.28 C. Furthermore, for the six DTL tanks, the waste heat loads range between 34.2 and 96 kw and require and 240 gpm of cooling water. A flat plate liquid-to-liquid heat exchanger was selected to 189

190 transfer the waste heat from the closed DTL water loop to chilled facility water. The chilled facility water supply temperature was specified to be 7.2 o C. Pressure drop through the hot side loop of the heat exchanger at extreme operating conditions was limited to 5 psi. This value was selected based upon engineering judgement. For the facility side of the heat exchanger, a 10 psi limit was imposed. Selection of the appropriate heat exchangers for the SNS Linac water cooling systems is critical for successful operation of the Linac. The size and quantity of plates are the only two variables used in evaluating the heat transfer coefficient of potential heat exchanger models. To proceed in sizing of the heat exchangers, the plates sizes were set at 10 x 20 which allowed the heat exchanger sizing study to be based on the quantity of plates. Added to the complexity of this analysis is the desire to minimize the number of unique heat exchangers. This section demonstrates the process and reasoning behind the selection of heat exchangers. The selection process is not trivial, however, and many variables, shown in Figure 4.5, must be considered. Tho R F Structure Qin Thi By-Pass Proportional Control Valve Tmix Variable-Speed Pump FR total FR hx Qout P hs P cs Heat Exchanger Size FR cs Tco Facility Chilled Water Outlet Facility Chilled Water Inlet Figure 4.5. Water skid flow loop and the variables that influence the heat exchanger size. Tci 190

191 The block diagram in Figure 4.6, shows the steps necessary to identify the correct heat exchangers for each closed loop system. This method is trial-and-error approach that is time consuming and requires set parameters to reach a conclusion. SELECTING A HEAT EXCHANGER STEP 1 Determine all cases to study STEP 2 Determine cold side pressure drop STEP 3 Is cold side pressure loss less than 10 PSI? NO YES STEP 4 Select heat exchanger to study Consult manufacturer's data sheets for particular cases and create a relationship for overall heat transfer coefficient vs. heat exchanger flow rate STEP 5 Input relationship, heat load, and flow rates into Sinda/Fluint model and plot results STEP 6 Determine hot side pressure drop for each case at Tmix = 14 deg. C. STEP 7 Is the hot side pressure drop determine in STEP 6 less than 5 psi? NO YES All preliminary requirements met for specific case Eliminate case STEP 8 Select next heat exchanger to study Figure 4.6. Block diagram showing heat exchanger sizing procedure. 191

192 For this study, a heat exchanger was sized for DTL tank 3. It is appropriate to study tank 3 since it represents the "worst case" (highest waste heat to cooing water ratio 95 kw/ 240gpm) situation for cooling in the DTL. The results from the tank 3 study were adapted to size all other heat exchangers employed in the DTL. The heat exchanger sizing study is outlined in the following eight steps: Step 1: Determine all case for sizing the heat exchanger. Result: Table 4.2 displays all cases considered for sizing a heat exchanger for DTL tank 3. Note that the heat exchanger size is characterized by the number of plates. Cold Side Outlet Temperatures Table 4.2. Cases considered for sizing the DTL tank 3 heat exchanger Number of Plates < <10 <10 <10 <10 <10 <10 <10 < >16 >16 >16 >16 >16 >16 >16 >16 Step 2: Determine cold side pressure drop for each case. Results: See Figure 4.7 for cold side pressure drop information. Figure 4.7 shows that the pressure drop in all cases becomes too large after a cold side flow rate of approximately 8.5 kg/s. 192

193 Heat Exchanger Cold Side Pressure Drop Pressure Drop (psi) plate 40 plate 50 plate 60 plate 70 plate 90 plate Flow Rate (kg/s) Figure 4.7. Pressure loss across cold side of heat exchanger vs. cold side flow rate for each heat exchanger size. Step 3: Eliminate cases by determining which cases do not meet the 10 psi pressure drop restriction imposed on the cold side. Result: Table 4.3 Table 4.3. Evaluation of cases based upon cold side pressure drop criteria. Cold Side Outlet Temperatures Number of Plates < <10 <10 <10 <10 <10 <10 <10 < >16 >16 >16 >16 >16 >16 >16 >16 Cold Side Pressure Drop Criteria Not Met Hot Side Pressure Drop Criteria Not Met Both Hot and Cold Side Pressure Drop Criteria Not Met All Criteria Met Have Not Been Studied 193

194 Step 4: Select a specific heat exchanger to study Result: Selected a 70 plate heat exchanger with cold side outlet temperatures of 11, 12, 13, and 14 deg. C and created relationships between the overall heat transfer coefficients and hot side flow rates. Refer to Figure Overall Heat Transfer Coefficient y = x x x x x y = x x x x x y = x x x x x y = x x x x x Flow Rate (kg/s) Figure 4.8. Plot of the heat transfer relationships for a 70 plate heat exchanger. Step 5: Input relationships into Sinda/Fluint model and vary the amount of water sent to the heat exchanger to determine Tmix. Result: Figure

195 70 Plate Heat Exchanger for DTL Tank Tmix (deg. C) Flow Rate (kg/s) Figure 4.9. Tmix vs. hot side flow rate for various cold side outlet temperatures. Step 6: Determine hot side pressure drop for each case at Tmix = 14 deg. C. Results: See Figure 4.10 for hot side pressure drop information. Figure 4.10 shows that the pressure drop in all cases become too large after a flow rate of approximately 5.75 kg/s. Hot Side Heat Exchanger Pressure Drop Comparison Pressure Drop (psi) plt 60 plt 90 plt 40 plt 50 plt 70 plt Flow Rate (Kg/s) Figure Hot side pressure loss vs. flow rate for various heat exchangers. 195

196 Step 7: Eliminate cases by determining which cases do not meet the 5 psi pressure drop restriction imposed on the hot side. Use appropriate data inferences to further eliminate or accept other cases (i.e. if a 70 plate at Tco =12 deg. C is acceptable, then so must a 90 plate at Tco =12deg. C) Results: Table 4.4. Table 4.4. Evaluation of cases based upon hot side pressure drop criteria. Cold Side Outlet Temperatures Number of Plates < <10 <10 <10 <10 <10 <10 <10 < >16 >16 >16 >16 >16 >16 >16 >16 Cold Side Pressure Drop Criteria Not Met Hot Side Pressure Drop Criteria Not Met Both Hot and Cold Side Pressure Drop Criteria Not Met All Criteria Met Have Not Been Studied Step 8: Repeat steps 4-9 for a 90 plate heat exchanger. Results: Table

197 Table 4.5. Final results of heat exchanger size elimination. Cold Side Outlet Temperatures Number of Plates < <10 <10 <10 <10 <10 <10 <10 < >16 >16 >16 >16 >16 >16 >16 >16 Cold Side Pressure Drop Criteria Not Met Both Hot and Cold Side Pressure Drop Criteria Not Met Hot Side Pressure Drop Criteria Not Met All Criteria Met Have Not Been Studied Table 4.5 shows that a heat exchanger with 60 or more plates satisfies all criteria. Note however that a 60 plate heat exchanger does not have very much flexibility. Therefore, it is recommended that at least a 70 plate heat exchanger be employed for use in cooling DTL tank 3. Rather than continuing through the costly and time consuming selection process for every DTL tank, a 70-plate heat exchanger was selected for use on all DTL tanks. A quick check was performed on tank 1. For tank 1, both a 30 and 90 plate heat exchanger were modeled with the same cold side flow rates to determine the effect that an oversized heat exchanger had on temperature. Figure 4.11 displays the results. 197

198 Tank 1 Heat Exchanger Comparison t =12 90plt t =14 90 plt t = 12 30plt t = 14 30plt Tmix (deg. C) Hot Side Flow Rate (kg/s) Figure Tank 1 heat exchanger comparison. Figure 4.11 shows that there is not a significant difference between the curves. Therefore, all DTL tanks may employ the same heat exchanger as that used on tank 3. The heat exchanger sizes required for the DTL water cooling systems are summarized in Section of this report. As a final note on the heate exchanger design, fouling of the heat exchanger surfaces is of significant concern due to losses in heat transfer efficiency. In the case of the linac water cooling systems, the hot side of the water skid s heat exchanger is kept clean of any potential deposits by the use of filters and a high quality water purification system. However, this is not the case for the cold side flow from the facility supply. Therefore, there is some concern about the effects of fouling on the cold side of the heat exchangers. a) Definition-The definition of heat exchanger fouling is deposition of an insulating material on the heat transfer surfaces. These deposits can be biological, precipitation of dissolved substance, accumulation of finely divided and suspended solids, and chemical reactions [4.12]. Corrosion is also another form of fouling that can occur. These types of fouling can occur separately or simultaneously. 198

199 b) Effect-The effect of insulated deposits on the heat exchanger surfaces is to reduce the heat transfer and increase the pressure drop through the heat exchanger. Heat transfer is impeded due to an added layer of material that must conduct the heat. Fouling reduces the overall heat transfer coefficient by adding an insulating deposit that increases the thermal resistance. If the flow area is significantly reduced and the surface is roughened due to fouling deposits, this can cause the pressure drop to increase. However, the velocity increases due to the reduced flow area and its effect increase is directly proportional to velocity squared. c) Fouling Factors-Fouling factors are multipliers of thermal resistance and as they increase the thermal resistance increases. The authors of reference 1 present a table that shows that fouling factors can range from m 2 K/W to m 2 K/W for river water. Cooling tower treated makeup ranges from m 2 K/W for treated makeup to m 2 K/W for untreated makeup. City or well water can range from m 2 K/W to m 2 K/W depending on the velocity. Reference [4.12] shows that a heat exchanger in a fouled condition can increase the pressure drop by 70%. d) Flat Plate Heat Exchangers-Extraction of heat deposited in the cooling water by the CCL and DTL structures will be removed by flat plate heat exchangers. The cold side supply will be chilled water at 7.2 C and may have outlet cold side temperatures that range from 10 C to 17 C. Reference [4.12] discusses the performance of flat plate heat exchangers and points out that these types of heat exchangers are used in processing of foodstuffs where frequent cleaning is required. The corrugated and torturous path leads to high heat transfer coefficients. The turbulence reduces the potential for fouling of the heat transfer surfaces. Fig shows the data presented by Cooper et al on a flat plate heat exchanger using cooling tower water [4.13]. This data shows the relationship of fouling resistance as a function of velocity and temperature. The point where the cold side water flow for the linac structures are likely to be located is also shown on Fig The temperature for the heat exchangers is expected to be approximately 284 K and the velocity to be approximately 0.26 m/s. This shows that the fouling resistance will be much less than for the temperature and velocity in the RF structure cooling loops. Therefore, it can be concluded that heat exchanger cold side fouling is not likely to be a problem. However, if it does become a problem a provision is included to flush the heat exchanger with an acid solution. 199

200 Fouling in a plate heat exchanger 334 K midpoint surface temperature Asymptotic Fouling Resistance (m2k/w) K 321 K cooling tower water Ref: Cooper,Suitor, and Usher, CoolingWater Fouling in Flat Plate Heat Exchangers, Heat Transfer Eng., Vol 1, No 3, 1980 CCL HX Data (284 K, V=0.26 m/s) Velocity (m/s) Figure Effect of water velocity on fouling factor Control Valves Two control valves are required for the water system. On the warm side of the heat exchanger will be a 3-way electronically actuated valve. This is the primary control valve that divides the water flow between the hot side of the heat exchanger and the bypass line. On the cold side of the heat exchanger will be a 2-way electronically actuated valve. This valve will help maintain a desired chilled water flow rate through the cold side of the heat exchanger. Both control valves shall meet applicable ANSI and ISA requirements. The primary control valve will be a 3-way diverging valve located at the heat exchanger-to-heat exchanger bypass intersection. The valve will provide true linear proportioning and a smooth gradual flow reduction when flow adjusting. The valve will have stable transitioning when switching ports to prevent valve slamming and pipeline water hammer. All wetted surfaces will be 316 stainless steel and packing/sealing made from Teflon or Viton. Teflon is not recommended for a radiation environment however 200

201 the water skid, located in the Klystron Gallery, is well away from the potential damaging radiation found in the Linac Tunnel. Additionally, this is a static seal, which will prevent any significant wear. The valve will require a 100% duty cycle. The valve shall have at least 200 incremental steps in it setting position to allow for sufficient flow control resolution. The valve will have sufficient actuation speed so as to move across its full range of motion in less than 60 seconds. A 2-way electronically actuated flow control valve will be required prior to the entrance on the cold side of the heat exchanger. This is facility-chilled water and is not deionized. This valve does not require 316 stainless steel, Teflon or Viton as construction or housing materials. However, it is required to be compatible with the tubing material and must be highly sturdy and reliable. The valve will require a 75% duty cycle. The valve shall have at least 200 incremental steps in it setting position to allow for sufficient flow control resolution. The valve will have sufficient actuation speed so as to move across its full range of motion in less than 60 seconds. The valves will be electronically actuated and will provide a position feedback signal. Each will operate with a two-wire, 4 to 20 ma signal for both the input command and the position feedback signals. The available input power will be 24 Vdc. Position accuracy will be ± 1% of full actuator travel for the 3-way valve and ± 5% of full actuator travel for the 2-way valve. The actuator housing for each valve will be NEMA type 4 requirements Heater The inline water heater has been sized accoriding to the following analysis. The primary use of the heater will be to heat the water and RF structures/magnets when RF heating is not available. For example, during the alignment phase of the linac, the structure will need to be at its mean operating tempertature, which is several degrees Centigrade above room temperature. By heating the cooling water, the RF structure can be brought up to temperature, and the alignment technicians can still have access to the linac tunnel (which is not the case if high levels of RF energy are present). 201

202 The heater sizing was based on the following calculation. Assuming that a typical water loop contains about 300 gallons of water, and that the water needs to be heated 5 C in less than one hour, determine the size of the water heater to accomplish this task. q = mc p ( T/dt) where q = 20 kw, m = 300 gallons, T = 5 C 20,000 W = 300 gal/264.2 gal/m 3 x 1 m 3 x 1000 kg/m 3 x 4180 m 3 /kg x 5 C/dt solving for dt dt = 1185 seconds < 20 minutes for a 20 kw inline heater dt = 1972 seconds < 33 minutes for a 12 kw inline heater Based on actual usage, cost, size, and a reasonable time required for heating the water, an inline between 12 kw and 20 kw will be sufficient. As a final note, the desired water connection ports on the heater unit, will be flanged for ease of removal. The ports shall be a minimum of 1 diameter and optimally 3 in diameter to reduce pressure drop Water Purification System The Water Purification/Filtration System will be hard mounted directly to the water skid structural frame. It will contain filters, mixed bed canisters, carbon bed canisters, a flow meter, and will provide water purity status to the PLC. The purification system will draw off a small portion (1-5%) of the water from the primary flow path, treat and clean the water, and return this newly purified water back to the primary flow loop. This purification loop functions within the overall closed loop system. For more detail regarding the water purification system, refer to Section 5 of this report System Performance The water cooling system performances are basedd on the cooling requirements and thermal/fluid modeling as described in Sections 1 and 3 of this report. The critical components in the water cooling systems are the heat exchanger and pump. Sections and describe the detailed process in selecting the pumps and heat exchangers, respectively. In particular, the heat exchanger has many intricacies that required detailed examination as well as many performance variables. Under steady state conditions, the heat exchanger will have a performance range, controlled by the 3-way 202

203 electronically actuated bypass control valve, that readily covers the temperature performance requirements as specified in the associated SNS Linac Water Cooling and Resonance Control Systems Description Document [1.2]. Based on steady state analysis, the estimated number of different water skids, based on pump and heat exchanger sizes, is three (See Table 4.6 below). Pump size estimates were made using a 25% extra capacity in the pump pressure, over that predicted in the analyses of Section 3 of this report. The increased pressure drop capacity will account for any uncertainties that may exist in the SINDA/FLUINT modeling, future increase in the temperature range of operation, and allow for increased potential cooling capabilities. Table 4.6. Summary of heat exchanger and pump sizes for the DTL, CCL, and SCL water cooling systems. FLAT PLATE FLOW 125% OF DELTA P ESTIMATED PUMP UNIQUE SKID HEAT EXCHANGER (GPM) (PSI) HEAD (Feet) MOTOR (Hp) IMPELLER SIZE EFFICIENCY SKID % DTL-1 FP10X20-70(2"MPT) x 1.5 x 6 70% A DTL-2 FP10X20-70(2"MPT) x 2 x 6 53% B DTL-3 FP10X20-70(2"MPT) x 2 x 8 60% B DTL-4 FP10X20-70(2"MPT) x 2 x 8 58% B DTL-5 FP10X20-70(2"MPT) x 2 x 8 55% B DTL-6 FP10X20-70(2"MPT) x 2 x 8 56% B CCL-1 FP10X20-90(2-1/2"MPT) x 2 x 6 74% C CCL-2 FP10X20-90(2-1/2"MPT) x 2 x 6 73% C CCL-3 FP10X20-90(2-1/2"MPT) x 2 x 6 73% C CCL-4 FP10X20-90(2-1/2"MPT) x 2 x 6 73% C CCL-MAG FP10X20-70(2"MPT) x 1.5 x 6 60% A SCL-MAG FP10X20-70(1-1/2"MPT) x 1.5 x 6 61% A In addition to the cooling water temperature and flow rate ranges supplied by the selected heat exchangers and pumps, cooling performance and capacity can be influenced significantly with the use of the inline water heater, and the two control valves located on either side of the heat exchanger. Three off-normal operational conditions, which would take advantage of the flexibility in the water cooling system, are the follwing: RF power off, increase structure temperature 1] Turn power on for the inline heater. 2] Close manual inline heater bypass valve. 3] Adjust 3-way control valve to bypass 100% of flow to the heat exchanger. 203

204 RF power on, increase cooling to structure beyond system design temperature 1] Adjust 3-way control valve to increase flow to the heat exchanger. 2] Verify 2-way control valve on heat exchanger cold side is fully open. 3] Increase variable speed motor on pump. RF power on, increase warming to structure beyond system design temperature 1] Adjust 3-way control valve to bypass 100% of flow to the heat exchanger. 2] Turn power on for the inline heater. 3] Close manual inline heater bypass valve. The previously described scenarios are examples of how easily this system can be adjusted to meet a variety of conditions. The system design was created with great flexibility. As a final note, the number of unique water skids, based on heat exchanger and pump size, is three. These skids are identified in the last column in the Table 4.6. The SNS program will benefit by a reduction of required spare components with only three unique skids. 4.6 Parts Database and Naming Convention The DTL and CCL water system parts data bases were created to keep track of the multitude of cooling system components, and serve as a library of information for the entire design team. The parts databases were developed from the P&ID s and continuously updated to reflect the present design. These databases serve several functions; as a documentation source for the hardware sizes determined from the Sinda/Fluint numerical modeling, defining components (manufacturer/model number) and their device names, providing a data base for component purchasing, and device listing to facilitate PLC and global programming for the sub-systems controls. Device names for the data base were selected using the SNS Device and Signal Naming Convention and form the bases for overall system software development by identifying names from actual device through user interface screens. 204

205 The naming convention is mainly oriented to facility operations and has been developed for the purpose of naming signals which come into the SNS integrated control system which are then manipulated, displayed, archived, etc. Because signals are always associated with equipment or devices, the signal names include the names of devices with which they are associated. This syntax can also be used to name SNS equipment that does not have associated signals in which the signal part of the name is simply omitted. form: The general format of the SNS Device and Signal Naming Convention has the SystemPart:DevicePart:SignalPart The three parts together constitute the complete signal name. The SystemPart is made up of a System Name and optional Subsystem Name. The DevicePart is made up of the Device Name and the System/System Name. The SignalPart is made up of the System Name, Device Name and the Signal. Therefore: System Name=System/SubSystem, Device Name=System/SubSystem+Device, Signal Name=System/SubSystem+Device+Signal Example using the first CCL RCCS water cart: CCL_RCCS1:TT1:T CCL is the system- RCCS1 is the subsystem; 1 indicating #1 RCCS system- TT1 is a temperature transmitter of the water cart- T is the signal; temperature What is shown is the basic organizational scheme using the P&ID and associated spreadsheet database lists for the various systems. Since all the water carts are the same in component count, the System and Subsystem denotes which system is referenced at the control center or in the field during maintenance or trouble shooting periods. Signal names were left off the excel database at this time but will be incorporated after the control system is developed, then integrated into the data base. The controls group will be developing a signal list for the many different signals types and provide the appropriate signals for the excel database. The final excel data base will be incorporated 205

206 into the project configuration database using Oracle and will allow devices and signals to be located throughout the linac. The DTL parts database is contained in Appendix E. 206

207 5.0 Water System Purity 5.1 Introduction Pure water is a necessary commodity demanded by nearly all of the World s industries. It is required to produce items that are bought and used everyday by millions of people. Items such as intravenous injections by the medical industry, hardware by the computer industry, and even usable energy by the power generation industry, are all produced using pure, contaminant free water of different grades. Although many industrial processes require water that is pure and contaminant free, the level of purity and the type and quantity of contaminants present in the water can vary greatly. This implies that the pure water standards of one industry may not meet the pure water standards of another. Factors determining water purity for a particular application include permissible impurities, corrosion or erosion of wetted materials susceptibility, water availability, quantity, cost, etc. Each industry must define, implement and maintain a specific level of water purity to ensure both product quality and efficiency at a reasonable cost. Typical parameters used to measure or quantify water purity include ph, electrical conductivity, total suspended and dissolved solids, dissolved oxygen content, and radioactivity. Parameters of a specified value can be achieved by employing various purification techniques and equipment. Common techniques such as microfiltration, ultrafiltration, reverse osmosis, carbon adsorption, deoxygenation, ultraviolet radiation, and ion exchange can be employed to purify water. Figure 5.1 illustrates a typical water treatment system. Many of these technologies will be used to produce contaminant free water for use in the SNS Linac cooling systems. 207

208 Raw Water Pretreatment Activated Carbon Reverse Osmosis Drain Storage Pump UV Polishing Loop Mixed Bed Deionization Ultrafiltration Drain Figure 5.1. Generic water treatment system (courtesy of Cartwright [5.3]). Microfiltration 5.2 Water Purification Techniques This section describes several different water purification techniques commonly used in industry. Microfiltration Microfiltration is a purification process employed to remove contaminant materials from water including suspended solids, bacteria, colloids, etc., which are typically larger than 0.02 microns. The microfiltration process can occur in the polishing loop (see Figure 5.1) using either a dead end or cross flow filtration process. In the dead end filtration process, contaminant particles too large to pass through the filter are trapped. Heavy filter contaminant buildup will eventually occur and the filter will need to be replaced. Unlike the dead end filtration process, the cross flow filtration process utilizes a tangential water flow scheme to continuously wash out contaminants from the water system. Due to the continual loss of water, cross flow filtration should not be employed in completely closed water systems. Figure 5.2 illustrates both microfiltration processes. 208

209 Filtered Water Filtered Water Key Filters Contaminants Water Flow a) Dead end b) Cross Flow Figure 5.2. Microfiltration processes. Ultrafiltration Ultrafiltration is a process, which utilizes membranes to remove non-ionic contaminant particles ranging roughly from.002 to.02 microns in size. Ultrafiltration is most effectively used for the removal of microorganisms, high molecular weight contaminants, and colloidal material [5.3]. The ultrafiltration process requires a crossflow filtration scheme (similar to what may be used in microfiltration Figure 5.2(b) to continually remove contaminants from the system. Ultrafiltration usually occurs in the polishing loop of the water system as seen in Figure 5.1. Reverse Osmosis Reverse osmosis (RO) employs the properties of semi-permeable membranes to purify water. Only selective materials, such as water and water-similar molecules (based on size and molecular weight), may transcend the membrane. Typically, reverse osmosis systems can remove 90 to 98 % of ionic contaminant [5.3]. The reverse osmosis process usually occurs in the pre-treatment portion of the purification system. While effective, due to its requirements for an additional pump and drain an RO system will not be employed in the hot model purification loop. Carbon Adsorption Another commonly used process in water purification is carbon adsorption. Carbon adsorption utilizes activated carbon (usually in powder or granular form) to remove high molecular weight organic contaminants from water systems. Another attractive characteristic of this technology is the ability to remove chlorine and traces of certain 209

210 heavy metals from the water system. Since chlorine is harmful to most membranes used in reverse osmosis, carbon adsorption is usually one of the initial purification processes. Ultraviolet Radiation Ultraviolet radiation exposure is another technology often used to reduce the number of microorganisms present in water systems. By taking advantage of the fact that microorganisms, such as bacteria, have little or no resistance to intense ultraviolet radiation, a simple, low-tech process can be employed to inhibit the propagation of microorganisms. Deoxygenation Deoxygenation is a process of removing excess amounts of dissolved oxygen from water. In removing oxygen from the water system, both the corrosion and the microorganism growth rates are decreased significantly [5.4]. Several methods exist for removing oxygen gas from liquid. One deoxygenation method utilizes resins that act as scavengers to remove oxygen from water. Another method employs a hydrophobic microporous membrane. A fluid passes over the membrane a vacuum on the other side of the membrane pulls the gas out of solution. Critics claim that the removal of excess dissolved oxygen from water systems will have much better results in reducing biological growth than ultraviolet radiation [5.4]. Ion exchange Ion exchange is a purification process that employs special resins to remove positively and negatively charged ions from solution. Resins are synthetic materials composed of small beads and can be of either the cation or anion type. In a water purification system, cation resins exchange hydrogen ions for unwanted cations, while anion resins exchange hydroxyl ions for unwanted anions. Cation and anion resins can be used individually or can be combined in a mixed bed system to purify water. Typically, high water quality (resistivity of 18 MOhm-cm) can be achieved when an ion exchange system is employed. Ion exchange is an economical water purification technology since resins can be repeatedly regenerated after they have become fouled. 210

211 Figure 5.3 compares the various purification technologies. Colloids Particles Contaminants Retained Contaminants Passed Through Organics Salts Bacteria Distillation Deionization Ultrafiltration Reverse Osmosis Figure 5.3. Comparative purification technologies ( Courtesy of Cartwright [5.5]). 5.3 Particle Accelerator Specific Issues A water purification system will be responsible for maintaining high quality cooling water for the SNS linac. Water treatment is a necessary process for retaining high and consistent system efficiency. The formation of deposits, scale, biological growths, corrosion and activation can be of significant threat to the performance of the SNS linac water cooling system. Corrosion Corrosion must be limited in the SNS Linac water cooling system to maintain cooling efficiency and minimize damage to accelerator comonents. Corrosion is the dissolution of a solid in a fluid (in this case, metal in water). Corrosion in the cooling passages promotes material build-up, thus reducing the heat transfer rate. One form of corrosion, oxidation, occurs when dissolved oxygen in the water reacts with metal flow-passage walls. Galvanic corrosion occurs when two or more dissimilar metals come in contact. It 211

212 may be reduced by several techniques. Minimizing dissolved oxygen and dissolved salts in the water, employing compatible building materials, and providing a galvanic insulation between dissimilar metals can significantly reduce the amount of corrosion. At any rate, corrosion can cause a number of problems to occur within the SNS linac cooling system, including flow rate restrictions, increased head loss, reduced heat transfer, pitting and leakage. Scaling Scaling is the formation of deposits, including calcium and silica salts, on metal surfaces. The formation of scale on pipe walls occurs as a result of large temperature changes. Large temperature changes affect solubility, minerals and salts precipitate out of solution and eventually build up on exposed metal surfaces. Scaling can cause problems in water systems by reducing flow rates and heat transfer rates. In the case of the SNS linac cooling system, large temperature variations are not expected. Consequently, scaling should not be a major concern in the SNS Linac water cooling system. Biological growth Biological growth in water systems is very common. A water system is an ideal place for microorganisms to grow and reproduce. If large amounts of microorganisms are present in a water system, problems such as increased corrosion, water leaks, head loss, and a reduction in heat transfer are likely. It is very important that biological growths within the SNS cooling system remain at a minimal level. This can be accomplished by reducing the dissolved oxygen content in the water to a level, at which microorganisms cannot survive, or by passing the water through an ultraviolet light source [5.4]. Pipe design also plays a roll in reducing bacterial growth. Piping will be designed as to eliminate stagnation areas, or dead-legs, which are areas that have little or no flow. The lack of motion or kinetic energy in the fluid provides a breeding ground for bacteria, both due to the lack of fluid motion as well as a place to trap dissolved oxygen. Eliminating dead-legs makes it much more difficult for bacteria to thrive. This will be done by designing the piping such that there is flow into each run whenever possible. Carbon 212

213 based populations can be measured by taking water samples, and are counted as Total Organic Carbon (TOC) represented as mg/l and Heterotroph Plate Counts (HPC) represented as (Colony Forming Units) CFU/L. Activation During operation of the SNS accelerator, scattering of the proton beam may allow the cooling water to be subjected to direct spallation and activation. Be-7 is a radionuclide produced through the direct spallation of oxygen and is largely responsible for the activity in the cooling water. In addition, spallation neutrons activate corrosion products present in the water and in turn generate long-lived radionuclides including Co-60, Zn- 65, and Mn-54 [5.6]. Radioactivity in the cooling system is of concern due to the potential for contamination of hardware and personnel. Safe accelerator operation demands that the quantity of activated water, caused primarily by Be-7, be minimized. Deionization has proven to be a very effective process for the removal of Be-7 [5.6] from accelerator cooling water. Entrapment of radionuclides in the resins used in a Linac cooling system cannot be regenerated. Instead, they must be removed from the cooling water system and replaced. The old resins will need to be dried and properly disposed as low-level radioactive waste, unless surveys deem otherwise. Radiation levels are not expected to be high in the purification system and therefore shielding will not be required. 5.4 Operating Parameter Specifications Features from current water treatment systems on particle accelerators were considered for use in the SNS cooling water system. These included the Los Alamos Neutron Science Center (LANSCE), the Accelerator Production of Tritium (APT), and the Advanced Photon Source (APS). The Linac water-cooling purification system was designed with the intent of minimizing erosion, corrosion, scaling, biological growth, and hardware activation. Each component was selected to target the removal of a specific impurity, and in some cases, multiple impurities. 213

214 The basic mechanical design of the cooling loop has helped to minimize erosion and scaling. Water flows in the cooling systems will be kept below 2.5 m/s on surface impingement areas such as tees and elbows, and less than 5 m/s in straight sections to reduce the effects of erosion. The narrow temperature band of the cooling water, 10 to 25ºC, reduces scaling. Other critical parameters, which have been defined and will be controlled, include electrical resistivity, pressure, ph, and dissolved O 2 content. Typically an electrical resistivity value above 6 MO drastically reduces scaling [5.9]. However, it is important to keep the resistivity below 15 MO, particularly in copper structures and piping. Due to the polar nature of ultrapure water, a very high resistivity tends to strip away ions from the metal surface of piping, particularly copper when dissolved oxygen is present [5.10]. Maintaining the resistivity below 15 MO minimizes this effect. Analog pressure gauges will be used to not only indicate system pressure, but also to indicate filter-loading information for filter replacement. Approximately 1-5% of the main water flow will be diverted to the water purification loop, the quantity being monitored with a flow meter. The flow in this loop requires an oxygen content of less than 20 ppb to minimize corrosion [5.1]. When copper is exposed, the corrosion is embodied as insoluble particles of CuO and Cu 2 O, which amass in the system after removal from the parent surface. Filtration is helpful in reducing this effect, however minimizing oxygen in the system proves more effective [5.1]. Table 5.1 summarizes the water quality to be obtained from the water purification system. Table 5.1. Water purification parameters. Parameter Recommended Value Ref. Flow rate 1-5 % of total flow [5.7] (through purification tanks) ph 8 ± 1 [5.2] Electrical Resistivity MO [5.10], [5.11] Dissolved Oxygen Content < 20 ppb [5.3] Particulate size = 1 micron 214

215 5.5 Water Purification System Design The water purification system design is shown in Figure 5.4. It was designed to meet the water purification specifications for the SNS linac system. Considerations included minimizing the adverse effects of scaling, flow blockage, biological growth, activation, and other forms of contamination, much of which is covered in section 5.2. Component selection was made carefully to develop a lower maintenance passive system. Reverse Osmosis was discarded for this application. While very useful and effective at removing a variety of contaminants, it is a higher maintenance item with more moving parts than other components. A 1-5% portion of the flow will be diverted from the main loop into the purification or polishing loop. A 5-micron pre-filter sieves out the largest contaminants. Following the filter, the water will pass through an oxygen scavenger resin canister, similar to the system used in the APS Linac Water Cooling System [5.1]. The carbon canisters remove hydrocarbon contaminants, including residual petroleum products remaining in the piping from manufacturing processes. After passing through the carbon adsorption canisters, the water then passes through the deionization canisters. Here any free ions remaining in the system are removed, which left untreated could contribute to scaling. Next, an ultraviolet light source is used to kill surviving bacteria, which are then filtered out by a 1-micron filter. At this point in the system, the water is quite pure. The pure or polished water will then be returned to the main loop. The ultraviolet light source may be redundant. It is believed that the oxygen removal mechanism will be able to deprive oxygen supply to effectively eliminate organic growth. At the time of this writing, the CCL hot model water cooling system is being used to study this very issue. Main flow Filter 5-µm Oxygen Removal Carbon Deionization UV Filter 1-µm Figure 5.4. Process flow through the water purification hardware for the DTL/CCL water cooling systems. 215

216 The piping material will consist of stainless steel, which serves as an effective barrier between the atmosphere and the cooling water, minimizing O 2 permeation. Due to the corrosive nature of deionized water, brass and carbon steel components are not acceptable. Flexible tubing will be used as jumper or transfer lines to avoid the need for high tolerances in the water line designs and to serve as vibration isolators. Several flexible tube options were explored. Material selection was based on DI water compatibility, low oxygen permeation rates, and resistance to radiation environments. A comparison of various flexible tubing materials based on these parameters, is presented in Appendix H. Many materials examined were found to meet one or two of our needs, but only a few were found to meet all three. Those that met all three of our requirements include Viton, Nylon, Hypalon, and Buna,. Tubing manufactured from the first two materials, may not possess a suitable pressure rating. A review of the various materials found that neoprene does possess adequate radiation tolerance and provide adequate pressure ratings, however, it does not provide an adequate barrier to O 2 permeation. Consequently, it may be advantageous to consider using a multi-ply hose or tubing with Viton, Nylon, or Hypalon as the wetted surface, with a neoprene sheath since the neoprene sheath can provide the needed pressure rating and radiation resistance. Multiply hose and tubing of this type may be obtained from sources such as Goodyear, under various trade names, as well as Boston Nyall and Thermoid/HBD Industries. Various types of instrumentation will be used to monitor and record the performance of the water purification system. Instrumentation will be provided to measure electrical resistivity, pressure, flow rate, ph, and dissolved O 2 content. Measuring each of these parameters will give feedback concerning the overall quality of the water as well as alerting operators to component failure. Much of the water quality data can be obtained from the instruments incorporated within the system. However, parameters that cannot be easily monitored with sensors or their capital costs make them prohibitive include particulate, bacteria, Total Organic Carbon (TOC), heterotroph plate counts (HPC) and trace elements. Particulate sampling will verify the filters are removing the desired particulate size. TOC and HPC testing will indicate bacterial and carbon based populations. Testing for trace elements such as 216

217 iron, copper, and zinc, elements used in the piping system will indicate negative effects of the DI water on the piping system. A water sample should be taken after initial startup, after water has circulated through the purification loop. Samples should also be taken periodically and after major water system hardware configuration changes. A draft procurement specification has been produced for the water purification systems (see Appendix J). This procurement specification will be incorporated with the over-all procurement document for the water skids. 5.6 Prototype Design and Testing Prototype Design To test the performance of various water purification techniques and hardware, and thus optimize the design of the water purification system for the SNS linac, a prototype water-pumping skid with the water purification system features shown previously in Figure 5.4, has been developed for the CCL Hot Model. In regards to water quality, experimental data was taken to optimize the performance and minimize the cost of the purification system. The prototype purification system has been sized for a closed-loop application utilizing ¾ piping for flow rates up to 3 gpm and is comprised of the following components: A mixed-bed canister and a cation canister provide deionization for the cooling water. The mixed bed canister contains resins for removing both cations and anions. As mentioned earlier the SNS accelerator will require a Be-7 removal mechanism. A cation resin, Amberlite IR-120 in H + form, which has been proven effective at removing the Be-7 nuclide, was used [5.7]. Although Be-7 was not generated in the hot model experiment, the Amberlite IR-120 was incorporated into the hot model to observe its performance as a cation resin. Two carbon canisters eliminated dissolved organics, such as soldering fluxes or machining oil, and decrease biological growth build-up. An ultraviolet light source was provided to kill bacteria. A 5-µm filter upstream of the purification loop was used to remove large particulates before they enter the purification loop. A 1-µm filter removed smaller particulate 217

218 matter, some large bacteria, and any resin material, which may have entered the loop from the canisters. Two Liqui-Cel contactors were required to achieve the dissolved oxygen content requirement in the cooling water. The contactors, which are comprised of a large number of very fine mesh polypropylene cross flow filters, allow only gases to pass through. A small vacuum pump was used to pull dissolved gases out of the cooling water along with a nitrogen sweep gas. The basic removal mechanism is shown in Fig The Liqui-Cel unit removed dissolved oxygen in the cooling water, minimizing corrosive effects, and limiting bacterial growth. Nylon3 hoses with PVC reinforcement were used to connect the canisters where flexible lines are needed. Nylon derivatives are compatible with DI water and minimize the diffusion of O 2 from the outside environment [5.8]. Short runs of Teflon and polyethylene tubing were used on data acquisition insturmentation sampling ports. Data acquisition hardware was an integral part of the purification loop on the hot model. Resistivity monitors, a ph meter, and a dissolved oxygen analyzer, Orbisphere 3660, measured dissolved oxygen content in the system. Sensor model #2952A, was selected in conjunction with the analyzer, which would also be a good choice for the facility due to its maximum dose limit of 10 8 rad. 218

219 Figure 5.5. Gas removal mechanism in the Liqui-Cel contactor [5.8] The prototype CCL Hot Model water-pumping skid, which includes the purification system can be seen in Figure

220 Figure 5.6. CCL hot model water-pumping skid with water purification hardware in the forefront. 220

221 Approximate hardware cost figures for the CCL Hot Model water purification system is presented in Table 5.2. Table 5.2. Water purification hardware costs. System Component Manufacturer Cost Water purification unit CLW Systems Inc. $9,200 Degassifier Liqui-Cel by Celgard $3,000 Dissolved O 2 monitor Orbisphere 3660 $6,600 PH monitor Omega $550 Water testing Assaigai Analytical Laboratories $1,500 Prototype Testing This section outlines the procedures used to operate and test the water purification system on the SNS CCL Hot Model experiment. The focus of this effort was to manipulate various configurations of the water purification system to obtain the desired purity while minimizing required hardware and costs in the facility design. All tests began by using ultrapure electronics grade water that was transported from a clean room water purification system. The order and purposes of the 7 tests (A through G) are summarized in Table 5.3. As mentioned above, the first goal as to see if in fact the system could obtain and maintain the purity specifications listed in Table 5.1, with various configuration changes. The final test, which still needs to be performed, will demonstrate how the purification system performs when tap water is used to charge the water cooling system. If this latter test is successful, it will preclude the need for high purity water being trucked to the SNS site, and possibly allow the use of on-site municipal supplies. Another goal is to verify that the purification system is providing adequate water quality for the testing of the copper CCL accelerating structures. Each test recorded the parameters specified in Table 5.1, namely ph, electrical resistivity, and dissolved oxygen. 221

222 Evidence of corrosion and bacterial growth in the system was investigated through laboratory testing of water samples. Table 5.4. Water purification test summary. Test A B C D E F G Test Description Initial water quality check after system was fully operational Test A was rerun to verify the results were repeatable Flow rate increased to 3 gpm to see if the DO concentration could be reduced Flow rate further increased to 3.5 gpm, looking for DO concentration reduction With the UV system off, the system was run and water samples were collected to see how the lack of a UV system impacted the bacterial growth Observe the effect on the carbon and bacterial growth counts when the UV and one carbon bed is removed from the system System water will be replaced with tap water and data will be compared with previous samples. Initial Start-Up and Water Quality Verification Testing This section outlines the initial start-up and regular testing of the water purification system. Tests A and B are included as part of the initial quality verification. This procedure was followed the first time the system was brought on-line, and should be used if the system has been inactive (water is not circulating through the system) for a long period of time, typically more than a couple of months. Additionally the system should be inspected for bacterial growth, as well as all sensors should be examined and cleaned or replaced as necessary. 1) A cursory check was made for loose wiring, including power and data acquisition lines, and any evidence of leaks, or loose fittings. The CLW system was pressure tested at the factory. 2) Extra care should be taken if the system has not been circulated for a long period of time or lines have been broken due to the addition of components 222

223 or soldering operations. Turn off valves to isolate the purification system from the rest of the loop. Run water through the rest of the system to flush out any large particles, fluxes, metal fragments from machining operations etc. Volume will be determined by the appearance of the water. If the system has been dormant for less than a couple of months, skip steps ) Switch position of the valves to direct water into the purification system. 4) Water was added to fill the system, and use air bleeds as necessary to get as much gas out of the system as possible. Reducing the air in the system reduces the amount of oxygen trapped in the piping, reducing bacterial growth. Then the system is back-filled with nitrogen to purge the system of any remaining oxygen not in solution. 5) The power was switched on to the purification system as well as all monitors, ph, resistivity and O 2 concentration. 6) Initially the time as well as all pressures from the pressure gauges, in particular those that indicate the pressures across the filters were recorded. Also the readings from the ph, resistivity and O 2 monitors were noted. This process was automated with the advent of LabView software to record all needed data. 7) Measurements were taken every 15 minutes initially, until measurements appeared to stabilize or the trend began to slow. 223

224 System Performance Testing After the initial start-up was been completed, various water purification hardware configurations were tested, these included Tests C, D, and E. Tests F and G were not completed at the time of this writing. It is suspected that water purification parameters can be met without the UV source and possibly one carbon canister. In addition to instrumentation monitoring, several water samples were taken and sent out for complete chemical analysis. All data should indicate that the water quality is within acceptable parameters prior to proceeding with another test. Tests A and B (repeatability) were performed to observe the ability of the water purification system to reduce and maintain the water s dissolved O 2 concentration to a value of 20 ppb or less. As mentioned earlier, research indicated that a concentration above 20 ppb could be a catalyst of corrosion on copper structures. Tests A and B each consisted of a 2.5 gpm flow rate through the water purification system, the UV system was switched on, and the Nitrogen sweep gas and vacuum pump for the degasifying system was on as well. Based on the data obtained, it was conclude that the system in this configuration could repeatedly obtain dissolved oxygen concentration to an acceptable level, as shown in Figure

225 Data A & B O2 Comparison at T= O2 Concentration (ppb) Data A Data B Goal Time (min) Figure 5.7. O 2 concentration versus time for two test runs (tests A and B) To further enhance the performance of the oxygen degassing system, the flow rate of the influent passing through the purification system was varied at 3 discrete values of 2.5 (Tests A and B), 3.0 (Test C) and 3.5 gpm (Test D). It was desired to see if the degassing system performance was dependent on the flow rate through the water purification hardware. Figure 5.8 shows the results after running the system for 120 minutes. 225

226 O2 Concentration vs. Flow Rate: B,C,D Data Sets T=120 O2 Concentration (ppb) Data B gpm Data C gpm Data D gpm Time (min) Goal Figure 5.8. O 2 concentration vs. time for various water flow rates (Test C). In reviewing the data, a small reduction in the dissolved O 2 concentration was observed, however it is too small to suggest there is any significant effect of water flow rate on dissolved O 2 concentration. Although the dissolved O 2 concentration is slightly above the design goal of 20ppb, water test samples showed no evidence of copper corrosion. The current laboratory water sampling data can detect copper oxide levels down to 10ppb, and thus far, have not detected any copper corrosion products. In tests A through E, the ph was maintained just below 7, which is consistent with the desired operating range specified in Table 5.1. The electrical resistivity values recorded for each of the tests were between 16.3 MO and 17.2 MO, slightly higher than the desired upper limit of 15 MO specified in Table 5.1. This high electrical resistivity was no doubt an artifact of the clean water that was used to fill the system, which had an initial electrical resistivity around 18.1 MO. It is speculated that the water which will be used to fill the SNS water cooling systems, will not have such high purity, and thus the electrical resistivity range specified in Table 5.1 should able to be met. Although the specific tests were not yet performed on the prototype 226

227 water purification system, it is speculated that the electrical resistivity will be able to be adjusted by controlling the water flow rate through the purification hardware. As a final note, water test samples showed no evidence of copper corrosion as a result of the electrical resistivity values obtained in the current set of tests. Chemical analysis data from the source water and the tests completed thus far, are included in Table 5.5. The HPC s and TOC s are more of a relative measure. It is more of a concern if these values show a rapid increase throughout the progress of testing. This would indicate growth of the bacteria or heterotrophs. Thus far, the HCP s and TOC s have been reported at relatively low values. The Test E data set was essentially invalidated since the vacuum pump required service during the test and had to be shut down. It should also be mentioned that the water in the SNS Accelerator would be flowing for long periods of time. The prototype system water does not run continuously for a long period of time. It is frequently shut down, which gives bacteria the opportunity to repopulate. Table 5.5. Water purification data summary for Tests A-E. ph Electrical Dissolved Temp. Cu Resistivity Test ºF (mg/l) (MO) O 2 (ppb) Heterotroph Plate Count (CFU/mL) Total Organic Carbon (mg/l) Source ND ND ND A ND B C D ND 3 ND E* ND 97 ND * Vacuum pump required service during test, degassing system not fully functional ND = Not Detected Additional tests are currently underway and include: 1) The functionality and necessity of the UV system will be studied. It may be possible to eliminate bacteria by simply minimizing the O 2 content of the water and eliminating the UV system. 227

228 2) The size requirements of the carbon canisters will be studied to see if one canister can be eliminated and thus reduce the size, cost, and complexity of the system. 3) Tap water will be placed in the prototype water purification system to determine the ability of the system to deal with a relatively impure water source. This will help to determine the consequences of using relatively impure water to fill the SNS Linac water cooling systems. 4) The dependence of the ph, electrical resistivity, and dissolved oxygen on the water flow rate through the purification system hardware will be studied in more detail. 5.7 Facility-related Issues Initial Start-Up and Water Quality Verification Testing Initial start-up is expected to be similar to the prototype water purification system. The water skid/purification system manufacturer will provide, a detailed start-up procedure and a troubleshooting matrix to be used for diagnosing potential system failures. The purification system will have been tested to provide the appropriate water quality as well as an examination of workmanship for safety and quality, and will be pressure tested. It is recommended that the piping of the main loops be flushed with tap water prior to connecting and operating the purification system, to remove possible debris remaining from the manufacturing process. Once complete, the entire system should be brought on-line, and after a period of approximately 24 hours filters should be replaced and water samples should be taken to verify purity. It has been suggested to use tap water to fill the system as a cost reduction measure, rather than having water trucked to the facility. Past experience on the LANCSE accelerator and the studies done with the APT testing has shown this to be an effective source for water supply [5.13]. Samples will need to be taken of the source to ensure the manufacturer can provide a system compatible with the water. This particular test has not been completed on the prototype system to compare water quality. In addition to the manufacturers procedures, the following general guidelines are recommended to the start-up of the purification system. 228

229 1) A cursory check was made for loose wiring, including power and data acquisition lines, and any evidence of leaks, or loose fittings. 2) Extra care should be taken if the system has not been circulated for a long period of time or lines have been broken due to the addition of components or soldering operations. Turn off valves to isolate the purification system from the rest of the loop. Run water through the rest of the system to flush out any large particles, fluxes, metal fragments from machining operations etc. Volume will be determined by the appearance of the water. If the system has been dormant for less than a couple of months, skip steps ) Switch position of the valves to direct water into the purification system. 4) Use air bleeds as necessary to get as much gas out of the system as possible. Reducing the air in the system reduces the amount of oxygen trapped in the piping, reducing bacterial growth. Then the system should be back-filled with nitrogen to purge the system of any remaining oxygen not in solution. 229

230 System Operation and Maintenance In general, the purification system is intended to operate with little or no operator assistance after being brought on-line. The system has been designed to require maintenance annually, with maintenance schedules in the manufacturer s maintenance manual included with each purification system. Although very little or no levels of radiation are anticipated in the water or to accumulate in the system components, it is still recommended that each bottle be surveyed for activation. A general resin disposal procedure is outlined in Appendix K, which will need to be revised to reflect ORNL s procedures. Conclusion Although prototype testing has not been completely finalized at the time of this writing, it is believed the water purification system design will meet all of the functional requirements needed for the SNS Linac water cooling systems. Preliminary prototype tests show that the water purification system maintains the desired operational levels of ph, electrical resistivity, and dissolved oxygen content. Water samples do not show any significant levels of corrosion or bacterial growth. A procurement specification for the water purification system has been drafted and will be modified as deemed necessary following completion of the prototype tests. Finally, operational and maintenance activities for the water purification system, including handling and disposal procedures of the water purification hardware, have been documented in this report. 230

231 6.0 Instrumentation and Controls 6.1 Local Controls The DTL and CCL water cooling and resonance control systems will employ a control system that can be operated by a local, programmable logic controller, interfaced through a touchscreen interface, or it can be operated through the SNS global control system network. This section discusses the features of the local control system, while the next section discusses the global control system and interfaces Introduction and Design Requirements There are two types of water cooling systems associated with the DTL and CCL structures. The first is the Resonant Control Cooling System (RCCS), which serves to keep the DTL tanks and CCL cavities in resonance by removing the RF waste heat from the copper cavity structures. The resonance control is accomplished by manipulation of the DTL drift tube/tank wall and CCL cavity dimensions (expansion/contraction) by adjusting their wall temperatures with the RCCS. The DTL and CCL RCCS not only performs resonance control of the RF structures, but, in addition, provides performance assessment and diagnostics of the water cooling system and safety interlocking. The second type of water skid, termed the Quarupole Magnet Cooling System (QMCS), serves the CCL quadrupole electro-magnets that are located between each CCL segment along the beam line. It is very similar to the RCCS except that the magnets simply require constant temperature water. One QMCS water skid will serve all quadrupole electro-magnets along the CCL. The CCL QMCS is responsible for removing electrical waste heat from the magnet coils and maintaining the magnets at an acceptable operating temperature. The control loop for the QMCS will be similar to the RCCS except that it will be responding to water temperature measurements only instead of both RF frequency error and water temperature. The proceeding section will first discuss the types of instrumentation and control hardware that will be used on the DTL and CCL RCCS systems. Next, the logic behind the control system in maintaining the resonance of the DTL and CCL structures is reviewed. Next, the safety features incorporated in the Resonant Control Cooling System 231

232 are presented and reviewed. Finally, the signal and device naming conventions used for the control system will be discussed Instrumentation and Control System Architecture The RCCS will be responsible for monitoring the performance of the water cooling system, maintaining resonance of the RF structures, diagnosing the water cooling system in case of off-normal operation, and providing proper safety interlocks in the event of system failure. The final design of this system is shown in the layout of Figure 6.1, while a more detailed control system block diagram is shown in Figure 6.2. The main architectual features are discussed below. Figure 6.1. Schematic of the DTL/CCL water cooling control system. 232

233 Figure 6.2. Block diagram of the DTL and CCL water cooling and control systems. Programmable Logic Controller The water cooling control system is being designed for local stand-alone operation and for interface with the SNS global control EPICS system. The heart of the local control system will consist of an Allen Bradley ControlLogix Programmable Logic Controller (PLC) and a rack-mounted touchscreen operator interface. The PLC will be programmed with Allen Bradley s RSLogix5000 ladder code programming toolkit. 233

234 The RCCS control system will use the Allen Bradley ControlLogix architecture. This equipment was selected from the SNS Control Standards Handbook. The controller hardware will consist of the following modules: 1756-L1M1 Processor with 512K memory 1756-CNB Ethernet communications module 1756-IF16 16 Channel Analog Input module 1756-OF8 8 Channel Analog Output module 1756-IB32 32 Channel Digital Input module 1756-OB32 32 Channel Digital Output module 1756-IR6I 6 Channel RTD (temperature) module The ControlLogix equipment will be installed in the RCCS equipment rack. For each RCCS system, 2 PLC chassis will be used. The first chassis will contain the modules for instrumentation and devices in the water skid. Since all DTL, CCL, and SCL water skids are essentially the same, this chassis configuration will be the same for all 12 (6 DTL, 5 CCL, and 1 SCL) water cooling systems. The second chassis will contain the modules for the devices in the tunnel. This configuration varies for each RCCS system. The appropriate module selection and chassis size will be implemented based on the device requirements. For example, DTL#1 will have 2 chassis with the configuration shown in Figure

235 DTl#1 Chassis DTl#1 Chassis Close up of Chassis A Figure 6.3. ControlLogix Modules for a typical water cooling control system. 235

236 To simplify wiring and reduce installation costs, pre-wired cables and DIN rail mounted Allen Bradley Interface Modules (IFMs) will be used to connect to the field wiring. The IFMs contain screw terminal, and may optionally contain fuses and diagnostic LEDs. The field wiring will connect to the screw terminals of the IFMs, and the signals will then reach the ControlLogix modules via the pre-made interface cables. The IFMs will be mounted in the back of the rack for easy access and maintenance. See Figure 6.4 for a typical IFM module installation. Figure 6.4. Typical Interface Module (IFM) Instrumentation Digital signals will be 0/24VDC, with 0 V normally representing the OFF or CLOSE state and 24V representing the ON or OPEN state. Analog signals will be 4 to 20 ma. All temperature signals will be 3 wire RTDs. Instrumentation will be provided on each water skid and in the water supply manifolds for control, diagnostic, and safety interlock purposes. In particular, the following instrumentation will be employed: Pressure transducers: Monitor local static pressure before and after the heat exchanger and pump, within the reservoir/expansion tank, and at the inlet and outlet heat exchanger manifolds. 236

237 RTDs: Monitor water temperature before and after the heat exchanger and pump, within the reservoir/expansion tank, and at the inlet and outlet heat exchanger manifolds. Flow meters: Monitor water flow rates through the heat exchanger, out of the RF structure, in the water purification and pump by-pass loops. Water purification: Monitor the water s electrical conductivity, ph, and dissolved oxygen content. Liquid low-level switch: Monitor the water level in the expansion tank. Equipment Rack The water cooling control system will be installed in a 19 equipment rack. The re will be one rack for each water cooling system. The rack follows the SNS Basic Order Agreement (BOA) # for rack procurement. The rack provides inches of vertical equipment space, and contains a Lexan front door. For more details, refer to the standard rack specification in the WBS 1.9 Integrated Control System Control Standards Handbook, Section 3.4. The prototype RCCS rack is shown in Figure

238 Figure 6.5. Prototype RCCS equipment rack. The equipment rack will contain the Allen Bradley ControlLogix equipment, power supplies, PanelView 1000E local operator interface, internal cabling, IFM modules, and a cooling fan. Total heat generation of the rack is approximately 400 watts. A single 6 fan will provide air circulation. Cabling Multiconductor cables will be used to carry signals from the RCCS rack to the water skid and to the tunnel junction box. These cables will be tray-rated. Digital signal cables will have an overall outside shield. RTDs and analog signals will have individual shields. Table 6.1 provides a listing of the types of cables used. 238

239 Table 6.1. Cabling descriptions for the DTL and CCL water cooling control systems. Trade Number Manufact. Conductors Description Usage AWG Jacket Code Rating Dia. (in) Weight (lbs/1000ft) 1065A Belden 8 pair twisted pair, overall digital signals 18 PVC NEC type TC shield tray cable 1067A Belden 16 pair twisted pair, overall digital signals 18 PVC NEC type TC shield tray cable 1050A Belden 8 pair twisted pair, analog signals 18 PVC NEC type TC individual shield tray cable 1052A Belden 16 pair twisted pair, analog signals 18 PVC NEC type TC individual shield tray cable 1094A Belden 8 triad twisted triad, individual shield RTD 18 PVC NEC type TC tray cable Belden 1 pair twisted pair, overall digital and analog 20 Tefzel n/a shield signals Belden 1 triad twisted triad, overall shield RTD 20 Tefzel n/a Belden 1 pair twisted pair, overall digital and analog 22 PVC NEC type shield signals PLTC tray cable EXPP- 3CU-24S Omega 1 triad standard RTD extension wire RTD 24 Polyvinyl n/a Local Operator Interface Based on the product evaluation described in the SNS DTL Water Cooling and Resonance Control System Prelimiary Design Report, an Allen Bradley PanelView 1000E industrial operator terminal (Figure 6.6) will be installed in the equipment rack to provide a local operator interface. This terminal uses a 10 color LCD touchscreen. It is programmed using the Allen Bradley Panel Builder software development package. For the prototype control system, the terminal will communicate with the ControlLogix system using ControlNet. An Ethernet version of the terminal should be available from the manufacturer by the middle of Based on this, the production RCCS control systems will use the Ethernet version of the terminal. This will eliminate the need for a ControlNet communications module in the ControlLogix chassis. 239

240 Figure 6.6. PanelView 1000E operator terminal. The RCCS system will normally be operated in the global control mode via the SNS EPICS control system. An IOC (Input/Output Controller) will be used as the interface between the RCCS PLC and EPICS. The EPICS developers will provide a software driver to read values to and from the PLC memory, via Ethernet or ControlNet, and pass it along to EPICS Control Methodology and Logic Resonance Control As discussed previously, one of the primary functions of the Water Cooling and Resonance Control System is to aid the LLRF control system in maintaining the resonance of the DTL and CCL RF structures. Refer to Section 1.3 for more details on the resonance control philosophy. The LLRF Control and the RCCS share the responsibility of the resonance control of the DTL and CCL. Consider the DTL as an example. From system start-up, when RF power is gradually introduced to the DTL tank, to full-on steady-state accelerator operation, there are many complicated thermal, fluidic, structural, and electrical 240