Impact Evaluation of 2004 Compressed Air Prescriptive Rebates

Impact Evaluation of 2004 Compressed Air Prescriptive Rebates May 15, 2006 Prepared for: National Grid USA Service Company P.O. 0000027684 DMI# 05006.520 Prepared by: DMI 450 Lexington Street Newton, MA 02466 (617) 527-1525

Impact Evaluation of 2004 Compressed Air Prescriptive Rebates Introduction The purpose of this study was to evaluate the energy savings achieved by selected prescriptive compressed air measures installed in 2004. The scope of this study was to provide annual energy savings, summer and winter peak demand reduction, and percent of energy savings that occurs during on-peak hours realizations for each of the compressed air measures included in the study. National Grid will use these results to adjust the existing savings realization factors that are associated with the program. Description of Evaluation Methodology National Grid USA Service Company contracted with DMI to evaluate the savings of twenty prescriptive incentive applications. National Grid did the sample selection and provided information on the sites to DMI. DMI developed the methodology for the evaluation of variable speed compressors, load / no load compressors, a variable displacement compressor, cycling dryers and a variable speed dryer. Before the Site Visit For each application being evaluated, DMI received a copy of the application package and any additional information available from National Grid. We reviewed the application and attached documentation to develop an understanding of the project and of the equipment and assumptions that were made. We developed in-house calculation methods and templates, and created an interview guide to collect information necessary to complete the analysis. Before visiting the site, DMI contacted the site representative to collect information on equipment make/model, where the equipment was located, the facility operating hours and changes in production/operations over the past year. We also contacted the customer s account manager to inform them that we scheduled a visit to their customer s site. At the Site Two of the original sites had suffered equipment damage and were not in operation during our metering period. Replacement sites were provided by National Grid. DMI was unable to make contact with one of the twenty sites. Therefore, a total of nineteen out of twenty sites were metered and are included in this study: twelve sites have compressors only; three sites have dryers only; and four sites included both compressor and dryer projects. To confirm the information obtained during the pre-visit phone conversations, the customer was interviewed during the site visit about current operations, hours of use, and the base or pre-retrofit condition. During each site visit, we took spot-meter power and volt-amp measurements and noted production variables such as throughput, compressor operating speed (when applicable), pressure, and/or flow rate. We reviewed and collected customer data such as hours of use, operators log sheets, and other available data pertinent to the specific application. 1 (JTF)

Data Analysis DMI used the metered data and collected information to develop an estimate of annual energy savings, percent of energy saved on-peak, and summer and winter peak demand reduction for each measure at each site. We utilized Microsoft s Excel 2003 as our principal calculation tool in the evaluation analyses. A calculation template was created for each type of measure. The next five sections describe the analysis process for each type of measure that occurred in this study. Variable Frequency Drive Compressors For variable frequency drive compressors, the compressor package amperage was metered and logged every thirty seconds for two weeks. The package amperage is the total amperage including the compressor motor, variable frequency drive, cooling fan and electronics. These thirty second intervals were then averaged into one-hour time bins for each hour of metering. The average percent-load was determined by dividing the hourly-averaged amperages by the full-load amperage data provided by the manufacturer, provided that the manufacturer s data agreed with the metered data. The averaged hourly amperage was also used in conjunction with the spot metered voltage and power factor to calculate the package kw for each hour bin. The drive burden is not subtracted from these calculations because the percent-load is calculated using the full-load package amperage (including drive burden). It is assumed that, with variable frequency drives, the power factor remains relatively constant throughout the operating range. To calculate the flow rate for each hour, the hourly percent full-load and the manufacturer s full-load flow rate were multiplied together. This averaged-hourly flow rate was then used to look up the coincident power for the base case compressor (modulating with blowdown), using an in-house air compressor model. To capture the summer, winter, on-peak and off-peak periods, the day type and hour of each time bin was considered. Since the metering occurred between the months September through November, the summer and winter peak periods were extrapolated from the day and hour data only. All compressors included in this study are positive displacement compressors. Therefore, weather variability does not affect power consumption. Next, to calculate the percentage of each period that the compressor was running, an on/off threshold was set at 10% (adjustable) of the full-load flow. If a compressor operates less than 10% for any hour, the base case and installed case are both considered off for that entire hour. This becomes important for the lookup kw of the base case compressor. Since the power of the base case compressor at minimum flow is 25% of full-load power, a false savings would become a significant part of the total savings associated with a variable frequency drive compressor loaded at 10% or less. After the compressor state for each hour is determined, the total number of hours on divided by the total number of on-peak and off-peak hours separately, gives the percent of the time that the compressor is running for those periods. For the base case and installed case energy use, these on percentages are multiplied by the number of total hours and the average kw (power) in each period to give the total kwh for each period (kw*hours = kwh). The energy savings for each period is the difference in base case and installed case kwh calculations. The hours for each period are shown for National Grid s old and new time period definitions in Table 1 below. The percent on-peak for each peak period is calculated by dividing the on-peak kwh savings by the total kwh savings for the year. The reduction in summer and winter peak demand is the difference in the average kw for the base case and installed case multiplied by the percent-on for each period. This method includes 2 (JTF)

both a diversity factor and a coincidence factor. The diversity factor is included when the kw is averaged for each period. The coincidence factor is the percent of time that the compressor is running for each period. New Definition Old Definition Summer Peak June-August 3:00pm - 5:00pm June-September 11:00am - 3:00pm Winter Peak January 5:00pm - 7:00pm December-February 5:00pm - 7:00pm On Peak January-December 6:00am - 10:00pm January-December 8:00am - 9:00pm Table 1 National Grid Peak Hours (New and old Definitions) Load / No Load Compressors Similar to the variable frequency drive compressors, the load / no load compressors were metered using a current transducer and the full package amperage was logged every thirty seconds for two weeks. In addition, a state logger with an adjusted current switch was included to verify how often the compressor loaded and unloaded. During the site visit, the compressor loading and unloading power curve was recorded with a Fluke power quality analyzer. This recording is compared with the loaded and unloaded power information provided with the manufacturer s data sheet. Information to calculate the compressor loaded and unloaded power factors are also gathered during the site visit. Using the manufacturer s listed loaded amperage, the logged amperage is tagged loaded (2), unloaded (1) or off (0). The off threshold is adjustable and determined by the manufacturer s data sheet. If the loaded amperage and/or the off threshold can not be derived from the manufacturer s information, the power curve that was recorded during the site visit was used. With the thirty-second amperage data and tags, the percent loaded, percent unloaded, and percent off is calculated on an hourly basis by summing the number of minutes each state occurs. Since a load / no load compressor is only supplying compressed air during its loaded state, this loaded percentage is used to calculate the average delivered CFM for each hour during the metering period. This hourly CFM is used to look up the coincident base case compressor power, which is calculated with an in-house compressor model. The base case compressor is a modulating compressor with blowdown. The unloaded power of the metered compressor is calculated by using the averaged unloaded amperage, the metered voltage and the metered unloaded power factor. Even though the compressor is not supplying the system with air, it continues to consume energy while it is unloaded. The percent-unloaded multiplied by the average unloaded power gets added to the percent-loaded multiplied by the average loaded power to calculate the total hourly kw for each hour of metering. Equation 1 shows how the hourly installed compressor power was calculated. Hourly _ kw = % * kw + % * kw Equation (1) Loaded Loaded Unloaded Unloaded Each averaged hourly total kw is categorized for demand and energy as summer peak or winter peak and on-peak or off-peak, as appropriate. These tags were used to determine the average kw and percent time-on for each period. The percent on (loaded plus unloaded) and percent 3 (JTF)

off are also averaged for each period. These averaged percents are used to calculate the number of hours that the compressor is on during each period. The number of hours on multiplied by the average total power equals the total energy for each period. The difference in total energy of the base case and installed compressors gives the energy savings for each period. The percent on-peak and demand reduction are calculated using the same methodology as the variable frequency drive compressors. Variable Displacement Compressor One of the evaluation sites uses a variable displacement air compressor. This type of compressor has a variable power factor and its power must be metered directly to accurately evaluate the energy use. For this site, we used a Synergistic Controls Systems Inc. Model C- 140 power recorder. The power of the air compressor was logged every five minutes for twenty days. The data reduction template for the variable displacement compressor is similar to the variable speed compressor template. The recorded power data is averaged into one-hour time bins. Every twelve data points are averaged to get an overall power average for each hour. This averaged power is used in conjunction with a variable displacement compressor model to lookup the average airflow, in CFM, for each hour. The returned CFM is then used as a reference value in the base case compressor model to give the base case power. The base case compressor is a modulating compressor with blowdown. The on-peak and off-peak savings, percentages and demand reductions are calculated using the same methodology as the variable frequency drive compressors. Cycling Air Dryers The cycling air dryers were metered using state loggers. A current switch was attached to one phase of the input power and adjusted to send a signal, along with a time stamp, to the logger every time the dryer would cycle on (1) or off (0). During the site visit, the dryer power was recorded during a cycle-on cycle-off period. This power recording gave the power consumption for both on and off cycle states. The dryer voltage and power factor were also recorded during the site visit. A Microsoft Excel template was designed to sort the dryer state data by hour using the date/time stamp and the cycle-on cycle-off label. The minutes on and minutes off were computed on an hourly basis. The percent-load was calculated by dividing the number of minutes the dryer is cycled on during each hour by sixty minutes. Similar to the variable speed and load / no load compressors, a threshold is set at 10% (adjustable) to interpret when the base case dryer would be off for the entire hour. This threshold was determined to be a point where if at least 10% of the capacity flowrate was required from the compressor, the dryer would also be required to remove moisture out of the air. The power for each hour of the installed cycling dryer is calculated using the metered power and the percentage that the dryer is cycled on. The cycleoff electricity for each dryer was insignificant relative to the overall savings and was not included within the savings calculations. The base case dryer uses a National Grid compressed air dryer CFM vs. kw/cfm look-up table. This table is derived from the manufacturer s data sheets of non-cycling dryers with a 4 (JTF)

flow range of 50-800 CFM. The full capacity flow of the installed dryer is used as the look-up value and a corresponding power/cfm is used to calculate the base case dryer energy use. The base case energy use is calculated by using the percentage of hours that the installed dryer was on during the metering period. This percentage is multiplied by the number of hours in each period (on-peak, off-peak) and then by the base case dryer s full load power. The on-peak and off-peak savings percentages and demand reductions are calculated using the same methodology as the variable frequency drive compressors. Variable Frequency Drive Dryer One of the evaluation sites installed a variable speed dryer. This dryer works in a similar fashion to a variable frequency drive compressor. The variable frequency drive changes the speed of the refrigeration compressor to match the compressed air load while maintaining a constant dew point. This reduces the power consumption compared to a non-cycling dryer. This dryer was metered by recording the full-package amperage every thirty seconds for fifteen days. A power and voltage profile was also recorded during the site visit with a Fluke power quality analyzer. The spot-meter data was used to calculate the power factor through the range of operation. The data reduction for the variable frequency drive dryer is handled with a template that combines hourly amperage averaging, a percent-loaded threshold, the metered full-load power and the National Grid compressed air dryer design curve. The amperage is averaged into hourly bins and the percent-load is calculated based on the averaged amperage divided by the full-load amperage. This percent-load is used with the adjustable off threshold to determine which hours the base case dryer would be interpreted as on. The hourly power was calculated by multiplying the percent-loaded by the metered full-load power. Finally, the base case dryer is studied just as in the cycling dryer analysis. The on-peak and off-peak savings percentages and demand reductions are calculated using the same methodology as the variable frequency drive compressors. Explanation Key An explanation key has been derived to aid in reconciling discrepancies between the tracking and evaluated calculations. Some impact variables share common explanations. For example: total kwh savings and on-peak savings may both be skewed by a difference in application operating hours versus metered operating hours. Table 2 shows the Discrepancy key and an explanation of each entry is given after the table. The key is arranged into 4 categories (Load Profile, Operating Hours, Over-sized Equipment, and Peak Hours). 5 (JTF)

LP-1 LP-2 OE-1 OH-1 OH-2 PH-1 PH-2 Tracking/Evaluation Discrepancy Key Application load profile vs. Metered load profile Equipment operating overnight at light-load Over-sizing of new equipment Application operating hours vs. Metered operating hours Overnight equipment operation Change in peak hours from previous year Summer/Winter/Off-Peak production profiles Table 2 Explanation Key for Tracking/Evaluation Differences LP-1 Application vs. Metered Load Profile Each application lists the actual or expected equipment load information and estimated operating hours. This information is then used to make the calculations for the tracking savings. A load/no load compressor achieves most of its energy savings over the base case compressor while at a load profile between 45% and 90%. If a site s load profile falls outside this band, the savings are reduced significantly. This similar situation can also happen with variable speed compressors and cycling dryers. For these two cases, if the load profile is near 100%, the savings will be reduced. LP-2 Equipment Operating Overnight at Light-Load This condition can occur when there is a compressed air demand during the nighttime hours of greater than 10% but below any significant usage. Typically this occurs with the cycling dryer calculations. When the loading of a cycling dryer remains above 10% in any hour overnight the base case dryer is modeled as on for these hours. This base case dryer is considered fullyloaded and consumes significantly more energy. This may also occur with variable speed compressors since the minimum power draw of the base case compressor is 25%. OE-1 Over-Sizing of New Equipment During the Design 2000 application process, the vendor will work with the customer to determine the pressure and volume of compressed air along with the storage capacity that would best suit the customer s needs. These estimates can become conservative to the point where the compressed air system is significantly oversized. The results of over-sizing the system is commonly seen in variable speed compressor applications. When a variable speed compressor load profile falls below 60% for all of the hours it is on, a dramatic increase in savings occurs over the base case compressor of the same size. This savings is reduced for a VSD compressor sized to operate between 60% and 100% of its full-load capacity. OH-1 Application vs. Metered Operating Hours The tracking gross and peak energy savings is calculated using the number of hours listed on the application. These hours are approximate operating hours each week. Even though a site may not be in a production mode, some compressed air equipment may be at a minimum of 10% loading or more and thus be counted as operating. This will increase the actual hours that equipment is on and also increase the overall kwh savings. OH-2 Overnight Equipment Operation For equipment that is considered on during the overnight hours, overall total operating hours increases. This has a significant effect on the overall kwh savings and the amount of the 6 (JTF)

savings that occurs during off-peak hours. Increasing the off-peak kwh savings will affect the percent on-peak. This is different than operating overnight at low-load (LP-2) because this is only taking into account the increased hours and not the savings from efficient equipment or lightly loaded equipment. PH-1 Peak Hour Change from Previous Year This year, the peak hours for National Grid are Monday through Friday 6:00AM 10:00 PM. Previous years, the peak hours were Monday through Friday 8:00 AM 9:00 PM. This increases the number of peak hours each day by three hours (12.5%). For sites that have early operations, hours that were previously considered off-peak may now be on-peak. PH-2 Summer/Winter/Off-Peak Production Profiles Savings will accumulate unevenly for sites that have little or no production during a certain peak or off-peak period. For example: if production only occurs between the hours of 8:00 AM 5:00 PM, no winter-peak or off-peak savings will accumulate. This affects the winter-peak demand reduction and percent on-peak kwh savings. Results DMI evaluated the energy savings of nineteen National Grid customers that received incentives for upgrading or installing compressed air equipment that conserves energy and reduces peak demand. The five major impact variables studied were annual kwh savings, summer and winter peak kw reduction, percent peak savings and operating hours. For each applicant, these evaluated savings criteria were compared with the estimated savings. By comparing these evaluated results with the tracking analysis that was performed at the time of the incentive application, a realization rate for each variable is determined. These realization rates will allow National Grid to better approximate the energy savings and peak demand reductions on future compressed air applications. Whenever possible, the sources of discrepancies between tracking and evaluated results are described and the impacts are quantified. Supporting appendices include a site-specific summary for each measure. Each impact variable comparison table uses the key in Table 2 to provide likely reasoning for discrepancies. The summary pages in the appendix are listed in ascending order by application number. Several sites have both a compressor and a dryer. For these sites, the summary sheets are next to each other. The summary sheets show what information was gathered to make the annual energy calculations and savings estimates. These sheets also show how the savings accumulated differently between the old definition of on-peak and summer peak hours (8am- 9pm and 11am-3pm Mon-Fri) and the new definition of on-peak and summer peak hours (6am- 10pm and 3pm-5pm Mon-Fri). The old peak hour definition results are compared to the tracking results in the same manner as the new peak hour definitions. To avoid confusion, these results using the old definitions can be found in the appendix. Table 3 compares the existing realization rates with the realization rates calculated in this study for both the old and new peak hours definitions. The differences in summer kw and on-peak savings can be linked to the change in the summer coincident hours and on-peak hours. Also, several of the sites that were evaluated had little or no operations during the winter kw hours. This significantly reduced the evaluated winter kw realization rate. The kwh savings for the evaluated dryer applications are high mainly because two of the sites were operating the dryers 24 hours a day where the application only accounted for part-year operation. 7 (JTF)

All Evaluated Compressors All Evaluated Dryers Existing Old Definition New Definition Old Definition New Definition Realization Rates Realization Rates Realization Rates Realization Rates Realization Rates Summer kw 0.9 0.78 0.69 1.47 1.37 Winter kw 0.9 0.47 0.45 1.11 1.10 kwh 1.0 1.06 1.06 2.36 2.40 On-Peak % Savings 0.5 1.12 1.36 1.31 1.44 Operating Hours - 1.34 1.34 1.70 1.70 Table 3 Existing and Evaluated Realization Rates Also calculated was the average kw savings/hp for compressors and average kw savings/cfm for dryers. While quantifying savings on this basis can be helpful for quick calculations, caution should be taken. The kw savings is highly dependent on the load profile for each site. Using the results from this evaluation, the average kw savings/hp for compressors is 0.152 and the average kw savings/cfm for dryers is 0.00430. Table 4 shows a comparison between the existing National Grid savings rates and the rates calculated with the evaluation sites. These calculated savings for compressors were significantly lower than what National Grid is currently using, while the average for dryers is higher. However, the sample sizes for each category are relatively small. Given the small sample size, it is recommended that evaluation results be reflected in a change to the realization rate, rather than changing the savings factors themselves. NGRID kw Reduction Evaluated Reduction Measure Type kw Savings/hp kw Savings/hp Sample Size L/NL Compressor 0.102 0.0733 4 Variable Displacement Compressor 0.116 0.0424 1 VFD Compressor 25hp and Greater 0.206 0.1908 11 VFD Compressor 15hp- 25hp 0.207 N/A 0 Measure Type kw Savings/CFM kw Savings/CFM Cycling Dryer 0.00329 0.0044 6 VFD Dryer 0.00329 0.00373 1 Table 4 Gross Savings Factors Comparison Compressor Results Gross Annual Energy Savings The gross annual energy savings is a comparison between the base case compressor and the application installed compressor. These installed compressors are variable speed, load / no load or variable displacement. The base case for each site is a modulating compressor with blowdown of the same horsepower as the installed compressor. The overall gross annual kwh savings realization rate for all compressors is 1.06. A range between 32% and 309% was seen throughout the compressors. 8 (JTF)

2005 Compressed Air Compressor Tracking Results Evaluation Results Evaluated / Tracking Explanation Compressor Evaluation Size/Type Energy Savings Energy Savings Energy Savings Code Site Application Number kwh kwh % D2 503296 40 hp L/NL 9,120 19,276 211% OH-1 D2 504532 60 hp VSD 55,620 112,281 202% OH-1, OE-1 D2 504385 75 hp VSD 33,990 105,132 309% OH-1, LP-1 D2 500028 20 hp L/NL 6,080 4,221 69% OH-1, OE-1 D2 501875 50 hp VSD 28,325 27,143 96% OE-1 EI 500159 75 hp Var Disp 28,884 9,281 32% LP-1, OH-1 D2 500485 50 hp VSD 71,250 34,782 49% LP-1 D2 503894 30 hp VSD 15,650 27,143 173% LP-1 D2 500124 40 hp L/NL 21,341 19,559 92% OH-1 EI 500135 50 hp VSD 51,953 29,485 57% LP-1 D2 502470 40 hp VSD 99,864 59,792 60% LP-1 D2 503559 60 hp VSD 82,268 77,828 95% - D2 504072 15 hp L/NL 2,850 1,564 55% LP-1 D2 500767 25 hp VSD 49,020 47,085 96% - EI 500181 50 hp VSD 89,610 62,411 70% LP-1 D2 501435 50 hp VSD 95,760 36,204 38% LP-1, OH-1 Total/Average 741,585 673,187 106% Table 5 Compressor Energy Savings Comparison Peak Energy Savings Overall, the peak savings proved to be higher than the 50% average that was assumed during the time of the application submittal. This increase in peak savings is mostly due to the profile of each customer s operating hours (PH-2). Facilities that operate routinely on first shift will see a majority of their savings during peak hours. The new peak hours, weekdays 6:00 AM to 10:00 PM, were used to calculate the peak savings. 2005 Compressed Air Compressor Tracking Results Evaluation Results Evaluated / Tracking Explanation Compressor Evaluation Size/Type Peak Savings Peak Savings Peak Savings Code Site Application Number % % % D2 503296 40 hp L/NL 50% 48% 96% - D2 504532 60 hp VSD 50% 62% 123% OE-1, OH-1, PH-2 D2 504385 75 hp VSD 50% 52% 104% PH-2 D2 500028 20 hp L/NL 50% 99% 199% PH-2 D2 501875 50 hp VSD 50% 100% 200% PH-2 EI 500159 75 hp Var Disp 50% 90% 181% PH-2 D2 500485 50 hp VSD 50% 46% 91% OH-2 D2 503894 30 hp VSD 50% 100% 200% OH-1, PH-2 D2 500124 40 hp L/NL 50% 35% 70% PH-2 EI 500135 50 hp VSD 50% 66% 132% PH-1, PH-2 D2 502470 40 hp VSD 50% 49% 98% - D2 503559 60 hp VSD 50% 66% 133% PH-1, PH-2 D2 504072 15 hp L/NL 50% 99% 198% PH-1, PH-2 D2 500767 25 hp VSD 50% 48% 95% - EI 500181 50 hp VSD 50% 69% 138% PH-1 D2 501435 50 hp VSD 50% 61% 122% PH-1, PH-2 Total/Average 50% 68% 136% Table 6 Compressor Peak Savings Comparison 9 (JTF)

Peak Demand Reduction The summer and winter peak demand reduction averaged below the tracking calculations. Overall, the summer demand reduction was closeer to the tracking value than the winter demand reduction. Several sites had little or no second shift operation (PH-2). Sites that showed a significant increase in peak reduction often had oversized equipment (OE-1) or the listed application profile was different than the metered profile (LP-1). 2005 Compressed Air Compressor Tracking Results Evaluation Results Compressor Evaluation Size/Type Coincident Demand Savings Coincident Demand Savings Site Application Number Summer kw Winter kw Summer kw Winter kw D2 503296 40 hp L/NL 4.6 4.6 2.2 2.2 D2 504532 60 hp VSD 12.4 12.4 9.4 11.6 D2 504385 75 hp VSD 15.5 15.5 15.9 17.0 D2 500028 20 hp L/NL 1.5 1.5 1.6 0.0 D2 501875 50 hp VSD 10.3 10.3 5.6 0.0 EI 500159 75 hp Var Disp 7.0 7.0 1.1 0.0 D2 500485 50 hp VSD 11.4 11.4 3.4 4.3 D2 503894 30 hp VSD 6.8 6.8 12.5 0.0 D2 500124 40 hp L/NL 4.6 4.6 1.6 1.6 EI 500135 50 hp VSD 10.3 10.3 3.9 1.6 D2 502470 40 hp VSD 11.4 11.4 7.2 7.2 D2 503559 60 hp VSD 12.4 12.4 13.6 13.5 D2 504072 15 hp L/NL 1.1 1.1 0.1 0.0 D2 500767 25 hp VSD 5.7 5.7 5.4 5.4 EI 500181 50 hp VSD 10.3 10.3 9.6 11.3 D2 501435 50 hp VSD 11.4 11.4 5.0 0.0 Totals 136.6 136.6 98.3 75.7 Table 7 Compressor Peak Demand Reduction Results 2005 Compressed Air Compressor Evaluated / Tracking Explanation Compressor Evaluation Size/Type Coincident Demand Savings Code Site Application Number Summer % Winter % D2 503296 40 hp L/NL 49% 48% D2 504532 60 hp VSD 76% 94% LP-1, OE-1 D2 504385 75 hp VSD 103% 110% LP-1, PH-2 D2 500028 20 hp L/NL 106% 1% OE-1, PH-2 D2 501875 50 hp VSD 54% 0% PH-2 EI 500159 75 hp Var Disp 16% 0% LP-1, PH-2 D2 500485 50 hp VSD 30% 37% LP-1, OE-1 D2 503894 30 hp VSD 183% 0% OE-1, PH-2 D2 500124 40 hp L/NL 36% 35% LP-1 EI 500135 50 hp VSD 38% 16% LP-1 D2 502470 40 hp VSD 63% 63% LP-1 D2 503559 60 hp VSD 110% 109% LP-1, PH-1 D2 504072 15 hp L/NL 13% 0% PH-2 D2 500767 25 hp VSD 96% 95% LP-1 EI 500181 50 hp VSD 94% 110% PH-2 D2 501435 50 hp VSD 44% 0% PH-2 Average 69% 45% Table 8 Compressor Peak Demand Reduction Comparison 10 (JTF)

Operating Hours Table 9 shows the difference in application operating hours versus metering hours. Overall, the metered operating hours were higher. This could be due to equipment left on overnight or a compressed air demand of 10% or more was required during the overnight hours. 2005 Compressed Air Compressor Tracking Results Evaluation Results Evaluated / Tracking Explanation Compressor Evaluation Size/Type Code Site Application Number Hours Hours Hours % D2 503296 40 hp L/NL 2000 8760 438% D2 504532 60 hp VSD 4500 6779 151% OH-1 D2 504385 75 hp VSD 2200 6961 316% OH-1 D2 500028 20 hp L/NL 4000 2201 55% OH-1 D2 501875 50 hp VSD 2750 2607 95% - EI 500159 75 hp Var Disp 4150 2920 70% OH-1 D2 500485 50 hp VSD 6250 8760 140% OH-2 D2 503894 30 hp VSD 2288 2607 114% OH-1 D2 500124 40 hp L/NL 4680 6177 132% OH-1 EI 500135 50 hp VSD 5044 5162 102% - D2 502470 40 hp VSD 8760 8760 100% - D2 503559 60 hp VSD 6656 7430 112% OH-1 D2 504072 15 hp L/NL 2500 1647 66% OH-1 D2 500767 25 hp VSD 8600 8760 102% - EI 500181 50 hp VSD 8700 8760 101% - D2 501435 50 hp VSD 8400 4276 51% OH-1 Total/Average 134% Table 9 Compressor Operating Hours Comparison Compressed Air Dryer Results Six of the seven dryer measures were cycling dryers. Site 502100 has a variable frequency drive dryer. The base case for all dryers is a fully-loaded non-cycling dryer of the same capacity as the installed dryer. Gross Annual Energy Savings Overall, the energy savings realization rate for the dryer measures was 2.40. This rate is skewed by sites 503286 and 504074. These two sites had significantly higher kwh savings due to lowload profiles (LP-1) and the dryers operating more hours that indicated on the application (OH- 1). The dryer at site 504074 also operated during the overnight hours at more than 10% (LP-2). This increases the savings over the base case dryer. Based on the hourly averages during the metering period, four of the dryers appeared to be oversized (OE-1). A low-load profile prevailed throughout (LP-1). 11 (JTF)

2005 Compressed Air Tracking Results Evaluation Results Evaluated / Tracking Explanation Dryer Evaluation Energy Savings Energy Savings Energy Savings Code Site Application Number kwh kwh % D2 503296 1,264 7,934 628% LP-1, OH-1 D2 503894 4,294 996 23% OE-1, OH-1 D2 502470 6,920 12,352 179% LP-1 D2 502100 9,466 13,065 138% OE-1, LP-1 D2 504072 889 839 94% LP-1,OE-1, OH-1 D2 504074 1,077 5,880 546% LP-1, LP-2 OE-1, OH-1 D2 501094 1,867 1,405 75% LP-1, OH-1 Total/Average 25,777 42,471 240% Table 10 Dryer Gross Energy Savings Comparison On-Peak Energy Savings The on-peak energy savings for the dryer measures closely matched the tracking assumption of 50% at four of the seven sites. The three sites with above average peak energy savings were the result of the majority of the operating hours during peak periods only (PH-2). The change in peak hours to include the hours 6:00 AM through 8:00 AM and 9:00 PM to 10:00 PM also increased the overall peak usage and savings. 2005 Compressed Air Tracking Results Evaluation Results Evaluated / Tracking Explanation Dryer Evaluation Peak Savings Peak Savings Peak Savings Code Site Application Number % % % D2 503296 50% 52% 104% - D2 503894 50% 100% 200% PH-1, PH-2 D2 502470 50% 47% 94% - D2 502100 50% 49% 99% - D2 504072 50% 100% 200% PH-1, PH-2 D2 504074 50% 54% 108% - D2 501094 50% 100% 200% PH-1, PH-2 Average 50% 72% 144% Table 11 Dryer Peak Savings Comparison Peak Demand Reduction The reduction in peak demand was higher than the calculated tracking value on average. Overall, with the exception of two sites (503894 and 504074), both winter and summer averages were in agreement with each other. This shows that while the loading profile was constant between the summer and winter, it was much lower than the tracking assumption (LP-1). Another cause of increased demand reduction is that the equipment is oversized compared to the system requirements (OE-1). Oversized equipment at low loads will increase the average kw difference from the base case. 12 (JTF)

2005 Compressed Air Tracking Results Evaluation Results Evaluated / Tracking Explanation Dryer Evaluation Coincident Demand Savings Coincident Demand Savings Coincident Demand Savings Code Site Application Number Summer kw Winter kw Summer kw Winter kw Summer % Winter % D2 503296 0.63 0.63 1.05 1.00 167% 159% LP-1 D2 503894 0.50 0.50 0.56 0.00 112% 0% PH-2, OE-1 D2 502470 0.79 0.79 1.37 1.39 174% 176% LP-1 D2 502100 1.09 1.09 1.54 1.57 142% 144% LP-1, OE-1 D2 504072 0.36 0.36 0.27 0.00 76% 0% LP-1, OE-1 D2 504074 0.36 0.36 0.77 0.78 214% 216% LP-1, OE-1 D2 501094 0.47 0.47 0.33 0.35 72% 76% LP-1 Total/Average 4.19 4.19 5.91 5.09 137% 110% Table 12 Dryer Peak Demand Reduction Comparison Operating Hours The operating hours realization of the facility was again skewed by the increase in metered hours versus application hours (OH-1) of sites 503296 and 504074. Metered data showed that these two dryers, along with site 501094, were operating during overnight hours (OH-2). The average of the other five sites is 97%. 2005 Compressed Air Tracking Results Evaluation Results Evaluated / Tracking Explanation Dryer Evaluation Code Site Application Number Hours Hours Hours % D2 503296 2000 8760 438% OH-1, OH-2 D2 503894 8544 2447 29% OH-1 D2 502470 8760 8760 100% - D2 502100 8700 8760 101% - D2 504072 2500 2738 110% OH-1 D2 504074 3000 7926 264% OH-1, OH-2 D2 501094 4000 5944 149% OH-1, OH-2 Total/Average 170% Table 13 Dryer Operating Hour Comparison 13 (JTF)

Recommendations for National Grid Technical Reviewers 1. The major source of discrepancy in evaluated results from original savings estimates are differences in actual operating schedules from those predicted in the evaluation. The applicant should specifically provide a more detailed (actual and expected) schedule of operation. This information should include actual daily production hours. Certain processes require the compressed air equipment to remain online overnight. This information should be taken into consideration when making annual operating hour and average load assumptions. Better knowledge of operation hours will provide more accurate tracking calculations for all of the impact variables. Overall, the more complete the air demand information on the application, the more accurate the savings calculations. 2. More accurate loading profiles of existing equipment will increase tracking accuracy. Customers may be able to work with equipment vendors to track actual compressed air demand. This will help correctly size new equipment. If equipment is to remain online overnight, an overnight load profile may increase savings estimate accuracies. A more accurate load profile will also increase the accuracy of all of the impact variables. 3. Peak savings should be derived from the actual schedule of operation. Customers that primarily operate during first and second shifts will realize a majority of their savings during peak hours only. 4. Peak demand reduction should also be derived from the actual schedule of operation. Facilities that only operate on first shift will not see any demand reduction during the winter peak period 5. It appears that the gross savings factors for compressors that are currently being used are significantly greater than the savings factors calculated for this study. The factors for dryers currently being used are lower than the factors calculated savings for this study. Given the small sample sizes in the study, it is recommended that evaluation results be reflected in a change to the realization rates, rather than changing the savings factors themselves. However, with the evaluated realization rates of the compressor and dryer measures being significantly different, it is suggested that separate realization rates be used for compressors and dryers. Recommendations 1 through 4 can be completed with a table or chart within the application with the daily hours of interest (on-peak, off-peak, summer and winter demand) grouped together. The applicant can list percent-on and average percent full-load for each period. This will give an approximate distribution of peak energy savings and demand reduction. Table 14 is an example of a chart that, if filled out completely, will provide the equipment profiles to accurately estimate savings. 14 (JTF)

Comp 1 Comp 2 Comp 3 Compressor Model Compressor Type Compressor Size (hp) Daily Hours % On % Loaded % On % Loaded % On % Loaded 12am-6am 6am-3pm 3pm-5pm 5pm-7pm 7pm-10pm 10pm-12am Table 14 Example Chart for Prescriptive Compressed Air Application Currently, there are no guidelines to follow for listing these profiles. There is only a location for annual operating hours and average percent-load per machine. Increasing this level of detail will increase the savings estimates accuracy. Please feel free to contact me if you have any comments or questions associated with this study. Sincerely, John T. Forester Demand Management Institute 450 Lexington St. Newton, MA 02466 617-527-1525 x109 15 (JTF)