1、International Journal of HeatingJentilating, Air-conditioning and Refrigerating Research Volume 1, Number 1, January 1995 1894-95 1 7994-95 International Journal of Heating, Ventilating, Air-conditioning and- Refrigerating Research Editor Raymond Cohen, Ph.D P.E., Professor of Mechanical Engineering
2、 and Herrick Professor of Engineering, Purdue University, U.S.A. Associate Editors Arthur E. Bergles, Ph.D P.E., John A. Clark and Edward T. Crossan Professor of Engineering, Department of Mechanical Engineering, Aeronautical Engineering and Mechanics, Rensselaer Polytechnic Institute, U.S.A. Univer
3、sity of Oxford, United Kingdom Fire Research Laboratory, National Institute of Standards and Technology, U.S.A. Arthur L. Dexter, D.Phil., C.Eng University Lecturer, Department of Engineering Science, David A. Didion. D.Eng., P.E., Leader, Thermal Machinery Group, Building and Rdph Goldman, Ph.D Sen
4、ior Consultant, Arthur D. Little. Inc U.S.A. Hugo Hens, Dr.Ir., Professor, Department of Civil Engineering, Laboratory of Building Physics, Katholieke Universiteit. Belgium Ken-Ichi Kimura. Dr. Eng., Professor, Department of Architecture, Waseda University and President, Society of Heating, Air-cond
5、itioning and Sanitary Engineers of Japan, Japan Horst Kruse, Dr.-Ing., Professor, Institut fr Kltetechnik und Angewandte Wrmetechnik, Universitt Hannover, Germany Jean J. Lebrun. Ph.D., Professor, Laboratoire de Thermodynamique, , Universit de Lige, Belgium John W. Mitchell. Ph.D P.E., Professor, Me
6、chanical Engineering. University of Wisconsin-Madison, U.S.A. Dale E. Seborg. Ph.D Professor, Chemical Engineering, University of California, Santa Barbara, U.S.A. Poiiy Committee Ronald J. Kessner. chair Frank M. Coda Eugene Stamper Fritz W. Steimle W. Stephen Comstock Raymond Cohen Editorial Assis
7、tant Jenny OtletJakovljevic Fbiisher Frank M. Coda W. Stephen Comstock ASHRAE Editorial and Production Staff Robert A. Parsons, Handbook Editor Adele J. Brandstrom Christina D. Tate Ron Baker Nancy F. Thysell i Publishing Director a1995 by the American Society of Heating, Refrigerating and Air-Condi
8、tioning Engineers. Inc 1791 Tullie Clrcle. Atlanta. Georgh 30329. All rights reserved. a peer-reviewed archlval research journal for the F (2) the effect of different oils on transport phenomena; and (3) the effect of heat exchanger enhancements. New energy technologies, assuming that the fossil fue
9、l era will fade away. Second, our editors highlighted some of the newest control science in the WAC consequently, they are not different from single component refrigerants for all practical purposes. The more common occurrence is the formation of zeotropic (or non-azeotro- pic) mixtures that have te
10、mperature and composition changes during evaporation or condensation processes. There are potential advantages to using zeotropes, such as increased cycle performance, capacity control, and a wide choice of desired properties. However, there are also possible disadvantages to using zeotropic mixture
11、s, including system design changes, difficulty in charging and servicing, and composition changes during a leak (Didion 1990). Min SOO Kim is an assistant professor at the Department of Mechanical Engineering, Seoul National University, Korea. Dr. Kim was a visiting researcher and David A. Didion is
12、 the leader of the Thermal Machinery Group, Building Environment Division of the Building and Fire Research Laboratory at the National Institute of Stan- dards and Technology. Gaithersburg. Maryland. 3 4 HVAC liand Vindicate those of the liquid phase. The overall mass fraction change is also shown i
13、n this figure, as the marked curve connecting the point 4 to UJ As the temperature is lowered, the vapor, liquid, and overall mass fraction changes are greater. The mass fraction change during the liquid leak of the R-32/125/134a mixture is shown in Figure 9 as a function of mass percentage leaked o
14、ut. The vapor and liquid VOWME 1, NUMBER 1. JWm 1995 11 a n I g 8 e n 35 3.0 2.5 2.0 1.5 1 .o 0.5 T-50 “C 0.0 I I I I Mass fraction of R-32 Figure 6. Pressure and overall mass fraction changes during the isothermal vapor and liquid leaks of the R-32/134a mixture at different temperatures Wand Lrepre
15、sent vapor and liquid leak processes, and i andfrepresent initial and hal states.) O 20 40 60 80 975 loo Mass percentage leaked out, % Figure 7. Mass fraction change during the isothermal vapor leak of the R-32/125/134a mixture as a function of mass percentage leaked out 12 R-125 Figure 8. Mass frac
16、tion change during the isothermal vapor leak of the R-32/ 125/ 134a mixture at several temperatures (u and I represent vapor and liquid phases, and t andfrepresent initial and final states.) mass fractions of the most volatile refrigerant in the cylinder decrease slightly because of the pressure dec
17、rease caused by the leak. The mass fractions in the leaking liquid are represented by X, The overall mass fraction of the most volatile refrigerant (, o 4 8 I I O 5 10 15 20 25 28.7 30 Mass percentage leaked out, % Figure 11. Mass fraction change during the adiabatic vapor leak of the R-32/134a mixt
18、ure as a function of mass percentage leaked out 1.5 1 .o 0.5 T-lO”C T - 41.1”C 0.0 0.0 0.2 0.4 0.6 0.8 Mass fraction of Fi32 1 .o Figure 12. Pressure and mass fraction changes during the adiabatic vapor leak of the R-32/134a mixture 16 HVAC R-32/ 134a (30/70) and R-32/ 125/ 134a (30/ 10/60). The mas
19、s fraction change in the system is presented as a function of the mass percentage leaked out of the system. In the isothermal leak process, both vapor and liquid mass fractions of the most volatile component decrease during the vapor and liquid leaks, and the overall mass fractions of this component
20、 decreases during the vapor leak, but increases during the liquid leak. As the temperature is lowered, the overall mass frac- tion change becomes greater. In the adiabatic leak process, the liquid mass fraction of the most volatile component decreases while the vapor mass fraction of this component
21、increases. The temperature and pressure inside the cylinder decrease drastically for the adiabatic vapor leak and drop slightly for the adiabatic liquid leak. The results of this study show that the refrigerant mixture left in the system remains in a nonflammable region during the isothermal vapor l
22、eak for both cases of binary and ternary mixtures. The highest fraction of the flammable component, R-32, in the mixture is obtained in the vapor phase at the initial state. In the adiabatic leak process, the highest fraction of R-32 is obtained in the vapor phase at the final state of the leak proc
23、ess. Since the simulation in this study is based on REFPROP 4.0, which uses the Cama- han-Starling-DeSantis equation of state, its precision can be no better than this equa- tion represents the vapor-liquid equilibrium data. Experience so far has indicated two possible problem areas. For pure compon
24、ents, predicted values above a reduced tem- perature of about 0.95 (i.e. within about 15 - 2OoC of the critical temperature) may be suspect, particularly for derived data (for example, enthalpy or entropy). Mixtures, too, have this problem since the predicted mixture data is based on individual comp
25、onent data. Thus one must be careful that the mixture data is not within 5% of the critical point for any component. In addition, the mixing coefficients, when predicted, may be suspect. REFPROP states when they are predicted and when they have been measured. Examples in this paper specifically avoi
26、ded the above problem areas. In reality, a refrigerant charging process is close to the adiabatic leak process. If the system is charged with a liquid refrigerant from the bottom of a charging cylinder, the mass fraction change for a zeotropic mixture is negligible. One of the problems associ- ated
27、with a leak is how to measure the amount of refrigerant leaked out of the system. If the initial pressure, amount of charge, and the composition of the zeotropic refrigerant mixture are known, the amount leaked out of the system can be estimated by measur- ing the pressure of the system and applying
28、 the Equations described in this paper. ACKNOWLEDGMENTS The authors thank Mr. Terry G. Statt, Project Manager at Electric Power Research Institute (EPRI), and John Ryan of the U.S. Department of Energy fot their sponsorship, and the Korea Science and Engineering Foundation (KOSEF) for its financhi s
29、upport of M.S. Kims stay at NIST. The authors recognize the invaluable contributions of Dr. Gra- ham Momson of the Thermophysics Division at NIST. He passed away in August, 1993 and his support and guidance will be greatiy missed. Finally, the authors appreciate the collaboration with Dr. Donald B.
30、Bivens and Dr. A. Yokozeki of Du Pont for suggestions made in the early stages of the models development. 20 NOmNCLATURE h, N nt “u P Q T V X n U U X enthalpy of escaping refrigerant, W/kmol total number of components total number of moles, km01 number of moles in the liquid phase, hol number of mol
31、es in the vapor phase, km01 pressure, MPa molar quality, Q = nu/n temperature, “C internai energy, W/kmol system volume. m3 molar volume, m3/km0i, u = v/n , liquid mass fraction liquid mole fraction. xt = 7 = - - n, z.1 J=l Y vapor mass fraction n“, nu, y vapor mole fraction. Y = = - nu 2 overall ma
32、ss fraction J = 1 c nu, z overall mole fraction E Subscripts and Superscript L initial state, i-th component f hal state 1 liquid phase u vaporphase escaping fraction of the total moles, E = An/n property at the next step after a leak REFERENCES ASHFUE. 1992. ANSI/ASHRAE Standard 34-1 992, Number de
33、signation and safety classificaffon of refrigerants. Atlanta: ASHRAE. Blaise, J.C., T. Dutto, and J.L. Ambrosino. 1988. First industrial application of non-azeotropic mixture. Int. J. Re c w P = UDD confidence limit of I t-66 t-46 t+26 t+46 t+66 t+96 time of last reading t-26 lt Time Figure 2. Schem
34、atic of energy per sampling internai for a demand internai with 15 time steps and a forecast length of 9 time steps L al a x al P iz t-46 t-36 t-26 i-6 t t+6 t+26 t+36 t+46 time of last reading / Time Figure 3. Sliding window metering with four readings per demand internai. Figure 2 depicts the prob
35、lem of finding the load shedding requirements for a demand interval that ends in 9 sampling intervals when there are 15 samples per demand inter- val. Figure 2 shows the target energy consumption per sample interval and the energy consumed during previous sampling intervals. In this paper, we presen
36、t an algorithm for controlling energy consumption below a target level by shedding non-essential loads. Load shedding requirements are determined from the upper confidence limit of the uncontrolled electric consumption. 24 HVAC+.+oj+.+o; n Substituting Equations (9) and (12) into Equation (1 1) give
37、s o1 U, = C,+St+za;J1+2+ .+j+.+ n = c,+s,+zao,J (12) Equation (13) relates the upper confidence limit of the uncontrolled electric energy consumption per time step for the next n steps with the following: energy used and shed VOLUME 1, NUMBER 1. JANUARY 1995 27 during previous time step, standardize
38、d normal deviate, standard deviation of one-step ahead forecast errors and n. Equations (6), (), and (8) can be used to adaptively esti- mate the standard deviation of one-step ahead forecast errors and a table that relates the confidence level (1 - a) to the standardized normal deviate, %, can be f
39、ound in Spiegel (1961). Box et al. (1978) or Wadsworth (1990). Energy Balance to Determine Load Shedding Requirement The objective of demand-limiting control is to maintain electrical energy used over the demand interval below a target level T. An energy balance is used to determine projected load s
40、hedding requirements from the energy used during the current demand interval, the target energy level per demand interval and the upper confidence limit of the uncon- trolled electric consumption. For a demand interval that ends in n time steps, the energy balance is N- n-1 The first term on the rig
41、ht-hand side of Equation (14) is the projected energy con- sumption for the next n time steps and the summation on the right-hand side of Equa- tion (14) is the electrical energy consumed over the last N-n time steps. Equation (14) is based upon maintaining energy consumption over the demand interva
42、l below a target level. This equation does not restrict energy consumption on an individual time step basis. Thus, the energy consumed for a single sampling interval may be greater than the target energy level divided by the number of samples in a demand interval. For example, the energy consumption
43、 can be greater than the target energy level per sample interval shown in Figure 2. Solving Equation (14) for loads that need to be shed per time step until the demand interval ends, 5, , gives Equation (15) relates projected load shedding requirement per time step to the upper confidence limit of t
44、he uncontrolled electric consumption, the target energy level, the number of samples until the end of the demand interval, and the electrical energy con- sumed for the demand interval of interest. The upper confidence limit of the uncon- trolled electric consumption can be determined from Equation (
45、13). Determine Additional Loads to be Shed or Restored The load shedding algorithm should contain a table of the following information: devices that can be turned off, electric consumption of load shedding devices, minimum time that a device can be turned on, minimum time that a device can be turned
46、 off, maximum time that a device can be turned off, time before a load can again be consid- ered for shedding after it is restored, and order in which loads are to be shed. This infor- mation and the projected shedding requirement are used to determine additional loads to be shed or restored during
47、the next time step. The projected shedding requirement per time step is determined from 28 HVAC&R RESEARCH for fixed window metering for sliding window metering 5, is determined from Equations (6). (7). (81, (131, and (15) and s, is determined from estimates of S, and Equation (1). If the summation
48、of electrical loads shed over the past time step that can remain shed during the next Ume step is greater than the projected shedding requirements, then no additional loads need to be shed and loads that are being shed can be turned on (i.e., loads can be restored). If the summation of loads which a
49、re off and can remain off is less than the projected shedding requirements, then additional loads need to be shed. Additional loads to be shed or restored can be determined from: where L, +off = loads which are being shed and can remain shed ZL, off = additional loads that need to be shed = additional loads to be restored . ELo, Equations (17) and (18) require the building control system to know when loads are going to be restored due to the maximum Ume iimit for shedding loads. Knowledge of when loads are restored helps maintain the electric consumption below the t
