Printer Friendly

Analysis of measurement and verification methods for energy retrofits applied to residential buildings.


In 2007, a detailed evaluation was performed to assess the effectiveness of energy conservation measures and renewable energy technologies for 30 low income housing units in Center, CO, operated by the Center Housing Authority (CHA) (Guiterman, 2007). In particular, energy audits were performed including walk through audits, blower door tests and infrared scans. Historical utility data for all units was pre-screened and analyzed to determine baseline energy use. Moreover, six representative units were modeled using detailed whole-building energy simulation tool and calibrated against the historical data.

The calibrated building energy models were used to simulate numerous energy conservation measures (ECMs). Recommended ECMs included reducing hot water heater set points, installing programmable thermostats, replacing light bulbs with CFLs, adding additional roof insulation and repairing existing insulation, insulating existing water heater tanks/pipes, installing faucet aerators, replacing furnaces with high efficiency units and installing tankless water heaters in 20 units. Reducing infiltration through leaky mechanical closets in the one- and two-bedroom units was given a high priority and was achieved by installing direct vented furnace and water heater systems in these units.

Ground source heat pumps (GSHP) and solar hot water panels were also included in the original analysis. Due to funding constraints none of these options were implemented. Nearly all of the ECMs were installed by the Energy Resource Center (ERC) of Colorado (ERC, 2010), an agency under contract to the state to implement the federally-funded weatherization assistance program (WAP). The tankless water heaters were installed by a private plumbing contractor. The ECMs were implemented in June 2007 and are shown in Table 1.
Table 1 Installed Energy Conservation Measures

No. of Bedrooms            1    2    3     4
Area ([ft.sup.2])        600  750  950  1100
Number of Units           10    8   10     2
New Furnace Installed      X
Existing Furnace Tune-Up        X    X     X
Programmable Thermostat         X    X     X
Tankless Water Heater      X    X
CFL Bulbs                  X    X    X     X
Roof Insulation Repaired   X    X    X     X
Roof Insulation Added      X    X    X     X
Reduce DHW Temp to 120F    X    X    X     X

Measurement and verification (M&V) of the energy savings follows the ASHRAE Guideline 14-2002, Measurement of Energy and Demand Savings protocol (ASHRAE, 2002). This protocol was chosen due to the clarity of the analysis procedure and broad acceptance of the guideline within the energy efficiency community.

The measurement and verification (M&V) of the energy savings due to the implemented ECMs began in 2008 with on-site data collection and was completed in 2010 with two full years of utility data. Post-retrofit blower door tests were completed and data logging captured temperature settings in several units. On-site verification and occupant/staff interviews confirmed the installations of the specified measures.


Blower door tests were performed on one- and two-bedroom units before and after the retrofits. Table 2 summarizes the results of the leakage tests and indicates that infiltration was reduced by 61% and 51% for the one- and two-bedroom units, respectively.
Table 2 Results of Pre- and Post-Retrofit Blower Door Tests
(Air Changes per Hour)

Unit      Pre-Retrofit (ACH)  Post-Retrofit (ACH)  % Change

Unit 1-A                0.89                 0.35       61%
Unit 2-A                0.67                 0.33       51%

Temperature data loggers were placed in four (4) units as well as two (2) exterior locations for a period of approximately 40 days from February 14, 2008 through March 25, 2008. Two exterior loggers were placed for redundancy in case of equipment failure or other unforeseen problems. Temperature monitoring was completed in February and March.

Figure 1 provides an example of a weekly temperature profile for a one-bedroom unit. This unit has no programmable thermostat and the monitoring data shows that the tenant maintains a very warm and constant temperature of approximately 78 [degrees] F


Figure 2 shows the weekly temperature profile for a two-bedroom unit that received a programmable thermostat as an ECM. The tenant maintains the thermostat at a constant temperature of 70 [degrees] F at all hours of the day and night, indicating that none of the automated capabilities of the thermostat are being employed.



Two M&V analysis procedures were utilized in this study and include the calibrated simulation approach and the whole building approach, both described in ASHRAE Guideline 14. Two methods are employed for the whole building approach, both using regression analysis of pre- and post-retrofit utility data against government-reported weather data. The first method correlates energy use to average outdoor temperature, referred to as the temperature-based method, and the second correlates energy use to heating degree days, referred to as the degree day-based method.

Calibrated Simulation Approach

eQuest software was utilized in the original pre-retrofit analysis to determine the most cost effective ECMs to recommend for implementation. Seven units were modeled to reflect the range of unit types. Three are one-bedroom units in a triplex building, with individual electric meters but one shared gas meter. Two two-bedroom units and two three-bedroom units were also modeled.

Table 3 shows the annual calibration errors for electricity and gas use for the six units modeled. All models are calibrated to two-years of pre-retrofit utility data. Annual errors are within 5% of the utility data and the coefficients of variation of the root mean squared error (CV (RMSE)) are less than 15%, the maximum value specified in ASHRAE Guideline 14-2002. The energy model for triplex unit was challenging to calibrate due to the fact that only one of the three occupants of the triplex was available for pre-retrofit interviews so there is more uncertainty in the occupancy behavior and schedule patterns for two of the units. The whole building gas meter also adds additional complexity to the calibration process, as there are three separate models that need to be independently adjusted to match the utility data.
Table 3 Calibration Errors for Modeled Units

                Electricity           Natural
Unit ID         Annual Error    CV    Annual   CV (RMSE)
                              (RMSE)  Error

1-A                    -1.9%    6.3%      N/A        N/A
1-B                     3.5%    6.2%      N/A        N/A
1-C                    -2.3%    6.7%      N/A        N/A
Triplex                  N/A     N/A    -4.6%      10.6%

(1-A/1-B/1-C) (1)
2-A                    -1.2%    6.3%    -2.0%       6.3%
2-B                    -3.5%    5.9%    -1.9%      10.1%
3                       0.2%    3.0%    -0.4%       7.2%
4                       2.2%    3.8%    -3.8%       8.7%

All available information from utility bills, occupant interviews, walk-through audits and blower door tests were utilized to accurately calibrate the energy models. Figure 3 shows the monthly natural gas consumption for the energy model compared to historical utility data.


Each ECM was simulated individually using the energy models and all ECMs were simulated as one "bundled" package of ECMs to arrive at the total estimated savings and to capture any interactive effects (such as an increased heating load due to reduced electricity use with CFL bulbs).

Whole Building Approach

The temperature-based method provides information about the base water-heating load, the fuel utilized for space heating, the building load coefficient (BLC) and the balance temperature of the house (Krarti, 2010). This method was also used in the original pre-retrofit analysis to estimate the base water heating load and calibrate the energy models. For the M&V analysis, it is used to estimate natural gas savings from ECM implementation, to provide a calculated balance temperature for the pre- and post-retrofit periods and to calculate the BLC for the pre- and post-retrofit periods for comparison.

The degree day-based method is based on the relationship between natural gas use and heating degree days, resulting in a linear model that can be used to project energy use assuming no retrofit took place and a model to project post-retrofit gas use. This method is commonly referred to as the PRISM (Princeton Scorekeeping) method (Princeton, 2010).

Temperature-Based Method. The temperature-based method utilizes monthly outdoor average temperature readings for Alamosa, Colorado sourced from the National Oceanic and Atmospheric Agency (NOAA) National Climactic Data Center (NCDC) online databases (NCDC, 2010). Monthly utility data is normalized over the days per month to derive a "monthly daily" fuel use per day in MMBtu/day (Million Btu/day). This daily fuel use is then plotted against the average outdoor temperature for each month.

An example of the results of the regression analysis is shown in Figure 4 for both pre- and post-retrofit periods. Baseload gas use for water heating is reduced for this unit, and the more gradual slope of the heating line in the post-retrofit case shows that demand for heating has decreased due to the retrofit measures.


The intersection of the regression line for the winter gas use with the base-load represents the balance temperature of the house. The balance temperature represents the outdoor temperature below which space heating is required. The slope of the winter regression line allows for the BLC to be calculated using Equation 1 and the annual fuel used for heating is then determined from Equation 2 (Krarti, 2010).

BLC = Slopex [eta.sub.heating]/24

E.sub.heating = x BLCx d [d.sub heating]/[eta.sub.heating]

Energy savings are calculated for the temperature-based method using the following equations:

%SAVINGS = 1-Epost/epre %Svings = Epost/epre (3)

Epre = [(24xBLCprex DD/ [eta.sub.PRE) + DH[W.sub.Summerprex12]) (4)

[] = [(24xBL[]xDD/[])] + (DHW summerpostx12)] (5)

Degree Day-Based Method. The DD-Based method plots fuel use per day vs. daily heating degree days (monthly total divided by number of days per month) for each month of the pre- and post-retrofit periods. Unlike the temperature-based method, all months including winter, summer and swing months are used in the analysis. The pre- and post-retrofit period degree days are a function of the balance temperature selected. The calculated balance temperatures resulting from the temperature-based method described above provide a starting point for the regression analysis. The balance temperature can then be adjusted to ensure the highest [R.sup.2] value for each regression. This is also referred to in Kalinic (2009) as the "best" balance temperature. Figure 5 shows an example of the natural gas regression analysis for the degree day-based method.


The slope and intercept from the linear equation describing the correlation between energy use and heating degree days for each period provides the inputs to the model that projects baseline energy use and post-retrofit energy use as shown in Figure 6.


The degree day-based method allows for the calculation of energy savings following the ASHRAE Guideline 14 (ASHRAE, 2002) protocols, by subtracting the projected post-retrofit energy use from the projected baseline energy use, as shown in the equation below.

Energy Savings = [E.sub.projected baseline]--[E sub.projected post]

Uncertainty Calculations

The method utilized for quantifying uncertainty in calculating the energy savings follow the protocols described in ASHRAE Guideline 14 for fractional uncertainty in savings measurements (ASHRAE, 2002). Guideline 14 requires that fractional uncertainty be less than 50% for the whole building approach at a 68% confidence level. The savings calculations presented in this analysis all meet the fractional uncertainty limit at the 68% confidence level as well as at the 95% confidence level. The key equations used in calculating uncertainty in both the whole building approach models and calibrated simulation approach models are shown below.

CVRMSE = {[SIGMA]X[y.sub.i] [sub.2]}/(n-p)] {[sup.2/1]}/y (7)



Figure 7 shows average savings estimates for each unit type and M&V analysis method. For the one-bedroom units, the temperature-based method under-predicted savings relative to the degree day-based method and the calibrated simulation approach by 2% and 10%, respectively. Savings estimates range from 20% to 29%.

For the two-bedroom units, the temperature-based method and the degree day-based method estimated savings at 25% and 27%, respectively. Both methods over-predicted savings relative to the calibrated simulation approach by approximately 4% to 6%. Savings estimates for all methods for the two-bedroom units range from 21% to 27%.

The one-bedroom units received new tankless water heaters and furnace replacements in addition to general weatherization. The facility manager elected not to install programmable thermostats in the one-bedroom units, which are reserved exclusively for elderly tenants, due to concerns about tenants' abilities to utilize the new devices properly.[3] ECM costs were over $2,700 per unit, and taking the degree day-based estimate of a 22% reduction in annual natural gas use, simple paybacks are nearly 18 years. Figure 8 shows the estimated gas savings by method for the one-bedroom units (three gas meters serving ten units).


The two-bedroom units received new tankless water heaters in addition to general weatherization and programmable thermostats. Utilizing the degree day-based estimate of 27%, these units save an average of approximately $200/year on gas costs and received just over $2,000 of ECMs per unit, resulting in a simple payback of 10 years. It is interesting to note that these units saved more than the one-bedroom units which received new furnaces but not programmable thermostats. Figure 9 shows the estimated gas savings by method for the two-bedroom units.


The three-bedroom unit analysis reveals that the temperature-based method and the degree day-based method estimated savings at identical levels, 17%. Both methods over-predicted savings relative to the calibrated simulation approach by approximately 6%. These units received general weatherization including programmable thermostats. The three-bedroom units obtained the most cost-effective results, with savings of 17% of pre-retrofit natural gas use and ECM costs of only $361/unit, resulting in a payback just over 2 years. Figure 10 shows the estimated gas savings by method for the three-bedroom units.

Figure 11 shows the estimated savings by each method averaged over all CHA units. The calibrated simulation approach and whole building approach methods estimated similar levels of total gas savings for all units, with all three approximating 19% - 21% savings, representing a total range of only 2%. The temperature-based method and degree day-based method both predicted savings within one approximately percent of the other, at 20% and 21%, respectively, while the calibrated simulation approach estimated savings of 19%.


Based on these findings, verified or actual natural gas savings across all unit types at the Center Housing Authority are between 17% and 24% at a 95% confidence level for the two whole-building approach methods. Verified natural gas savings for the calibrated simulation approach are between 14% and 24% at the 95% confidence interval.

The results indicate that units receiving programmable thermostats achieved cost-effective savings, as the three-bedroom units received inexpensive measures yet reached average savings of 17%. The energy savings from the new furnaces may not be fully realized as these units did not also install programmable thermostats. It would be worthwhile to install programmable thermostats in a select number of one-bedroom units and monitor energy use to compare to the units that did not receive new thermostats. Tankless water heaters are clearly expensive systems that can generate significant energy savings, yet payback periods for the units with tankless heaters are still greater than 10 years.



Simulation models in this analysis are performed using typical meteorological year (TMY2) weather files (NSRDB, 2010). The calibration for the models was completed with approximately two years of pre-retrofit data, which typically can help in reducing the impact of the variability in weather patterns from year to year.

To compare the results of the models using TMY data to actual weather data, a weather file was created for the calendar year 2008 based on Alamosa weather data. A comparison of the predicted solar radiation using the weather file and the measured solar radiation is shown in Figure 12, which validates the weather file as an accurate depiction of actual weather conditions.


Calendar year 2008 was the first full year that the ECMs were in place, since the retrofits were completed during summer 2007. The simulation model energy use represents the consumption with all ECMs included and is compared against actual utility data for those months. The energy models were not re-calibrated using the 2008 (post-retrofit) actual weather data. These models were calibrated to TMY weather data and include all the ECMs described earlier.

Table 4 shows the average error across all the units when comparing the simulated results using 2008 weather data to the actual utility billing data. The utility data is taken as the real value, so a negative value indicates that the simulation results under-predicted consumption relative to actual utility data.

Overall, the simulation models estimated electricity use at 18% below the actual utility data [(Guiterman, 2010) sup.5] Gas consumption is over predicted by the model by an average of 9% compared to actual utility data, indicating that the utility data shows greater savings for this year than the savings estimated by the energy models.. The average energy use intensity (kBtu/f [t.sup.2]) for the units was over predicted by the calibrated models by approximately 3%.
                         1    2    3

Temperature-Based      20%  25%  17%
DD-Based               22%  27%  11%
Calibrated Simulation  29%  21%  11%
Figure 7 Estimated Gas Savings by Analysis Method and Unit Type

                       118 Triplex-North  Fourplex  156 Triplex-South

Temperature-Based                    17%       29%                15%
DD-Based                             19%       28%                21%
Calibrated Simulation                29%       27%                29%

Figure 8 Gas Savings by Method - One-Bedroom Units

       Temperature-Based  DD-Based  Calibrated Simulation

710                  12%       17%                    21%
716                  32%       28%                    21%
163-A                43%       43%                    21%
163-B                20%       21%                    21%
696 B                27%       28%                    26%
696 A                18%       27%                    16%

Figure 9 Gas Savings by Method - Two-Bedroom Units

     Temperature-Based  DD-Based  Calibrated Simulation

765                39%       31%                    11%
362                24%       23%                    18%
143                 9%       10%                     4%
159                26%       22%                    11%
107                11%       13%                    11%
103                25%       24%                    11%
767                16%       17%                    11%
358                 6%        6%                    11%
125                 8%       11%                    11%
139                 9%       13%                    11%

Figure 10 Gas Savings by Method - Three-Bedroom Units (4)

Temperature-Based      20%
DD-Based               21%
Calibrated Simulation   19

Figure 11 Estimated Gas Savings by Analysis Method for All Units

Table 4 Relative Error Comparing Simulation of 2008 Weather to
Actual Utility Data

                Annual Electricity  Annual Gas    EUI
                        kWh             MMBtu     kBtu/SF
Relative Error         -18%                9%       3%

Figure 13 shows an example of the comparative analysis of the simulated results using 2008 weather data to the actual utility billing data using the monthly gas consumption for one of the units.



The measurement and verification analysis for the energy conservation measures installed at the Center Housing authority reveals that the units experienced significant natural gas savings. The calibrated simulation approach and the two whole-building approaches, the temperature-based and degree day-based methods, all estimated natural gas savings within a range of 2%, from 19% to 21%.

The calibrated simulation approach underestimated gas savings for the units receiving general weatherization and new tankless water heaters, yet overestimated savings for the units receiving new furnaces. Calibrated simulation is a costly and time-intensive analysis method, and the energy savings estimates are within the error bounds of both of the whole-building approach methods.

The temperature-based method requires the least amount of input weather data and processing to estimate energy savings, and yields results consistent with the degree day-based method for units receiving both HVAC and weatherization measures. It is therefore a viable and effective methodology of verifying the savings from energy conservation measures and would likely save time and reduce cost compared to both the degree day-based method and calibrated simulation approach.

The two-bedroom units achieved the highest savings, more than the one-bedroom units that received new 90% efficient, condensing furnaces. The three-bedroom units achieved the most cost-effective savings, receiving only general weatherization and programmable thermostats at a cost of approximately $360/unit and saving 17% of natural gas use or $160/year. Simple paybacks for these units are just over 2 years. The one- and two-bedrooms received over $2,000 per unit in retrofits and achieved savings ranging from 22%-27% ($155-$200/year) of pre-retrofit natural gas use, respectively. Paybacks ranged from 10years for the two-bedroom units to 18 years for the one-bedroom units.

Programmable thermostats can help tenants achieve significant energy savings when their capabilities are utilized properly. The temperature monitoring shows that some units clearly do not take advantage of the automated scheduling for the thermostats. However, the energy savings achieved by the three- and two-bedroom units prove that many tenants do utilize the setback capabilities. Tankless water heaters are shown to be an expensive measure that yields high paybacks, typically over 10 years.


K[eta]_heating = efficiency of heating system

DD_heating = the number of heating degree days at the balance temperature for this location

[E.sub.pre] = Pre-retrofit energy use [] = Post-retrofit energy use = Heating system efficiency

DD = Heating degree days at specified balance temperature

DH[W.sub.sumer] = Average fuel use for domestic hot water for June, July and August

24 = Conversion factor (hours/day)

12 = Conversion factor (months/year)

CVRMSE = Coefficient of Variation of Root Mean Squared Error

y = Measured monthly energy consumption

y^ = Projected monthly energy consumption

n = Number of pre-retrofit months

p = Number of model parameters (3-Whole Building; 1-Calibrated Simulation) y~ = Average monthly energy consumption for analysis period

U = Fractional uncertainty

t = T-statistic (per ASHRAE)

F = Percent savings estimated by model

m = Number of post-retrofit months

(1) This triplex consists of units 1-A, 1-B and 1-C, and has only one gas meter shared by all three units. The units have individual electric meters.

(2) Out of the thirty units in the CHA, there was one fourplex and two triplexes containing ten units reserved for elderly tenants. Each of these single level buildings shared one common gas meter but individual electric meters (e.g., 3 gas meters and ten electric meters). The graphs shown in this paper are derived from the analysis of the fourplex and triplex gas meters.

(3.) The current manual thermostats in these units were oversized and designed to assist those with vision difficulties. The facility manager was concerned that the tenants would not be able to properly use the new thermostats and this would result in a high frequency of service calls and tenant frustration. CHA was willing to consider installing programmable thermostats designed for the elderly at some point in the future.

(4) There is a wide scatter in savings estimates for the temperature based and degree day based methods, yet both simple averages and averages weighted by contribution to total energy savings result in average savings across all units of approximately 17%

(5) For brevity, analysis of verified electricity use is not included in this paper. The CFL replacements were the only measure directly impacting electricity use and the analysis found a wide and inconsistent variation in post retrofit electricity use. Several units showed higher electricity use and further investigation (including tenant and facility manager interviews) revealed no clear reason for the increases. For all applicable units, gas savings exceeded electricity increases by two to three times when calculated as BTU's of source energy use.


ASHRAE. 2002. ASHRAE Guideline-14, Measurement of Energy and Demand Savings. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Energy Resource Center (ERC). 15, October 2010.

Guiterman, T. 2007. Center Housing Authority Project: Energy Conservation Measures and Renewable Energy Analysis. Independent Study Report.

Guiterman, T. 2010. Center Housing Authority Project: Energy Efficiency Implementation and Measurement and Verification of Savings. Final Master's Project

Kalinic, N. 2009. Measurement and Verification of Savings from Implemented Energy Conservation Measures. Final Master's Project.

Krarti, M. 2010. Energy Audit of Building Systems, Second Edition. Boca Raton: CRC Press, 170-173. National Climactic Data Center (NCDC). 1, March 2010.

National Solar Radiation Data Base (NSRDB). 1, March 2010.

Princeton University. 2010. Princeton Scorekeeping Method. Program on Science and Global Security. Princeton.

Tim Guiterman, LEED [R] AP

Associate Member ASHRAE

Moncef Kararti, PhD., PE. PE, LEED [R] AP


Moncef Krarti is a professor in the Department of Civil, Environmental and Architectural Engineering, University of Colorado, Boulder, Colorado. 7LP Tim Guiterman holdsanMS from the Building Systems Program of the Civil, Environmental and Architectural Engineering, University of Colorado, Boulder and is a Senior Consultant at Navigant Consulting, Burlington, Vermont.

ASHRAE Transaction
COPYRIGHT 2011 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2011 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Guiterman, Tim; Krarti, Moncef
Publication:ASHRAE Transactions
Article Type:Report
Geographic Code:1U1VT
Date:Jul 1, 2011
Previous Article:Update procedures for commercial building energy audits.
Next Article:Photovoltaics on landfills in Puerto Rico.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |