Velocidad Variable En Compresores.pdf

396

IEEE Transactions on Energy Conversion, Vol. 7, No. 3, September 1992.

A Study and Evaluation of Power Electronic Based Adjustable Speed Motor Drives for Air Conditioners and Heat Pumps with an Example Utility Case Study of the Florida Power and Light Company Alexander Domijan, Jr., Member, IEEE

Omar Hancock

Florida Solar Energy Center 300 State Road 401 Cape Canaveral, Florida 32920

Florida Power Affiliates & Power Electronics Consortium Department of Electrical Engineering University of Florida, Gainesville, FL 3261 1

Abstract - In this paper is investigated an emerging technology called “power electronics” with an evaluation of its potential for air conditioning and heat pump applications in Florida. The study also addresses the power savings to be expected for the customer and for the utility in its management of load. Furthermore, data and a computer flow chart are provided so that the information herein may be transferred and used by engineers in other geographic regions for their specific utility systems. Four heat pump models are developed and their performance analyzed and compared with each other specifically in regard to Florida climates. These models are a conventional single-speed, two-speed, variable-speed, and Oak Ridge National Lab variable-speed systems. Some of the results indicate that generally the major advantage - “greater efficiency” - of power electronic based variable-speed variable-capacity heat pumps is due to the higher coefficients of performance and energy efficiency ratios resulting from oversized heat exchangers, and the decrease of cycling losses that result from on/off operation which must occur when capacity exceeds load. Key Words: Adjustable Speed Drive (Variable Speed Drive), AC and DC Motors, Variable Capacity, Air Conditioner, Heat Pump, Energy Efficiency Ratio, Peak Load Reduction, Annual Energy Savings, Energy Conservation, Bin Method, Power Electronics, Coefficient of Performance, Cycling Capacity, and Efficiency Analysis.

I “ Power Electronic (PE) based inverterAiven variable speed (also culled adjustable speed drives) variable-capacity (VS/VC) air conditioners and heat pumps will soon be common products in the North American market place. They have been sold in Japan since 1982 and the U.S. manufacturers began selling made-in-the-USA versions in 1988. Previous work by Domijan [ 11 placed into context the overall and very diverse aspects of PE based conversion technologies in terms of a strategy for energy development and utilization in the southeast region of the USA. This paper, in contrast, presents an evaluation, analysis and comparison of several specific PE based, and non-PE based, air conditioner and heat pump technologies, and their potential impact for energy savings on the Florida Power and Light Company (FPL) system. FPL is a firm with approximately three million utility customers. The work reported herein was conducted as part of a grant from FPL (see acknowledgments) [2]. This work may be of value to other utility engineers in evaluating energy development and usage strategies in utilities whose systems are being increasingly penetrated by PE based air conditioning and heat pump equipment. Since air conditioning accounts for about 30% of electrical energy consumption, or about 30 GWh per year in Florida, any means for 92 WM 028-1 EC A paper recommended and approved by the IEEE Energy Development and Power Generation Committee of the IEEE Power Engineering Society for presentation at the IEEE/PES 1992 Winter Meeting, New York, New York, January 26 - 30, 1992. Manuscript submitted August 22, 1991; made available for printing December 18, 1991.

Craig Maytrott

improving air conditioning efficiency would offer significant energy savings to the state, its utilities and consumers. Furthemore, if this energy reduction could be effected in a manner that would reduce peak loads, utilities may be able to defer or avoid altogether construction of additional generating plants. The first part of this paper deals with the assumptions made in the study. The second part provides the technical background needed to understand the study. The third part briefly evaluates retrofit as an option. The fourth part presents the four drive models and an efficiency analysis. The fifth part illustrates the program flow chart. Lastly, the sixth part of this work discusses selected results. ASSUMPTIONS AND PROJECTED EFFECTS ON FPL As will be seen in later sections of this paper, the use of PE to drive and control air conditioners and heat pumps can benefit the end user. But what effect does the use of such appliances have on an electric utility? Making an appliance more efficient in its use of electricity is not necessarily the best solution. Probably the single most important question from the utility’s perspective, is: How much reduction of peak loading will result from the use of PE in FPL’s service area? Capacities and numbers of air conditioners and heat pumps, were provided from FPL’s 1986 home energy survey. As provided, the data is lumped system-wide and is not broken down into climatic regions, nor does it include energy use data, and peak (power) loads. Calculations were made to estimate annual energy savings and peak load reductions resulting from the use of PE for both cooling and heating, using the FPL data and the scenarios of table 1. Calculation results are presented in table 2. Table 1 Assumptions for Estimating Energy Savings and Peak Load Reductions for FPL Straight Air 6 Heat Pump (cooling) Average capaclty, kBtulh Average EER Average Ilk of unit, years Total number 01 units: Replacement rate, unltslyr Variable-speed Varlable-capacity penetratlon rate, percentlno. units 1st year 2nd year 3rd year 4th year 5th year 6th year 7th year and subsequent Compressor-hours per year Average energy savings, percent

-

Window units Central systmms 12 9 10 574,241 57,424

24 10 12 1,694,676 141.223

101 5,742 ~0111.4a5 30/17,227 40122,970 5ona.712 5012a.712 5ona.712 1.200 15

Heat Pump (heating mode) Average capacity, kEtuh Average €ER Average life of units, years Total number of units Replacement rate, unitslyear Variable-speed varlablccapaclty penetration rate, percentlno. units 1st year 2nd year 3rd year 4th year 5th year 6th year 7th year and subsequent Compressor-hours per year Average energy savings. percent

0885-8969/92$03.000 1992IEEE

12 9 10 7a.156 7,816 101 782 2011,563 3012,345 4013,126 50l3.9oa 5013.90a 5013.90a 1,440 15

24 10 12 220,071 1~1,339

a i 1.467 161 2.934 241 4,401 321 5 . m 401 7.336 481 11.803 56110,270 1,440 ia

I

II

391

Table 2 Estimated Energy Saving and Peak Power Reduction Using Power Electronic ,n the FPL System

Annual Energy Savings, MWh *During 2nd year *During 11th year *During 20th year

7,200 340,000 480,000

1,100 46,000 57,000

Peak Load Reduction, MW *During 1lth year *During 20th year It should be noted that the summer demand peak occurs at the same time as the peak air conditioning need. The same sequence of events occurs in the winter, but since the winter heating peak is associated with resistance heaters being energized due to reduced capacity of the heat pumps, this peak can be modified by V S N C heat pumps. VSNC heat pumps can be over-speeded in heating mode, resulting in greater capacity, thus reducing or eliminating the need for resistance heaters. The over and under-speed capability of VS/VC heat pumps and air conditioners can have another beneficial effect on the utility grid. Some V S N C heat pumps are already being marketed with remote controls. Thus it becomes possible to control the device from a centralized location by the utility for a number of load control strategies. Annual energy savings (AES) were calculated for both air conditioners and heat pumps by the following method: AES = [(v)(w)(x)(y)l/z [MWhl where, v =capacity rating of the average air conditioner (heat pump), Btuh = 12,000 (24,000) w = number of compressor-hours per cooling (heating) season = 1,200 (1,440) x =performance improvements due to the use of power electronics = 15%, air conditioners (18%, heat pumps) y = cumulative number of VS/VC air conditioners and heat pumps that have replaced single-speed units. (Note that credit is taken for performance improvements for heat pumps operating in the cooling mode.) z = (EER)( 1 million Watts per MW); Note that the “EER,”or energy efficiency ratio, is equal to the seasonal BTU’S provided by the heat pump divided by the total electrical energy into the unit, in watts, during that season. The approach to estimating Peak Load Reduction (PLR) is different for cooling and for heating. For cooling, central air conditioners and central heat pumps are lumped together and credit is taken only for assumed oversizing. If there is no oversizing there is nothing in VS/ VC systems to perform peak-shaving. Also, it was assumed that window units tend to be undersized for cooling. Window units may have been chosen properly, initially, but most installations try to cool a larger area than was originally intended. For the summer cooling season, the PLR was calculated by: PLR (cooling mode) = [(m)(v)(y)]/z [MWI where, m = Average over-sizing for cooling load (8%) v, y, z = as per AES while the PLR for the winter heating season (heat pumps, heating mode) used the expression: PLR (heating mode) = (r)(s)(t) [MWI where, r = elecmc strip heat avoided, adjusted for diversity, KW = 2 for window units; 5 for central systems s = conversion factor = 1,000 K W N W t = number of VSNC heat pumps replacing single-speed units

TECHNICAL BACKGROUND: POTENTIAL USE OF PE To improve the system efficiency of air conditioners and heat pumps, it is necessary to reduce the power used by the three motors driving the three major elements: compressor, inside blower, and outside fan. These three motors consume more than 90% of the required power, with the largest user being the compressor, which uses about 80% in residential systems.

-

Modem heat pumps use a sealed motor-compressor unit, which contains an induction motor and a reciprocating compressor. The unit is often termed “the compressor,” but both parts must be considered in any effort to lower power consumption. The most common motor used in the compressor unit is a 230-volt, single phase, 60Hertz induction motor, running at 3500 or 1750 rpm. A few compressors are built to switch between these two speeds on command, but most run at one speed only. To change the speed significantly, we must change the frequency input or change the motor windings. PE offers the means to vary the motors speed. While the input to the PE based converter is in most residential cases single phase, the output is three phase, which is advantageous since a three phase motor needs no starting apparatus. The only disadvantage lies in the fact that almost all residential air conditioners and heat pumps now in service have single phase compressor motors, so it is not reasonable to try to install a PE package as a retrofit. The PE speed change circuit asks for an efficient three phase motor. Since compressor motors are usually several HP in size, while the fans use fractional HP motors, it is best to concentrate on the compressor motor. Table 3 summarizes several motor efficiency figures. The efficiency figures in table 3 assume that power to the motors is a 60 Hz pure sine wave. For the variable speed heat pump, we need to also consider the motor efficiency at lower frequencies (which will Table 3 Motor Efficiency Comparison for 3-Hp Induction Motors

No. of Phases 1

Ii

Motor Type

Efficiency Nom. Value

I

Capacitor Start Two-value Capacitor Standard Efficiency Energy Efficient

74% 78%

be used for most of the operating time) and with higher frequencies (for those occasions when the heat pump is called upon to work at 150% rated capacity). This kind of information is not available from manufacturers’ literature, and remains a topic for further research. The other option is a brushless dc motor. Previous work by Cathey [3] has shown that greatly improved efficiencies at wide ranging speeds can be obtained from brushless dc motors, compared to high efficiency induction motors. For example, it was found testing a high efficiency three-phase induction motor that the efficiency was 83.3% at full load and rated speed. The same motor, slowed to 30% of rated speed and supplying half of rated torque has a measured efficiency of 63.0%. For the brushless dc motor, the comparable figures were: full speed, full torque gave 84.05%efficiency, while 30% of full speed and half of rated torque gave 75.09% efficiency. Even at 20% of rated speed and 25% of rated torque, the brushless dc motor achieved about 65% efficiency. The standard efficiency induction motor was down to 41.85% and the high efficiency induction motor was down to 47.87% at this light load condition. The importance of these efficiency comparisons lies in the design strategy behind the variable speed heat pump. If the complete system is undersized for the building’s load, the user may be dissatisfied when there is a period of either inadequate heating or cooling. The customer may have bought a deluxe heat pump, but in July it leaves the room tempera-

398

ture at a warm 8 1 degrees F (withoutside temperature at 95 degrees F), and for a few cold days in January the house is a cool 68 degrees F, because that is the best the undersized unit can do. So the system should be adequately sized for the load that will be imposed. This means that most of the time the unit will be running at less than 60 Hz (60 Hz supplied converter). Especially in spring and fall, it will be running 70 to 90% of the time at 30 to 35 Hz. It is important to note that the efficiency of the compressor motor at these speeds will contribute at least as much to annual energy cost as will the efficiency of the motor on those occasions when it is running at 60 Hz or above. The efficiency of a three-phase induction motor when fed from a PE converter producing a typical six step waveform has not been fully explored, compared to the same motor’s efficiency when fed from a pure sine wave. The large harmonic content in the six step waveform permits sizable currents to flow in the motor windings, but with no useful torque produced. A course appraisal of this situation can be gained by examining the conventional induction machine steady state equivalent circuit. The useful torque produced by the motor comes from the rotor current flowing in the rotor resistance divided by the slip. Torques that harmonics of the fundamental frequency attempt to produce fail to add to the useful mechanical output, but losses associated with the resistances represent real power lost. The input leakage reactance governs the size of harmonic currents that result from harmonic voltages. Since that reactance varies linearly with frequency, the input will accept much more current from a lowfrequency harmonic than from a high frequency harmonic. This explains briefly why the pulse width modulation based PE type converter, which uses a high carrier frequency, is preferred to the usual six-step design. However, many of the adjustable speed drive systems that are sold for general application to industry produce the sixstep waveform. It is also necessary to consider other elements in the system to SUCcessfully integrate the system. The compressors in use today have all been designed to work well at one speed, either 3500 rpm or 1750 rpm in the U.S. How does such a compressor perform at lo00 rpm? What becomes of its efficiency? Does it get proper lubrication to last 10 years of typical duty? Will its bearings survive if the speed is pushed to 120% or 150%of design speed. These questions prompt consideration of different compressor designs: reciprocating, rotary including scroll, screw, etc. The compressor must be efficient, reliable, inexpensive over a range of speeds; if these requirements can not be satisfied fairly well, the PE based variable speed systems loses its appeal. The scroll design seems best suited to variable speed operation, and has been chosen by several Japanese companies producing variable speed heat pumps. Production in Japan is now one million heat pumps per year, and the variable speed units dominate the market, accounting for over 2/3 of the annual production. Between the single-phase, 60 Hz ac power line and the compressor motor, various PE based links may be chosen. Figure 1 shows most of the possibilities, where a system consists of a rectifier, a filter, and an inverter. How the circuits accomplish their functions and how accurate a replica of a sine wave results is what distinguishes one circuit from another, and is described in detail by the author in [4]. Briefly, the rectifier (elementson the left offigure I ) convert singlePhase ac to roughly dc. In figure l a diodes are used (an uncontrolled rectifier). In the other rectifier circuits, two (or four) of the diodes are replaced by thyristors (controlled rectifiers), permitting control of the dc signal from zero to full value. The filter is aimed at smoothing the pulsating unidirectional dc that comes from the rectifier. In figure la, a relatively constant voltage results, so the inverter is said to be voltage fed. In figure Ib, there is a current fed input to the inverter. The filter in figure I C includes a chopper, which turns dc voltage on/off at a high rate; when this is smoothed out, a variable dc voltage is presented to the inverter. The inverter converts dc voltage or current into three-phase ac power. The six-step inverter has been

phase

6-step inverter

Motor

A-C 230 V.

a

Pulse width modulalor

inverter

b

1

b

Figure 1 Rectifier-filter-inverter mode options: Motors (a) and (b) are high efficiency three-phase induction motors, and motor (c) is a brushless dc motor (BDCM) very popular in the past and remains the basis of many inverters. In a transistor based six-step inverter, the control inputs are fed to the transistor base leads in acyclic way so as to fix the desired frequency and to assure the proper timing sequence of the six transistors. The line-tdine voltage applied to the motor is a wave which is rich in harmonics. Fourier analysis shows that a voltage waveform having a fundamental frequency of an amplitude of 100 units also contains a fifth harmonic of 20 units, a seventh of 14.3 units, an eleventh of 9.1 units and so on. Missing are all even harmonics and integers divisible by three. The important question is: what do these harmonics do to the performance of the motor? The fundamental frequency produces a torque in the forward direction and a speed slightly slower than synchronous speed, i.e., (1-Slip)(Synchronous speed). The harmonic voltages attempt to produce different speeds in both forward and reverse directions. They fail to affect the motor’s rotation appreciably, but they do cause losses in the motor. As a result, the motor heats up and the efficiency drops. To counter the heating effect, one solution is to oversize the motor. To reduce the efficiency drop, an inverter is needed which emulates a sinusoidal waveform more closely than does the six-step design. The pulse-width modulation (PWM) scheme in figure l b synthesizes a sine-like waveform from rectangular pulses of constant voltage amplitude but varying timing. The pulses are sparse at points where a sine wave is small and dense near the positive and negative amplitude peaks. The greater the number of pulses used to synthesize the sine wave, the higher order of harmonics closest to the fundamental. For example, with 12 pulses per cycle, the first harmonic is the 11th, and its amplitude is 40% of the fundamental. If 24 pulses per cycle are used, the lowest-ader harmonic is the 23rd, again 40% the size of the fundamental. The motor has resistance and inductance in its circuit model. The higher the frequency of a harmonic, the less current it will cause in the motor. Hence, moving the lowest harmonic up in frequency cuts the motor losses for a given fundamental. By switching the output devices at 4,000 hertz, for example, the harmonics in the motor waveforms are so high as to result in only 1% or 2% loss in motor efficiency, and this only at or near 100% torque.

7 N R L I ITRY The inside blower, the fan, and control circuitry must also be adapted. The inside blower should change speed as the capacity of the unit changes with compressor speed. This can be done either with a multi-speed motor (single phase) or an ASD with a small

three-phase motor. The condenser fan should slow when the compressor slows down. If this fan is driven by a three phase motor, the same variable frequency source can be fed to both compressor and condenser fan. The control circuitry must be “smarter” than in conventional air conditioners. It must frequently measure temperatures, pressure and relative humidity to “tell” the PE section what voltage and frequency to create for the compressor and fans. This takes very little power, but does suggest enough complexity to warrant using an integrated circuit, such as a microprocessor with built in memory. / E T P NTR L Controls on a conventional air conditioner are very simple. A thermostat monitors the inside temperature and compares it to the thermostat set point. The compressor and inside air handler are either on or off, although there is usually provision for running the blower alone. The primary controls for a PE based air conditioner or heat pump must be more complex by necessity. The function of the controller is to protect the components of the heat pump from damage; to provide the greatest human comfort in the interior space; and to minimize the power required to do the job. To take full advantage of the variable speed unit, a number of temperatures must be sensed and a mounted thermistor would allow computation of the relative humidity in the air stream. A system used by Mitsubishi, for example, in their line of room air conditioners uses four temperature and two current levels for measurement, which are then fed to the main microprocessor (see figure A). This chip also creates the pulse width modulation signal that creates variable frequency, three-phase power for the motor. The conventional air conditioner wastes energy by cycling on and off when serving a load below its full capacity. With variable frequency power supplied by the inverter, the new unit does not shut down, but rather slows down, thereby reducing its capacity. For a light load, it may run at 30 Hz or whatever other frequency is necessary to keep the unit in continuous operation. The Coefficient of Performance (COP) will be higher at low speeds, for the condenser and evaporator coils are oversized at this condition of reduced capacity. (Note:The COP is a dimensionless ratio of output to input. Units of power or energy may be used. It is analogous to efficiency; however, unlike efficiency it may exceed 100%. COP is used to rate the efficiency of heat pumps or air conditioners. It can be taken on any level - component, subsystem or sjstem - and may include losses or parasitics.)

CW.“”,

prr

:

,n*

Figure A Variable Speed Heat Pump Control (from Mitsubishi): Temperatures are: T1 Condenser, T2outside ambient, T3 inlet tube, T4 outlet tube of compressor. CT1 and CT2 monitor dc and ac currents

MODELING FOR AN EFFICIENCY ANAL Y m In order to determine the power savings to be expected from heat pumps and air conditioning systems using PE, four heat pump mathematical models were developed and their performance analyzed specifically in regard to Florida climates. The models reflected single-speed, twespeed, variable speed, and Oak Ridge National Laboratory’s (ORNLJvariable speed systems. An examination of each model now follows.

C & E D This model is best explained (and is completely characterized) by reference to its assumed capacity and COP vs temperature (outside dry bulb) curves depicted in figure 2. This conventional heat pump model and its load are characterized by these following factors: l.Loads (kBtulhr) which are linear functions of temperature as shown. The load curves are defined by the two set temperatures THSET (heating mode) and TCSET (cooling mode) and the points LOAD17 (load at T = I7 degrees F ) and LOAD95 (load at 95 degrees F). salislied cooling load

J CAP47 CAP87 Since there are advantages to PE in air conditioners and heat pumps, CAP95 one goal should be to retrofit this equipment to existing units. But C there is a serious problem in attempting this. The motor is almost certainly single-phase and designed for 240 volts ac, with an incorporated starter. The superior variable speed units employ a threephase motor. Further almost all residential units use a sealed unit that combines the compressor and the motor that drives it, so one I ,Balance temp. \ ,halance temp.L i cannot simply replace a compressor motor. Thus, the only possibility of retrofit at this level is to replace the entire sealed compressor-motor unit with another having a three-phase i Cycle Cycle motor inside. In addition, the complete control circuitry in a conven! Steady slate I tional unit is totally inadequate to the requirements of a variable speed heat pump. The advantages of variable speed are lost unless the unit is advised of its correct speed under a variety of conditions. Thus the idea of retrofit should be abandoned. There have been built, however, several million variable speed units that incorporate threephase motors for the compressor and sophisticated integrated circuits to control the operation of the entire heat pump. The best Note Heavy lines are ellechve syslem curves. course seems to be to allow the current air conditioners and heat pumps to serve out their time. When they fail, hopefully a modern 17 42 47 87 95 Temp.‘F design of variable speed heat pump will be on the market at a sensible price, will offer cooling, dehumidification and heating at a high effiFigure 2 Characteristic Curves of Single Speed Heat Pump: ciency, and will provide superior comfort compared to the convenCapacity and COP as a function of Temperature tional units in use in the USA today.

400

2.Capacities (kBrulhr) which are also linear functions of temperature except for discontinuities in the heating mode at T = 42 degrees F. Heating capacity in the range of 17 to 42 degrees F is presumed degraded from the otherwise calculated value by an amount which is a linear function of the difference (T - 17 degrees F), varying to a maximum degradation at T = 42 degrees F. The base capacity curves are defined by the points CAP17 (hear pumps capacity at I7 degrees F ) and CAP47 (heating mode) or CAP87 and CAP95 (cooling mode). Although, there may appear at first glance to be a discrepancy involving T = 42F, CAP47 and COP47, this is not the case - the only deviation is the defrost region degradation which is explained and labeled.. 3.Resistive backup heat (Kbtulhr) = 3.412 times the parameter RESESTPOWER (KW) for temperatures below the heating balance point. 4.Linear (with temperature) heating and cooling steady-state COP curves defined by the four values COP17, COP47, COP87, and COP95 as shown, except that the heating COP’s are assumed degraded by a variable amount in the defrosting region 17
tion at low-speed otherwise. The authors acknowledge that this presumed mode of operation will yield pessimistic results; a more realistic two-speed model would presumably cycle between low and high speeds in the inter-balance-point temperature region and thus avoid some cycling loss. 4.A low speed COP (steady state) curve lies above the high speed curve. Each of the four curves is defined by the points shown. 5.Potential unsatisfied loads, make up heat, and defrost and cycling losses are identical in form to those discussed for the single speed unit. 6.Effective (actual)COP’s are defined by the heavy curves in the figure, to include discontinuities at 42 degrees F and at the two low-speed balance point temperatures. Note that for all results and simulations presented herein the highspeed characteristics (i.e.,capacity and COP) of the two-speed heat pump have been taken as identical to the characteristics of the single-speed heat pump. The identical statement will apply to the variable-speed heat pump to be discussed. In eitner text or graphics the term “max speed” will be used interchangeably with high-speed in discussing the variable speed heat pump. VARIABLE-SPEED HEAT PUMP MODEL The variable-speed heat pump is characterized in figure 4. The principal difference in this heat pump model and that of the two-speed model is that it is assumed that, by modulating compressor (and perhaps fans) speed, the heat pump can generate any capacity between the extremes defined by the high-speed and low-speed limits.

I

TWO-SPEED MODEL The two-speed model is similar to the single speed model. The heat pump model is characterized in figure 3; it shows two capacity curves in each mode of operation and, similarly, two COP curves. Implicit in figure 3 are these features: 1.The low speed capacity is presumed to be less than the high speed capacity 2.At any given temperature the low speed capacity is presumed to be a constant times the high speed capacity. 3.Because of constraints imposed by the temperature-bin method used in the annual performance simulation code, the resulting pump capacity is presumed, given by the heavy lines in figure 3, i.e., operation at high speed for all temperatures below/ above the low speed heating/cooling balance point temperatures; cycling operation at high speed between those and the high speed heating/cooling balance point temperatures; and cycling operaHigh speed capacl

n 0

Note: Heavy lines are effectivesystem CUNes. 17

42

47.

E7 95

Temp.OF

Figure 4 Characteristic Curves of a Variable-Speed Heat Pump: Capacity and COP as a Function of Temperature

9s

a 0

\ Cycling tosses’ Note: Heavy h e r are elfective ryatem curves. 17

42

47

87

95 Temp.’F

Figure 3 Characteristic Curves of Two Speed Heat Pump: Capacity and COP as a Function of Temperature

Moreover it is further assumed that it will indeed operate (because ot its inherent “smart” controller) at a capacity identically equal to the load. Thus the capacity curves are those defined by the heavy lines in the figure 4 plot of capacityfload. The resulting effective COP curves are also given by the heavy lines on the COP plot of figure 4. Implicit there is the approximation that the COP at intermediate speeds (capacities) varies linearly between the minimum speed and the maximum speed levels. Note that, for loads less than the minimum speed capacity, the variable speed heat pump will cycle on/off at the minimum speed level, and there will be a resulting COP degradation in both heating and cooling modes. d P MODEL The results of an extensive test of a variable speed heat pump were made available recently in [ 81. The nature of the performance data

40 1

available dictated a somewhat different model be created for the ORNL unit, specifically: 1.Heat pump COP was found to be a strong, nonlinear function of capacity as indicated in figure 5. In the model the data points shown were fitted to a quadratic polynomial, then linearly modified with temperature deviation from the 40 and 82 degree F points. 2.The heat pump data just discussed were obtained by modulating indoor fan speeds to yield optimal performance. 3.Consult [5] for additional information on this model.

I

then analysis of the model gives the results shown in table 4 for cycle period and degradation factor Cd (effectiveCOP = COPsstCd). The terminology used is defined as follows for purposes of clarification: 1.CAPss is the steady state capacity at the balance temperature. 2.2 is a dimensionless ratio of load to CAPss. 3.2 prime is a dimensionless ratio of equivalent run time at steady state capacity at the minimum run time to the minimum cycle time. 4.Cd is a variaole degradation factor as a function of cycle timing which in turn is a function of load. 5.Effective COP is the steady state COP divided by the degradation factor. Table 4 Cycle Time and Degradation Factor, Linear Rampin Capacity

I H.nng

I Cycle time (period)

z=

mod.

MINRUN

2

8 L

2a

3

;

1 -

coolhlg mod. I

I

I

I

I

I e 1 2’ (light load)

I

Degradation factor, CO

-2

I

MiNRUN MINRUN

-

Z(MIN0FF) +

TAU -

Z(MIN0FF) +

TAU

2 2 Z‘(hesvy load) 1-2

The model yields reasonably conservative degradation factors; for example, if MINRUN is 5 minutes and TAU is 100 seconds then at light loads Cd =6/5 (the effective COP would be 516 of the steady state COP). An examination of the light-load Cd shows that both the minimum running time and the capacity buildup time constant TAU have approximately equal adverse effects on performance: K Change in CO

“h Change in TAU

- - ‘h Change in CO

-

% Change in MINRUN

-

-

TAU 2 MINRU~

2

PERFORMANCE SIWLATION PROGRAM The simplified overall program flow chart is shown in figure 7. The annual performance simulation uses the “temperature bin” method: hourly data from [ 1I] for Jacksonville, Fort Myers, and Eglin Air Force Base (all in Florida) are grouped into 22 bins at 5 degree F intervals, with the initial bin center temperature at 2 degrees F. The temperature bin method uses multiple calculations across the range of outside hourly temperature bins, in association with the corresponding equipment part load characteristics. More detailed flow charts, including the heating and cooling modes of the models developed herein, and complete computer codes for the heat pump specific models are provided in [2]. The code is written in IBM basic, and has been executed on an IBM PC.

n

I,

N Y N U

a

I0 RUNT~YE.+.OFFT~ME

Figure 6 Model Generic Heat Pump Capacity During On/off Cycling complex function, of course [9, lo]; for the comparative purposes of this study the model used is adequate. The model assumes the following: l.A linear r a m p u p in capacity from zero to the steady state value CAPss, reaching steady state at the time constant, TAU. 2.A minimum prescribed running time, once started, of MINRUN time units. 3.A minimum prescribed off time of MINOFF time units. 4.That the input power to the pump attains and maintains its steady state value immediately at start up. Given-the four constraints and defining Z by:

LOAD

2:

and 2

CAPu

=

MINRUNMINRUN + MINOFF

SELECTED m U L T S FROM THE ST UDY In this section data from a limited exercising of the three “generic” and the ORNL variable speed models are presented. Except where explicitly stated otherwise all results shown are for the following nominal parameter values: l.TAU, the cycling time constant is 100 seconds [121. 2.MINRUN (the heat pump’s minimum allowable running time, once activated) is six minutes. 3.MINOFF (the heat pump’s minimum allowable off time, once shut off) is also six minutes. 4.The maximum (at42 degrees F) degradation in capacity due to defrost operations/cycles, DELCAP, is 15%. 5.The maximum (at 42 degrees F ) degradation in COP due to defrost operations/cycles, DELCOP, is 20%. 6.For the two generic multi-speed heat pumps, the ratio [(lowspeed COP)/(high-speed COP)] versus capacity curve is modeled as in figure 8. The model depicted is fairly conservative. Theoretical considerations and data from [8,13], as well as several Japanese sources 114, 151, indicate that low-speed COP could exceed high-speed COP by

~

402 Single speed

Select pump load, location Fill temp. bins GOSUB 20000 GOSUB 30000 Fill load curves GOSUB 15000 Fill other Daramelers (AIIOWS uier interaction) Find ORNL COP Poly. Cocls. GOSUB 21000 Find ORNIJTVA Driveq Poly. Coelr. GOSUB 22000 Set T z 2OF

--

ringlcspeed model

F n a l varlabie rpeed

11000- Generic

I ORNL model

I

I '

--

Go lo pump specilic routine GOSUB 18000 All pumps ORNL GOSUB 16000 6 19000

I

Update cumuiativevalues: 300 Resistive (backup) energy Input healing/cooiing energy Unsatlslied haalinglcooilng load Healing/cooling running Ilme No. of heatingkooling on/olf cycles Healinglcooling load

I GOSUB 40000 GOSUB 42000

CAPFAC .5

1

- results on monitor

- results on printer

;

I

ORNL with TVA BDC drive

10

SEER Caic. ORNL Ian power

C,, T cyclr, elc.

Figure 9 Cooling Seasonal Energy Efficiency Ratio (SEER), five generic heat pumps matched to load, Ft.Myers weather data l.A variablexapacity heat pump with an inefficient drive is probably not worth consideration, and; 2.Performance of the generic variable-speed unit is close to that of the ORNL unit. To investigate the effects of changes in capacity and COP due to defrost degradation, all four heat pump models were exercised under matched load for the Jacksonville, Florida weather. The maximum degradation (at 42 degrees F ) was cut by 50%;the effect on Annual EER's are shown in table 5. Table 5 Effects of Improvements in Defrost Performance on EER

U 19000- Calc. Mod. ORNL COP

-

I I

I

1

I Quantitv cut

Percent improvement in "annual" EER

Poly. Coats.

Output lo printer

Determine load lines

Figure 7 Simplified Overall Simulation Program Flow Chart about 40%. Additional information may be found in [8,15] with regard to this topic. 7.Interior temperatures TCSET and THSET were, in all cases, 73.5 and 64.4degrees F, respectively; thus none of the hours in the bins centered at 67 and 72 degrees F affected any of the results. 8.For the ORNL variable speed heat pump, the COP'S are determined as discussed previously, based on ORNL data as shown in figure 6. Figure 9 shows the cooling season performance (rhe Energy EDciency Ratio (EER) during that season), for Fort Myers, Florida weather, of the three generic heat pumps and two versions of the ORNL model: one with a brushless DC drive and one with an AC drive. The implications of this one simulation are: S I Z I !+ 1.2

p-1

,

816

1.0

I

2

I

ORNL with TVA AC drive

Determine heating or cooling mode Find no. of hours in bin Find load al this Iemperalure

,

Nominal 2-speed

I

c

two-speed model

I

CAPFAC = .5

.5

.6

1

Fractional capacity

Figure 8 Model low-speed COP improvement vs fractional capacity, generic heat pumps

DELCAP (degradation in capacity at 42' F) DELCOP (degradation in COP @ 42O F)

1

l-speed

Variable -speed

-.3

+.5

+1.6

I

+1.9

2-sp-d

+.2

-.2

I

+1.99

I

OANL D C drive

I

+1.96

I

Additionally, the single-speed unit model was run at matched load with Ft. Myers weather; as expected the performance was insignificant: a 50% cut in capacity degradation at 45 degrees resulted in a 0.08% decrease in annual EER; the same magnitude of change in COP at 42 degrees F improved annual EER by 0.2%. Not unexpectedly, insignificant improvements in defrost performance in the Florida climate did not result in significant returns in annual performance. improvement. Figure 10is a plot of the cooling Season EER (SEER) at Jacksonville, as a function of the capacity-buildup time constant, TAU, for the three generic heat pump models (all subsequent data will concern only the generic models). The plots for the conventional (single-speed) heat pump confirm that undersizing is to be highly preferred to oversizing, and show that one can expect a one point loss in seasonal EER for every minute of delay in capacity build up. As expected, the cycling penalty is not as great for the multiple speed units, but is still significant. With respect to the variable-speed heat pump, the data show that: 1.If cycling losses are small (TAU < I minute),an oversized heat pump gives highest seasonal performance (a greater percentage of the time is spent ut lower speeds, hence higher COP'S). 2.If TAU is greater than one minute, cycling losses begin to dominate the benefits which accrue from the higher low speed COP. The results for the two-speed heat pump are, at first inspection, surprising and seemingly at odds with the other results; but are explainable: if the cycling penalty is large (s > I), it is clearly better to cycle

403

ously variable capacity unit versus a two-speed unit if the minimum capacity is over 60% of the maximum capacity. Plots of annual hours of operation of the three generic heat pumps at two of the three locations investigated are given in figure 12 for matched/25% oversized/25% undersized units. These data are also for the case TAU = 1 minute and CAPFAC = 0.5. The multi-speed units clearly operate for substantially more hours than the single speed unit, with the difference being nearly insensitive to under/ oversizing. The substantially greater operating hours of the multispeed units may be viewed by some as an indicator of decreased reliability, but this is not likely the case. There is probably a strong advantage to the increased cooling season running hours: dehumidification performance will be improved.

12

11

10

9 E U

4m-

8

7

3SOo

-

3000

-

2500

-

2000

-

t 3 I * Has some unseUs11.d load CAPFAC = 0.S

=d

A 0

1

2

-2 -a c

1

TAU(M1n)

d .

Figure 10 Cooling Season EER vs “Time Constant” TAU for Jacksonville, Florida (CAPFAC = 0.5) at low-speed, where the COP is 20% better than it is at high speed. Note, of course, that initial cost does not enter into these results. In all cases shown herein, it should be. noted that the results for an undersized heat pump are slightly misleading in that the simulation code gives an upward biased result when there is an unsatisfied load (because a lesser percentage of operating time is spent at the lower COP extreme temperatures). Figure 11 is a plot of the three generic heat pumps annual EER versus the parameter CAPFAC for Jacksonville’s weather, with TAU, the capacity build-up time constant, a nominal 100 seconds. The plot indicates that a 25% oversized variable speed unit can yield twice as much improvement as can the equivalent two-speed unit if the variable-speed unit can maintain its COP advantage down to 20% of high speed capacity. Further, there is little advantage to a continu-

CAPFAC = 0.5

I

IS+

-r

I

I

I

Matched

25%

25% Ovwsked

Undenlred

-

Sire

Figure 12 Annual Hours of Opexation at Jacksonville and Ft. Myers, Florida versus heat Pump Size Figure 13 is a plot of the normalized energy cost for space conditioning at the three locations as a function of heat pump sizing for all three generic heat pumps. The data are for the case TAU = 1 minute

11 L

TAU = 00 seconds CAPFAC = 0.S

10

6 ul

-

0.12

-

0.11

-

0.10

0

2-speed

Verhblq

8

FLMyen

.I

.2

.3

.1

.I .o .7 CAPFAC

.8

.9

1.0

Figure 11 Annual Seasonal EER at Jacksonville vs CAPFAC (capacity factor) for TAU = 100 seconds

2% Overshe

-

Matched @%S

1

25% Undushe Some un~llslled ked

Figure 13 Normalized Energy Cost vs Heat Pump Sizing

404

and CAPFAC = 0.5. In cooling season dominated Ft. Myers costs for the single-speed unit increase with size, but in the other two locations are offset by less resistive makeup heat. For the variable-speed unit costs decrease with size at both Eglin Air Force Base and Jacksonville, but are insensitive to size at Ft. Myers. Results for the t w e speed unit are similar, the peak at the matched load point is probably noise caused by the large bin size. All undersized heat pump cases involve some unsatisfied loads; both heating and cooling at the two northern locations, but only cooling at Ft. Myers. CONCLUSIONS The advantages which power electronic based variable-speed variable-capacity air conditioners and heat pumps offer the consumer will insure them a substantial share of the market, and thus they will be penetrating the electric utility grid in increasing numbers. The major advantage, greater efficiency (and,possibly, lower electricity bills) is due primarily to two mechanisms: 1.Higher COP’S(and EER’s)resulting from “oversized” heat exchangers. Oversizing occurs when the refrigerant flow rate is decreased from full capacity flow rate; and 2.Decrease of the cycling losses that result from on/off operation which must occur when capacity exceeds load. This decrease will contribute to even higher COP’s and EER’s. There are also two disadvantages to PE based V S N C heat pumps. The first is certain, while the other is mentioned as a word of caution, as further study is required. 1.There will be a greater first cost. After PE based VSNC units achieve a degree of market maturity, their costs should not be greater than 10% more than the costs of single-speed units. Additional costs will result from the use of additional and more costly components: the outside heat exchanger may be larger, more sensors and at least one microprocessor will be required. Also, the home-owner’s control interface will likely be more sophisticated. 2.Greater care in the design may be required and field installation tc protect the PE devices and integrated circuits from the effects of lightning. However, since these or similar devices are being used successfully in other applications, suitable protection devices and/or schemes will probably be developed. Pulse-width-modulated PE inverters are currently the best available motor drives, as low frequency harmonics and their attendant losses can be easily reduced. Some of them are utility interactive and have essentially zero harmonic content below the 20th harmonic. High harmonic content motor drives are lossy and, consequently, poor choices for VSNC PE based heat pumps and air conditioners. ACKNOWLEDGMENTS The authors wish to acknowledge the support of FPL under contract number 31727-87874 for the funding to the Florida Solar Energy Center which permitted this work to be performed. The authors also express their appreciation to Dr. J.F. Schaefer, Mercer University, and Dr. R.L. Walker, University of Central Florida, for their participation in this study. [l]A. Domijan, “Formation of a Strategy for Energy Development and Utilization with Power Electronic Conversion Technologies: A Southeast U.S.A. Regional Approach,” In Press for publication in the IEEE Transactions on Energy Conversion,and 1991 IEEE PES Summer Meeting, Paper No. 91SM328-5-EC (Sponsored by Energy Development and Power Generation Committee), July 29, 1991. [2]Power Electronics for Air Conditioners and Heat Pumps, FSECCR-222-88, Florida Solar Energy Center, Aug.5, 1988, pp. 1-105. [3]J.J. Cathey, “Efficiency Comparison of Inverter-Fed Induction Motor Adjustable Speed Drives and Brushless DC Motor Drives,” TVA/OP/ED&T - 86/84, April 1987, TVA, Division of Energy Demonstrations and Technology, Chattanooga, TN. [4]A. Domijan, and E. Embriz-Santander, “A Summary and Evaluation of Recent Developments on Harmonic Mitigation Techniques

Useful to Adjustable Speed Drives,” In Press for publication in the IEEE Transactions on Energy Conversion, and 1991 IEEE PES Summer Meeting, Paper No. 91SM333-5-EC, July 30, 1991. [5]J.W. MacArthur, and E.W. Graid, “Prediction of Cyclic Heat Pump Performance with a Fully Distributed Model and a Comparison with Experimental Data,” ASHRAE Transactions, Vol. 93, Part 2, 1987, pp. 1159-1 178. [6]D.A. Didion, and G.E. Kelly, “New Testing and Rating Procedures for Seasonal Performance of Heat Pumps,” ASHRAE Journal, Vol. 21, No. 9, September 1979, pp. W . [7]U. Bonne, et al, “Electric-Driven Heat Pump Systems: Simulations and Control 11,” ASHRAE Trans.,Vo1.86,1980, pp. 687-705. [8]W.A. Miller, “Steady-State Refrigerant Flow and Airflow Control Experiments for a Continuously Variable Speed Air-tc+Air Heat Pump,”ASHRAE Trans., Vol. 93, Part 2,1987, pp. 1191-1204. [9]P. Domanski, and D. Didion, Computer Modeling of the Vapor Compression Cycle with Constant Flow Area Expansion Device, NBS 155, May 1983, National Bureau of Standards, Center for Building Technology, Washington, DC, pp. 1-148. [ 10lS.K. Fisher, and C.K. Rice, The Oak Ridge Heat Pump Models: A Steady-State Computer Design Model for Air-to-Air Heat Pumps, ORNL/CON-80/Rl, August 1983, Oak Ridge national Laboratory, Oak Ridge, TN, pp. 1-187. [ 11lU.S. Department of Air force, Engineering Weather Data - Facility Design and Planning, AFM 88-29, July I, 1978, Departments of the Air Force, the Army and the Navy, Washington, DC. [ 12lM. Khattar, “Power Electronics Applications to Air Conditioners,” Internal Memorandum, Florida Solar Energy Center, Cape Canaveral, FL, January 1986, pp. 1-5. [ 131Lennox Indusaies, HS 14 Landmark IV Condensing Units, Product Literature, March 1, 1981, pp. 1-22. [ 14lY. Shimma, et al, “Inverter Control Systems in the Residential Heat Pump AC,” ASHRAE Trans.Vo1. 91,1985, pp. 1541-1554. [15]T. Senshu, et al, “Annual Energy Saving Effect of Capacity Modulated Air Conditioner Equipped with Inverter Driven Scroll Compressor,” ASHRAE Trans.,Vo1.91,1985, pp. 1569-1584.

BIOGRAPHIES was born in 1958, and raised in Boonton, N.J. Dr. Domijan obtained his B.S. degree in electrical engineering from the Univ. of Miami, M.Engr. degree in electric power engineering from RPI, and the Ph.D. in EE from the U. of Texas at Arlington. He has been a Postdoctoral Fellow with the Energy Sys. Research Ctr. at UTA, an engineer with GE, and aconsultant with Florida Power Corp.He joined the EE faculty at University of Florida in 1987, and is presently Director of the Florida Power Affiliates (UF center for electric power engineering). He has received 15 grants as PI in power system modeling and simulation, educatioc, instrumentation, finite elements, power quality, and PE since being at UF. He is a member of several IEEE societies. MavtroU was born in 1956, and raised in Vineland, N.J. Mr. Maytzott obtained his B.S. degree in Engr. from Widener College. He has been a reliability engineer with Harris Corporation in Palm Bay, FL. He is presently a senior electronics engineer at FSEC working on PV system battery charge controller evaluations, electric vehicles, PE and harmonics. Omar Hancock is a retired Professional Engineer, and is formerly a project manager with FSEC. He holds a B.A. degree in physics from Baylor University. He was FSEC’s lead specialist in studying the potential of hydrogen produced by electricity from solar PV as a means for storing solar energy.