A Case Study of Control Problems and Resolutions in a High rise

Transcription

1 A Case Study of Control Problems and Resolutions in a High rise Office Building U.S. General Services Administration, Mark O. Hatfield U.S. Courthouse, Portland, Oregon Program Sponsors California Energy Commission Iowa Energy Center April 2008

2 TABLE OF CONTENTS Introduction 2 Facility and Mechanical System Description 4 Control Problems 7 Chilled Water Plant and Distribution System Control 7 Heating Water Plant Pumping Control 19 Heating Water System Control 20 Hot Deck Static Pressure Reset Control 22 AHU1 and AHU2 Static Pressure Reset Control 23 Cold Deck Supply Air Temperature Reset Control 23 Economizer Control 24 Summary 25 References 28 INTRODUCTION Since its inception in 2001, a fundamental premise of the National Building Controls Information Program (NBCIP) has been that properly functioning control systems are a significant contributor to building energy efficiency, and problems associated with building controls and operation are a primary cause of energy inefficiency. Several scoping activities were undertaken in NBCIP s first year to help understand the link between control systems and inefficient energy use in buildings. These activities included: A literature review of case studies that reported control problems leading to energy waste; 1 Roundtable discussions with building control experts aimed at identifying and categorizing the most common control problems that impact building energy use, and identifying information needs to address these problems; 2,3 and An analysis of the 1995 Commercial Buildings Energy Consumption Survey database to characterize building controls and other energy efficiency measures for the purpose of identifying their prevalence and their impact on energy consumption in the nation s building stock. 4 These scoping activities provided a good perspective on the nature of control problems in buildings; however, believing that the only way to fully understand building controls is through the experience of troubleshooting them in real buildings, NBCIP partnered with Portland Energy Conservation, Inc. (PECI) to document control problems and resolutions in a building undergoing retrocommissioning. The case study building is a 592,000 square foot, multi story office and courthouse building located in Portland, Oregon. PECI was engaged by the Bonneville Power Administration on behalf of the U.S. General Services Administration to conduct the retrocommissioning effort on this building. After the initial study to identify low cost retrocommissioning measures, PECI served as the general contractor and provided oversight during implementation of the measures, with a local controls company performing most of the control modifications. The retrocommissioning process resulted in significant improvements in building comfort and systems operations and produced estimated annual energy cost savings of $56,000 for all of the measures implemented. This represents a 10% reduction in the annual utility expenditures and includes a 10% adjustment factor that lowers the savings estimate to account for the interaction of individual measures. There were eleven control related measures identified and NBCIP

3 implemented during the retrocommissioning process, which accounted for approximately $12,364 of the $56,000 overall annual cost savings. A significant fraction of the total cost savings associated with the retrocommisisoning project was attributed to the following: Fixing building envelope issues to minimize infiltration and improve building pressurization; Optimizing courtroom VAV box operation; and Trimming pump impellers. No attempt was made to quantify the cost savings attributed to the non energy benefits of the control improvements. However, experience suggests that the reduction in maintenance costs and extended equipment life due to improved equipment operation associated with the control changes could be significant. This report highlights each of the control related problems identified by PECI in the course of retrocommissioning the case study facility. These problems have been classified by NBCIP staff using a categorization scheme originally applied to the results of the literature review and later used as the basis for the roundtable discussions. 1,2,3 These categories are described in the sidebar Classification of Control Problems. The report begins with a description of the case study facility and select mechanical systems. The control problems are then described, with particular attention given to a control problem that resulted in unstable operation of the chilled water plant and distribution system. The description of each control problem also includes a discussion of how the problem was resolved, estimates of energy savings and discussion of non energy benefits, and classification of the problem according to the categories and subcategories described in the side bar. Finally, the summary section relates the findings from this case study to the previous scoping activities of NBCIP. Classification of Control Problems Previous NBCIP studies have identified three main categories of control problems, namely, hardware, software, and human factor problems. Each of the main categories is further divided in four subcategories. The subcategories and descriptions of the types of problems that fall into the various subcategories follow: Hardware Category Input device Refers to problems associated with sensors, transducers, switches, relays, and related devices used for measuring or indicating some condition. Controller Refers to problems associated with the hardware device that receives sensor input data, executes control logic on those data, and causes an output action to be generated. Controlled device Refers to problems with the device that receives output signals from controllers and changes the state of an end device. Examples of controlled devices include valve operators, damper operators, electric relays, and variable speed drives. Communications Refers to problems associated with the hardware necessary for transmitting analog signals, digital signals, and network communications between components of the control system. Examples of communication hardware include wiring, cabling, communications interfaces, and gateways. Software Category Input/Output implementation Refers to problems arising from the configuration of input and output points that occur prior to turning over the direct digital control (DDC) system to the building operator. Programming Refers to problems arising from incorrect or inappropriate control logic and parameters that produce output to control heating, ventilating, and air conditioning (HVAC) equipment. Operation system Refers to problems associated with the operation of the DDC system software and its interface to the computer operating system. Data management Refers to problems associated with producing information from data including data monitoring, display, alarming, and logging. Human Factor Category Operator error Refers to problems associated with changes to the control system made by the operator during routine operation and maintenance, as well as failure to perform routine operation and maintenance, which unintentionally result in improper operation of a system. Operator unawareness Refers to problems arising from an operator s lack of understanding or familiarity with the control system due to inadequate training. Operator interference Refers to problems associated with intentional changes to the control system made by the operator causing interference with the normal operation of the system. Operator indifference Refers to any number of control problems stemming from an operator s apathy toward operation and maintenance. NBCIP

4 Facility and Mechanical System Description The case study building is a 592,000 square foot, multi story office and courthouse building located in Portland, Oregon. Construction was completed in 1997, and the building went through an abbreviated construction phase commissioning process before the building was occupied. Operations and maintenance (O&M) for the building is contracted to an outside firm. Heating and cooling for the building are provided by dual duct, variable air volume (VAV) air handling systems, with central chilled and heating water plants. The air handling systems consist of four cold deck air handling units (AHU1 and AHU2 serve the basement to 8th floors through a common supply duct, and AHU3 and AHU4 serve the 9th through 16th floors through a common supply duct) and two hot deck air handling units (AHU5 and AHU6 serve the entire building through a common supply duct). Conditioned air is delivered to the zone level via more than 500 terminal units. Approximately 50% of the terminal units are cooling only VAV boxes, 38% are standard dual duct VAV boxes, and 12% are series fan powered dual duct VAV boxes. The supply air temperature setpoint for AHU1/AHU2 and AHU3/AHU4 is reset between 55 F and 70 F based on the deviation from zone temperature setpoint for selected terminal units serving the respective areas of the building. The condenser water system serving the chilled water plant consists of three condenser water pumps for CH1 and CH2, and two pumps for CH3. Similar to the primary chilled water pumps, CH1 and CH2 have dedicated condenser water pumps and share a backup pump. The two condenser water pumps serving CH3 are arranged in a lead/backup configuration. The condenser water system also serves stand alone computer room units via a heat exchanger. There are three cooling towers serving the condenser water loop, each with two speed fans that are staged to meet condenser water temperature setpoint. The chilled water plant is comprised of three chillers, two 450 ton centrifugal units (CH1 and CH2) and one 115 ton reciprocating unit (CH3). Each chiller has a dedicated, constant volume primary pump. CH1 and CH2 share a single backup pump and CH3 has a dedicated backup pump. The primary pumps serving CH1 and CH2 are rated at 675 gpm each, and the primary pump serving CH3 is rated at 180 gpm. Chilled water is delivered to the building through a de coupled secondary loop, with three distribution pumps each rated at 765 gpm and controlled by variable frequency drives to meet system demand by maintaining a differential pressure setpoint across the suction and discharge headers serving the secondary distribution pumps. The pump differential pressure setpoint is reset based on maintaining a constant differential pressure across the cooling coil with the largest pressure drop (most remote coil). A schematic diagram of the chilled water plant and distribution system is shown in Figure 1. The original chilled water plant and distribution system sequence of operation is provided in the side bar Sequence of Operation Chilled Water System Chiller Control (page 6). NBCIP

5 CH1 450 ton CH2 450 ton CH3 115 ton Manual Isolation Valves Throttle Valve Throttle Valve Throttle Valve Throttle Valve Throttle Valve CHWP 1 CHWP 2 CHWP 3 CHWP 4 CHWP 5 Bypass Water Temperature Remote Coil DP Sensor Typical AHU Cooling Coil Two way Control Valve (typical of 2 valves in parallel) Bypass Line Primary CHWS Temperature Primary CHWS Flow Meter Figure 1. Chilled water system diagram. VFD Control DP Sensor VFD CHWP 6 Throttle Valve VFD CHWP 7 Throttle Valve VFD CHWP 8 Throttle Valve Secondary CHWS Temperature Secondary CHWS Flow Meter NBCIP

6 Sequence of Operation Chilled Water System Chiller Control 1 Chiller 1,2,3 Control: Each chiller has the following objects: 1. Remote chill water setpoint control. 2. Remote Demand limiting control. 3. Remote Start/Stop control. 4. Remote General Alarm. 5. Remote Indication of KW usage or % of full load amps The building control system provides lead lag selection that allows and of the chillers to act as the lead machine. CHWS Reset: The CHWS temperature is reset based on cooling coil valve position greater than 90% open at AHU 1,2,3,4. Three 90% = 42F, Two 90% = 44F One 90% = 46F and no valves at 90% = 48F. The reset is overridden when the return air RH rises above 50%. Demand Limiting: delays between stages are adjustable to limit the demand. Graphics: System graphics included. Chiller Start up: The chiller system is enable upon the following conditions: 1. Chiller system is schedule on. 2. Outside air temperature is greater than 56F (adjustable) 3. There is a call for cooling as determined by any of the cooling valve 15% open 4. The chiller sump temperature is great then 40F. 5. Emergency Stop switch is Normal (off) Chiller Pumps (CHWP 1,2,3,4,5): Chiller start sequence: On a system start up the secondary loop is enabled. Upon status that a secondary pump is running a chill water pump is started. On chill water pump status the condenser water pump is started. On pump status of both pumps, the chiller is enable to start. Chilled water pumps operate continuously when its associated chiller is selected as the lead machine. The lag pumps start in sequence whenever the secondary CHW flow exceeds the primay CHW flow as determined flow meters and the lead chiller is 95% loaded. A run time delay is used so that reverse flow must be maintain to a minimum of 20 minutes before the lag pumps are started. The last lag chiller and pump stops whenever the primary CHW flow exceeds the secondary CHW flow by the capacity of the last scheduled chiller. A time delay on is used so that excess flow must be maintain for a minimum of 20 minutes before stopping chiller and lag pump. Calculated load: The building control system calculates the system load based on GPM and temperature. Pump Failure: An Alarm is sent to the building control system on pump failure detected via current switch. Secondary Chilled Water Pump (CHWP 6,7,8): 1. The secondary pump system to operate when the outside air temperature is above 56F (adjustable) and any cooling supply fan unit operating and associated chilled water valve is at least 15% open. 2. As the differential pressure across the pumps falls, the lead pump increases the speed to maintain DP setpoint. If the pump GPM exceeds the pump design condition the lag pump is started and the two pumps will modulate their speed in unison to maintain DP setpoint. If the GPM exceeds the GPM of both pumps and the DP falls 2 psi the second lag pump is started and all pumps modulate the their speed in unison to maintain DP setpoint. 3. As the DP increases, the pumps will reduce their speed until GPM equals one or more pumps at which time the lag pump will stop and remaining pumps modulate to maintain DP setpoint. If the DP continues to increase, the pumps reduces speed until the GPM equal one pump at which time the lag pump will stop and the remaining pump modulates to maintain DP setpoint. 4. There are adjustable time delays between lead lag pump starts and stops. Operator can select lead lag pump sequence. 5. Discharge Differential setpoint at the secondary pumps is reset from remote DP sensors in the field. 6. Building control system monitors the RPM and GPM of each secondary pump and allow any pump to operate at or beyond its published End of Curve 7. Variable Speed Drives: Alarm is sent on drive failure. Units report RPM to the building control system 8. Secondary Pumps run under normal control and all cooling valves to open with operation of respective AHU when smoke exhaust system EF 10 is in operation and outside air temperature is below 40F. 1 The sequence of operation was reproduced as written in the building control system documentation, except the name of the building control system manufacturer has been removed. NBCIP 6

7 CONTROL PROBLEMS The retrocommissioning process identified numerous control problems and opportunities for energy savings in the case study building. This report highlights some of these problems, with particular attention focused on a control problem that resulted in unstable operation of the chilled water plant and distribution system. The detailed discussion of the chilled water plant and distribution system control problem is followed by summaries of six additional control problems that were identified during the retrocommissioning process. A complete list of the control problems described in this report follows: Chilled water plant and distribution system control problem; Heating water plant pumping control problem; Heating water system control problem; Hot deck static pressure reset control problem; AHU1 and AHU2 static pressure reset control problem; Cold deck supply air temperature reset control problem; and Economizer control problem. Chilled Water Plant and Distribution System Control Original Operation The three chillers CH1, CH2, and CH3 were originally programmed to operate in a lead/1 st lag/2 nd lag configuration, with any of the three chillers configured as lead chiller and either of the two remaining units designated as the 1 st lag unit, with the remaining unit becoming the 2 nd lag unit by default. The configuration of the lead/1 st lag/2 nd lag units was changed on a regular basis to ensure equal run time of the equipment. This is a common strategy for controlling multiple chillers and is effective if all chillers have equal capacity; however, it becomes problematic with chillers of unequal capacity ratings because the combination of units selected for operation at any given time may be unable to meet the load effectively. This is especially true if turn down limitations of the lead unit make it unable to match a low load operating condition. This is the problem encountered in the case study facility, and it was the primary factor leading to unstable operation of the chiller plant and distribution system that propagated out to the cold deck air handling units. AHUs try to satisfy their respective supply air temperature setpoints, causing the cooling coil valves to be commanded 100% open. In addition, there is a large pull down load on the chiller plant to remove the residual heat from the water in the piping network. Figures 2 5 illustrate a representative day (September 30, 2003) when the 450 ton chillers (CH1 and CH2) are selected as the lead and 1 st lag units. On this day, CH1 is the lead chiller and CH2 is the 1 st lag unit. Figure 2 shows the chiller operation, the chilled water supply, return and setpoint temperatures, and the maximum chilled water valve control signal of the four cold deck AHUs (labeled chilled water valve control signal high select). Figure 3 shows the primary and secondary chilled water flow rates and the operation of the secondary loop chilled water pumps. Figure 4a shows the operation of the chilled water cooling coil valve and outdoor air damper for AHU2. The associated supply air temperature (measured at the discharge of the AHU) and setpoint, and the outdoor air temperature are shown in Figure 4b. Corresponding graphs for AHU4 are shown in Figures 5a and 5b. The description of the operation on this day follows: System start up occurs at 5:00 am. At start up, the AHU chilled water valves go to the full open position and secondary loop pump CHWP7 comes on and ramps up to 100% speed. When CH1 is commanded ON at a little after 5:00 am, it ramps up quickly to 100% capacity to satisfy the load. Because the cooling coil valves are 100% open, secondary loop pumps CHWP8 and CHWP6 are brought online in succession in an effort to maintain the differential pressure across the supply and return headers at setpoint. With the cooling coil valves 100% open and the secondary pumps all operating at 100% speed, the water flow through the secondary loop exceeds the chilled water flow provided by the primary pump associated with CH1. The combination of excess secondary flow and the lead chiller operating above 95% full load amps causes the staging timer to start counting. Once these conditions have persisted for 20 minutes (the typical staging timer delay), the 1 st lag chiller (CH2) is commanded ON to help meet the load. However, on a typical morning, the chilled water capacity produced by operating both CH1 and CH2 (even at maximum turn down, which is about 30% for these chillers) far exceeds that required to satisfy the load. During unoccupied hours when the chilled water plant is not operating, the air temperature in the AHUs, as well as the water in the piping system, reaches equilibrium at ambient temperature. When the system is started in the morning, the NBCIP 7

8 Percent Full Load Amps (%) Chiller CH1 Chiller CH2 Chiller CH3 Chilled Water Flow Rate (gpm) Primary Flow Secondary Flow 0 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20: :00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (a) Chiller operation. (a) Primary and secondary flow rates. Temperature ( F) and Valve Control Signal (% Open) 100 Chilled Water Valve Control Signal (high select) Chilled Water Supply Temperature Chilled Water Return Temperature Chilled Water Supply Temperature Setpoint 0 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 Commanded Pump Speed (%) CHWP6 CHWP % CHWP % Note: For better graphical presentation, 50 the commanded speeds for CHWP7 and CHWP8 have been adjusted by 100% and 200%, respectively. 0 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (b) Chilled water temperature control and AHU chilled water valve control. (b) Secondary distribution pump operation. Figure 2. Original operation of chilled water plant on September 30, Figure 3. Original operation of chilled water distribution system on September 30, The operation of the second chiller causes the secondary loop chilled water temperature to decrease from 55ºF to 44ºF in 20 minutes. As a result of this rapid temperature decrease and the fact that the cooling coil valves are 100% open, the supply air temperatures of the AHUs drop below their respective setpoint values, causing the cooling coil valves to close rapidly. As the cooling coil valves close rapidly, the differential pressure across the supply and return headers in the secondary distribution system increases rapidly. This causes a rapid decrease in speed of the secondary distribution pumps and flow rate of chilled water through the secondary loop. The rapid decrease in pump speed can be seen in Figure 3b occurring at about 6:15 am. Excess primary chilled water now flows through the bypass line and returns directly back to the chillers, lowering the return water temperature entering each chiller. The cold return water temperature causes both chillers to turn OFF with the load apparently satisfied. Although CH1 and CH2 are now OFF, the primary chilled water pump associated with each remains ON for 20 minutes. Thus, the primary pumps for CH1 and CH2 circulate return water through deactivated chillers (with the exiting chilled water temperature quickly increasing to approximately 60 F) and directly into the secondary loop, where it is distributed to the cooling coils. Since the AHU supply air temperature setpoints can not be met with 60 F water, the cooling coil valves are commanded to the full open position again, causing the secondary pumps to ramp back up. When CH1 is commanded ON again, the chilled water produced by CH1 blends with the return water that continues to circulate through CH2 due to the excessive time delay on the primary pump operation (CH2 is still NBCIP 8

9 Control Signals (% Open) Outdoor Air Damper Cooling Coil Valve Control Signals (% Open) Cooling Coil Valve Outdoor Air Damper 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (a) Cooling coil valve and outdoor air damper control signals. 80 (a) Cooling coil valve and outdoor air damper control signals :00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 Temperature ( F) Supply Air Temperature Supply Air Temperature Setpoint Outdoor Air Temperature Temperature ( F) Outdoor Air Temperature Supply Air Supply Air Temperature Temperature Setpoint 40 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (b) Supply and outdoor air temperatures. Figure 4. Original operation of AHU2 on September 30, :00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (b) Supply and outdoor air temperatures. Figure 5. Original operation of AHU4 on September 30, commanded OFF). The blending of the chilled and return water is significant enough that CH1 is unable to satisfy the load. CH1 ramps back up to 100% full load amps, the stage up timer expires, and CH2 is commanded ON again. With both CH1 and CH2 operating, the chilled water capacity again far exceeds the load, resulting in the same sequence of events described previously. The cooling coil valves close, the secondary loop pumps slow down, and excess primary chilled water is returned directly to the chillers through the bypass line. This causes the return water temperature to decrease rapidly and the chillers to shut down with the load apparently satisfied. This operating scenario continues until the actual cooling load at the AHUs is large enough that the cooling coil valves begin to gradually modulate closed. When this happens, the 1 st lag chiller shuts down and stays OFF with the load being met by the lead chiller. System operation becomes unstable again late in the day as the cooling load continues to increase (indicated by the steady increase in the percent full load amps of CH1). Figure 3a shows a rapid increase in the primary chilled water flow rate at approximately 3:30 pm. This is caused by the primary pump for CH2 being commanded ON despite the fact that the chiller plant sequence of operations indicates the lag pump should be commanded ON only if the lead chiller exceeds 95% of full load amps and the secondary chilled water flow rate exceeds the primary chilled water flow rate for 20 minutes. It is evident from the flow rates in Figure 3a that this latter condition is not met prior to the start up of the lag pump. The data are not adequate to determine how this event triggers the onset of the unstable behavior that ensues, but it is evident that the two are linked. This was not an isolated incident. The data collected during retrocommissioning included at least one other occurrence of this type of behavior. NBCIP

10 Early in the retrocommissioning process, the cause of the instability was not evident. The fact that the control system frequently changed the role of lead/1 st lag/2 nd lag between the three chillers made it difficult to identify patterns in the operational behavior. Furthermore, the interaction of the control loops is such that an instability in one loop can propagate through many other loops. Initially it was thought that the source of the instability might be oversized or poorly tuned cooling coil valves. Instability in the cooling coil valves would lead to instability in the differential pressure being controlled by the secondary loop pumps and this could cascade throughout the chilled water distribution system and chiller plant. However, it is just as plausible that an unstable chilled water supply temperature could result in unstable control at the cooling coil valves and that this could cascade throughout the distribution system and chiller plant. One of the first actions taken to try to alleviate the unstable control was to adjust the control loops for the cooling coil valves to slow down their response; however, the instability persisted. The original sequence of operations stated that the variable frequency drives were to be controlled by maintaining a differential pressure setpoint at the most remote coil, but the actual setpoint value was not defined. During retrocommissioning it was observed that the system was trying to maintain a differential pressure of approximately 30 psig (69 ft.wc.) across the furthest coil. This value is high compared to actual system configuration requirements and contributed to excessive pump operation. Improved Control The primary reason for the chiller plant instability arose from a mismatch of the chiller operation and the existing load (i.e., less than optimal chiller selection and staging). The original sequences rotated chiller selection for lead/ 1 st lag/2 nd lag operation on a time basis without regard for the actual load within the building. During the retrocommissioning process, the chiller staging was revised to more closely match the operation of the three chillers (a 115 ton unit, CH3, and two 450 ton units, CH1 and CH2) to the cooling load. The chiller staging now consists of five stages, as shown in Table 1. The building control system provides lead/lag selection that allows the small chiller to act as the lead machine, with either of the two larger units staged as necessary as the 1 st lag/2 nd lag machines. Selection of CH1 and CH2 for 1 st lag/2 nd lag operation is rotated automatically by the building control system. The revised sequence of operations for the chillers and the primary and secondary chilled water pumps are outlined in the sections that follow and summarized in Figure 6 (page 11). Table 1. Revised chiller staging. Stage Chiller Operation Capacity Range 1 CH3 0 to 115 tons 2 CH1 or CH2 135 to 450 tons 1 3 CH3 + CH1 or CH2 450 to 565 tons 4 CH1 + CH2 565 to 900 tons 5 CH1 + CH2 + CH3 900 to 1015 tons 1 The 135 ton value is a 30% turn down for a 450 ton chiller. Note that there is still a small amount of excess capacity when CH1 or CH2 operates with a load between 115 and 135 tons. Chilled Water System Enable The chiller system is enabled when all of the following conditions are satisfied: Chiller system is scheduled ON; Outdoor air temperature is greater than 56 F; Any of the cooling coil valves is at least 15% open; Chiller oil sump temperature is greater than 40 F; and Emergency stop switch is set to normal (OFF). Chilled Water System Disable The chiller system is disabled when any of the following conditions are satisfied: Outdoor air temperature is less than 56 F; All of the cooling coil valves are less than 15% open; or System is scheduled OFF. Chilled water supply temperature reset The chilled water supply (CHWS) temperature is reset between 42 F and 48 F based on the cooling coil valve positions of AHU1, AHU2, AHU3, and AHU4. The reset schedule is as follows: IF three or four valves are greater than 90% open: CHWS temperature setpoint = 42 F IF two valves are greater than 90% open: CHWS temperature setpoint = 44 F IF one valve is greater than 90% open: CHWS temperature setpoint = 46 F IF none of the valves is greater than 90% open: CHWS temperature setpoint = 48 F Chiller Staging Chiller operation is separated into five stages to maximize system performance and match building loads more closely. Individual chillers are not allowed to start until a positive status signal is received to verify operation of their respective primary chilled water and condenser water pumps. The staging is performed as follows: NBCIP 10

11 CHILLED WATER SYSTEM DISABLED Outdoor air temperature < 56 F OR All cooling coil valves < 15% open OR System is scheduled OFF Chiller system scheduled ON AND Outdoor air temperature > 56 F AND Any cooling coil valve > 15% open AND Chiller oil sump temperature > 40 F AND Emergency stop switch is Normal (OFF) CHILLED WATER SYSTEM ENABLED STAGE 1 CH3 enabled Secondary chilled water flow rate < CH3 chilled water flow rate AND stage down timer has expired CH3 %FLA > threshold OR Secondary chilled water temperature > CH3 chilled water temperature setpoint AND stage up timer has expired STAGE 2 CH3 disabled AND CH1 or CH2 enabled Secondary chilled water flow rate < 608 GPM AND stage down timer has expired CH1 or CH2 %FLA > threshold OR Secondary chilled water flow rate > primary chilled water flow rate AND stage up timer has expired STAGE 3 CH3 enabled AND CH1 or CH2 enabled Secondary chilled water flow rate < 770 GPM AND stage down timer has expired CH3 %FLA > threshold AND CH1 or CH2 %FLA > threshold OR Secondary chilled water flow rate > primary chilled water flow rate AND stage up timer has expired STAGE 4 CH3 disabled AND CH1 and CH2 enabled Secondary chilled water flow rate < 1215 GPM AND stage down timer has expired CH1 %FLA > threshold AND CH2 %FLA > threshold OR Secondary chilled water flow rate > primary chilled water flow rate AND stage up timer has expired STAGE 5 CH3 enabled AND CH1 and CH2 enabled Figure 6. Logic for enabling and disabling chiller stages. NBCIP 11

12 Stage 1: In Stage 1, CH3 is enabled. Stage 1 is enabled when the chilled water system is enabled. Stage 1 is disabled when the chilled water system is disabled. Stage 2: In Stage 2, CH3 is disabled and either CH1 or CH2 is enabled depending on which unit is selected by the building control system as the 1 st lag unit. Stage 2 is enabled when the measured percent full load amps (%FLA) of CH3 exceeds its threshold (~100%), OR when the secondary loop supply water temperature is greater than the chilled water temperature setpoint of CH3 AND the stage up timer has expired (~20 minutes). Stage 2 is disabled when the secondary chilled water flow rate is less than the chilled water flow rate of CH3 AND the stage down timer has expired (~20 minutes). Stage 3: In Stage 3, CH3 is enabled and either CH1 or CH2 is enabled depending on which unit is selected by the building control system as the 1 st lag unit. Stage 3 is enabled when the measured %FLA for CH1 or CH2 exceeds its threshold (~90%), OR when the secondary chilled water flow rate is greater than the primary chilled water flow rate AND the stage up timer has expired (~20 minutes). Stage 3 is disabled when the secondary chilled water flow rate is less than 608 gpm (approximately 90% of water flow rate from the primary pump for CH1 or CH2) AND the stage down timer has expired (~20 minutes). Stage 4: In Stage 4, CH3 is disabled and both CH1 and CH2 are enabled. Stage 4 is enabled when the measured %FLA for CH3 and CH1 or CH2 (whichever unit is selected as the 1 st lag unit) exceed their respective thresholds, OR when the secondary chilled water flow rate is greater than the primary chilled water flow rate AND the stage up timer has expired (~20 minutes). Stage 4 is disabled when the secondary chilled water flow rate is less than 770 gpm (approximately 90% of combined water flow rate from the primary pumps of CH3 and either CH1 or CH2) AND the stage down timer has expired (~20 minutes). Stage 5: In Stage 5, CH3, CH1, and CH2 are all enabled. Stage 5 is enabled when the measured %FLA for CH1 and CH2 exceed their respective thresholds (~90%), OR when the secondary chilled water flow rate is greater than the primary chilled water flow rate AND the stage up timer has expired (~20 minutes). Stage 5 is disabled when the secondary chilled water flow rate is less than 1215 gpm (approximately 90% of the combined water flow rate from the primary pumps for CH1 and CH2) AND the stage down timer has expired (~20 minutes). Primary Chilled Water Pump Operation The primary chilled water pumps are constant volume. Operation and staging of the primary chilled water pumps is executed as described below: On a Chilled Water System Enable command calling for Stage 1 mechanical cooling and a positive status signal to verify operation of the secondary pump(s), primary chilled water pump CHWP4 or CHWP5 is commanded ON. CHWP4 and CHWP5 are dedicated to CH3 (Stage 1 chiller) and are configured for lead/stand by operation. Selection of CHWP4 and CHWP5 for lead/ stand by operation is rotated automatically by the building control system. The stand by pump is automatically commanded ON if a positive status signal is not received to verify operation of the lead pump. A positive pump status signal is necessary before the chiller will be allowed to start. Upon calls for Stage 2, 3, 4, and 5 mechanical cooling, the primary chilled water pump associated with the enabled chiller(s) is commanded ON as required per the chiller staging sequences described previously. CHWP1 and CHWP3 are dedicated to CH1 and CH2, respectively, and both pumps are configured for lead/ stand by operation with CHWP2. If the lead pump fails to start, an alarm is generated at the operator work station. In this case, the isolation valve(s) in the distribution piping associated with CHWP2 must be opened manually and the stand by pump must be commanded ON by the facility operator. A positive pump status signal is necessary before the associated chiller will be allowed to start. A pump failure alarm is sent to the operator work station if a pump is commanded ON but a positive status signal is not received to verify pump operation. Pump failure is detected via a current switch mounted on the electrical power leads to the motor. Secondary Chilled Water Pump Operation The three secondary chilled water loop pumps are variable flow and are controlled via variable frequency drives to match system load requirements. Operation and staging of the secondary chilled water pumps is executed as described below: On a Chilled Water System Enable command calling for Stage 1 mechanical cooling, secondary chilled water pump CHWP6, CHWP7, or CHWP8 is commanded ON depending on which pump is selected as the lead pump. Any of the three pumps can be configured as the lead pump, and either of the two remaining pumps can be designated as the 1 st lag pump, with the remaining pump becoming the 2 nd lag pump by default. The configuration of the lead/1 st lag/2 nd lag pump is rotated automatically by the building control system. The 1 st lag pump is commanded ON if a positive status signal is not received to verify operation of the lead pump. NBCIP 12

13 A positive secondary pump status signal is necessary before the primary chilled water pump(s) are commanded ON. Flow control of all secondary pumps is based on maintaining the differential pressure across the supply/ return header serving all of the pumps at the setpoint value. As the load within the building increases and cooling coil valves begin to open, the differential pressure in the secondary distribution loop decreases. In response, the variable frequency drive (VFD) increases the speed of the lead pump to maintain the differential pressure setpoint. If the secondary flow exceeds the near design flow rate for the lead pump (~700 gpm), the 1 st lag pump is started and the speed of both pumps are modulated in unison to maintain the differential pressure setpoint. If the loop differential pressure continues to drop and secondary flow continues to increase until chilled water flow exceeds the near design flow of both the lead and 1 st lag pumps (~1400 gpm), the 2 nd lag pump starts and the speed of all pumps are modulated in unison to maintain the differential pressure setpoint. As the load within the building decreases and cooling coil valves begin to close, the differential pressure in the secondary distribution loop increases. In response, the VFDs decrease the speed of the pumps to maintain the differential pressure setpoint until the secondary flow equals the near design flow of two pumps (~1400 gpm). At this point the 2 nd lag pump will stop. The speed of the remaining two pumps will continue to be modulated in unison to maintain the differential pressure setpoint. If the load continues to decrease, the chilled water flow rate will continue to decrease as the VFDs decrease the speed of the pumps in order to meet the differential pressure setpoint. If the secondary flow rate decreases to the point where it equals the near design flow of one pump (~700 gpm), the 1 st lag pump will stop. The speed of the lead pump will then be modulated as necessary to maintain the differential pressure at the setpoint value. The lead pump is shut off when the Chilled Water Enable command is OFF. As described in the first bullet, the building control system automatically rotates each pump for lead, 1 st lag, and 2 nd lag operation. The operator can also manually select the lead/1 st lag/2 nd lag pump sequence. The adjustable time delay between pump start/stop is approximately 20 minutes. The secondary loop differential pressure setpoint is reset based on remote differential pressure sensors. A pump failure alarm is sent to the operator work station if a pump is commanded ON but a positive status signal is not received to verify pump operation. Pump failure is detected via a current switch mounted on the electrical power leads to the motor. Variable frequency drives: An alarm is sent to the operator work station when a VFD fails. Drives report speed in revolutions per minute to the building control system. During a fire/smoke event, all cooling coil valves will be commanded 100% open if the outdoor air temperature is below 40 F. The secondary pumps will run under normal control to circulate water through the coils to prevent freezing. Figure 7 illustrates system operation using the revised sequences on a representative day (May 16, 2006). The chiller staging is shown in Figure 7a and the chilled water supply and return temperatures are shown in Figure 7b. From Figure 7a it is apparent that the chiller control is far superior to that before the modifications. The large chillers, CH1 and CH2, do not experience any of the short cycling behavior that was common prior to the modifications. The chilled water temperature control has also improved, although considerable variation in the supply and return water temperature is still observed in Figure 7b. This stems from the disparity in the equipment capacity at different stages. For instance, for Stage 4 operation, CH1 and CH2 are enabled and the nominal flow rate in the primary loop is 1350 gpm. In Stage 3, CH3 and CH1 or CH2 are enabled and the nominal flow rate in the primary loop is 855 gpm. In the afternoon of the day considered, while the plant is operating in Stage 4, the secondary water flow rate is approximately 350 gpm (i.e., 1000 gpm less than the primary flow rate) and 1000 gpm of chilled water at approximately 45ºF passes through the bypass line. The bypass water mixes with return water from the loads, which reaches nearly 71ºF during Stage 4 operation, producing return water to the chillers at approximately 52ºF. When the transition to Stage 3 occurs, one of the large chillers is taken off line and CH3 is brought back on line. During the transition period when CH3 is being ramped up to meet chilled water temperature setpoint, warm return water circulating through CH3 is blended with the chilled water still being produced by the large chiller, causing the temperature of the water supplied to the loads to suddenly increase to 56ºF. This in turn causes the valves at each AHU to begin opening again, resulting in an increase in secondary flow that continues until operation eventually transitions back to Stage 4. Thus, physical limitations stemming from the equipment sizes place practical constraints on just how well the chilled water plant and distribution system can be controlled. NBCIP 13

14 Stage ON Stage ON CH3 CH2 OFF OFF ON CH1 OFF :00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 Control Signal (% Open) Outdoor Air Damper Cooling Coil Valve 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (a) Chiller staging. (a) Cooling coil valve and outdoor air damper control signals Chilled Water Return Temperature 70 Outdoor Air Temperature Temperature ( F) Temperature ( F) 60 Supply Air Temperature 40 Chilled Water Chilled Water Supply Supply Temperature Setpoint Temperature 30 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 50 Supply Air Temperature Setpoint 40 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 (b) Chilled water temperatures. (b) Supply and outdoor air temperatures. Figure 7. Improved operation of chilled water plant on May 16, Figure 8. Improved supply air temperature control for AHU1 on May 16, Impact on AHU Supply Air Temperature Control As seen previously, the instability in the chiller plant caused the chilled water supply temperature to fluctuate significantly. Figure 2b shows the chilled water supply temperature cycling between 40ºF and 64ºF. This, in turn, caused the chilled water valve for each air handling unit to hunt in an attempt to maintain the supply air temperature at the setpoint value. The revised chiller staging and the resultant improvement in the chiller plant stability affected the air handling unit control of the supply air temperature. The operation of the cooling coil valves is now more stable as a result of a more stable supply chilled water temperature. Figure 8 illustrates the improved supply air temperature control for AHU1. The impact of the changes to the chiller plant control can be seen by evaluating parameters associated with the processes being controlled. For instance, actuator travel (how much a valve or damper moves while attempting to maintain control), process error (difference between the variable being controlled and its setpoint), and the actuator reversals (number of times the direction of movement of a valve or damper changes) can all be employed to quantify control performance. Better control is characterized by less actuator travel, smaller process errors, and fewer actuator reversals. In this study, the average absolute value of the supply air temperature error and cooling coil valve reversals for the four cold deck AHUs (AHU1, AHU2, AHU3, and AHU4) were used to quantify control performance. NBCIP 14

15 The average absolute value of the supply air temperature error, e, is calculated as follows: where T measured,i and T setpoint,i are the i th sampled values of the measured and setpoint supply air temperature, respectively, and n is the total number of samples considered. For the purpose of this study, an acceptable temperature error of 1 F is defined since the HVAC system is serving an office space. The acceptable error may be much smaller, however, for critical applications like operating rooms, labs, or processing facilities. To account for the 1 F acceptable error, the error was taken to be zero whenever the measured supply air temperature was within ±1 F of the setpoint. Furthermore, temperature errors in excess of ±1 F were reduced in magnitude by 1 F, so that a 2 F error (T measured,i = 57 F, T setpoint,i = 55 F) and a 2 F error (T measured,i = 53 F, T setpoint,i = 55 F) both become a +1 F error when the absolute value of the respective errors is taken. There are different ways to calculate the number of valve reversals that occur over a period of time. For the purpose of this study, a valve reversal occurred whenever the middle value of three sampled data points was either greater than or less than the data points that immediately preceded and followed it. Additional explanation of the calculation of the control performance parameters is provided in the sidebar Control Performance Calculations. Results from the comparison of the average absolute value of the supply air temperature error e and the number of cooling coil valve reversals before and after retrocommissioning are shown in Tables 2 and 3, respectively. In each case, calculations are based on data collected at two minute intervals. Two weeks of data were used for the temperature error calculations before retrocommissioning, although data monitoring and archiving limitations associated with the building control system resulted in the data being collected over two separate two week periods (one period for AHU1 and AHU3, and a second period for AHU2 and AHU4). Three weeks of data were used for the temperature error calculations after retrocommissioning. In this case, since the monitoring was targeted to collect the data needed for these calculations and those in Table 3, it was possible to collect all the necessary data over the same period. The valve reversal calculations are based on one week of data before and one week of data after retrocommissioning. Once again, two separate one week periods of data were needed for the calculations before retrocommissioning due to monitoring and archiving limitations. Control Performance Calculations To understand how the average absolute value of the supply air temperature error is calculated and how it can be used to quantify control performance, consider the example in Figure 9, which uses artificial data for illustrative purposes. The cycling supply air temperature (Tmeasured) and constant setpoint (Tsetpoint) values are shown in Figure 9a. The error between these values e = (Tmeasured Tsetpoint) and absolute value of the error e = Tmeasured Tsetpoint is shown in Figure 9b. The curve corresponding to the error e (black dashed curve) is hidden at times by the curve for e (red solid curve). The cumulative sum of e and e are shown in Figure 9c and are computed from the following equations: where Tmeasured,i and Tsetpoint,i are the measured and setpoint values of the supply air temperature at a particular time designated by subscript i, and the cumulative errors Err k and Err k are computed by summing the k previous errors. Figure 9c shows that Err k is always increasing (except at t = 5 minutes, at which time the supply error temperature is equal to the setpoint), whereas the cycling supply air temperature results in the cumulative error Err k that alternately increases and decreases (except again at t = 5 minutes), but on average remains near zero. If the average error is calculated from the two cumulative errors after the 30 minute period illustrated in Figure 9, the result using Err k is e = 0.05ºF and the result using Err k is e = 0.9ºF. 2 Thus, e gives the false impression that the control is very good, whereas e clearly identifies that the supply air temperature is not being maintained at the setpoint value. Note, however, that from e alone it is impossible to determine whether the error results from a constant offset between the supply air temperature and its setpoint, from cycling as shown in this example, or from some other behavior. (CONTINUED ON PAGE 16) 2 This is based on sampling at 10 second intervals, resulting in 181 samples for a 30 minute period in the example. NBCIP 15