ASHRAE IJHVAC 15-5-2009 HVAC&R Research《《HVAC&R研究》》.pdf

《ASHRAE IJHVAC 15-5-2009 HVAC&R Research《《HVAC&R研究》》.pdf》由会员分享，可在线阅读，更多相关《ASHRAE IJHVAC 15-5-2009 HVAC&R Research《《HVAC&R研究》》.pdf（173页珍藏版）》请在麦多课文档分享上搜索。

1、 VOLUME 15, NUMBER 5 HVAC they no longer need to rely solely on a self-appointed technical elite to provide policy advice on technical issues. So what is the role of American Society of Heating, Refrigerating and Air-Conditioning Engi-neers, Inc. (ASHRAE) in this brave new world? Like other professi

2、onal societies, ASHRAE develops consensus standards based on more detailed technical handbooks that are continuously updated and vetted by researchers and expert practitioners at semiannual meetings of the mem-bership. No matter what means are available for handling information in the future, this w

3、ill remain an important domain and function of professional societies. And, in the case of ASHRAE, there is a unique capability. ASHRAE is perhaps the only professional society that raises funds for and manages its own research projects that foster the understanding of science and technology support

4、ing HVAC.Established nearly a century ago, the ASHRAE research program has played a vital role in the development of materials incorporated in the ASHRAE Handbooks, design guidelines, and stan-dards. The societys $2.5 million/year investment is augmented by matching funds from others, and guided by

5、the Research Strategic Plan which is periodically updated by the members. The Clark Bullard is a professor emeritus in the Department of Mechanical Science and Engineering at the University of Illinois at Urbana-Champaign, Urbana, IL. Jelena Srebric is an associate professor in the Department of Arc

6、hitectural Engineering at Pennsylvania State University, University Park, PA. Reinhard Radermacher is a professor in the De-partment of Mechanical Engineering and director of the Center for Environmental Energy, University of Maryland, Col-lege Park, MD. 2009, American Society of Heating, Refrigerat

7、ing and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC accepted February 20, 2009This paper proposes a method for establishing improved guidelines regarding the use of refriger-ant blends that contain a flammable component. This approach was verified in a test chamber under rea

8、listic refrigerating equipment leak scenarios. First, the paper provides an overview of the safety classification described in ANSI/ASHRAE Standard 34-2007, Designation and Safety Classification of Refrigerants (ASHRAE 2007) and its international equivalent ISO Standard 817-2006, RefrigerantsDesigna

9、tion and Safety Classification (ISO 2006). The refrigerant blends examined were A1, as formulated, and A2, at the “worst-case fractionated formulation,” as defined by ASHRAE Standard 34-2007. The blend is prepared in a cylinder and is then slowly leaked inside a test chamber of 1 m3(35.31 ft3). At t

10、he beginning of the leak, the gas phase occu-pies 10% of the cylinder volume, corresponding to one of the leak scenarios featured in ISO Stan-dard 817-2006. The concentrations inside the test chamber are measured by gas chromatography. In addition, this paper provides the algorithm of calculation of

11、 a computer code named Room_leak, which simulates leaks of any refrigerant blends in realistic scenarios.BACKGROUNDNew hydrofluorocarbon (HFC) refrigerants have been developed to stop the production of ozone depleting refrigerants such as chlorofluorocarbons and hydrochlorofluorocarbons. The develop

12、ment of new low global warming potential (GWP) refrigerants are necessary to limit radiative forcing of most of the HFCs developed since 1990. The compromise among energy efficiency, GWP, and flammability is difficult to establish for many applications, and the use of mildly flammable refrigerant ha

13、s to be carefully analyzed to increase the number of choices when it comes to refrigerants. It may be possible to reduce the GWP of refrigerant blends by allowing the use of larger quantities of flammable, low GWP refrigerant ingredients, as long as this can be done without increasing the flammabili

14、ty risk associated with the refrigerant.ANSI/ASHRAE Standard 34-2007, Designation and Safety Classification of Refrigerants (ASHRAE 2007) and its Appendix B, as well as its international equivalent ISO Standard 817-2006, RefrigerantsDesignation and Safety Classification (ISO 2006), define safety cri

15、teria for toxicity and flammability limits of refrigerants. Several definitions that are essential to this discussion are provided in the Annex at the end of this paper. Because of the worst-case frac-tionated formulation (WCFF) some refrigerant blends are classified as mildly flammable only for the

16、 worst-case fractionation, while others are nonflammable as initially formulated, even when taking into account the worst case formulation (WCF). ISO Standard 817-2006 utilizes a dual classification system, which classifies the blends as 1/2, 1 (for the WCF), and 2 (for the Dennis Clodic is director

17、 of research and laboratory manager and Youssef Riachi is project leader at the Center for En-ergy and Processes, MINES Paris Tech, Paris, France. 2009, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC 0.7812/0.2096/0.0092) at stan-d

18、ard conditions (101.325 kPa and 25C 14.69 psi and 77F), the air and refrigerant masses, mole numbers, and concentrations inside the test chamber are recalculated at each time-step along the leak process. These concentrations vary depending upon the remaining components in the cylinder. This step of

19、the program is the very different compared to step seven in the REFLEAK VERSION 3.1: NIST Leak/Recharge Simulation Program for Refrigerant Mix-tures Database (NIST 2005).8. The mass of refrigerants remaining in the cylinder is the difference between the initial mass of each component and the mass of

20、 each component that has leaked at a given time step. The remaining mass of refrigerants inside the cylinder are then taken as initial masses and calcu-lations continue until there is no liquid remaining in the cylinder. When the vapor volume quality ( ) is larger than 0.999 the calculation stops. F

21、igure 1 summarizes the algorithm to calculate concentration changes over time.Simulation ResultsConsidering a 10 105m3(0.006 ft3) cylinder that was evacuated (10 Pa abs 14.5 104psia) and then filled with 90 g (0.198 lb) of R-134a and 75 g (0.165 lb) of R-32, the mass concentration of R-134a was 54.5

22、% (45.5% of R-32). The refrigerant blend leaking from the cylinder in the vapor phase was released inside the 1 m3(35.31 ft3) test chamber initially filled with air at standard conditions.The test chamber was assumed to be air tight. A 1 g (2.2 103lb) leak sample was removed during the vapor phase a

23、t each sampling step. Calculations ended when the vapor volumetric quality in the cylinder were larger than 99.9%. The initial mass filled was 165 g (0.364 lb). At the end of the leak process, 7 g (0.016 lb) of the vapor phase remained in the cylinder. So, the entire mass leaked from the cylinder an

24、d released into the test chamber was 158 g (0.348 lb), which can be seen in Figure 2. The overall molar and mass concentrations inside the cylinder along the leak process are also shown in Figure 2. At the beginning of the leak, the vapor phase showed the highest concentration of R-32, which then co

25、ntinuously decreased.The molar concentration variation during the leak scenario in the vapor and liquid phases inside the cylinder are presented in Figure 3. The vapor phase was richer in the more volatile component (based on molar concentration). At the end of the leak, the vapor remaining in the c

26、ylinder contained 70% of R-134a.The mass concentration variation during the leak scenario in the vapor and liquid phases inside the cylinder are presented in Figure 4. The liquid phase was always richer in R-134a (based on mass concentration). When 82 g (0.181 lb) had leaked, the mass concentrations

27、 of R-32 and R-134a in the vapor phase became equal. At the end of the leak, the vapor remaining in the cylinder was composed of more than 80% of R-134a (in mass concentration).As shown in Figure 3, the WCFF was met at the beginning of the leak process and was well above the lower flammability limit

28、 (LFL). So, the refrigerant blend was be classified as 1/2 (on a scale where 2 means mildly flammable). The following results (shown in Figure 5) show that none of the concentrations met in the equivalent room, which was the test chamber, were flam-mable. All of them were under the RCL value of R-32

29、, which is 20% of the LFL, according to ISO Standard 817-2006 (ISO 2006). In ASHRAE Standard 34-2007 (ASHRAE 2007), the RCL is 25% of the LFL.The molar concentrations of refrigerants inside the test chamber are presented in Figure 5. As previously mentioned, the test chamber was initially filled wit

30、h air at standard conditions. The number of air moles at these conditions was 40.87, which remained constant throughout the leak Qgv824 HVAC accepted May 12, 2009Model predictive control applied to commercial buildings requires short-term weather fore-casts to optimally adjust setpoints in a supervi

31、sory control environment. Review of the litera-ture reveals that many researchers are convinced that nonlinear forecasting models based on neural networks (NNs) provide superior performance over traditional time series analysis. This paper seeks to identify the complexity required for short-term wea

32、ther forecasting in the context of a model predictive control environment. Moving average models with various enhancements and (NN) models are used to predict weather variables seasonally in numerous geographic locations. Their performance is statistically assessed using coefficient-of-variation and

33、 mean bias error values. When used in a cyclical two-stage model predictive control pro-cess of policy planning followed by execution, the results show that even the most complicated nonlinear autoregressive neural network with exogenous input does not appear to warrant the additional efforts in for

34、ecasting model development and training in comparison to the simpler MA models.INTRODUCTION Model predictive control (MPC) of commercial buildings is highly appealing as it holds the promise of energy and cost savings through intelligent building operation. Progress toward the ultimate goal of real-

35、time optimal control of buildings is made through a systematic inquiry of each process component. Although a wealth of research has been conducted in building physics, systems, and component modeling, there remains a great opportunity for the proper integration and control of each of these elements

36、from a supervisory control perspective. Among the obsta-cles is the accuracy by which the stochastic local weather processes can be predicted. It is neces-sary to investigate suitable forecasting models to uncover the complexity required for short-term prediction of local weather, while observing th

37、e applicability within a MPC framework. This paper strives toward this goal through a systematic evaluation of various climates, weather vari-ables, and forecasting models.Time series prediction of local weather is crucial for many aspects of energy conservation, eco-nomic operation, and improved th

38、ermal comfort in commercial buildings. In particular, signifi-cant motivation exists for HVAC supervisory control applications (e.g., building thermal mass control to respond to utility pricing signals, increased free cooling through superior economizer control, and thermal load prediction for unifo

39、rm chiller loading and improved part-load perfor-mance). The desire is to optimize operation based on a short-term prediction of weather and building utilization to yield energy and cost savings through the minimization of an objective function over the prediction horizon. Identifying a MPC strategy

40、 and determining the forecasting Anthony Florita is a doctoral student and Gregor Henze is a professor in the Department of Civil, Environmental and Architectural Engineering at the University of Colorado at Boulder, Boulder, CO. 2009, American Society of Heating, Refrigerating and Air-Conditioning

41、Engineers, Inc. (www.ashrae.org). Published in HVAC the availability of accurate dynamic building energy simulation programs further promotes future application of MPC in commercial buildings.In terms of the weather forecasting task investigated in this study, it is well known that govern-ment and p

42、rivate institutions alike provide forecasting services involving complex meteorological models and supercomputer computation. The arguments against relying on these services are the five main concerns: availability of all variables of interest, service interruption, service availability, system inte

43、gration, and the data-driven perspective. First, the presented models can be applied to predicting local solar data such as global horizontal and direct normal insolation, which is not com-monly offered by commercial service providers. Second, service interruption could seriously negate the benefits

44、 of model predictive control. For example, with electrical demand charges typi-cally being levied over monthly bill periods, and ratchet clauses extending this charge to future billing periods, any failure of demand limiting strategy caused by a forecasting malfunction could have large cost implicat

45、ions. Third, the availability of hourly (or subhourly) weather forecasts is not pervasive. That is, forecast updates required in realtime, necessary for error minimization, are not available through most forecasting services. Error can additionally be introduced through localized deviations (e.g., t

46、he grid point for which the forecast was generated does not accurately represent local climatic observations). Fourth, system integration concerns stem from the authors belief that the smart building benefits from an on-site weather station for appropriate model predic-tive supervisory control. Fift

47、h, a practical, data-driven (on-site) perspective is favored over devis-ing complex (physical) probability models of, say, solar irradiation and/or cloud cover. Therefore, a range of weather forecasting models for MPC have been evaluated, from simple models that only capture the shape of the weather

48、 processes to complex models that attempt to capture the stochastic processes and their underlying dynamics. PREVIOUS WORKA number of relevant case studies with weather prediction aspects were reviewed. Many of the studies came from the very active research domain of electrical power forecasting use

49、d to opti-mally dispatch central power generation equipment, with weather forecasting typically influencing the unit commitments. Despite overlaps in forecasting strategies and algorithms, the present con-cerns limit the literature review to studies involving commercial buildings with weather forecast-ing and/or prediction features to corroborate the unique investigation presented herein. In both domains, a significant fraction of researchers are convinced that nonlinear forecasting models b