· web view27 techniques and tips for retrocommissioning energy calculations, presentation at the...

A Conceptual Plan forEnergy-Related Measurement

& Verification of AdvancedRetro-Commissioning Technology Demonstration Projects

Craig Wray, Jessica Granderson, Guanjing Lin, Xiufeng PangBuilding Technology and Urban Systems Division

Lawrence Berkeley National Laboratory

Prepared for:

DOE Building Technologies Office

June 25, 2015

DISCLAIMERThis document was prepared as an account of work sponsored by the United States Government. While this document is believed to contain correct information, neither the United States Government nor any agency thereof, nor the Regents of the University of California, nor any of their employees, makes any warranty, express or implied, or assumes any legal responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by its trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof, or the Regents of the University of California. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof or the Regents of the University of California.

ACKNOWLEDGEMENTThis work was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Building Technology of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

Much of the material in this document is derived from or was provided by the International Performance Measurement and Verification Protocol, and documents published by the Federal Energy Management Program (FEMP) of the U.S. Department of Energy, and Quantum Energy Services and Technology (QuEST).

TABLE OF CONTENTS

1. Introduction.................................................................................................................................................... 42. Measurement and Verification: An Overview.....................................................................................53. M&V Options: Summary of Approaches.............................................................................................8

3.2 Option B: Retrofit Isolation With All Parameter Measurement...............................................................103.2.1 Approach to Option B........................................................................................................................................................10

3.3 Option C: Whole-Building Data Analysis........................................................................................................103.3.1 Approach to Option C........................................................................................................................................................113.3.2 Data Collection....................................................................................................................................................................11

3.4 Option D: Calibrated Simulation.........................................................................................................................123.4.1 Approach to Option D.......................................................................................................................................................133.4.2 Simulation Software...........................................................................................................................................................143.4.3 Model Calibration...............................................................................................................................................................14

4. Developing Regression Models for M&V...........................................................................................164.1 Independent Variables.............................................................................................................................................164.2 Choosing a Model.....................................................................................................................................................16

5. Selecting an M&V Approach: IPMVP Options A-D.......................................................................185.1 M&V Considerations for Option A....................................................................................................................185.2 M&V Considerations for Option B....................................................................................................................195.3 M&V Considerations for Option C....................................................................................................................195.4 M&V Considerations for Option D....................................................................................................................20

6. Developing an M&V Plan...................................................................................................................... 227. Considerations for Technology Demonstration Projects................................................................26References......................................................................................................................................................... 27List of Acronyms.............................................................................................................................................. 29Other References............................................................................................................................................. 30Bibliography..................................................................................................................................................... 33Appendix A: Site Selection Questionnaire.................................................................................................37Appendix B: M&V Data Analysis Tool......................................................................................................43

Integration with Universal Translator.......................................................................................................................44Modeling Method.............................................................................................................................................................45Uncertainty Method.........................................................................................................................................................45Other Tool Algorithms and Routines........................................................................................................................46Summary............................................................................................................................................................................. 46

1. IntroductionBuilding commissioning is a systematic process that can be used to reduce building energy use and to improve indoor environmental conditions for occupants. Retro-commissioning (RCx) is specific to existing buildings and includes identifying performance goals as well as deficiencies and improvement opportunities. It also includes implementing changes through tuning, low cost repairs, and more capital-intensive retrofits as needed; and using measurement and verification (M&V) techniques to verify that changes indeed improve building operation. Examples of advanced strategies and technologies include using utility data to identify candidate buildings, integrating building automation systems with fault detection tools to inform stakeholders of opportunities, and implementing automated control system adjustments.

This document describes a conceptual approach for validating the benefits of RCx-related technology innovations relative to current practice, which can be used as a foundation for developing subsequent site-specific M&V plans. More specifically, it provides an overview of M&V practices that could be used to assess energy savings associated with implementing these technologies and includes a methodology for comparing post-installation performance to baseline data. It also includes project hypotheses, technical objectives, specific indicators of success, and criteria needed to select optimum implementation sites for technology demonstrations. Specific criteria include at least the following: facility size and characteristics, number of locations required to develop generalizable conclusions, and required performance data. A demonstration site questionnaire that addresses these criteria is included as an Appendix to support follow-on site selection efforts.

2. Measurement and Verification: An OverviewEnergy savings themselves cannot be measured directly because they represent the absence of energy use. Instead, savings are determined by comparing the energy use before and after the installation of energy conservation measure(s). The “before” case is called the baseline; the “after” case is called the post-installation or performance period. Proper determination of savings includes adjusting for changes that affect energy use, but that are not caused by the implemented measure(s). Such adjustments account for changes in weather, occupancy, or other such factors that might differ between the baseline and performance periods. Equation 1 describes the general equation used to calculate savings:

Savings = (Baseline Energy - Post Installation Energy) ± Adjustments Eqn. 1

Equation 1 can be restated so that no adjustments need be made for the post-installation energy measurements. In that case, a regression model is developed from baseline energy use measurements and independent variables are used to determine “what baseline energy use would have been” under post-installation conditions. Similarly, both baseline and post-installation energy use may be restated to some set of conditions other than baseline or post-installation conditions.

M&V protocols to determine savings in energy conservation projects have existed since about 1995. Notable protocols include the International Performance Measurement and Verification Protocol (IPMVP) – Volume I, the Federal Energy Management Program (FEMP) M&V Guidelines for Federal Energy Projects, and ASHRAE Guideline 14.

IPMVP Volume I (2014) provides guidance in the form of a conceptual framework for measuring, computing, and reporting savings achieved by energy or water efficiency projects in buildings. It defines key terms and outlines issues that must be considered when developing an M&V plan, but does not provide details for specific measures or technologies. Developed through a collaborative effort involving industry, government, financial, and other organizations, the IPMVP document provides four M&V options, and addresses issues related to the use of M&V in third-party-financed and utility projects.

The FEMP M&V Guideline (2008) contains specific procedures for applying concepts originating in the 2007 version of IPMVP Volume I. The Guideline represents a specific application of the IPMVP for federal projects. It outlines procedures for determining M&V approaches, evaluating M&V plans and reports, and establishing the basis of payment for energy savings during the contract. These procedures are intended to be fully compatible and consistent with the IPMVP document.

ASHRAE Guideline 14 (2002) is a reference for calculating energy and demand savings associated with M&V activities. In addition, it sets forth instrumentation and data management guidelines and describes methods for accounting for uncertainty associated with models and measurements. It specifies three engineering approaches to M&V. Compliance with each approach requires that the overall uncertainty of the savings estimates be below prescribed thresholds. The three approaches presented in Guideline 14 are closely related to and support the options provided in IPMVP Volume I. Guideline 14, however, does not discuss issues related to performance contracting.

In general, M&V activities include site surveys, metering energy and independent variables, engineering calculations, and reporting. How these activities are applied to determine energy savings depends on characteristics of the energy conservation measures (ECMs) being implemented and balancing accuracy in energy savings estimates with the cost of M&V itself.

IPMVP Volume I lists four M&V protocol options that enable one to apply a range of suitable techniques for a variety of applications (summarized in Section 3 of this document, with excerpts from Chapter 4 of the FEMP Guideline):

Option A (Retrofit Isolation with Key Parameter Measurement).

Option B (Retrofit Isolation with All Parameter Measurement).

Option C (Whole Building).

Option D (Calibrated Simulation).

A simple graphical representation of the savings impact demonstrated through application of M&V is shown in Figures 1 and 2. These data were collected as part of a monitoring-based commissioning project (MBCx) at UC Berkeley’s Soda Hall. In this IPMVP Option B approach, energy use by the HVAC systems (chillers, pumps, air handlers) was measured for three months prior to improving the operational efficiency of these systems. As Figure 1 shows, a baseline energy model was developed with a simple linear regression of daily energy use versus ambient dry-bulb temperature.

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

40 45 50 55 60 65 70

Ambient Dry-Bulb Temperature

Dai

ly k

Wh

Use

Baseline Data Base Model Post Data Post Model

Figure 1. Scatter plot of daily HVAC energy use vs.ambient temperature (2-parameter model).

The resulting statistical indices (i.e. root-mean square error - RMSE) are shown in Figure 2. Ambient temperature and energy use data continued to be collected in the post-installation period, and the baseline energy use projected into the post-installation period using the model. Figure 2 shows the savings over the post-installation measurement period as the difference between the projected baseline model and the measured post-installation use.

Figure 2 is a powerful visualization of the M&V concept, and highlights its strengths in demonstrating savings to project and program sponsors. It is simple to understand, and provides several useful insights into the HVAC system energy use:

Demonstrates the dependence of HVAC energy use with ambient temperature.

Demonstrates magnitude of energy savings each day.

Creates new baselines for additional projects.

Compares post-installation model (orange line in post installation period) with post-installation energy use to provide an indication when energy performance is slipping, and can be used for on-going tracking purposes.

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

2/9/20

06

2/11/2

006

2/13/2

006

2/15/20

06

2/17/2

006

2/19/2

006

2/21/2

006

2/23/20

06

2/25/2

006

2/27/2

006

3/1/20

06

3/3/20

06

3/5/20

06

3/7/20

06

10/31

/2006

11/2/20

06

11/4/

2006

11/6/

2006

11/8/

2006

11/10

/2006

11/12

/2006

11/14/2

006

11/16

/2006

11/18

/2006

11/20

/2006

11/22/2

006

11/24

/2006

11/26

/2006

11/28

/2006

Date

Dai

ly k

Wh

Use

HVAC Daily kWh Usage Baseline Post-Install Model

Date Break

Baseline Model:kWh = 79.9*OAT + 1129RMSE = 136 kWh

Baseline Period Post-Installation Period

Post-Install Model:kWh = 44.1*OAT - 336RMSE = 213 kWh

Figure 2. Representation of the M&V concept over time.

3. M&V Options: Summary of ApproachesM&V approaches are divided into two general types. One type (retrofit isolation) considers only the affected equipment or system independent of what occurs in the rest of the building. The other type (whole-building) considers the total energy use and de-emphasizes specific equipment performance. IPMVP Options A and B are retrofit isolation methods; Option C is a whole-building method; Option D can be used as either, but is usually applied as a whole-building method. One primary difference between these approaches is where the boundary of the ECM is defined. All energy used within the boundary must be considered.

The four generic M&V options are summarized in Table 1. Each option has advantages and disadvantages based on site-specific factors and the needs and expectations of the stakeholders.

Retro-commissioning efforts (RCx) usually focus on HVAC operations (although lighting is sometimes included), with improvements spanning many pieces of equipment, and have been shown to generate median savings of 16%. Therefore, IPMVP Options B, C, and D are most useful for quantifying RCx energy savings. If lighting measures are implemented, Option A is often used to determine end-use specific savings.

Table 1: IPMVP M&V Options

M&V Option Performance and Usage Factors Savings Calculation Option A:Retrofit Isolation with Key Parameter Measurement

This option is based on a combination of measured and estimated factors when variations in these factors are not expected. Measurements are spot or short-term and are taken at the component or system level, both in the baseline and post-installation cases. Measurements should include the key performance parameter(s) that define the energy use of the ECM. Estimated factors are supported by historical or manufacturer’s data. Savings are determined by means of engineering calculations of baseline and post-installation energy use based on measured and estimated values.

Direct measurements and estimated values, engineering calculations, and/or component or system models are often developed through regression analysis. Adjustments to models are not typically required.

Option B:Retrofit Isolation with All Parameter Measurement

This option is based on periodic or continuous measurements of energy use taken at the component or system level when variations in factors are expected. Energy or proxies of energy use are measured continuously. Periodic spot or short-term measurements may suffice when variations in factors are not expected. Savings are determined from analysis of baseline and reporting period energy use or proxies of energy use.

Direct measurements, engineering calculations, and/or component or system models often developed through regression analysis. Adjustments to models may be required.

Option C:Utility Data Analysis

This option is based on long-term, continuous, whole-building utility meter, or sub-metered energy data. Savings are determined from analysis of baseline and reporting period energy data. Typically, regression analysis is conducted to correlate with and adjust energy use to independent variables such as weather, but simple comparisons may also be used.

Based on regression analysis of utility meter data to account for factors that drive energy use. Adjustments to models are typically required.

Option D:Calibrated Computer Simulation

Computer simulation software is used to model energy performance of a whole-building (or sub-building). Models must be calibrated with actual hourly or monthly billing data from the building. Implementation of simulation modeling requires engineering expertise. Model inputs include building characteristics; performance specifications of new and existing equipment or systems; engineering estimates, spot-, short-term, or long-term measurements of system components; and long-term whole-building utility meter data. After the model has been calibrated, savings are determined by comparing a simulation of the baseline with either a simulation of the performance period or actual utility data.

Based on computer simulation model (such as EnergyPlus) calibrated with whole-building or end-use metered data or both. Adjustments to models are required.

3.2 Option B: Retrofit Isolation With All Parameter MeasurementM&V Option B uses periodic or continuous metering of all energy quantities, or all parameters needed to calculate energy, during the performance period. This approach provides the greatest accuracy in the calculation of savings, but increases the performance-period M&V cost.

Option B is typically used when any or all of the following conditions apply:

For simple equipment replacement projects with energy savings that are less than 20% of total facility energy use as recorded by the relevant utility meter or sub-meter (Option C is not applicable).

When energy savings values per individual measure are desired.

When interactive effects can be estimated using methods that do not involve long-term measurements.

When the independent variables that affect energy use are not complex and excessively difficult or expensive to monitor.

When operational data on the equipment are available through control systems.

When sub-meters already exist that record the energy use of subsystems under consideration (e.g., a separate sub-meter for HVAC systems).

3.2.1 Approach to Option BOption B procedures rely on the physical assessment of equipment change-outs to ensure that the installation meets specifications. The potential to generate savings is verified through observations, inspections, and spot/short-term/continuous metering of energy or proxies of energy use (e.g., using variable frequency drive speed as a proxy for motor power). Baseline models are typically developed by correlating metered energy use or proxies with key independent variables. Depending on the ECM, spot or short-term metering may be sufficient to characterize the baseline condition, and the continuous metering of one or more variables may occur after retrofit installation. It is appropriate to use spot or short-term measurements in the performance period to determine energy savings when variations in performance are not expected, and may support some normalized savings approaches though adjustments to the baseline and/or the performance period model(s). When variations are expected, as in the case of retrocommissioning, it is appropriate to measure factors continuously. Continuous monitoring of information also can be used to improve or optimize equipment operation over time, thereby improving the performance of the retrofit.

3.3 Option C: Whole-Building Data Analysis Energy savings under Option C are estimated by developing statistically representative models of whole-building or sub-metered energy consumption (i.e., therms and/or kWh). This method confirms total energy savings, but does not measure the savings from individual components.

In general, Option C should be used with complex equipment replacement and controls projects for which predicted savings are relatively large, i.e., greater than about 10% to 20% of the building’s energy use, on a monthly basis. Depending on the building’s predictability, availability of more granular interval meter data, and uncertainty requirements, savings of less than 10% may be measurable using Option C. Option C regression methods are valuable for

measuring interactions between energy systems or determining the impact of projects that cannot be measured directly, such as insulation or other building envelope measures. Regression analysis requires experienced, qualified analysts, and Option C methods should be employed only for projects that meet the following requirements:

Savings are predicted to be greater than about 10% to 20% of the overall consumption measured by the utility or sub-meter.

At least 12 and preferably 24 months or more of pre-installation data are used to calculate a baseline model.

At least 9 and preferably 12 months of performance period data are used to calculate annual savings.

Adequate data on independent variables are available to generate an accurate baseline model, and procedures are in place to track the variables required for performance period models.

Significant operational or other changes are not planned for the facility during the performance period, and procedures are in place to document changes that do occur at the site.

Again, these guidelines may be relaxed if interval meter data are available, and depending on building predictability, and project uncertainty requirements.

3.3.1 Approach to Option CWith Option C, regression models are developed to predict energy use based on the appropriate independent variables for the project. Although simple mathematical techniques utilizing utility bill comparison are sometimes used, they tend to be unreliable and are not recommended. Regression models can account for impacts of weather and other independent variables on energy use, whereas simple utility bill comparison techniques cannot.

The Option C approach includes developing an appropriate baseline model, which relates the baseline energy use to key independent variables, and continuously measuring the performance period energy use and the key independent variables. Savings are often calculated by comparing the energy use predicted by the baseline model using measured conditions with the actual energy use of the performance period. Alternately, performance period models may be developed if the baseline and performance period models are to be adjusted to typical conditions prior to comparison. Performance period models may also be needed if there is not a full year’s worth of data available for the performance period.

3.3.2 Data CollectionCollecting, validating, and properly applying data are important elements of using utility data analysis. Option C techniques utilize three types of data: utility billing data, independent variables, and information on unrelated changes at the site.

Utility billing data provide the basis for savings calculations by allowing a comparison of adjusted baseline models with performance period energy use. Regardless of the type of utility data used, a key to properly applying the data is ensuring that all start and end dates of the utility data are aligned with those of the independent variables. Collecting data on independent variables more often than collecting billing data can help align time frames. Billing data can be:

Monthly billing data. Billing data should be measured at least once a month. There are typically two types of monthly billing data: total usage for the month and usage aggregated by time-of-use periods. Although either type of data can be used with a regression model, time-of-use is preferable because it provides more insight into usage patterns. In many cases, the peak demand is also recorded.

Interval demand billing data. This type of billing data records the average demand (or energy use) for a given interval (e.g., 15 minutes) associated with the billing period, and typically includes peak demand charges.

Stored energy billing data. Inventory readings or delivery information can be used to determine historical consumption for resources such as fuel oil, although sub-metering is preferred.

One of the challenges in applying Option C is accounting for factors beyond the ECM that affect overall site energy use, such as changes in square footage or loads. These are referred to as ‘non-routine adjustments’ (see Section 2 and Equation 1). Tracking site changes provides a means for accounting for changes in energy use not associated with ECM installation. Adequately tracking the information needed to make these non-routine baseline adjustments can be a challenging task for long-term contracts and sites that have significant operational changes.

3.4 Option D: Calibrated Simulation Option D involves using calibrated computer simulation carried by out experienced, qualified analysts to model the building and mechanical systems in order to predict building energy use both before and after the installation of ECMs. The accuracy of the models is ensured by using metered site data to adjust the model to match baseline and/or performance period conditions. Carefully constructed models can provide savings estimates for several different ECMs on a project. More elaborate models generally improve the accuracy of savings calculations, but increase costs.

Option D methods should be used only for projects that meet any or all of the following requirements:

For complex equipment replacement and controls projects with too many ECMs to cost-effectively use retrofit isolation methods.

When interactive effects between ECMs are too complex for retrofit isolation approaches, but need to be quantified.

When the Option C approach is not viable due to expected savings on the order of less than 10-20% of metered use.

When complex baseline adjustments are expected during the performance period.

When energy savings values per individual measure are desired.

When new construction projects are involved.

When savings levels are sufficient to warrant the cost of simulation.

When either baseline or performance period energy data, but not both, are unavailable or unreliable.

Option D is especially useful where a baseline does not exist (e.g., new construction or major building modification) or the factors responsible for savings are not easily measured (e.g., reduced solar gain and heat loss through new windows).

Situations for which computer simulation is not appropriate include:

Analysis of ECM savings that can be more cost-effectively determined with other methods.

Buildings that cannot be adequately modeled, such as those with complex geometries or other unusual features.

Building systems or ECMs that cannot be adequately modeled, such as radiant barriers or demand-response control algorithms that are important in comparing baseline and performance period scenarios.

Projects with limited resources that are not sufficient to support the effort required for data collection, simulation, calibration, and documentation.

3.4.1 Approach to Option DM&V Option D for an existing building typically follows five general steps: 1) collect data; 2) input data and test baseline model; 3) calibrate the baseline model; 4) create and refine the performance period model; and 5) verify performance and calculate savings. These steps are summarized below. Detailed technical guidance and considerations for each step can be found in the Federal Energy Management Program’s M&V Guidelines: Measurement and Verification for Federal Energy Projects. ASHRAE Guideline 14 specified a more detailed 8-step approach.

Collect data

Input data and run the baseline model

Calibrate the baseline model

Create and refine the performance period model

Verify performance and calculate savings

The methodology followed for new construction projects is somewhat different, and is detailed in IPMVP Volume III. One primary difference between the methods used for existing and new buildings is the availability of utility data. For new construction, the performance period model could be calibrated to utility data, whereas the baseline model could not due to lack of data, although comparisons with similar buildings can be made. This approach would also apply to an existing building that does not have reliable baseline energy data.

3.4.2 Simulation SoftwareThe most frequently used type of energy simulation program for energy analyses are whole-building tools that involve customized models of buildings and their systems, and employ hourly weather data to predict energy use. Two of the most common public domain programs of this type are eQUEST and EnergyPlus.

These building simulation programs require extensive input data to accurately model the energy use of a building. Recently, user interfaces have been improved that simplify the input process with graphical formats, and libraries of typical building components have been added to facilitate model development.

Simulation programs acceptable for Option D should have the following characteristics:

The program is commercially available, supported, and documented.

The program has the ability to adequately model the project site and ECMs.

The model can be calibrated to an acceptable level of accuracy.

The program allows the use of actual weather data in hourly format.

3.4.3 Model CalibrationModel calibration for existing buildings is accomplished by linking simulation inputs to actual operating conditions and comparing simulation results with whole-building and/or end-use data. The simulation may be of a whole building or just for the end use or system affected by the ECM. Both baseline and performance period models should be calibrated wherever possible. Model calibration is typically an iterative process of adjusting model inputs and re-comparing the results to measured data. A model is considered to be calibrated when the statistical indices demonstrating calibration have been met. Expected calibration requirements should be specified in the project-specific M&V Plan. These requirements should be adjusted as required to meet the needs of the project.

For most models, there are multiple levels of calibration that can be performed:

System level calibration with hourly monitored data.

Whole-building level calibration with monthly utility data.

Whole-building level calibration with hourly utility data.

Determining the level of calibration that is needed depends on the value of the project, the availability of data, and the need for certainty in the savings estimates. All models should be calibrated to monthly data at a minimum. Simulation models that focus on specific systems should be calibrated to system level data. Also, calibrating the models to hourly data will help

ensure accuracy, especially for determining peak demand savings. Calibrating a computer simulation to measured utility data necessitates that actual weather data be used.

4. Developing Regression Models for M&VAll M&V options utilize some form of a model to predict the baseline or performance period energy use based on the behavior correlated with appropriate independent variables (e.g., weather, occupancy). An independent variable is a parameter that is expected to change regularly and has a measurable effect on the energy use of a system or building. The models used to predict energy use, with the exception of Option D which utilizes simulation software, are often mathematical equations derived through regression analysis that incorporate key independent variables. The IPMVP document and ASHRAE Guideline 14 both provide additional details about developing models by means of regression analyses as well as techniques for validating these models.

4.1 Independent VariablesData on independent variables may be from a third party or may be tracked using onsite data collection, depending on their nature. Weather data are typically more reliable when supplied by an independent source, but should be validated with site data to ensure applicability.

Once the data have been collected, the mathematical model that is used to predict the baseline (or performance period) energy use is developed. The model should make intuitive sense: the independent variables should be reasonable and the coefficients should have the expected sign (positive or negative) and be within an expected range or magnitude.

4.2 Choosing a ModelThere are various forms of models used in standard statistical practice. Examples of multi-variant regression models are included in the IPMVP document and ASHRAE Guideline 14.

One example of a linear multi-variant regression model for a weather-dependent ECM is shown in Equation 2 below. For models using weather data, it can be beneficial to define a custom temperature base for calculating HDD and CDD data based on the actual behavior of the building.

E = B1 + (B2×Ti−Ti−1)+(B3×HDDi)+(B4×CDDi)+(B5×X1)+(B6×X2)+(B7×X3) Eqn. 2

where:

E = energy use

i = index for units of time for period

B1-7 = coefficients

T = ambient temperature

HDD = heating degree days using a base temperature of 60ºF

CDD = cooling degree days using a base temperature of 65ºF

Xn = independent steady-state variables

It is important that the best model type be used, which in turn will depend on the number of independent variables that affect energy use and the complexity of the relationships. Finding the best model often requires testing several models and comparing their statistical results. The number of coefficients should be appropriate for the number of observations. Similarly, the form

of the polynomial should be suitable to number of independent variables. Additionally, the independent variables must be truly independent of one another. The model should be tested for possible statistical problems (e.g., auto-colinearity) and corrected.

Validation steps should include checks to make sure that statistical results meet acceptable standards. The statistical requirements outlined in Table 2 are examples of validation standards for mathematical models using typical statistical indicators, as provided in the FEMP M&V Guidelines. ASHRAE Guideline 14 also provides guidelines for statistical model validation, based on the resolution of the data, e.g., monthly or hourly. Specific goals should be set for validating mathematical models used in each project based on suitable levels of effort and should be specified in the site-specific M&V Plan. Many analysis tools provide some of these statistical results, while others will need to be calculated separately.

Table 2: Statistical Validation Guidelines

Parameter Evaluated Abbreviation

Suggested Acceptable

Values PurposeCoefficient of determination

R2 > 0.75 Indicates model’s overall ability to account for variability in the dependent variable. Lower R2

values may indicate that independent variables may be missing or additional data are needed.

Coefficient of variation of root-mean squared error

CV(RSME) < 15% Calculates the standard deviation of the errors, indicating overall uncertainty in the model.

Mean Bias Error MBE +/- 7% Overall indicator of bias in regression estimate. Positive values indicate that regression over-predicts actual values; negative values indicate that it under-predicts values.

t-statistic t-stat > 2.0 Absolute value >2.0 indicates independent variable is significant.

5. Selecting an M&V Approach: IPMVP Options A-DThe level of certainty and thus effort required to verify both a demonstration project’s potential to perform and its actual performance will vary from project to project. The site-specific M&V plan should be prepared with consideration of what M&V requirements, reviews, and costs will be specified. Four key factors, outlined below, should be considered when choosing the M&V options and techniques to use for each project.

Value of ECM in Terms of Projected Savings and Project Costs: M&V efforts should be scaled to the value of the project so that the value of the information provided by the M&V activity is appropriate to the value of the ECM and the project itself.

Complexity of ECM or System: Complex projects may require more complex (and thus more expensive) M&V methods to determine energy savings. In general, the complexity of isolating the savings is the critical factor. For example, a complicated chiller measure may not be difficult to assess if there are energy sub-meters and monitoring systems dedicated to the chiller system.

Number of Interrelated ECMs: If multiple ECMs are being installed at a single site, the savings from each measure may be, to some degree, related to the savings resulting from other measures or other non-ECM activities at the facility. Examples include interactive effects between lighting and HVAC measures or between envelope improvements and a chiller replacement. In these situations, it may not be possible to isolate and measure one system in order to determine savings. Thus, for multiple, interrelated measures, whole-building Options C or D may be the most appropriate.

Other Uses for M&V Data and Systems: Often, the instrumentation installed and the measurements collected for M&V can be used for other purposes, including commissioning and system optimization. Data and systems are more cost-effective if they are used to meet several objectives, and not just related to the M&V plan. In addition, savings could be quantified beyond the M&V phase. This information could be useful for allocating costs among different tenants, planning future projects, or allocating research.

The four IPMVP M&V options can be applied to almost any type of ECM. However, the rules-of-thumb listed below generally indicate the most appropriate M&V approach for an application.

5.1 M&V Considerations for Option AOption A is intended for projects where the potential to generate savings must be verified and actual savings can be determined from short-term measurements, estimates, and engineering calculations. Some considerations when using Option A approaches include:

Option A methods can vary in the level of accuracy in determining savings and verifying performance. The level of accuracy depends on the validity of estimates, the quality of the equipment inventory, and the measurements that are made.

Verifying proper ongoing operation and potential to perform is an important aspect of Option A.

Option A is appropriate for relatively simple ECMs whose baseline and post-installation conditions (e.g., equipment quantities and ratings such as lamp wattages or motor kW) represent a significant portion of the uncertainty associated with the project.

Option A methods are not suitable for ECMs whose performance is uncertain or unpredictable.

In general, comprehensive retro-commissioning projects will not lend themselves well to Option A approaches.

5.2 M&V Considerations for Option BSome considerations when using Option B approaches include:

All end-use technologies can be verified with Option B; however, the degree of difficulty and costs associated with verification increases as metering complexity increases.

The task of measuring or determining energy savings using Option B can be more difficult and costly than that of Option A. However, Option B results are typically more precise than for Option A.

Periodic spot or short-term measurements of factors are appropriate when variations in loads and operation are not expected.

Performing continuous measurements will account for operating variations and will result in closer approximations of actual energy savings. Continuous measurements also provide long-term persistence data on the energy use of the equipment or system.

Data collected for energy savings calculations can be used to improve or optimize the operation of the equipment on a real-time basis, thereby improving the benefit of the retrofit. For constant-load retrofits, however, there may be no inherent benefit of continuous over short-term measurements.

5.3 M&V Considerations for Option CThe following points should be considered when conducting Option C utility data analyses for M&V:

All independent variables that affect energy consumption must be specified, whether or not they are accounted for in the model. Critical variables can include ones such as weather, building occupancy, set points, and time of day. The most common variable for many types of ECMs is outdoor air temperature.

The form and content of any separate performance period model(s) (if used) should be specified, along with the statistical validation targets. Statistical validity of the final regression model(s) must be demonstrated.

Independent variable data need to correspond to the time periods of the billing meter reading dates and intervals. A plan for data collection, including sources and frequencies, should be specified.

It is best to develop models using data in whole-year sets (12, 24, 36, or 48 months) so that any seasonal variations are not overstated.

It is necessary to specify how site changes unrelated to the installation of the ECMs will be tracked over the performance period and how these data will be used to perform savings adjustments.

If baseline energy use needs to be adjusted to incorporate minimum energy or operating standards (such as minimum ventilation rates or lighting levels), any modification to the model needs to be detailed.

5.4 M&V Considerations for Option DMany issues must be considered and addressed in developing a site-specific M&V Plan using Option D. Some of the more common issues are outlined below:

Use an experienced building modeling professional. Although new simulation software packages make much of the process easier, a program’s capabilities and real data requirements cannot be fully understood by inexperienced users, and resulting models may not be accurate.

Determine the availability of utility bill data.

Determine whether hourly or monthly billing data are available and whether meters can be installed to collect hourly data. Calibrations to hourly data are generally more accurate than calibrations to monthly data because there are more points to compare. Hourly energy or demand data, however, are generally only available for a utility’s largest customers or need to be collected using the EMCS or portable data loggers. If only monthly billing data are available, conducting additional short-term monitoring of building sub-systems can improve the accuracy of the model.

Use actual equipment performance data in the simulation models. Many software packages have libraries of HVAC equipment that closely match actual system performance. Be cautious and investigate the library HVAC description to be sure it is a good representation of the real system and consider developing user-defined equipment performance curves based on field measurements or manufacturer’s data.

Specify spot measurements and short-term monitoring of key parameters for both the baseline and performance period models. Spot and short-term measurements augment the whole-building data and more accurately characterize building systems. It is recommended that an end use be monitored over a period that captures the full range of the equipment’s operation (e.g., spring and summer for cooling systems. The data must also be collected in a way that facilitates sub-system level calibration. Careful selection of spot measurements and short-term monitoring is necessary because it can add significant cost and time to the project.

Use trend data to determine actual control performance. Sequencing of building controls is difficult to interpret from interviews, site surveys, manufacturer’s data, and spot measurements. The best way to ascertain actual sequences is through trend data. Sometimes, the EMCS systems can be utilized to determine actual operating scenarios. However, the capability for data collection and storage in many systems may be limited.

Specify model calibration procedures that will be followed for monthly, hourly, or subsystem data for both the baseline and performance period models. Prescribe statistical calibration requirements based on the accuracy required for the project.

Specify the simulation program and version and the source of weather data used (onsite, local weather station, or typical weather data).

Clearly explain how savings will be calculated after the first year. Keeping models up to date can be expensive. For projects without substantial site changes expected, an Option C utility billing analysis approach may be viable. Regardless of how savings are calculated each year, the ongoing performance of the measures needs to be verified periodically.

6. Developing an M&V PlanRegardless of the M&V option chosen, for M&V to be useful, one needs to confirm that:

The baseline conditions are accurately defined.

The proper equipment/systems are installed and properly commissioned.

The equipment/systems are performing to specification.

Although confirming these three items may appear simple, a structured approach is helpful. The following describes six steps for this purpose:

Step 1: Allocate Project Responsibilities

Site-specific M&V Plans require that responsibilities for key activities be assigned to the appropriate stakeholders involved. For example, for an Energy Savings Performance Contract (ESPC), a number of financial, operational, and performance issues must be considered when allocating risks and responsibilities. The distribution of responsibilities will depend on each stakeholder’s resources and preferences, and on their capability to provide and act upon information about factors related to these issues.

A few fundamental principles can be applied to the allocation of responsibilities:

Logic and cost-effectiveness drive the allocation of responsibilities.

The responsible party predicts its likely tasks and associated costs to fulfill its responsibilities, and makes sure these are covered in the project budget.

Stipulating certain parameters in the M&V Plan can align responsibilities, especially for the items no one controls.

Step 2: Develop a Site-Specific M&V Plan

A site-specific M&V Plan may be the single most important item in determining energy savings. The plan defines how savings will be calculated and specifies activities that will occur during M&V.

Although the M&V Plan is usually developed during contract negotiations, it is important that the stakeholders agree upon general M&V approaches to be used prior to starting any Investment Grade Audit (IGA). The M&V method(s) chosen will determine to a large extent what activities are conducted during the audit, and will affect its cost and duration.

The project-specific M&V Plan includes project-wide items as well as details for each energy-conservation measure (ECM).

Project-wide items include:

Overview of proposed energy and cost savings.

Schedule for all M&V activities.

Agency witnessing requirements.

Utility rates and the method used to calculate cost savings.

O&M reporting responsibilities.

ECM-level items include:

Details of baseline conditions and data collected.

Documentation of all assumptions and data sources.

Details of any engineering analyses performed.

The method by which energy savings will be calculated.

Details of any O&M or other cost savings claimed.

Details of proposed energy and cost savings.

Details of post-installation verification activities, including inspections, measurements, and analysis.

Details of any anticipated routine adjustments to baseline or reporting period energy.

Content and format of all M&V reports (post-installation and periodic M&V).

Step 3: Define the Baseline

Typically, the entity carrying out M&V defines the baseline as part of the IGA. Baseline physical conditions (such as equipment inventory and conditions, occupancy schedule, nameplate data, equipment operating schedules, energy consumption rate, current weather data and control strategies) are determined during the IGA through surveys, inspections, spot measurements, and short-term metering activities. Utility bills are often used to verify that the baseline has been accurately defined. Baseline conditions are established for the purpose of estimating savings by comparing the baseline energy use with the post-installation energy use. Baseline information is also used to account for any changes that may occur during the performance period, which may require baseline energy use adjustments. If a whole building metering or calibrated simulation approach is used, it is important to document the baseline energy use of all end uses, not just those affected by the retrofit.

It is important to recognize that in some cases after the ECM has been implemented, one cannot reevaluate the baseline unless the ECM can be disabled to return the building to its pre-retrofit state. Therefore, it is important to properly define and document the baseline conditions. Deciding what needs to be monitored (and for how long) depends on such factors as the complexity of the measure and the stability of the baseline, including the variability of equipment loads and operating hours, and the other variables that affect the load.

Step 4: Install and Commission Equipment and Systems

Commissioning installed equipment and systems is considered industry best-practice. It ensures that systems are designed, installed, functionally tested in all modes of operation, and are capable of being operated and maintained in conformity with the design intent (e.g., appropriate lighting levels, cooling capacity, comfortable temperatures). Commissioning is generally completed by the technology implementer and witnessed by building representatives. In some cases, however, it is contracted out to a third party.

Commissioning activities include inspections and functional testing. These activities are specified in a Commissioning Plan, and their results are documented in a Commissioning Report. More specific information on commissioning for technology implementation projects is provided in Section 8 of the FEMP Guideline.

Commissioning usually requires performance measurements to ensure that systems are working properly. Because of the overlap in commissioning and post-installation M&V activities, it is easy to confuse the two. The difference is that commissioning ensures that systems function properly, whereas post-installation M&V quantifies how well systems work from an energy standpoint.

Step 5: Conduct Post-Installation Verification Activities

Post-installation M&V activities include surveys, inspections, spot measurements, and short-term metering. The Post-Installation Report includes:

Project description.

Detailed list of installed equipment.

Details of any changes between the proposed and as-built conditions, including any changes to the estimated energy savings.

Documentation of all post-installation verification activities and performance measurements conducted.

Performance verification - how performance criteria were met.

Documentation of construction-period savings (if any).

Status of rebates or incentives (if any).

Expected savings for the first year.

For projects using certain M&V methods (Option A), the post-installation verification is the most important M&V step, because any measurements to substantiate the savings are made only once. For some measures, where equipment performance and energy savings are not expected to vary significantly over time, post-installation measurements may be the primary source of data used in the savings calculations. Thereafter, inspections are conducted to verify that the potential to perform exists.

Step 6: Perform Regular-Interval Verification Activities

At least once a year, as part of ongoing M&V activities, best practice is to audit the project to ensure savings persistence. This step includes, at a minimum, verifying that the installed equipment/systems have been properly maintained, continue to operate correctly, and continue to have the potential to generate the predicted savings.

An Annual Report can be used to document annual M&V activities and report savings for the year. In many cases, however, more frequent verification activities are appropriate. More frequent monitoring and/or inspection ensures that the M&V monitoring and reporting systems are working properly, installed equipment and systems are operating as intended throughout the year, allows fine-tuning of measures throughout the year based on operational feedback, and it avoids surprises at the end of the year.

The Annual Reports should include:

Results/documentation of performance measurements and inspections.

Verified savings for the year (e.g., energy, energy costs, O&M costs).

Comparison of verified savings with the guaranteed amounts.

Details of all analysis and savings calculations, including commodity rates used and any baseline adjustments performed.

Summary of operations and maintenance activities conducted.

Details of any performance or O&M issues that require attention.

7. Considerations for Technology Demonstration ProjectsTechnology demonstration projects such as those conducted through the General Service Administration’s Green Proving Ground, and the Department of Energy’s High Impact Technology (HIT) Catalyst program are designed to accelerate the adoption of high-potential yet under-adopted technologies for improved building energy efficiency.

The project hypotheses, technical objectives, success indicators, and technology- and site-specific M&V plans are developed collaboratively between the sponsoring agency, technology vendor, and subject matter experts conducting the measurement and verification. In addition to energy savings, demand savings, utility cost savings, occupant comfort impacts, and benefits to operational and energy management staff may be of interest in assessing the technology. Moreover, as conclusions regarding market uptake and broad-scale applicability are important outcomes of these demonstration programs, factors such as installation, operations and maintenance and technology warranties are also of interest in the demonstration evaluation.

Given the desired outcomes of these programs, site selection is a critical component of each technology demonstration. Key site selection considerations are summarized below:

Engagement and willing partnership of operational staff

Ownership and management models, and the participation of property management companies.

Compatibility of systems and controls with the advanced RCx technology being demonstrated. For example, applicability to built-up systems vs. packaged systems, and requirements associated with the building automation system.

Degree to which the electrical infrastructure, and existing metering and monitoring systems can be leveraged to support the desired M&V approach.

Diversity of building types and climates across the demonstration cohort.

Site energy-savings potential.

Quality and completeness of building documentation, e.g., electrical one-line diagrams and panel schedules, control sequences, and mechanical and electrical equipment.

Appendix A contains a general questionnaire that can be customized for technology-specific site selection. Appendix B contains an overview of data analysis tools that can be used to streamline the M&V process.

ReferencesASHRAE. 2002. “ANSI/ASHRAE Guideline 14 Measurement of Energy and Demand Savings”. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

ASHRAE. 2007. “ANSI/ASHRAE Standard 90.1 Energy Standard for Buildings Except Low-Rise Residential Buildings”. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

ASHRAE. 2008b. “ANSI/ASHRAE Standard 111 Measurement, Testing, Adjusting, and Balancing of Building HVAC Systems”. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Claridge, D., M. Liu, and W. Turner. 1999. “Whole–Building Diagnostics”. Proceedings of the Diagnostics for Commercial Buildings: Research into Practice Workshop, June 16 & 17, Pacific Energy Center, San Francisco, CA.

Efficiency Valuation Organization (EVO). 2012. “International Performance Measurement and Verification Protocol (IPMVP)”. Available from www.evo-world.org; accessed on May 15, 2015.

FEMP. 2008. “M&V Guidelines: Measurement and Verification for Federal Energy Projects, Version 3.0”. Prepared for Us. Department of Energy, Federal Energy Management Program by Nexant, Inc., April.

Haberl, J.S, D.E Claridge, J.K. Kissock, and T.A. Reddy. 1993. “Emodel: A New Tool for Analyzing Building Energy Use Data”. Energy Systems Labs Report ESL-IE-93-03-33, Texas A&M University.

Kissock, J.K., Energy Explorer, software and user manual, available at: Available from http://academic.udayton.edu/kissock/http/RESEARCH/EnergySoftware.htm; accessed on May 15, 2015.

Kissock, J.K., J.S. Haberl, and D.E Claridge. 2002. “Development of a Toolkit for Calculating Linear, Change-point Linear and Multiple-Linear Inverse Building Energy Analysis Models”. ASHRAE Research Project 1050, Final Report, November. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Kreider, J.F. and J.S. Haberl. 1994. “Predicting Hourly Building Energy Usage: The Results of the 1993 Great Energy Predictor Shootout Identify the Most Accurate Method for Making Hourly Energy Use Predictions”. ASHRAE Journal, Vol. 35, No. 3, pp. 72– 81.

PECI. 2004. “National Strategy for Building Commissioning”. Portland Energy Conservation, Inc. Report to U.S. Department of Energy.

Piette, M., S. Kinney, and P. Haves. 2001. “Analysis of an Information Monitoring and Diagnostic System to Improve Building Operations”. Energy and Buildings, Vol. 33, pp. 783-791.

Subbarao, K., L. Yafeng, and T.A. Reddy. 2011. “The Nearest Neighbor Method to Improve Uncertainty Estimates in Statistical Building Energy Models”. ASHRAE Transactions, pp. 459-471.

Wright, J. 2011. “Chapter 11: HVAC Systems Performance Prediction”. In Building Performance Simulation for Design and Operation, J.L.M. Hensen and R. Lamberts, ed. New York, NY: Spoon Press.

List of Acronyms

ACM Alternative Calculation Manual

AHRI Air-Conditioning, Heating, and Refrigeration Institute

ASHRAE American Society of Heating, Refrigerating, and Air-Conditioning Engineers

BAS Building Automation System

CV Coefficient of Variation

EMCS Energy Management Control System

ESCOs Energy Service Companies

GUI Graphical User Interface

HVAC Heating, Ventilating, and Air-Conditioning

IPMVP International Performance Measurement and Verification Protocol

LBNL Lawrence Berkeley National Laboratory

M&V Measurement & Verification

QuEST Quantum Energy Services and Technology

R2 Coefficient of Determination

RCx Retro-commissioning

RMS Root Mean Square

RMSE Root Mean Squared Error

TMY Typical Meteorological Year

U.S. DOE United States Department of Energy

UT Universal Translator

VFD Variable Frequency Drive

Other References1 California commercial end-use survey (CEUS), available at www.energy.ca.gov/ceus, final report, p. 8.

2 Matthew, P., S. Kromer, O. Sezgen, S. Meyers, “Actuarial Pricing of Energy Efficiency Projects: Lessons Foul and Fair,” Energy Policy, 2004

3 Available from http://energyalmanac.ca.gov/electricity/index.html; accessed on May 15, 2015.

4 Available from http://gov.ca.gov/press-release/7689/; accessed on May 15, 2015.

5 S. Kromer, “Efficiency Valuation: How to ‘Plan, Play, and Settle’ Energy Efficiency Projects,” Strategic Planning for Energy and the Environment, vol. 27, no. 1, 2007

6 Available from www.evo-world.org; accessed on May 15, 2015.

7 Jump, D., M. Denny, and R. Abesamis, “Tracking the Benefits of Retro-Commissioning: M&V Results from Two Buildings,” proc. National Conference on Building Commissioning, Chicago, May 2-4, 2007, available from www.peci.org/ncbc; accessed on May 15, 2015.

8. The energy models referred to are empirical models developed from energy use and independent variable data collected from the building using statistical techniques. Some of the models to be incorporated into the tool are documented in ASHRAE Research Project 1050, available from www.ashrae.org; accessed on May 15, 2015.

9 “Verification of Savings Project: Definition of Objectives Report,” California Commissioning Collaborative, available from: http://resources.cacx.org/library/ ; accessed on May 15, 2015.

10 Available from http://www.uccsuiouee.org/index.html; accessed on May 15, 2015.

11 “Verification of Savings Project: Existing Methods Report,” California Commissioning Collaborative, available from: http://resources.cacx.org/library/, search on “Cx Guidelines and Reports,” with Subheading “Measurement and Verification.” ; accessed on May 15, 2015.

12 “California Energy Efficiency Evaluation Protocols: Technical, Methodological, and Reporting Requirements for Evaluation Professionals,” available from: www.calmac.org. (search database on ‘Evaluation Protocols’).

13 Energy Explorer is commercial software available from Prof. Kelly Kissock at the University of Dayton. Prof. Kissock was a principal investigator of the ASHRAE Research Project 1050 on inverse energy models, available from http://www.engr.udayton.edu/faculty/jkissock/http/RESEARCH/EnergyExplorer.htm; accessed on May 15, 2015.

14 Please see results for project 37.01 Soda Hall in the UC/CSU/IOU Partnership MBCx Impact Evaluation, p. 97 (the 37.01 project is actually a combination of two projects – Soda Hall and Tan Hall).

15 Please see evaluation results for the 04-05 Building Tune-Up Program, and the 04-05 UC/CSU/IOU Partnership MBCx Program, both available from www.calmac.org (search on the “Impact Evaluation” category and “SBW Consulting, Inc.” as author) ; accessed on May 15, 2015.

http://www.calmac.org/

http://www.engr.udayton.edu/faculty/jkissock/http/RESEARCH/EnergyExplorer.htm

http://www.calmac.org/

http://resources.cacx.org/library/

http://resources.cacx.org/library/

http://www.ashrae.org/

http://www.peci.org/ncbc

http://www.evo-world.org/

http://gov.ca.gov/press-release/7689/

http://energyalmanac.ca.gov/electricity/index.html

http://www.energy.ca.gov/ceus

16 Available from http://www.cacx.org/resources/rcxtools/spreadsheet_tools.html; accessed on May 15, 2015.

17 “2007 California Retrocommissioning Market Characterization”, California Commissioning Collaborative, April 2008

18. Emodel is proprietary software available only to subscribers of Texas A&M’s service-marked ‘Continuous Commissioning’ process, available from http://txspace.tamu.edu/handle/1969.1/2475; accessed on May 15, 2015.

20. Review of SuperESPC Measurement & Verification Annual Reports, Report for the US Department of Energy’s Federal Energy Management Program, available from http://ateam.lbl.gov/mv/docs/MV-Annual-Review-June-2002.doc#_Toc13035452 ; accessed on May 15, 2015.

21. Jump, D., M. Denny, J. Gallishaw, and D. Wiedwald, “After the (RCx) Party is Over – Unrequited or Life Long Savings?” proc. National Conference on Building Commissioning, Newport Beach CA, April 22-24, 2008, available from www.peci.org/ncbc; accessed on May 15, 2015.

22 Jump, D., K. Kinney, M. Denny, R. Abesamis. “Tracking the Benefits of Retro-Commissioning – An Integrated Measurement and Verification Approach”, proc. National Conference on Building Commissioning, San Francisco, April 19-21, 2006, available from www.peci.org/ncbc; accessed on May 15, 2015.

23 Jump, D, D. Johnson, and L. Farinaccio, “A Tool to Help Develop Cost-Effective M & V Plans”, proc. American Council for an Energy Efficient Economy (ACEEE), Summer Study on Energy Efficiency in Buildings, Asilomar, CA, 2000

24 Rubenstein, E., S. Schiller, and D. Jump, “Standard Performance Contracting: A Tool for Both Energy Efficiency and Market Transformation?”, proc. American Council for an Energy Efficient Economy (ACEEE), Summer Study on Energy Efficiency in Buildings, Asilomar, CA, 1998

25 Energy Charting and Metrics Tool, presentation at the National Conference on Building Commissioning, Newport Beach, CA, April 2008.

26 Easy to Use Tools for Performance Monitoring Data Analysis, presentation at the ASHRAE winter meeting, New York City, January 2008.

27 Techniques and Tips for Retrocommissioning Energy Calculations, presentation at the National Conference on Building Commissioning, Chicago, May 2007.

28 Application of Clustering Techniques to Energy Data to Enhance Analysts’ Productivity, Proceedings of the 2002 Summer Study on Energy Efficiency in Buildings, American Council for an Energy Efficient Economy (ACEEE)

29 Internet-Based Building Performance Analysis Provided As a Low-Cost Commercial Service, 2001 International Conference for Enhanced Building Operations

30 DOE 2.1C Model Calibration with Short Term Tests versus Calibration with Long Term Monitored Data, Proceedings of the 1992 Summer Study on Energy Efficiency in Buildings, ACEEE

http://ateam.lbl.gov/mv/docs/MV-Annual-Review-June-2002.doc#_Toc13035452

http://txspace.tamu.edu/handle/1969.1/2475

http://www.cacx.org/resources/rcxtools/spreadsheet_tools.html

31 “M&V Value Tool Alpha Version Specifications” D. Jump, available from http://ateam.lbl.gov/mv/index.htm; accessed on May 15, 2015.

32 Best Practices in HVAC Retrofit and Monitoring Based Commissioning (MBCx) – Shields Library, 2008, available from http://sustainability.calpoly.edu/tracks.html; accessed on May 15, 2015.

33 Best Practices in HVAC Retrofit and Monitoring Based Commissioning (MBCx) – Tan Hall, 2007, available from http://sustainability.ucsb.edu/conference2007/awards.php; accessed on May 15, 2015.

34 SBW Report No. 0606 - CALMAC Study ID QST0001.01, available from www.calmac.org; accessed on May 15, 2015.

35 1999 Nonresidential Large SPC Program Evaluation Study, available from www.calmac.org; accessed on May 15, 2015.

36 Nonresidential SPC M&V Case Study Report, April 30, 2002, available from www.calmac.org; accessed on May 15, 2015.

37 Mills, E., H. Friedman, T. Powell, et.al. “THE COST-EFFECTIVENESS OF COMMERCIAL-BUILDINGS COMMISSIONING, A Meta-Analysis of Energy and Non-Energy Impacts in Existing Buildings and New Construction in the United States,” 2004, available from http://eetd.lbl.gov/emills/PUBS/PDF/Cx-Costs-Benefits.pdf; accessed on May 15, 2015.

Bibliography

General M&V

ASHRAE Guideline 14-2002 Measurement of Energy and Demand Savings, ASHRAE ASHRAE Guideline Project Committee 14P, George Reeves, Chair, 2002.

International Performance Measurement & Verification Protocol—Concepts and Options for Determining Energy and Water Savings, Efficiency Valuation Organization, March 2002.

M&V Guidelines: Measurement and Verification for Federal Energy Projects Version 3.0, Prepared For U.S. Department of Energy Federal Energy Management Program, Lia Webster and James Bradford, Nexant Inc. April 2008.

A Best Practice Guide to Measurement and Verification of Energy Savings: A companion document to ‘A Best Practice Guide to Energy Performance Contracts’, The Australasian Energy Performance Contracting Association for the Innovation Access Program of AusIndustry in the Australian Department of Industry Tourism and Resources, Australia 2004.

Measuring Energy-Savings Retrofits: Experiences from the Texas LoanSTAR Program, Haberl, J., A. Reddy, D. Claridge, W. Turner, D. O’Neal, and W. Heffington. ORNL/Sub/93-SP090, 1996.

Review of Methods For Measuring And Verifying Savings From Energy Conservation Retrofits To Existing Buildings, Jeff s. Haberl, Ph.D., P.E. and Charles H. Culp, Ph.D., P.E. Texas A&M University Energy Systems Laboratory, September 2003, Revised April 2005.

Improving The Cost Effectiveness Of Building Diagnostics, Measurement And Commissioning Using New Techniques For Measurement, Verification And Analysis. Steve Blanc et al., for PIER, December 1999.

Inverse Model Toolkit

Development of a Tookit for Calculating Linear, Change-point Linear and Multiple-Linear Inverse Building Energy Analysis Models, ASHRAE Research Project 1050-RP Final Report. Kelly Kissock et al., ASHRAE November 2002.

Development of a Tookit for Calculating Linear, Change-point Linear and Multiple-Linear Inverse Building Energy Analysis Models, ASHRAE Research Project 1050-RP Detailed Test Results. Atch Sreshthaputra et al., ASHRAE. May 2001, Updated August 2001..

Literature Review Of Uncertainty of Analysis Methods (Inverse model toolkit), Jeff S. Haberl, Ph.D., P.E. and Soolyeon Cho, Texas A&M Energy Systems Laboratory, October 2004

Inverse Modeling Toolkit: Numerical Algorithms, John Kelly Kissock et al., ASHRAE Transactions, Volume 109, Part 2, 2003.

Inverse Model Toolkit: Application and Testing, Jeff S. Haberl et al., ASHRAE Transactions, Volume 109, Part 2, 2003.

Statistical Modeling

Effect of Time Resolution on Statistical Modeling of Cooling Energy Use in Large Commercial Buildings, ASHRAE Transactions 1995, Vol. 101, Part 2, Katipamula, S., Reddy, T.A. and Claridge, D. 1995.

A Development and Comparison of NAC Estimates for Linear and Change-Point Energy Models for Commercial Buildings, Energy and Buildings, Vol.

20, pp.87-95, Ruch, D.K. and Claridge, D.E., 1993.

A Baseline Model for Utility Bill Analysis Using Both Weather and Non-Weather Related Variables, Robert C. Sonderegger, Ph.D., ASHRAE Summer Meeting, Toronto, Canada, 1998

An Improved Cooling Tower Algorithm for the CoolToolsTM Simulation Model, Dudley J. Benton et al., ASHRAE Transactions 2002.

Development and Testing of the Characteristic Curve Fan Model, Jeff Stein and Mark Hydeman, ASHRAE 2004.

Development and Testing of a Reformulated Regression-Based Electric Chiller Model, Mark Hydeman et al., ASHRAE Transactions 2002.

Development and Application of Regression Models to Predict Cooling Energy Consumption in Large Commercial Buildings, in the Proceedings of the 1994 ASME-JSME-JSES International Solar Energy Conference, Katipamula, S., T. Reddy, and D. Claridge. 1994. San Francisco, CA, pp. 307-322.

Use of Simplified Models to Measure Retrofit Energy Savings, Katipamula, S., and D. Claridge. ASME Journal of Solar Energy Engineering 115: 57-68, 1993

Emodel: A New Tool For Analyzing Building Energy Use Data, IETC Conference, Kelly Kissock et al., March 1993

Monthly Variable-based Degree Day Template: A spreadsheet procedure for calculating a 3 parameter change-point model for Residential or Small Commercial Buildings, David S. Landman and Jeff S. Haberl, Texas A&M University Energy Systems Laboratory, August 1996.

Regression Analysis of Electric Energy Consumption of Commercial Buildings In Brazil. Fernando Simon Westphal and Roberto Lamberts, Energy Efficiency in Buildings Laboratory (LabEEE), Federal University of Santa Catarina (UFSC) Florianópolis - Santa Catarina – Brazil. Proceedings of Building Simulation 2007.

Uncertainty

Properties of NAC and CV[NAC] for Energy Models", Energy Systems Laboratory Technical Report, Texas A&M University, Ruch, D., May 1993.

Model Identification and Prediction Uncertainty of Linear Building Energy Use Models with Autocorrelated Residuals, ASME Transactions Journal of Solar Energy Engineering, Vol 121, No.1, pp. 63-68, Ruch, D.K., Kissock, J.K. and Reddy, T.A., February1999.

Bias In Predicting Annual Energy Use In Commercial Buildings With Regression Models Developed From Short Data Set, S. Katipamula et al., Pacific Northwest Laboratory, November 2004.

Uncertainty in Baseline Regression Modeling and in Determination of Retrofit Savings, ASME Journal of Solar Energy Engineering 120(3): 185-192., Reddy, T., J. Kissock, and D. Ruch. 1998.

Technique Of Uncertainty And Sensitivity Analysis For Sustainable Building Energy Systems Performance Calculations, Kotek Petr, Jordán Filip, Kabele Karel, and Hensen Jan, Department of Microenvironmental and Building Services Engineering, CTU Technical University in Prague, Czech Republic, and Building Physics & Systems, Technische Universiteit Eindhoven, Netherlands. Proceedings of Building Simulation 2007.

Metered and Monitored Data Analysis

Investigation of Metered Data Analysis Methods for Commercial and Related Buildings, McDonald and Wasserman, ORNL, May 1989

A Perspective on Methods for Analysis of Measured Energy Data from Commercial Buildings, ASME Journal of Solar Energy Engineering 120(3): 150-155., Claridge, D. 1998.

Lean Energy Analysis: Identifying, Discovering And Tracking Energy Savings Potential, Kelly Kissock and John Seryak, Proceedings of Society of Manufacturing Engineers: Advanced Energy and Fuel Cell Technologies Conference, Livonia, MI, Oct 11-13, 2004.

ASHRAE 1092-RP Final Report: Development of Procedures to Determine In-Situ Performance of Commonly used HVAC Systems, Mingsheng Liu et al., ASHRAE, February 2006.

Impact Assessment

Final Report Compilation for Impact Assessment Framework, Vernon A. Smith, Architectural Energy Corporation, and Michael Kintner-Meyer, Battelle Memorial Institute, for PIER, October 2003.

The California Evaluation Framework, Prepared for the California Public Utilities Commission and the Project Advisory Group, TecMarket Works et al., June 2004.

California Energy Efficiency Evaluation Protocols: Technical, Methodological, and Reporting Requirements for Evaluation Professionals, Prepared for the California Public Utilities Commission by TecMarket Works Team, Nick Hall et al., April 2006.

Quality Assurance Guidelines for Statistical, Engineering, and Self-Report Methods for Estimating DSM Program Impacts, CADMAC Study ID 2001M. Berkeley, CA: Pacific Consulting Services, Ridge, R., D. Violette, D. Dohrman, and K. Randazzo. 1997.

Model Energy-Efficiency Program Impact Evaluation Guide (Draft), National Action Plan for Energy Efficiency, Steve Schiller, July 2007.

Fault Detection and Performance Monitoring

Project 2.5 – Pattern-Recognition Based Fault Detection and Diagnostics, Task 2.5.2—Select Pattern-Recognition Techniques, R.S. Briggs, Battelle Northwest Division, for PIER, April 2001

High Performance Commercial Building Systems: Development of Whole-Building Fault Detection Methods, Element 5 - Integrated Commissioning and Diagnostics, Project 2.3 -

Advanced Commissioning and Monitoring Techniques, Mingsheng Liu et al., for PIER, June 2002.

Model-Based Performance Monitoring: Review of Diagnostic Methods and Chiller Case Study, Philip Haves and Sat Kartar Khalsa, Lawrence Berkeley National Laboratory, ACEEE 2000 Summer Study on Energy Efficiency in Buildings, Efficiency and Sustainability, August 20-25, 2000.

Methods for Fault Detection, Diagnostics, and Prognostics for Building Systems—A Review, Part I. Srinivas Katipamula, PhD and Michael R. Brambley, PhD, HVAC&R Research, ASHRAE January 2005.

Methods for Fault Detection, Diagnostics, and Prognostics for Building Systems—A Review, Part II. Srinivas Katipamula, PhD and Michael R. Brambley, PhD, HVAC&R Research, ASHRAE April 2005.

A Specifications Guide for Performance Monitoring Systems, Kenneth L. Gillespie, Jr. et al., Lawrence Berkeley National Laboratory, 2007.

Appendix A: Site Selection Questionnaire

Building commissioning is a systematic process that can be used to reduce building energy use and to improve indoor environmental conditions for occupants. Retro-commissioning is specific to existing buildings and includes identifying performance goals as well as deficiencies and improvement opportunities; implementing changes through tuning, low cost repairs, and more capital-intensive retrofits as needed; and using measurement and verification (M&V) techniques to verify that changes indeed improve building operation. Examples of advanced strategies and technologies include using utility data to identify candidate buildings, integrating building automation systems with fault detection tools to inform stakeholders of opportunities, and implementing automated control system adjustments.

The following are important considerations for technology demonstration site selection. Please provide as much information as possible.

SITE OVERVIEW

1) Site location (address):____________________________________________________________________________________________________________________________________________________________________

2) Climate zone1 and other location-specific weather-related factors:_____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

_

3) Commercial building type2 (check all that apply for the building): Education Lodging/Hospitality Fire Order/Safety Food Service Supermarket/Grocery Worship Hospital Office Warehouse/Distribution Center Restaurant Public Assembly Bank School Laboratory Multi-Family or Dorm Courthouse Retail Data Center

4) Annual Hours of Operation / Types and Number of Occupants / Occupancy Schedules:___________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

_

5) Year of Construction / Building Size (square feet) / Number of Floors:

1 http://apps1.eere.energy.gov/buildings/publications/pdfs/building_america/ba_climateguide_7_1.pdf

2 http://www.eia.gov/consumption/commercial/building-type-definitions.cfm and http://www.energystar.gov/buildings/facility-owners-and-managers/existing-buildings/use-portfolio-manager/identify-your-property-type-0

____________________________________________________________________________________________________________________________________________________________________

6) Building Annual Energy Use and Peak Demand (2013)Electricity (kWh and kW): ____________________________________________________________Natural Gas (cubic feet): _____________________________________________________________Total Site Energy (all fuels, Btu per square foot): __________________________________________

7) Heating and Cooling Sources: Central plant chillers Age: ________ Capacity: ________ Central plant boilers Age: ________ Capacity: ________ Direct-expansion (DX) packaged rooftop units Age: ________ Capacity: ________ District heating/cooling

8) Air Distribution System Types: Constant Volume (CAV) Age: ________ Airflow (cfm): ___________ Variable-Air-Volume (VAV) Age: ________ Airflow (cfm): ___________ Single-Duct Dual-Duct

9) Other technology(ies) under consideration for demonstration at this site (check any that apply): Advanced Compressor Rack and Refrigerant Systems HVAC (cooling, heating, ventilation, dehumidification) Open Refrigerated Display Case Retrofits Daylighting Systems and Controls External Shading Attachments Lead/Lag plug load strategies Renewables and Distributed Generation Submetering of building components or panels

10) Is there a commitment by a decision-making authority in the organization to facilitate the project through its conclusion by ensuring the availability of requisite staff and resources?a) ___ Yes. Provide the following:

Name: ____________________________________________________Title: ____________________________________________________Organization: ____________________________________________________Phone: ____________________________________________________Email: ____________________________________________________

b) ____ No.

11) Is there a strong advocate on or near the site who can serve as the primary point of contact, with access to the space, equipment, and operations? Having a local advocate will greatly facilitate addressing potential issues with both the space and its operation and occupants.a) ___ Yes. Provide the following:

Name: ____________________________________________________Title: ____________________________________________________Organization: ____________________________________________________Phone: ____________________________________________________

Email: ____________________________________________________b) ____ No.

12) Identify any factors below that could obstruct/create difficulty with using the site for a demonstration (check all that apply):a) ___ Site operates under energy savings performance contract.b) ___ Pending changes to physical space or usage that could disrupt the demonstration.

Typically, the demonstration project needs a consistent test environment for useful pre- and post-retrofit data collection.

c) ___ Safety or security constraints. These may come in the form of building or data access restrictions associated with the nature of the work in the space.

d) ___ Atypical/non-standard electrical layout or configuration (such as modifications that may make measurements difficult).

e) ___ Limited access for installation of technology and metering.f) ___ Other (explain):

___________________________________________________________

Explain the conditions for any of the boxes checked above:______________________________________________________________________________________________________________________________________________________________________________________________

13) Does the site have available historic energy use data?a) ___ Yes. Describe data period, interval (e.g., monthly, hourly, 15 min), and accessibility

(e.g., remotely accessible, archived in spreadsheets or PDF files)._________________________________________________________________________________________________

b) ___ No.

14) Are you able to access energy cost data?a) ___ Yes.b) ___ No, I don’t have that information.

15) Are there any problems in the building (e.g., high energy use, comfort or air quality complaints)?a) ___Yes.

Explain:_____________________________________________________________________________________________________________________________________________________

b) ___No. Explain why the site should still be considered:________________________________________________________________________________________________________________

16) Do existing controls operate the system properly?a) ___Yes.b) ___No. If no, can they be removed or re-configured?

i) ___Yes. Explain how modification or replacement of the controls might affect building operations:_____________________________________________________________________________________________________________________________________________

ii) ___No. Explain why the site should still be considered:_______________________________

______________________________________________________________________________

17) Are design and operation documents available: As-built drawings Equipment specifications Control sequences of operation

and set points Operations and maintenance

schedules Test and balance reports Commissioning reports Standard operating procedures

18) Have any renovations or retrofits been carried out since the design documents were prepared?a) ___Yes. Explain year and scope:

________________________________________________________________________________________________________________________________________________________________________________________________________________

b) ___No.

19) Is there stable space operation and function? Spaces where operations may be changing during the demonstration period should be avoided as this may introduce additional variability.a) ___Yes.b) ___No. Explain:

_____________________________________________________________________________________________________________________________________________________________________________________________________________________________

20) Does the site have an Energy Information System (EIS) or Building Automation System (BAS)?a) ___Yes. Make and model:

_________________________________________________________i) If yes, can the EIS or BAS be accessed remotely?

(1) ___Yes.(2) ___No. Explain restrictions:

_________________________________________________b) ___No.

21) Are there any known EIS/BAS limits (e.g., maximum number of points, data collection frequency)?(1) ___No.(2) ___Yes. Explain restrictions:

_________________________________________________

22) Does the site have whole-building and submetering through the same EIS/BAS?a) ___ Yes.b) ___ No.

Explain:______________________________________________________________

23) Are any standalone data loggers used (e.g., for temperature, electrical power, filter pressure drop)?(1) ___No.(2) ___Yes. Explain uses:______________________________________________________

______________________________________________________________________________

24) Is electric power measurement instrumentation installed that can be used for the demonstration?a) ___Yes.

Describe instrumentation (e.g., power meter types) and data collection process._______________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

b) ___No.i) Can electrical panels or junction boxes be made available for metering access along with

space for metering equipment?(1) ___Yes(2) ___No

ii) Do you anticipate any conflict with operations of the space or area?(1) ___No(2) ___Yes

iii) Is a qualified electrical worker available to install electric power instrumentation?(1) ___Yes(2) ___No

25) Is mechanical measurement instrumentation installed that can be used for the demonstration(i.e., air or fluid flow, pressure, relative humidity, speed, temperature sensors)?a) ___Yes.

Describe instrumentation (e.g., sensor types) and data collection process.__________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

b) ___No.i) Can the ducts, piping, mechanical rooms, and roof be made available for instrumentation

access along with space for measurement equipment? Explain any access restrictions: __________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ii) Do you anticipate any conflict with operations of the space or area?____________________________________________________________________________________________________________________________________________________________

26) Is an onsite weather station available?a) ___Yes. Explain parameters measured (temperature, RH, solar radiation, wind speed and

direction):_________________________________________________________________________________________________________________________________________________________________________________________________________________________________

b) ___No.

27) Describe accuracy and calibration frequency of monitoring equipment (EIS/BAS sensors, data logger, weather station). _______________________________________________________________

________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

28) Are any other advanced meters or smart meters used that can provide interval energy data readings?a) ___Yes. Explain parameters measured and meter types_________________________________

____________________________________________________________________________________________________________________________________________________________

b) ___No.

29) Are existing staff capable of carrying out M&V-related diagnostic tests and data collection?a) ___Yes. Explain skill level:

_____________________________________________________________________________________________________________________________________________________________________________________________________________________

b) ___No.

30) Other comments or notes: _______________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

Appendix B: M&V Data Analysis ToolTraditionally, data that can be used in M&V analysis have been difficult to obtain and come from multiple sources. The tasks of collecting data, merging data sets, pre-conditioning and preparing data for analysis are demanding enough to prevent practitioners from performing rigorous M&V. Practitioners need statistical analysis or simulation skills, plus software that is capable of manipulating large data sets. The time and materials costs to collect the required data for the required duration, prepare and run the analysis, and document the results can exceed the value of the savings obtained if these required resources are not in place.

M&V implementers also need to assess the ability of the baseline energy model to determine savings prior to finalizing the M&V approach. The annual savings must be larger than the uncertainty in the baseline model, in order to use the model effectively. Currently “rules of thumb” are used when analysis of the baseline data can validate the approach. This frequently results in discarding a perfectly appropriate and cost-effective M&V strategy, or a foregoing an M&V strategy altogether.

Commercial tools that are available both free of charge, and for licensing fees are increasingly beginning to offer support for ‘project tracking’ and M&V of energy savings. These tools may either fully or partially automate the process of creating a baseline model, and projecting the model to determine energy savings. For example, an M&V data analysis tool was recently developed by QuEST and integrated into Pacific Gas and Electric’s Universal Translator 3 (2013). The tool allows users to process data from different periods of operation and to determine non-routine adjustments. Another feature of the M&V tool is that enables estimation of the baseline model uncertainty as well as the savings uncertainty. Baseline model uncertainty indicates how well a particular regression model predicts measured baseline energy data. Savings uncertainty indicates the likelihood that the actual savings is within the confidence bounds described by the uncertainty estimate. Estimating savings uncertainty using regressions based on time-series data is an evolving field. Models developed from ordinary least-squares regressions are found to greatly underestimate uncertainties in both baseline energy as well as savings. This is primarily due to the assumption of independence of each data point: that the value of each data point does not depend on any other point. In buildings, this assumption does not hold for all points as energy use on an hourly basis does depend on the energy use of preceding hours. There is some dependence among points with daily time intervals as well. In ASHRAE Guideline 14, an approach based on fractional savings is used. While this approach allows ordinary least squares regressions, it makes allowance for the true number of independent data points being less than the actual number in its calculations of savings uncertainty.

There is a significant research need to develop more robust methods for computing uncertainty in the energy forecasts. Recent efforts include fractional savings (ASHRAE 2002) and nearest neighbor (Subbaro et al. 2011) approaches. Several issues must be addressed in using these methods, including the amount of data required, variations in building energy use not caused by the regressor variables, and data autocorrelation.

In the meantime, the M&V tool adopts a cross-validation approach to decrease the impact of auto-correlated energy data on uncertainty estimates. Among this method’s strengths is that it applies to most model development methods, not only those that assume residuals are normally distributed. Cross-validation is a method where a known data set is partitioned into several

equally sized subsets, and one subset is “held out” from the other data sets while the remaining datasets are used to “train” the statistical models. The model’s prediction results for the “held out” dataset (the “prediction” set) are used to calculate the modeling error. The process is repeated for all of the partitioned datasets and an average may be used to determine the generalized error of the model.

This general error term is typical of the amount of data in the subset. For the M&V tool, this subset was taken as one month of data, and thus the error is typical of a baseline month. How this error term propagates over multiple months, as would be required when calculating savings, is not known however. While further research is needed on this approach, the error is limited when using hourly or daily models.

The M&V tool facilitates assessment of this M&V approach prior to installation of measures. Tool users can compare the baseline model uncertainty with expected savings to understand how accurately the method will ultimately estimate savings. If the uncertainty exceeds user requirements, they may elect to pursue a different M&V method.

Providing high-level regression and savings uncertainty analysis in a tool that is integrated with a software platform where data are simple to upload and merge is the approach to reducing overall project costs and improve confidence in the results. Making the M&V tool freely available to any user reduces complicated data processing and analysis time. Making the project files portable should also speed project review and lessen overall project time.

PG&E’s UT provides a platform for uploading the data and conducting data quality checks. In particular, the UT has wizards that recognize data from different sources and file formats; users need only drag files across the screen to upload the data. Attributes of the data files can then be assigned, such as naming files, adding descriptions, and specifying time interval re-sampling rates (i.e., creating hourly or daily time intervals from the raw data). The UI also facilitates data set merging, filtering, applying functions, and charting. Each of these functions may be required prior to conducting the M&V analysis. All tool charts, data, and model outputs are exportable, so that the data and models may be used in other software or spreadsheets. Descriptions of the available functions and features of the UT are available on the website (http://utonline.org/cms/).

Integration with Universal TranslatorThe M&V analysis tool was designed according to typical M&V process steps. These steps include: data collection, merging, re-sampling, and quality control, which are each functions in the UT. Once the data are prepared, the user opens the M&V analysis module to proceed.

M&V analysis is not completed in one session with the tool. Typically, prior to installation of energy efficiency measures, baseline data are collected, prepared, and a baseline model is developed and assessed. The assessment compares the uncertainty of the baseline model with the expected savings. If the uncertainty is large, project sponsors can make adjustments to the M&V process or decide on an alternate M&V approach. The tool allows users to upload baseline data, develop and assess a model for the application, upload and append additional data, develop new models, and so on until a satisfactory model is established.

Following a project’s installation and waiting for a time to collect post-installation data, the tool allows users to upload, merge, re-sample, and check data quality, and proceed with the M&V analysis. All of the raw and processed data and analysis work performed is stored in a project

http://utonline.org/cms/

file. In addition, all raw and processed data and analysis results are available for export to other software.

Modeling MethodThere are many choices for developing regression models. Users may select a model type, of which there are four: Time and Independent Variable, Independent Variable only, Time Only, or Average Dependent. The user selects one of these types based on their own understanding of the major influences and available data that influence energy use in a building. For example, for a daily analysis time interval, and a building used continuously throughout the week, a temperature-only model may yield the best fit to the data. Other buildings may yield better fits by including time-of-week in the regression.

The tool allows users to also select the number of linear line segments for the independent variable. The choices are:

“Equal size linear segments”, where the range of independent variable data (minimum to maximum) are divided into a user-selected number of equal segments,

“Equal number of samples per segment”, where the user specifies the number of segments and the tool divides the number of data points by this number and establishes line segments with an equal number of data points per segment (line segments will be short for temperature ranges with a lot of data and longer where there are fewer data points)

“Optimize using 1 change point”, where the tool establishes two linear segments and finds the best location (i.e., independent variable point) where the segments meet

“Optimize using 2 change points”, where the tool establishes three linear segments and finds the two best locations (i.e., independent variable point) where the segments meet

“Quadratic”, where the tool assumes a quadratic relationship between energy use and the independent variable.

The energy data used for the independent variable model type are in the data portion that is not time dependent. The tool subtracts out the time dependent portion from the data prior to developing this relationship. The two change-point selections allow users to develop four-parameter and five-parameter change-point models (Kissock 2002) for models that do not have time dependence.

Uncertainty MethodUncertainties are estimated using a cross-validation method. This method requires the data from which the model is developed to be partitioned into multiple sets, and a model is developed using one set of data to predict the next set. The error between the model prediction and the actual data is then determined. This process is repeated for each of the sets, and an average of these errors, multiplied by the number of anticipated post-installation months, is taken as the model error. In the M&V tool, one month is used as the set unit. To have enough error points in the averages (e.g., for the baseline, avoided energy use, normalized savings cases), there must be at least six months of data for each model/case.

For avoided energy use (where savings are reported for the measurement period following an energy savings project), the savings uncertainty will be the number of post-installation months, times the average monthly baseline prediction error.

For normalized savings, when both baseline and post-installation energy are projected to annual conditions such as typical meteorological year (TMY) temperatures, and the baseline and post-installation model uncertainties are based on 12 months of prediction and the results are combined through normal propagation of error equations (square root of sum of squares of the error terms).

Other Tool Algorithms and RoutinesThe M&V tool allows users to calculate savings under two IPMVP scenarios, which have been mentioned above: avoided energy use and normalized savings.

Calculating avoided energy use requires a baseline energy model, and independent variable data from the post-installation period. Post-installation period data are uploaded into the UT and prepared in the same way as for the baseline data. Users select the Post-Installation option in the M&V analysis module to select the dependent and independent variables, and define the post-installation time period. No post-installation model development is required. For the “Avoided Energy Use” option, the M&V tool calculates avoided energy use upon selecting that option. A line chart shows the adjusted baseline energy use and post-installation use.

Normalized savings requires a post-installation energy model as well as a baseline energy model. Post-installation models are developed in the same way as baseline models, but in the Post-Installation tab instead. Once both baseline and post-installation models are developed, the user selects the Normalized Savings option. The user then selects the data file that has the conditions to which energy use is adjusted. The data file, usually a TMY weather file for temperature-dependent models, must have been previously uploaded into the UT. The Normalized Savings and savings uncertainty are then computed. Note that this allows savings to be “extrapolated” beyond the measurement period, which technically does not adhere to IPMVP principles, but is a common practice in the industry.

SummaryComplex regression modeling functionality has been programmed into the M&V analysis module. Regression models using time as well as an independent variable, usually ambient temperature, have been included in the M&V Tool. The user may use menu-driven selections to customize modeling types and apply them to filtered sets of data. This allows users to think more about how to model a building properly rather than about the mechanics of the regression process. Various charts and graphs allow the users to quickly assess model fit and performance in the development of appropriate models.

The ability of the M&V tool to set up analysis bins based on filtering the data in different time periods is useful to improve modeling accuracy. Occupied and unoccupied periods exhibit very different energy use behavior. One tool tester asserted that the tool should allow for automatic filtering of the data, as well as allowing the tool to set up a function to vary the start-stop operation in a building, as many buildings operate in this way. Although such features may be useful, they are beyond the scope of the current project. It was noted, however, that the users have begun to think about useful tool features that would make it more generally applicable.

Regarding the M&V analysis module, the decision to use a time-and-temperature based model rather than simple change point models was advantageous. Testing group members liked the accuracy the tool provides when using hourly time intervals, allowing them to visualize times throughout the day or week when savings are being achieved. This aspect of visualizing

projected baseline usage along with measured post-installation usage is helpful to owners and service providers to maintain savings over time.

Tool users also noted that the time-and-temperature model, which was developed by LBNL, has applications in demand response. Using 15-minute or hourly analysis time intervals, baseline models can be developed and used to estimate demand reductions when a utility calls a demand reduction event. While the M&V tool was not developed specifically for quantifying these benefits, the principles are the same. It is noted that future tool developments could include demand response applications.

Overall, the M&V tool provides users with a means to develop a rigorous M&V analysis in a streamlined process. It provides advanced regression modeling capability, and allows users to test multiple scenarios to get the best energy models. It also enables users to focus on the M&V approach and uncertainties in baseline modeling and savings at key points in an energy efficiency project, rather than on the actual analysis required to prepare data.

· web view27 techniques and tips for retrocommissioning energy calculations, presentation at the...

Documents