Pacific Energy Center's Retrocommissioning Workshops: Developing Expertise through Interactive Training

Transcription

1 National Conference on Building Commissioning: April 19-21, 2006 Pacific Energy Center's Retrocommissioning Workshops: Developing Expertise through Interactive Training Overview of Workshop Series Ryan Stroupe Pacific Energy Center Synopsis In May 2005 the Pacific Energy Center (PEC), a publicly-funded facility in San Francisco dedicated to promoting energy-efficient buildings, started an exciting and unusual approach to teaching retrocommissioning (RCx). The PEC began a workshop series to train building professionals interested in developing their commissioning skills, where attendees would get exposure to the planning, decision-making and diagnostic aspects of commissioning by retrocommissioning a real facility. While there are a growing number of sources for training on commissioning, most are introductory, focus on a process or are dedicated to a single aspect of commissioning. The format of this workshop series would allow students to experience the whole process and immediately apply what they ve learned through structured, hands-on activities under the supervision of qualified commissioning experts. The PEC is uniquely positioned to provide this type of training. Over the last five years we have offered a wide array of commissioning-related classes from introductory-level to advanced levels on a specific aspect of the commissioning process like test procedures or design intent. Through these classes we ve cultivated a group of building professionals with a fundamental knowledge of the subject, but eager for a more advanced and practical learning experience. Another advantage to holding the trainings at the PEC is that the facility was retrocommissioning in 2002 by Sherrill Engineering. This effort created documentation that would benefit the workshop series including a Systems Manual and problems log. Though Sherrill Engineering s work produced substantial energy savings (see Figure 1), the improvements included a minimal number of physical changes. Most of the work focused on optimizing the mechanical system controls. With documentation in hand from the RCx effort, the workshop series could follow through on some of the recommended measures. The genesis of the class occurred during a conversation between me and David Sellers of Pacific Energy Conservation, Inc. We were reviewing the electric billing data for the Energy Center (Figure 1) when we recognized that the energy usage was drifting upwards in We conceived of the workshop series as a means of providing advanced commissioning training for those attendees at our classes requesting it and as a means of returning the facility to the optimized operation that was established after the last retrocommissioning effort. In addition to having the commissioning documentation on hand there are other advantages for holding this type of training at the PEC. The facility has been the location for dozens of hands-on classes over the years. The staff and facility management are receptive to this type of activity and the mechanical equipment is readily accessible for demonstrations or for taking measurements. Stroupe: PEC s RetroCx Workshops Page 1 of 26

2 In addition the PEC has many systems common to large commercial buildings, and would thus expose the attendees to equipment they would see in future projects. The equipment list includes a built-up air handler with airside economizer and variable-speed supply air fan, direct digital controls (DDC), a direct-expansion (DX) chiller, and variable-air-volume (VAV) boxes with terminal re-heat. In addition to this somewhat standard list of equipment, the energy center also has thermal storage. The ice tanks provided an especially interesting aspect as we evaluated trends and looked at the sequence of operation. Many of the savings from the 2002 retrocommissioning effort can be traced to the thermal storage and the equipment it requires. We were convinced that the upward drift in energy use we witnessed in 2004 could also be traced to the thermal storage system. The first of the workshops was held on May 26, 2005 and was followed by monthly meetings. The class size was limited to 12 attendees because of the hands-on format. This ongoing class is composed of a group with a wide variety of backgrounds and expertise: architects, designers, controls experts, mechanical engineers, facility mangers and project managers. The diversity of experience and variety of perspective that these attendees provide has been an unexpected strength of the trainings. Aside from attending an introductory class on retrocommissioning they had little real exposure to building commissioning, but were eager to get hands-on experience. The class met once a month with all of the original students continuing to attend, an unprecedented level of commitment from the attendees. The class was immediately closed to additional students and a waiting list has been established for the next offering. A unique aspect of these workshops is that most of the planning and all of the findings occur in the context of the class with the students directly involved with critical decisions and the analysis of collected data. There were no pre-baked solutions here, nor were all the problems known in advance. At each session students were first taught basic engineering principals and shown how those principals apply to building systems. David Sellers served as the primary instructor and gave these lectures. This background enabled the students to take on structured tasks utilizing this new knowledge. The class divided into small groups for each task, each group with an instructor and a student leader selected on a rotating basis. The leader would be responsible for organizing group efforts and reporting out their findings and suggestions at the completion of each workshop. The groups rotated through the various tasks exposing everyone to different systems, measurement equipment and real data sets. Tasks included getting acquainted with resources available for retrocommissioning, surveying a project and collecting user input, developing a RCx plan, prioritizing activities, compiling a findings list, defining the work scope for the service contractor, setting up and analyzing trend data, benchmarking utility billing data and testing equipment. The second of these workshop series will begin in May of 2006 with a new group of 12 students. This class will benefit from the lessons learned during the first set of workshops. We have several changes planned. An additional instructor is a needed to help facilitate the class especially during the labwork. Kristin Heinemeier of PECI is already serving in this capacity. The agenda is defined in advance for the 12 sessions to insure that critical tasks are not skipped. Homework assignments will be given as a means of accomplishing work not completed during class time. A web page will be set up for posting all the relevant documentation including the monthly lectures, homework assignments, problem log and systems manual. A screening process will be instituted to cut the class size down from the 30 people wishing to attend to a manageable 12. In addition to attaining the pre-requisite, candidates for the class will take a test to demonstrate they have a fundamental knowledge of building systems. NCBC: PEC s RetroCx Workshops Page 2 of 26

3 What follows is a description of three of the workshop projects from attendees of the current class. The first is a look at the trend data that was collected for the workshops. This report is supplemented with data from utility bills and loggers. The second project is an analysis of the facilities heating hot water system. The third report describes a completed project including energy and cost savings: the trimming of a pump impeller. PEC Electric Data Before completion of RetroCx project Almost two years of persistance Building deviates from optimized operation kwh/day Average kwh/day during period of persistance RetroCx workshop series starts Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Figure #1 NCBC: PEC s RetroCx Workshops Page 3 of 26

4 Virginia Waik, City of Palo Alto Benchmarking & Trending Introduction Research indicates $18 Billion in annual energy savings can be obtained through retrocommissioning (RCx) of the existing US building stock. 1 Building retrocommissioning involves site assessments complemented by benchmarking and trending of energy use, operations, and building systems. Assessing the general magnitude of potential building energy savings can be achieved through benchmarking (comparing) one facility against others with similar characteristics. Trending annual utilities consumption can provide a useful means of comparing building performance over time. Trending of specific building systems and components can provide additional details required to make informed investment decisions. Together, benchmarking and trending can bridge the gap between problem identification, correction cost, and quantification of the benefits of correction, including increased tenant comfort, increased system efficiencies and operational savings. A brief summary is given below on the site assessment, benchmarking and trending performed on Pacific Gas and Electric Company s Pacific Energy Center (PEC) by students participating in an innovative retrocommissioning class. The approach is described below and includes specific examples on the process, problems encountered, and the decision-making during the site assessment, benchmarking and trending of the subject facility. Brief Summary of Site Characteristics The PEC is a two story interactive educational facility with 26,000 ft 2 of conditioned space and a lower deck parking garage (6000 ft 2 ) all of which are used for training energy professionals. It has operational and demonstration systems and components not normally found in a building of this size including thermal energy storage and sophisticated lighting and mechanical labs. In 1990 the original building was gutted and remodeled to provide high quality office spaces, a kitchen, meeting and exhibit space for up to 250 people (design capacity is 613). Conditioned spaces in the facility are kept between 72 0 F and 74 0 F for occupant comfort. The building generally operates between 7 am and 7pm weekdays with occasional evening and weekend events. Kitchen staff may come as early as 4 am or 5 am based on need. The system includes one main rooftop (AHU #1) which provides conditioned air to the building s variable air volume boxes and one indoor demonstration air-handling unit (AHU #2). A rooftop boiler serves heating hot water for the building. These systems are monitored and controlled by a direct digital control (DDC) system. The DDC system includes 8 programmable control modules (PCM) with ability to work with both analog and digital signals and 15 unitary control modules (UCM) and one energy management system (EMS) that supervises all the other control units. The EMS is used to perform scheduling, data collection, and provide equipment operation alarms. Innovative Educational Approach Recommissioning class instructors provided background information, tools, and acted as a resource while students were left on their own to discover what they could. This innovative teaching approach provided a rich hands-on environment with open access to all equipment. The building became the working lab. Students with little specific skill in commercial building analysis were encouraged to use common sense and apply it to the day s work. Students were 1 Investing a median cost of 27 /ft 2 for existing buildings and $1 for new buildings over the US could yield $18 Billion in annual energy savings potential. Evan Mills, LBL, December 2005, The Business Case for Commissioning New and Existing Buildings NCBC: PEC s RetroCx Workshops Page 4 of 26

5 rotated in small groups to gain first hand experience with obstacles typically encountered in the RCx process. Students identified problems, interpreted findings, and ascertained alternative approaches to solutions. That process was followed for all the examples in this report. This represented a distinctly different educational philosophy and approach than found in most engineering-type class settings where students have limited or no direct hands-on exposure to operating equipment and are sometimes provided with solutions before they are able to spend the time required to fully comprehend the overall complexity of a given problem. First Efforts Overall Process Here is an example of how that teaching style worked for the class in general, and the benchmarking and trending team specifically. Students were told to investigate and understand both building energy use and the magnitude of efficiency opportunities that might be available at the PEC. Occupant and owner interviews were complemented by site assessments including a review of original design intent, existing conditions, and how those conditions met or failed present needs. The team listed the problems discovered and used this prioritized list to focus the analysis of trend and utility billing data. Building Benchmarking The RCx team used Cal Arch, a web-based building comparison tool developed by Lawrence Berkeley National Labs, to obtain a quick and easy big-picture view of efficiency potential. PEC s building location, square footage, general activities, and annual utilities consumption for 2004, were obtained and entered into the program. Cal Arch accesses a database of buildings, pulls out those with similar characteristics, and then produces three frequency histograms (whole building, electricity & natural gas respectively). These diagrams (Figures 2, 3 & 4) plot the buildings in this subset based upon ascending energy use intensity (EUI). A green arrow highlights where the study building falls relative to the population of similar sites. The population chosen for comparison is filtered based on user input like location, size, or activity. Schools are compared to schools, and offices to other offices. The X-axis shows the values of the three frequency histogram bins (Kbtu/ft 2 -year, kwhr/ft 2 -year, & kbtus/ft 2 -year respectively). The Y-axis shows a percent frequency of all the buildings displayed in a given bin. In addition, users can choose to include a line that cumulates percent frequency displayed as percentile on the right hand axis. This line can tell you exactly what percent of buildings falls below or above yours in energy use intensity. PEC s total building EUI of 50 Kbtu/ft 2 -year beat the average of 60 Kbtu/ft 2 -year. When just comparing electrical use per square foot, PEC s score of 9 kwhr/ft 2 -year beat the average of 15 kwhr/ft 2 -year. Gas was a different story. PEC s score of 19 kbtus/ft 2 -year for gas was much higher than the average of 12 kbtus/ft 2 -year for similar buildings. Using just Cal Arch, it looked like most of the potential savings for the PEC might be in the gas area. The team decided to dedicate additional time and resources to investigate the existing gas uses in the building. Findings from the RCx team on the hot water system are given in a later section of this paper. NCBC: PEC s RetroCx Workshops Page 5 of 26

6 Figure #2 Figure #3 Figure #4 NCBC: PEC s RetroCx Workshops Page 6 of 26

7 Utilities Consumption Trending The Cal Arch numbers provided a snapshot view, but the student team wanted to look specifically at how the PEC was functioning over time. One of the likeliest places to obtain this information was to review building energy use history. The building owner provided three years of monthly utilities billing information to the RCx team. In general, experienced building experts have found that two years of billing data is sufficient, with four being optimal. The PEC had been retro commissioned in 2002 and students wanted to see what had happened since then. The information was entered into a simple spreadsheet program (Excel) and then graphed. What we saw would concern any building professional (see Figure 5): PEC Daily kwh kwh/day kwh/day 2003 kwh/day 2004 kwh/day 2005 kwh/day 200 Figure #5 0 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Month Utilities historical trends indicated that the building s first retrocommissioning did provide savings and those savings persisted for a few years (2002 and 2003). Experienced building energy auditors and consultants have stated that buildings, like automobiles run efficiently for a period after a tune-up, then the performance (and savings) degrade until the next tune up. This was the same case for the PEC. The tune up benefits persisted for a few years and then degraded in 2004 and Since there were no discernable changes in building activities over this period, students came to the conclusion that the building could be returned to its optimized, post-commissioning state with a lower annual energy usage. Building Automation System Findings Help Focus Trending Efforts The class accessed the Energy Management System (EMS) and reviewed the system s capabilities. Screen shots of various systems, operating conditions, and schedules were produced. The image (Figure 6) below shows a sample screen for the main air-handling unit (AHU1) with key temperatures, valve positions, and equipment status. Something looked wrong. The outside air (OSA) temperature indicated F. Students surmised that, if the EMS thinks it is very hot outside, the programming would activate unnecessary cooling. NCBC: PEC s RetroCx Workshops Page 7 of 26

8 Figure #6 Students decided to look at the sample data routinely gathered by the EMS reporting system to investigate how outside air temperature might impact specific pieces of equipment. Review of the EMS reports led the team to additional findings. For example, building temperatures were set between 60 F to 70 F regardless of building schedule or occupancy. See the sample EMS screen shot below (Figure 7) showing occupied and unoccupied cooling set points. These setpoint are all too low and would cause an overcooling of this space (and a significant energy waste). This screen shot also shows that the unoccupied heating setpoints are the same as the occupied values. Without a setback temperature, the space will continue to be heated during unscheduled hours. Potential efficiency improvements might be achieved by implementing more practical temperature setpoints in the EMS. Figure #7 Another finding revealed simultaneous operation of both the thermal energy storage system and the main air handler cooling on weekends and evenings when the building was unoccupied. These findings were gleaned through research of existing trend data sets that were difficult to interpret because of report order and configuration. Reports also included corrupted data. The student team decided to set up trending of both the hot and chilled water systems including the main air handler (AHU#1). Data loggers were used on both hot and chilled water pumps NCBC: PEC s RetroCx Workshops Page 8 of 26

9 since the EMS did not have those points monitored. In addition, students created a work order requesting replacement for the failed OSA sensor. The owner placed a work order for EMS schedules to be reconfigured to reduce unneeded equipment operation when the building was unoccupied. Good data is needed to make informed decisions about HVAC system efficiency. The EMS provided useful information (e.g. valve positions, ice level in storage tanks) that would have been difficult to obtain any other way. Available monitoring points were reviewed and students combined the reports to create truly useful trends correlating space temperatures with equipment operations (See A in Figure 8). The team then set up time-stamped data retrieval (See B in Figure 8), which was then directed to a common file for class access(c in Figure 8). Trend report creation is an iterative process. It may take several sessions before data is cleaned and formatted sufficiently to input into graphing tools. Input errors lead to time intensive corrections until good reports are achieved. It is useful to note that errors caused by incorrect attribution of time for 12 Midnight (AM) and 12 Noon (PM) readings can wreak havoc on trend reports (red circle on Figure 8 below shows the EMS input screen). The configuration of trends had to be meticulously consistent or data would not end up in the appropriate columns as tend files were appended. Careful attention had to be paid to required file naming formats and file locations or the data was lost. Mistakes in any of these areas can corrupt a truly useful data set. Correction time can be reduced if careful documentation is made after each review of the collected trend data before methodical corrections are made to the trend configurations in the EMS. A B C Figure #8 Some of the trend data needed to be reformatted so it could be graphed. For example, the text fields on and off were replaced respectively with 1 and 0. NCBC: PEC s RetroCx Workshops Page 9 of 26

10 The team produced useful graphs using a free web-based data management tool called the Universal Translator (UT). 2 The UT is especially good at processing a variety of trends and graphing these together regardless of their original interval periods. The UT easily combines hourly, minute-by-minute, or daily data samples into one congruent file. The student team cleaned up error, and formatted the data for transfer from Excel to the UT. Two samples follow. Trending provided verification of successful activation of desired fan operation schedules. The image (Figure 9) below shows the status of the control device operating the supply fan for AHU1. The trend shows the fan cycling up and down in operation, but never shutting off on a schedule until about Dec 20, 2005, when the service contractor corrected this issue. Figure #9 One of the most exciting moments in the trending process included the development and production of the trend graph below (Figure 10). The original intent was to look at how the cooling systems were operating, and to clarify possible measures that would correct (or clarify the cause for) some of the earlier findings that inspired trending efforts. All the key cooling components were gathered and graphed together. The graph below shows coincident trends for: the temperature of AHU1 supply air to the building, AHU1 valve position, AHU1 variable frequency drive operation, and thermal energy storage (TES) ice tank levels. AHU1 supply air temperature trend shows that supply air is being maintained between 55 0 F and 60 0 F (left Y axis shows temperature) continuously -- night and day, whether the building is occupied or not. The AHU1 fan speed (measured as a percent on the right axis) shows continuous fan operation. Note that this trend was taken prior to successful re-programming. 2 The Universal Translator can be found at: Note that input data to UT has to be in text format to work. NCBC: PEC s RetroCx Workshops Page 10 of 26

11 The EMS had a monitoring sensor to indicate ice levels in the TES tanks. The monitored point rises when ice is being made and falls when ice is being melted. The trend line shows that the TES system produces ice and then melts that ice to provide additional cooling to the building -- in the middle of the night. Because the system is in an ice making mode the chilled water (glycol) is extremely cold; unfortunately this cold water is transferred to the air handler unit cooling coil and causes it to send 30 air into the building. The chilled water supply temperature clearly shows a daily plunge to just above 20 F during the evening hours as ice is being made. As ice starts to melt, the chilled water temperatures increase in steps and hold at ice melting point of 32 F until the cycle terminates and reverts back to chilled water supply from AHU1. AHU1, as stated above, is providing cooling independent of building conditions. The final trend line of interest is the chilled water supply valve position from AHU1. This trend shows that the AHU1 CW valve stays within a bracket between 30 40% open for most of the operation time, but when ice is being made and melted, the valve modulates between the larger range of between 15 to 40%. Figure #10 The trend indicates that the EMS is in cooling mode all the time. It will be interesting to check this trend after the outside air temperature sensor is fixed to see if that repair helps the team reduce cooling energy consumption. NCBC: PEC s RetroCx Workshops Page 11 of 26

12 Conclusion The most significant conclusion from this study is that even beginners can make positive improvements to building performance when they are willing to learn about their building and make the commitment to trend and benchmark performance. Valuable information can be gained about the potential savings from buildings through simple and quick web-based building benchmarking tools. Historical energy consumption trends can refine understanding and show how a building has been performing over a period of time. Taking the time to learn about the systems that make the building function as designed and studying the capabilities of the tools available to monitor and trend system operations can reveal hidden energy savings and unanticipated areas of investment. If you have a chance to walk through it with a consultant, take it, because you may learn something you didn t know before. Ask questions about the different pieces the equipment, their function, and how they interact in the building s overall operation. Making a long-term commitment to both benchmarking and trending your key operations will allow you to make informed investments and increase building performance from average to excellent. NCBC: PEC s RetroCx Workshops Page 12 of 26

13 Rishabh Kasliwal, Cogent Energy Hot Water System Retrocommissioning The RCx team wanted to investigate the Hot Water (HW) System for potential retrocommissioning opportunities. Based on discussions with the owner and occupant interviews the team suspected that the HW system is unable to provide comfort conditions to all zones especially after long cool down periods like weekends and holidays. The team divided the RCx effort into the following three steps: 1. Understanding the HW system 2. Trend based functional testing 3. Manual functional testing 1. Understanding the Hot Water System The group decided to start with the design drawings to accustom themselves with the design, piping & layout of the existing HW system. The systems manual, developed as part of the previous commissioning effort, was another valuable resource available to the RCx team. The existing hot water (HW) system comprises of a 250 thousand BTU/hour (MBH) boiler with a primary HW pump designed for 13 gallons/minute 22 feet head. The building has 15 VAV boxes all of which have reheat coils. In addition the other hot water loads include one heating coil each at the first floor and roof lobby area. The education air handler (AHU 2) also has a heating coil. Reviewing the design drawing we noticed a discrepancy between the HW pump sizing and the coil capacities. The 18 HW coils put together are sized at 18.5 gpm whereas the pump and the loop are sized at 13 gpm. This discrepancy could be a result of the designer assuming 70% diversity on the design heating load or the HW system being undersized. The team next conducted a walk through to generate the actual system diagram. The boiler, HW pump and other associated equipment are located on the roof. Differences between the as-built drawings and the installed HW Loop system were apparent as soon as the team reached the roof. The installed boiler is different than the one specified in the design drawings. The HW pump, strainer and the circuit setter are located on the Hot Water Return (HWR) side though they were specified on the Hot Water Supply (HWS) side. The following graphics (Figures 11, 12 & 13) compares the design HW system diagram with the installed system diagram as well as the system graphic from the EMS. NCBC: PEC s RetroCx Workshops Page 13 of 26

14 Figure 11 Actual installed piping at boiler Balancing valves (circuit setter s) Drawings show HW pump, strainer, Balancing valve on HWS side Figure 12 Hot Water Loop graphic from EMS The team also observed that the current HWS and HWR temperatures were 162 o F and 137 o F respectively although the boiler is designed for HWS temp and HWR temp of 190 o F and 160 o F respectively. The team believes that such a depressed HWS temperature might be a result of boiler tube fouling and scaling reducing the heat transfer across the tubes. This may have resulted in condensation in the flue stack and corrosion of the boiler fire box. We observed that all but one of the 15 VAV boxes have 2-way HW valves. AHU2 also has 2- way valve while the HW valves at the two heating coils are 3-way valves. The HW loop bypass is located close to the boiler (see Figure 13) and is controlled by a circuit setter which has been NCBC: PEC s RetroCx Workshops Page 14 of 26

15 manually set at 0% open (no by-pass). The circuit setter upstream of the pump is manually set at 100% open (full flow to boiler). This is a less than optimal location of the circuit setter as in throttled position it could lead to potential cavitation at the pump. The coil capacity of the two heating coils and VAV coil 1-1 with 3-way valve is only 2 gpm. The team is concerned about the system performance and behavior when all or some of the 15 2-way valves stroke closed. Does the system dead head the pump or is there another bypass in the system? The team moved on to the testing phases to look for answers. Figure 13 Actual installed system diagram Bypass set at 0% flow (3) 3-way valves (15) 2-way valves 2. Trend-Based Functional Testing The team conducted trend-analysis-based functional testing which was aimed at establishing the programmed sequences of operation as well as investigate the comfort concerns. Trend data was collected during the 5 week period from December 2005 to January The team found that the HW pump is interlocked with the boiler and the system operates on the following schedule, Monday to Friday 7:00 am to 9:00 pm. There appears to be no OSA based reset applied to the HW Supply Temperature (HWST) set point (Figure 14). We believe the system tries to maintain a fixed HWST set point but is unable to do so successfully. The HWST rarely meets the design temperature of 190 o F although it consistently goes over 180 o F. The average differential between the supply and return temperatures is between o F against a design of 30 o F observed during the system walk through. NCBC: PEC s RetroCx Workshops Page 15 of 26

16 Figure 14 HWS SP HWS Temp with OAT OAT HWS Temp We also found that the peak HWST achieved on Monday s is significantly lower (up to 70 o F!) than that achieved during the rest of week. This is consistent with the occupant s complaints about the extended warm-up after a long down time (Figure 15). Figure 15 HWST & Pump status with time Low HWS Temp on Mondays HWS Temp (deg F) Pump status NCBC: PEC s RetroCx Workshops Page 16 of 26

17 By comparing average zone temperature s of all 15 VAV zones for a typical week, we found that the zones VAV 1-3 and 1-4 (in basement level) are maintained between o F (Figure 16) as against most other zones (72-74 o F). This is consistent with the cold comfort calls from the occupants in the basement. Lower zone temperatures for the two farthest zones VAV 1-3 and 1-4 from the HW pump suggests that the reheat coils may be getting starved as a result of the pump being undersized. Leaking 3-way valves at coils HC-1 and HC-2 (Figure 13), or an unknown hydraulic crossover between the supply and return piping may be causing mixing between HWS&R and thus aggravating the above problem. Figure 16 Avg. zone temp with time HWS temp. VAV 1-3 VAV 1-4 Avg. zone temp. (deg F) 3. Manual Functional Testing The team moved on to manual Functional testing to: a) Examine HW Pump size and flow capacity b) Investigate HW pump performance at variable flow (load) conditions c) Examine Discharge Air Temperature (DAT) at the most hydraulically remote coils from the boiler and the known trouble zones d) Determine boiler heating capacity and performance We designed the test so as to test the system at variable load (and flow) conditions. At each of the load conditions the following variables were measured. NCBC: PEC s RetroCx Workshops Page 17 of 26

18 Table 1 Component Measured Variable Equipment used Boiler Combustion efficiency Flue gas analyzer HW Pump Pump power, amps, motor speed, differential pressure (DP) to correlate to flow kw meter, amp meter, strobe meter, DP meter HW Loop HWST, HWRT Temperature sensor Zones Zone Set points, DAT at zones (diffusers) EMS, Temperature gun The following six conditions chosen for the test: 1) As found o document the as found operating condition of the system 2) Shut Off o to determine the impeller size and the pump curve 3) Full open o to determine boiler capacity with all 18 valves fully open 4) Full closed o to examine flow variation with all 18 valves fully closed 5) 1/3 rd open o to examine flow with 6 out of 18 valves fully open o to examine DAT in the known trouble zones 6) 2/3 rd open o to examine flow with 12 out of 18 valves fully open o to examine DAT in the known trouble zones The measured data was plotted on the pump curve in Table 2. Table 2. Test Number D Condition Design 1 As found 2 Shut Off 3 Full Open 4 Full Closed 5 1/3 rd Open 6 2/3 rd Open NCBC: PEC s RetroCx Workshops Page 18 of 26

19 From the HW pump test we found that there is little variation in flow. This could either be a result of an unknown bypass in the HW Loop or due to leaking valves at the coils. It should be noted that the due to the flat nature of the curve the DP measurements are very close to each other and some error is expected in correlating DP to flow. Little variation in pump power is consistent with the small flow variation observation. At the zone level, we verified the occurrence of depressed DAT at the zone level even during part load conditions (1/3 rd and 2/3 rd open). Figure 17 exhibits the depressed DAT occurring at the zones. Figure Known trouble zones DAT at most zones below 80 F! Temp (F) Heat Coil AHU Max. (red) and Min (blue) zone temp observed The Heating Water Return Temperature (HWRT) at the boiler was typically around 130 F, and thus the coils see at least 130 F water. Thus, we believe, that such depressed DAT values could be a result of poor heat transfer at the coils (scaling, dirty coils etc.) or due to an unknown bypass in the system resulting in a hydraulic crossover. Conclusions The HW Loop RCx the team has been able to confirm some suspicions relating to the thermal comfort issue. The boiler seems to be a little short on capacity, as observed from the extended Monday warm-up and less than design delta-t on the loop. There is little variation in the loop flow and pump power with variation in load. As there are only 3 known 3-way valves in the system, this could be a result of an unknown bypass or due to leaky 2-way valves. We verified, depressed DAT occurring in the zones farthest from the HW pump. We also verified that the reheat airflow is not excessive. This situation could be occurring due to poor heat transfer at the coils and/or due to an unknown bypass resulting in hydraulic cross over of the hot water. NCBC: PEC s RetroCx Workshops Page 19 of 26

20 Recommendations We recommend that more testing be done in order to resolve the remaining unknowns. The possibility of an unknown hydraulic crossover should be investigated. The reheat coils should also be examined for scale and leaking 2-way valves. NCBC: PEC s RetroCx Workshops Page 20 of 26

21 Blair Horst, Lawrence Livermore National Lab Chilled Water Pump Testing and Remedial Action Figure 18- Pacific Energy Center San Francisco Discovering the Problem One of the first activities of the workshop was to review the renovation mechanical design drawings. This was followed by a physical inspection of the facility, systems and components in an attempt to identify any discrepancies worthy of investigation. Immediately following the renovation both pumps were operated simultaneously, with circuit setters installed on each pump discharge, throttling flow to about 53-gpm each, matching the design intent. Following initial retrocommissioning activities, conducted by Ross Sherrill in 2002, pump control was changed to lead / lag, with a single pump serving the load adequately, without throttling. Investigation of the design and installed components indicated an anticipated requirement for a total chilled water flow of 106-gpm at a head of 110-Ft. water column (W.C.) with both pumps operating. A rough back-check of chilled water system pressure drop was estimated at about 55- Ft W.C. by determining losses in the chilled water piping system, allowing for fittings and adding pressure drops of coils, chiller and control valves, etc., in series. The design pump head is twice the estimated system head; this indicates the pumps may be considerably oversized. Thus, testing could confirm an opportunity to improve pump efficiency. Pump Test A pump test was executed to verify actual performance under normal building operating conditions of Chilled Water Brine Pumps (P-1 and P- 2) on November 10, 2005). The pumps are Bell & Gossett series ½ BC, are Base Mounted with 5-horsepwer (HP) motors, each rated at 53-gpm and 110-Ft. W.C. Chilled Water Pump P-1 s discharge pressure was tested twice at noflow and wide-open conditions. Test results are as follows: Figure 19 Retrocommissioning Workshop Team Test Description Test-1 Test-2 Average No-Flow, Shut-Off Test (Ft W.C.) Figure 20 - Pump Test 100% Flow, Wide Open (Ft W.C.) Analysis (Refer to Figure 21 - Pump Curve and Impeller Size Analysis) First, the existing impeller size is determined by performing a No-Flow, Shut-Off Test. The pump is run with discharge valves closed; the measured head (108.8-Ft W.C.) is compared to standard impeller sizes shown on the manufacturer s pump curve at 0-gpm flow. It appears the existing impeller size for this pump was a little less than 9½-inches, with the minor difference probably due to wear and/or wear-ring clearances. (Note, that during subsequent impeller trimming, it was found the pump is not fitted with wear rings.) The next step was to determine the first system curve point and flow rate using 100% Flow, Wide Open Test data. Enter the vertical axis at 0-gpm, follow horizontally from 93.5-Ft W.C. NCBC: PEC s RetroCx Workshops Page 21 of 26

22 (measured test data) until intersecting the 9½ Impeller Curve, then follow down to the horizontal axis to read the flow rate of gpm. The system performance curve is now developed based on 100% Flow, Wide Open Test data using the following equation based on the square law relationship derived from the Darcy- Weisbach equation): Pressure NEW = Pressure OLD x (Flow NEW Flow OLD ) 2. Results are plotted on Figure-21. (Refer to Note C.) This curve represents existing system performance at various flow and head conditions. A FT 9½ Impeller F Throttled Operations B Wide Open Test: 93.5 FT E Impeller Size: 7-3/4 D Required Head = 57-Ft Estimated Head = 55-Ft C System Curve Calculated Points (Typical) B Base Case Operations G Post-Impeller Trim No-Flow 70.2 Ft WC Wide-Open 58.7 Ft WC A 0-gpm D Design Flow = 106-gpm Read: gpm B Figure 21 - Pump Curve and Impeller Size Analysis A B C D E F G Find Impeller Size; No-Flow Pump Test: Measure head at 0-gpm; compare to impeller sizes: Ft 9½ Size. Base Case Wide-Open Flow Test: Determine flow rate from measured head (93.5-Ft W.C.), horizontally to right from 93.5 Ft W.C. to intersecting with 9½ impeller curve: follow down to flow rate; read gpm (Base Case). System Curve: Plot system curve based on Wide Open Test data using the square law relationship: P NEW = P OLD x (Flow NEW Flow OLD ) 2 Example: 93.5-Ft x (80-gpm 137-gpm) 2 = 31.6-Ft W.C Head at 80-gpm Find head required at design flow; extend design flow rate up to system curve, then left, to head = ~57-Ft W.C. Find Optimum Impeller Size: Extend line up to system curve from design flow. Interpolate impeller diameter at intersection (between 7½ and 8 impeller curves); indicates an impeller of ~7-3/4. Sketch a new impeller curve through the intersection of the system curve and the design flow point, to the vertical axis. Throttled Operations: represent throttling from the 100% Flow, Wide Open test condition to the design flow rate of 106-gpm, increasing pump head, but reducing energy use slightly. New, Post Impeller Trimming Curve: Pump test is repeated. No-Flow, Shutoff Test: 70.2 Ft W.C.; Wide Open Flow Test: 58.7 Ft W.C. New pump curve is slightly above predicted and will perform adequately. Determine the optimum impeller size by projecting from the design flow rate (106-gpm) up to the system curve, then, from the intersection, horizontally to the vertical axis to determine the head needed; about 57-Ft W.C. (Refer to Note D.) This is approximately the same head as determined in the rough estimate of system head (55-Ft W.C.) described above. Results indicate that the same pump with a smaller impeller could serve the building load more efficiently. Other solutions are possible, including operating the pump at a lower speed or throttling the pump. These alternatives are evaluated for their potential energy cost savings. NCBC: PEC s RetroCx Workshops Page 22 of 26

23 Alternatives Evaluations Existing operations (Base Case) and three alternative modifications are considered: 1. Trim the existing impeller from 9½ to 7-3/4 size 2. Change pump to a lower speed motor 3. Reduce rotational speed using a Variable Frequency Drive 4. Throttle flow of the existing pump to the design flow rate Replacement of the pump for a new pump-motor is ruled out due to expense. Electric power costs are determined for the Base Case and the remaining three alternatives. The pump s required break horsepower (BHP) is determined for each alternative condition by interpolation from Figure 21; with the exception of the variable speed drive alternative. Pumping energy requirements for the variable speed drive alternative are determined based on pump affinity laws. The resulting pumping energy loads (BHP) are divided by the motor efficiency and converted to electric power demand (kw). The results are multiplied by runhours determined from trend logs. Finally, annual pumping electric power costs are calculated based on the average power rate for the PEC during Results are tabulated and calculations explained below. Table 3 Summary Pump Modification Alternatives Parameter Base Case Alternative 1 Trim Impeller Alternative 3 Lower Speed w/ VFD 5 Alternative 4 Throttle Flow Impeller Size (Inches Diameter) 9½ 7-3/4 9½ 9½ Rotational Speed (RPM) 1,750-RPM 1,750-RPM 1,349-RPM 1,750-RPM Flow Rate (gpm) gpm gpm gpm gpm Pumping Head (Ft. W.C.) 93.5-Ft. W.C Ft. W.C. Not Calculated Ft. W.C. Pump Break Horsepower (BHP) BHP 2.63-BHP 2.13-BHP 4.65-BHP Pump Power Demand (kw) kW 2.18-kW 1.86-kW 3.85-kW Annual Power Use (kwh/yr) 3 18,115-kWH/Yr 9,529-kWH/Yr 8,161-kWH/Yr 16,847-kWH/Yr Annual Power Cost ($ / Year) 4 $2,898 / Year $1,525 / Year $1,306 / Year $2,696 / Year Notes 1. Values interpolated from pump curve, Figure 21. Alternative #2 savings were not calculated. 2. Electric demand (kw) is calculated assuming motor efficiency of 91.2%. 3. Annual power use is determined from 4,380-hours per year operations based on PEC trend data. 4. Annual electric power costs are based on the PEC s blended rate during 2005 of $0.16 / kwh. 5. Calculations include an added efficiency reduction of 5% due to internal losses of the VFD. Base Case Review of the pump curve in Figure 21, Note B, provides the following operational data for the existing pump: gpm and head of 93.5-Ft. W.C. with pump at 5.00-BHP and 63% efficiency 4.14 KW. NCBC: PEC s RetroCx Workshops Page 23 of 26

24 Alternative 1: Trim Impeller Review of the pump curve in Figure 4, Note E indicates the following operational data an impeller trimmed to 7-3/4 : 106-gpm and head of 57-Ft. W.C. with pump at 2.63-BHP and 58.5% efficiency 2.18 KW. Alternative 2: Change to Lower Speed Motor This option is not feasible; the pump and motor are provided as a unit. Replacement would require piping reconfiguration and would be prohibitively expensive. Alternative 3: Reduce Rotational Speed with Variable Frequency Drive Reduce motor rotational speed to preserve optimum pump efficiency; pump performance is predicted based on pump affinity laws, as follows: GPM NEW = GPM OLD x (RPM NEW RPM OLD ), Solving for RPM NEW: RPM NEW = RPM OLD x (GPM NEW GPM OLD ) 1750 x ( ) = 1,349 RPM BHP NEW = BHP OLD x (RPM NEW RPM OLD ) x (1,349 1,750) 3 x = 1.86 KW This alternative is not really practical for the installation; it requires purchase and installation of a variable frequency drive, reprogramming of the existing complex DDC control system followed by performance testing. Additional efficiency losses of about 5% will be incurred by the VFD and the complexity of the system will increase. The incremental annual cost savings is only $219 per year compared to trimming the impeller, which is a far less costly alternative without complicated electronic controls. Alternative 4: Throttle Flow of The Existing Pump Throttling flow causes the operating point to move up the pump curve of the existing pump, along the impeller line. The circuit setter located downstream of the pumps can provide the throttling function. Flow of 106-gpm against a head of Ft. W.C. requires 4.65-BHP and 61.8% efficiency 3.85 KW. Recommendation Trim impeller to achieve optimum performance. This option will save about 8,567 kwh per year representing an annual power cost savings of $1,371 compared to present operations. Trimming of the Impeller Impeller trimming was conducted on chilled water circulation pump P-1, by CalAir Corporation technicians Scott Estep and Joey De Martini on 19-January The trimming and pump repair activity was conducted at a fairly high cost of $2,476. However, the work included additional incidental tasks of replacing the shaft seal, grouting the pump bases (not done during the original construction), fixing leaking fittings and insulation, plus briefing students of this workshop regarding impeller trimming. Impeller trimming was performed without shutting down the system, thus, the first step in the procedure was proper isolation from operating equipment. The trimming process is summarized: NCBC: PEC s RetroCx Workshops Page 24 of 26

25 Isolation a. Adjust system controls to place chilled water circulation pump P - 2 in lead; pump P - 1 in lag. Assure pump P - 2 is operating and pump P - 1 is off prior to proceeding. b. Open circuit breaker for pump P - 1; perform standard lockand-tag-out procedure. c. Close pump P - 1 discharge and suction isolation valves. See Figure 22; the discharge isolation valve of pump P 2 is open and is closed for pump P - 1. Disassembly & Trimming d. Disassemble pump P-1; remove impeller. See Figures 23 and 24. e. The impeller was then taken to a machine shop equipped with a special, digitally controlled, grinding machine where the impeller was ground to a 7-3/4 diameter and dynamically rebalanced Reassembly and Start-Up f. Reinstall the trimmed impeller (Figure 25). g. Replace the pump seal and gasket; see Figure 26. Although the existing pump seal has about a 10-year life, with only half consumed, it makes sense to replace it since the pump is already opened. Pump life is extended. h. Reassemble the pump, assure there is no leakage. i. Follow procedure for chilled water distribution system shutdown. j. Reprogram lead-lag controls to place pump P-1 always in lead position with pump P-2 always in lag position. Since only the pump P-1 impeller was trimmed, this step will assure that the most efficient pump is always in lead position. k. Restart the chilled water distribution system following normal procedure. Verification / Development of New Pump Curve After allowing system operation to stabilize, the pump test was repeated (15-Feb-2006). The results lead to a new pump curve, slightly above the predicted performance (refer to Figure 4, Note G), confirming that the trimmed impeller performs essentially as predicted in the initial analysis. The head developed and power consumption were only slightly higher than predicted. Predicted Post Trim Re-Test Description Ft WC Power kw Ft WC Power kw No-Flow, Shut-Off Test N/A N/A % Flow, Wide Open P-1 Closed CHW Supply P-2 Open Figure 22 Isolate Pump Figure 23 Disassembly Figure 24 Pump with 9½ Impeller Figure 25 New 7-3/4 Figure 26 Replace Pump Seal NCBC: PEC s RetroCx Workshops Page 25 of 26

26 Measured Energy Savings Power demand was measured at 2.18 kw following impeller trimming. The predicted demand was the same, based on the pump curve (Figure 21). Pump electric demand savings is estimated at 1.96 kw. Based on annual operating hours and the rate PEC pays for electric power, annual cost savings are approximately $1,371 per year, resulting about a 1.81 year payback period for the $2,476 investment. Final Step Following completion of impeller trimming and verification of proper operation, as predicted, EMS controls must be programmed to accept and adapt to the changes. The pump motor sequences of operation are revised to make the newly trimmed pump (Pump P-1) the primary pump with the other pump (Pump P-2) the lag pump. NCBC: PEC s RetroCx Workshops Page 26 of 26