ASHRAE LV-11-C042-2011 Estimating Industrial Building Energy Savings using Inverse Simulation.pdf

《ASHRAE LV-11-C042-2011 Estimating Industrial Building Energy Savings using Inverse Simulation.pdf》由会员分享，可在线阅读，更多相关《ASHRAE LV-11-C042-2011 Estimating Industrial Building Energy Savings using Inverse Simulation.pdf（8页珍藏版）》请在麦多课文档分享上搜索。

1、Franc Sever is a research assistant, Kelly Kissock is a professor in the Department of Mechanical and Aerospace Engineering, University of Dayton, Ohio. Steve Mulqueen is a Project Engineer, Dan Brown is Vice President at Cascade Energy Engineering, Portland, Oregon. Estimating Industrial Building E

2、nergy Savings using Inverse Simulation Franc Sever Kelly Kissock, PhD, PE Dan Brown, PE Steve Mulqueen Student Member ASHRAE Member ASHRAE Member ASHRAE Associate Member ASHRAE ABSTRACT Estimating energy savings from retrofitting existing building systems is traditionally a time intensive process, a

3、ccomplished by developing a detailed building simulation model, running the model with actual weather data, calibrating the model to actual energy use data, modifying the model to include the proposed changes, then running the base and proposed models with typical weather data to estimate typical en

4、ergy savings. This paper describes a less time-intensive method of estimating energy savings in industrial buildings using actual monthly energy consumption and weather data. The method begins by developing a multivariate three-parameter change-point regression model of facility energy use. Next, th

5、e change in model parameters is estimated to reflect the proposed energy saving measure. Energy savings are then estimated as the difference between the base and proposed models driven with typical weather data. Use of this method eliminates the need for estimating building parameters, system perfor

6、mance, and operating practices since they are included in the inverse simulation model. It also eliminates the need for model calibration since the inverse model is derived from actual energy use data. The paper describes the development of statistical inverse energy signature models and how to modi

7、fy the models to estimate savings. Expected savings from inverse simulation are compared to savings predicted by detailed hourly simulation, and sources of error are discussed. Finally, the method is demonstrated in a case study example from the industrial sector. Limitations of the approach for com

8、plex building systems and the uncertainty of estimated savings are discussed. INTRODUCTION Estimating energy savings from retrofitting existing building systems is traditionally a time intensive process, accomplished by developing a detailed building simulation model, running the model with actual w

9、eather data, calibrating the model to actual energy use data, modifying the model to include the proposed changes, then running the base and proposed models with typical weather data to estimate typical energy savings. Moreover, the development of the detailed simulation model requires many assumpti

10、ons about building parameters, system performance, and operating practices. The unavoidable calibration error and the assumptions required to simulate energy use introduce uncertainty into the process. This paper describes an inverse simulation method of estimating energy savings in industrial build

11、ings using actual monthly energy consumption and weather data. The method begins by developing a multivariate three-parameter change-point regression model of facility energy use. Next, the change in model parameters is estimated to reflect the proposed energy saving measure. Energy savings are then

12、 estimated as the difference between the base and proposed models driven with typical weather data. Use of this method eliminates the need for estimating building parameters, system performance, and operating practices since they are included in the inverse simulation model. It also eliminates the n

13、eed for model calibration since the inverse model is derived from actual energy use data. This inverse simulation approach is appropriate for simple buildings, without simultaneous heating and cooling, and buildings that can be modeled as single zone buildings, such as many industrial facilities. In

14、 the sections that follow, development of the statistical inverse energy signature models and how to modify the LV-11-C042348 ASHRAE Transactions2011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 117, Part

15、1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAES prior written permission.models to estimate savings are discussed. Next, expected savings from inverse simulation are compared to savings predicted by det

16、ailed hourly simulation, and sources of error are discussed. Finally, the method is applied to a case study example from the industrial sector. Limitations of the approach for complex building systems and the uncertainty of estimated savings are discussed. OVERVIEW OF THE METHOD The method of regres

17、sing utility billing data against weather data used here builds upon the PRInceton Scorekeeping Method, PRISM, which regresses building energy use versus variable-base degree-days (Fels, 1986a). However, the method described here uses temperature change-point models instead of degree-day models and

18、can include other independent variables such as production. Temperature change-point models were described by Kissock et al. (1998) and Kissock et al., (2003). The temperature change-point model method was extended to include additional independent variables by Kissock et al. (2003) and Haberl et al

19、. (2003). The interpretation of regression coefficients, builds on early work by Goldberg and Fels (1986), Rabl (1988), Rabl et al. (1992) and Reddy (1989). Principle differences between this work and the aforementioned papers are that this work seeks to use inverse modeling proactively to estimate

20、energy savings from retrofitting industrial building systems rather than retroactively to measure energy savings. The method of estimating building energy savings using inverse simulation is accomplished in three steps. The first step is to develop a statistical multivariate three-parameter model of

21、 building energy use as a function of outdoor air temperature and production. Because this model describes the specific energy use pattern of a facility, it is called an “energy signature” model. The second step is to modify the energy signature model to simulate the performance of the building with

22、 the proposed energy efficiency measures. This model is formed by calculating the change in model coefficients to reflect the proposed energy saving measures. The third step is to drive both the base and proposed models with typical weather data to estimate the normalized annual consumption (NAC). T

23、his step can be accomplished using TMY2 (NREL, 1995) weather data, cooling degree hours (CDH) and heating degree hours (HDH), or binned temperature data. Typical energy savings are then calculated as the difference between the proposed models NAC and the baseline models NAC. The typical energy savin

24、gs are needed to evaluate the economic feasibility of the proposed energy saving measures. Description of Data and Software The method described here is demonstrated using monthly utility bills for energy consumption data because of their wide availability and accuracy. However, the method can be us

25、ed with higher time resolution data if they are available. When using utility billing data, the first step is to normalize the data to remove the effect of unequal days in the billing periods. The method uses both actual and typical weather data. Kissock (1999) posts actual average daily temperature

26、s for 157 U.S. and 167 international cities from January 1, 1995 to present on the internet. Typical weather is derived from TMY2 (or TMY3) data files from the National Renewable Energy Laboratory (NREL 1995). When using utility billing data, the average temperature during each billing period is cal

27、culated from the available temperature data The algorithms used to generate multi-variable change point models have been incorporated into the software designed for energy analysis (Kissock 2005), and the inverse simulations were performed using software for estimate energy savings (Sever and Kissoc

28、k 2009). Step 1: Energy Signature Models The first step is to derive a statistical energy signature model of a facilitys electricity or fuel use as a function of the actual outdoor air temperature over the same time period. The weather dependence of energy consumption can accurately be described by

29、three-parameter change-point models for most industrial facilities. Typical three-parameter heating (3PH) and three-parameter cooling (3PC) change-point models are shown in Figure 1. In the following discussion it is assumed that fuel is used for space heating and electricity is used for space cooli

30、ng. Electricity or fuel consumption is graphed on the vertical 2011 ASHRAE 349axis versus outdoor air temperature on the horizontal axis. The coefficients of a 3PC model are the weather-independent electricity use (Ei), the cooling change-point temperature (Tb,C), and the cooling slope (CS). The coe

31、fficients of a 3PH model are the weather-independent fuel use (Fi), the heating change-point temperature (Tb,H), and the heating slope (HS). In facilities using the same energy source for both heating during winter and cooling during summer, a five-parameter model with both heating and cooling slope

32、s can be developed (Haberl et al. 2003; Kissock et al. 1998; Kissock et al. 2003). Figure 1 (a) 3PC (Cooling) and (b) 3PH (Heating) regression models When electricity or the fuel is used in a production process in the facility, this method can be extended to include energy which is dependent on the

33、amount of production by adding another regression coefficient. This will further refine the results of the regression, increasing the accuracy of the disaggregation of weather-independent and weather dependent energy use. The production coefficient, (PD), represents the response of electricity or fu

34、el use with regards to the amount of production (P). Using these models, electricity consumption can be estimated as a function of outdoor air temperature (TOA) and production level, using Equation 1. Similarly, fuel consumption can be estimated as a function of outdoor air temperature (TOA) and pro

35、duction level, using Equation 2. The superscript + denotes that the parenthetic quantity equals zero when it evaluates to a negative value. (1) (2) Application of Model Coefficients to Building Systems One of the strengths of this method is that the model coefficients directly characterize the physi

36、cal properties of the envelope and operation of the facility. This eliminates the uncertainty associated with estimating the building parameters while calibrating a forward simulation model. Ei and Fi represent the electricity and fuel use that is not related to weather or production. For example, i

37、n industrial facilities, lighting electricity use may be unrelated to weather or production. Similarly, fuel use to make up heat lost through the shells of furnaces and ovens is often unrelated to weather or production. (Eger and Kissock 2007) have developed a lean energy analysis method that target

38、s these types of energy use for reduction. CS and HS represent the variation of energy consumption with outdoor air temperatures. CS and HS include the building cooling and heating loads and the cooling and heating system efficiencies. In simple industrial buildings, the sum of conductive heat gain/

39、loss through the building envelope and sensible heat gain due to ventilation and infiltration air dominate the cooling and heating loads of the building. Solar and latent cooling loads have been shown to be linearly related to outdoor air temperature (Ruch et al. 1993; Reddy et al. 1998), and are ac

40、counted for in these coefficients. Thus, the cooling and heating coefficients of the building, CC and HC, are given in equations 3 and 4 respectively. Where U is the overall building envelope conductance, A is the envelope area, is the sum of ventilation and infiltration flow rate, is the density of

41、 air and is the specific heat of air. CS and HS, are the quotients of the external cooling and heating coefficients and the overall efficiency of the space cooling or heating system efficiency, or . The balance-point temperature, Tb is defined as ECTO ATb ,CEiC S( a )FHTb ,HFiH STO A( b )350 ASHRAE

42、Transactionsthe temperature above or below which space conditioning begins. Tb is a function of the thermostat set-point temperature, Tset, the sum of the internal loads from electricity use, solar gain and occupants, Qi, and the CC and HC. (3) (4) (5) (6) (7) (8) Changes in building properties or o

43、peration would cause a change in the energy signature model, affecting one or more of the coefficients discussed above. Therefore, the savings potential of an energy saving measure can be estimated by adjusting model coefficients and parameters to reflect proposed retrofits. Step 2: Estimating Chang

44、es to Energy Signature Models This section describes the method of estimating changes to the energy signature models to estimate energy use after an energy efficiency retrofit. The method is demonstrated using a 3PH model. Equation 2 shows that there are four coefficients Fi, HS, Tb,H and PD and two

45、 variables Toa and P, that influence fuel use. Fi, PD and P are weather-independent and therefore will be assumed to remain constant and will not be considered further. Therefore, modifications are made to the coefficients HS and Tb,H to estimate savings of proposed retrofits. The proposed HS is cal

46、culated as the energy signature model derived HC adjusted for the estimated change in HC, divided by the proposed heating equipment efficiency. The proposed heating equipment efficiency is equal to the baseline efficiency adjusted for proposed equipment efficiency improvements. The calculation of th

47、e proposed HS is shown in Equation 9. is the sum of the expected change of the building heating load and the expected change of outdoor ventilation and infiltration air. This calculation is shown in Equation 10. (9) (10) Common examples of retrofits which would result in a are adding insulation to t

48、he envelope of a building and replacing exterior windows with units that have a higher insulation value. Common examples of retrofits which would result in a are closing outdoor air dampers during unoccupied periods and reducing infiltration by improving the sealing of the building envelope. Calcula

49、ting requires measuring or estimating the change in building envelope thermal resistance or the reduction of ventilation or infiltration air. The balance temperature also must be recalculated when proposed retrofits affect any of the following, Tset, Qi or HC. The proposed balance temperature calculation is shown in Equation 11. (11) Changes in Tset commonly result from lowering the space temperature set-point during the heating season, including night setback controls. Changes in Qi can result from retrofits whic