1

IECEC-98-I088

33rd Intersociety Engineering Conference on Energy Conversion

Colorado Springs, CO, August 2-6, 1998

A New Thermally Driven Refrigeration System with Environmental Benefits

Charles A. Garris, Jr., Woo Jong Hong, Catherine Mavriplis, Jeremy ShipmanThe George Washington University

School of Engineering and Applied ScienceWashington, DC 20052

202-994-6749garris@seas.gwu.edu, whong@seas.gwu.edu, mavripli@seas.gwu.edu, jshipman@seas.gwu.edu

ABSTRACTThe pressure-exchange ejector offers the

possibility of attaining a breakthrough in the level ofperformance of ejectors by means of utilizing non-dissipative non-steady flow mechanisms. Yet, the deviceretains much of the mechanical simplicity ofconventional steady-flow ejectors. If such a substantialimprovement in performance is demonstrated, itsapplication to ejector refrigeration will be veryimportant. Such a development would providesignificant benefits for the environment in terms of bothCFC usage reduction and greenhouse gas reduction.

The current paper will discuss in detail theconcept of pressure-exchange ejector refrigeration,compare it with existing technologies, and discuss thepotential impact that might be derived if cer tain levels ofejector performance can be achieved. Since the limitingissue on the system performance is in the fluid dynamicsof non-steady flow induction, research issues and recentprogress will be discussed.

EJECTOR REFRIGERATIONConvent ional reverse-Rankine cycle

refrigeration has three major drawbacks, all of which areassociated with the use of the compressor: i. The compressor is usually driven by high qualitymechanical power provided by means such as an electric

motor, mechanical power extraction from an engine, andthe like. When the electric power comes from fossil fuelpowered plants , or the mechanical power comes from acombustion engine, greenhouse gases and oth erpollutants are generated, and fuel is consumed. Thereare many applications where usable waste energy isavailable and is discarded and lost. One such exampleapplication is in automobile air conditioning where, inthe present state-of-the-ar t, a conventional r everseRankine cycle refrigerator is used such that thecompressor is powered directly by the engine, whilethermal energy in sufficient quantity to power the airconditioning system is lost through the engine exhaustand the cooling system. The possibility of usingautomotive waste heat for air conditioning in an ejectorrefrigerator is discussed in Balasubramaniam1, Hamner2

, Kowalski3.ii. Recent research has shown the chlorofluorocarbons(CFC’s), the refrigeran ts of choice for the reverseRankine cycle refrigeration system, have produced direeffects for the earth’s ozone layer. Alternate refrigerantshave been sought, particularly among thehydrochlorofluorocarbons (HCFC’s), however, concernshave been raised about other environmental problemsassociated with these refrigerants. Besides their adverseeffects on the ozone layer, Fischer 4 stated that “ . . .CFC’s have been implicated as a major anthropogenic

2

Figure 1 Ejector Refrigeration System Figure 2 Steady-Flow Ejector

cause of global warming.” Tur iel5 calculated that onepound of R-12 released to the environment has about15,000 times the global warming effect of the sameweight of CO2.iii. Conventional reverse Rankine cycle refrigeration isthe complexity of the compressor. This complexityincreases the cost of the unit, increases the maintenancerequired, and consumes excessive space or requires thatthe compressor be situated at inconvenient locations.

A possible remedy for all of these deficienciesis embodied in ejector refrigeration. A typical ejectorrefrigeration system is shown in Figure 1. Note that itoperates in a manner commensurate with reverseRankine cycle refrigeration with the modification thatthe mechanical compressor is replaced by a boiler-energized ejector and feed pump. A sketch of a typicalsteady-flow ejector is shown in Fig. 2. This device is

mechanically very simple, can directly convert thermalenergy into refrigera tion, and is compatible withenvironmentally friendly high specific volumerefrigerants such as water.

Common water is an excellent refrigerant whichhas a high latent heat of vaporization, a high specificheat, good heat transfer characteristics, non-corrosive,and is totally in harmony with the environment.However, since for normal ai r conditioning applications,the specific volume of water vapor under operatingconditions must be hundreds of times larger than thatof a system using CFC’s under comparableenvironmental conditions and design requirements.Water refrigerant, when used with a posit ivedisplacement type compressor, therefore, requires amuch larger compressor displacement volume. The needfor a much la rger compressor, and the associatedincreased cost, has rendered water to be less desirable asa refrigerant for conventional air conditioning. However, this problem can be overcome by the use of anejector. Ejectors are capable of transporting largevolumes of vapor within a relatively small space and ata low cost.

At the turn of the century when steam was moreabundant than electricity, the patent l iterature reveals aplethora of inventions based on ejector refrigeration. Inthis era, steam from locomotives was used to provide airconditioning in passenger trains. Steam ejectorrefrigeration enjoyed considerable popularity in the1930's for air-conditioning large buildings. It continuesto be used in industrial applications where steam isabundant.

Despite its substantial advantages, steam-ejectorair conditioning has not, in fact, been widely adoptedbecause of the low Coefficient of Performance (COP) thathas hitherto been attainable. Even when abundant wasteheat is available, such as in automotive applications, aconsequent resul t of the low COP is that not only mustthe evaporator heat absorbed by the refrigerant berejected in the condenser, but also the thermal energyused to power the ejector. If the COP is low, large

amounts of additional energy must be rejected at thecondenser which requires a condenser of substantiallyincreased size, weight, and cost. In automobiles, wherespace is a premium, this attribute would render steamejector refrigeration unacceptable.

The efficiency of the ejector is a limiting factorin determining the performance of the refrigerationsystem. The physical principal upon which the steady-flow ejector functions is that of entrainmen t of asecondary flow by an energetic primary flow by virtue ofthe work of turbulent shear stresses. Thus, the relativelylow energy secondary flow is “dragged” by the relativelyhigh energy primary flow through tangential shearstresses acting at the interface between the twocontacting streams. These turbulent stresses are a resultof mixing that occurs between primary and secondarystreams and the consequent exchange of momentum.While this mechanism is quite effective, and has beenwidely adopted in many applications, an inherentcharacteristic of mixing processes is to dissipate valuablemechanical energy. After mixing, these ejectors aredesigned2,6 so as to produce a strong normal shock priorto passing through a subsonic diffuser section. This

3

Figure 3 Radial-Flow Pressure-ExchangeEjector

strong shock produces further dissipation of energycontributing to a loss of efficiency.

Huang et al6 reports results which are typical ofthe state-of-the-art for ejector refrigeration with values ofCOP in the area of 0.20. Such levels of performance willrender steam-ejector refrigeration non-competitive for allbut the most specia lized applications. Since the limitingfactor on the COP is the ejector efficiency, it is clear thatin the absence of a major breakthrough in ejectorperformance, the benefits of steam-ejector refrigerationwill be unrealized on a lar ge scale. Such a breakthroughwith steady-flow ejectors is highly unlikely in view oftheir century old history of development.

One way in which the COP of the basic ejectorrefrigeration cycle can be increased is if the function ofthe ejector is r eplaced by a turbine-compressorequivalent as proposed by Rice7. However, such a systemwould be very expensive and would require excessivelylarge components if water were to be used as therefrigerant.

The present paper offers hope that a majorbreakthrough in steam-ejector refrigeration per formancemay be possible through a newly patented non-steadyflow ejector which util izes reversible means to achieveflow induction, while retaining much of the mechanicalsimplicity of conventional ejector refrigeration.Moreover, the device should be low in cost and compactin size. If such a breakthr ough in ejector performancecan be achieved, steam ejector refrigeration will have tobe considered in an entirely n ew light in view of itsoutstanding environmentally friendly attributes as wellas its high potential for energy conservation throughwaste-heat utilization.

THE PRESSURE EXCHANGE EJECTOR A detailed description of the pressure-exchangeejector is given in Garris8, Garris and Hong9 and Garriset al10. A sketch of a pressure exchange ejector is shownin Fig. 3. In the configuration under study, it is a simpleradial-flow turbomachine with one moving part: a free-spinning rotor. Integral with the rotor is a multiplicityof canted supersonic nozzles. Primary flow leaving thenozzles drive the rotor to free-spin, attaining angularspeeds possibly exceeding 50,000 rpm. The secondaryflow is directed towards the interaction space betweenadjacent fluid jets, between which it is trapped anddriven outward toward the periphery by virtue of thework of interface pressure forces. The work of pressureforces is inherently reversible and does not incur thedetrimental dissipation caused by the mechanism ofturbulent entrainment used in conventional ejectors.Furthermore, due to the radial geometry, the need for astrong normal shock in th e diffuser, a fur ther source of

dissipation in steady-flow ejectors, may be obviated.Since pressure exchange ejectors operate through anentirely different principal than steady-flow ejectors,there is a possibility that a major improvement inperformance can be obtained. Garris and Hong9

reviewed previous research on pressure-exchan ge ejectorsdesigned for thrust augmentation applications, and theuse of pressure exchange in wave engines. However, theidea of using it for ejector-compressor applications, aswell as ejector refrigeration applications, i s new.

EJECTOR REFRIGERATION CYCLE In the discussion that follows, we will attemptto place the per formance of ejector refr igerationequipment within the context of alternate technology.

Huang et al6 showed both theoretically andexperimentally what levels of performance might beexpected using real steady flow ejectors as characterizedin detail via their experimental results. In their verythorough investigation, the COP’s reported for a basicsystem, using R-113 rather that steam, were in the rangeof 0.06 - 0.25.

Since the ejector refrigeration system isdirectlyenergized by thermal energy, it is appropriate to

4

Figure 4 Mollier Chart of Steam-EjectorRefrigeration Cycle

compare it with absorption cycle refrigeration which isalso thermally energized. Herold et al11 report typicalvalues of COP from 0.5 to .75 for absorption chillers.While the current state-of-the-art ejector refrigerationsystem is much simpler, carries a lower initial capitalcost, lower maintenance, and a wider choice ofrefrigerants, absorption cycle refrigeration issubstantially more energy efficient.

For comparison with reverse Rankine cyclerefrigeration, one must also factor in the efficiency ofgenerating mechanical motive power for the compressorfrom a fuel source. In the case of electrica lly drivencompressors that receive their elect rical power fromfossil fuel powered plants, one must consider that thethermal efficiency of modern fossil-fuel fired plants isaround 35%-40% (Decher12). Transmission losses mustalso be taken into account. If one defines COP as theenergy removed from the cooling space divided by thethermal energy expended at the source in order to effectthat energy removal, reverse Rankine cycle refrigeratorswould have a COP in the vicinity of 1-1.5. Forautomotive applications13 where the maximum efficiencyof gasoline spark ignition engines is less than 30% , andfor Diesels between 35%-40%, the overal l COP forconventional automotive air conditioning systems wouldbe on the order of 0.75-1.0. It is the hypothesis ofthis paper that if the COP of ejector refrigeration couldbe brought up to about 0.75, it could be a cost-effect ivealter native to vapor compression refrigeration in manydomestic, automotive, and commercial applications. Itwould further provide immense benefits in reducing therelease of hazardous fluorocarbon-based refrigerants tothe environment, reduce the emission of greenhousegases by using waste heat or low greenhouse gasemitting fuels such as natural gas, and provide lowmaintenance, safety, and low initial investment.

Huang et al6 stated succinctly that: “It isobvious that the ejector is the heart of the jetrefrigeration system”. Although they did not correlatetheir results with adiabatic efficiency of the ejector,sufficient data were provided to permit such acalculation. Thr ee representative COP’s were extractedfrom the performance charts reported by Huang et al.6 :0.23, 0.125, and 0.05. The corresponding adiabaticejector efficiencies are computed to be 21.4%, 15.4%,and 8.4%, respectively. The result suggests the verystrong correlation of COP with ejector efficiency.

The efficiency of an ejector is closely connectedwith the entropy generation processes occurring within.The operating principles of a conventional steady-flowejector result in two major sources of entropy generation:turbulent mixing and strong normal shock waves. TheMollier chart of Fig. 4 suggests how entropy generation

in the ejector affect the thermodynamic cycle. Theprocesses indicated by ac and bc are reversible isentropicprocesses as would occur in an ideal turbomachineryanalog of the ejector with no shocks and no mixing. Theprocesses indicated by ac’ and bc’ represent theirreversible processes that occur in the primary andsecondary, respectively, of the ejector and includeentropy generation from both mixing and shocks.Depending on the amoun t of entropy generated, one cancompute the resulting COP of the cycle and the massflow ratio. One should remember that in performingthese exercises, energizing primary mass flow rate isadjusted to compensate for the dissipative processes so asto enable the operating temperatures and pressures toremain the same.

Garris et al.14 showed that as the entropy riseincreases, the system COP decreases precipitously. Itwas demonstrated tha t the COP can be improveddramatically if the entropy rise can be controlledeffectively. By means of the pressure-exchange ejectorwhich utilizes physical mechanisms which producemodest entropy rises, a substantia l improvement overconventional technology is possible. This study alsoshowed that when the entropy rise is small, only a smallamount of energetic superheated primary fluid is neededin order to drive the refrigeration system. However,when the entropy rise in the ejector increases, there is arapid drop in the mass flow ratio indicating that muchmore fluid is needed in the primary circuit in order todrive the system. By reducing the entropy rise in the

5

Figure 5 TASCflow Solution for CurvedSupersonic Nozzle

ejector, the pressure-exchange ejector minimizes theenergy requirement in the boiler/superheater whileminimizing the amount of heat to be rejected in thecondenser leading to a more cost-effective and spaceconserving system.

EXPERIMENTAL PROGRAMThe experimental program has the near-term

goal of understanding the behavior of the pressure-exchange ejector, and the long-term goal ofimplementing it in a refrigeration system. Recent workhas concentrated on the former.

The current test rig consists of a radial-flowpressure-exchange ejector having a 3.0" diameter rotorwith nine planar minimum-length Mach 2.5 supersonicnozzles canted at an angle of 20o from the radius. Air isprovided through a blow-down air supply. The system isinstrumented and pressures, temperatures, and flow ratesare measured in the primary and secondary flow circuits.As indicated in Fig. 3, the discharge flow is released tothe atmosphere through a vaneless diffuser. Staticpressure taps are placed at many radial locations in thediffuser to monitor diffuser performance. The rotor ismounted on a shaft with precision high- speed ball-bearings. Laser-fiber optic probes are used to measureangular speed, and a piezoelectric pressure transducer isplaced near the nozzle discharges to measurefluctuations in static pressure. A hot wire anemometer isemployed to measure velocity and turbulence in thediffuser. Vibration is also monitored. Data is compiledvia a computerized data acquisition system.

It must be appreciated that this type of ejector iscompletely new and that there is not any previous designexperience upon which to build. Before experimentscan be conducted, the ejector must satisfy unforgivingmechanical design constraints including bearing design,structural design, and dynamic sealing. The operatingparameters that m ust be studied include

primary/secondary stagnation pressure ratio,primary/secondary stagnation temperature ratio, ejectordischarge back pressure, and working fluid (air, steam,etc.). In order to obtain an optimum configuration, thegeometrical parameters th at must be studied include:Area ratio of supersonic primary nozzle (Mach Number),configuration of nozzle (planar, axi-symmetric,minimum-length, curved, etc.), number of nozzles, spinangle of nozzles (peripheral speed), diffuser spacing(distance between vaneless di ffuser surfaces), diffuserangle, secondary to primary throat area ratio, andaerodynamic design configuration.

A major goal of this research program is toexplore all of these aspects. However, to date, we havededicated much of our effort towards satisfying themechanical design constrain ts. In the process, we haveencountered difficulty with bearing failures, vibrationinstability, structural failure, and sealing problems. Wehave made much progress in this connection and are nowable to conduct experiments in order to explore thebehavior of the ejector subject to varying operatingparameters. The results to be presented constitute ourfirst attempt for one set of geometric parameters. Thusfar, no study of the important effects of the geometricparameters has been performed. The results to bepresented in the present paper, therefore, should in noway be construed as representative of the capabil ities ofthe pressure-exchange ejector. We expect to be able toprovide such information in future publications.

Test results, as reported in Garris et al.10,determin ed that rotor speeds exceeded 40,000 rpm. Inpractice, the rotor experiences bearing friction andwindage torque. Considerable effort was made to reducebearing friction and work on designing special airbearings is being conducted in parallel with theaerodynamic investigation of the ejector.

Figure 5 shows how the ejector suction(secondary pressure) varies with primary pressure forthree different secondary flow conditions ranging fromno-flow to maximum flow. The former is obtained byshutting the secondary flow valve, and the latter isobtained by opening it wide. The third condition shownis intermediate. It may be seen that the no-flowcondition corresponds to maximum suction.

COMPUTATIONAL SIMULATION PROGRAMTo model the characteri stics of this novel

pressure exchange concept computationally, thecommercial fluid flow analysis code TASCflow wasacquired. TASCflow solves the three dimensionalNavier-Stokes equations by utilizing a finite elementbased finite volume method over structured hexahedralgrids. It was chosen over other software packages

6

primarily for its ability to handle time periodic boundaryconditions and rotating frames of reference, making itwell-suited for modeling rotating machinery. Otherfeatures particularly suitable to this research include theabilities to accurately model real gas compressible flows,turbulence, heat transfer, shocks resulting from transonicand supersonic flows, and special fluids such as wetsteam and refrigerants.

To date, TASCflow has been used to model thetwo dimensional flow through the supersonic nozzles ofthe rotor used in the experimental rig. The current rotorconfiguration involves a plurality of minimum-length,Mach 2.5 nozzles canted at 20° with respect to the radiusof the rotor. In the previous work of Divinis15, theseminimum-length nozzles were design ed with a meth odof characteristics FORTRAN code developed by NASA16.

The TASCflow predictions of the boundarylayer growth were compared to that predicted with theNASA Code16. It was seen that both computationspredict comparable values of boundary-layer thicknessand are consistent.

TASCflow was further validated by comparisonof the work on curved nozzles of Divinis15 who utilizedthe method of characteristics for planar nozzles withdifferent degrees of curvature. These nozzles are ofinterest to the project because of their ease of packagingin the confines of a rotor . An efficient ejector willdepend on the ability of these nozzles to provideuniform, shock free flow. TASCflow is currently beingused to investigate the flow in these nozzles. Figure 6shows the solution for a Mach 2.5 nozzle with a 20°angle of curvature. While the flow is shock free, it isless than perfectly uniform and reaches the design Machnumber only at a small region.

CONCLUDING REMARKSIt is the hypothesis of this research program

that, through the work of interface pressure forces, amajor improvement in performance over conventionalsteady-flow ejector technology may be possible.Unfortunately, the resul ts herein reported do not exhibitthis. However, considering that this consti tutes our firstattempt, we are satisfied that the pressure-exchangeejector does function, and future efforts a t optimiza tionshould yield substantial improvements. Future plansinclude designing and testing a pressure-exchangeejector refrigeration system when sufficient informationis learned on optimal ejector design.

Steam-ejector refrigeration, known forgenerations, may finally come of age in the 21st Centuryas a result of the need for environmentally friendlyrefrigeration coupled with a breakthrough in ejectortechnology: the pressure-exchange ejector. Much workis needed in order to bring this vision to fruition.

Current performance levels are far below those of idealturbomachinery and, to date, improvements over steady-flow ejectors have not yet been demonstrated. While thepressure-exchange ejector is far simpler thanturbomachinery, being a direct fluid-fluid flow inductiondevice, optimization is quite a daunting task. Thenumber of geometrical and design parameters issubstantial and the device in unforgiving in terms ofsealing and bearing requir ements. On ce anunderstanding of this complex ejector is obtained,experiments and analysis on a complete pressure-exchange refrigeration system will be performed.

ACKNOWLEDGMENTSWe gratefully acknowledge sponsorship of this researchthrough grants from the National Science Foundationand the U.S. Environmental Protection Agency. We alsothank Messrs. Damon Hand and Sung Doug Yoon whoassisted with the experimental work and were supportedunder the NSF Research Experiences for Undergraduatesprogram, and to Mr. Charles A. Garris, III whostructurally evaluated the rotors with finite elementanalysis.

REFERENCES 1. Balasubramaniam, M.., “Combined Engine CoolingSystem and Waste-Heat-Driven Jet-Vapor-CompressionAutomoti ve Air Conditioning System”, Ph DDissertation, U.of Pennsylvania (1975).

2. Hamner, R. M. “An Investigation of An Ejector-Compression Refrigerat ion Cycle and Its Application toHeating, Cooling, and Energy Conservation’, Ph.D.Dissertation, U. of Alabama (1978).

3. Kowalsky, G. J., “Refrigeration System with Ejectorand Working Fluid Storage”, U. S. Patent 5,117,648(1992).

4. Fischer, S. K., “Total Equivalent Warming Impact:A Measure of Global Warming Impact of CFCAlternatives in Refrigerating Equipment” ,Rev.Int.Froid, v.16,n.6, pp 423-428 (1993).

5. Turiel, I., Levine, M. D., “Energy-EfficientRefrigerat ion and the Reduction of ChlorofluorocarbonUse”, Annual Review of Energy, pp.173-204 (1989).

6. Huang, B. J., Jiang, C. B. , Hu, F. L., “EjectorPerfomance Characteristics and Design Analysis of JetRefrigerat ion System”, J. of .Engr.for Gas Turbines andPower, ASME, v.107, pp 792-802(1985).

7

7. Rice, N. C., Davison, W. R., Biancardi, F. R..,“Environmental Control System”, U. S. Patent3,259,176(1966).

8. Garris, C. A., “Pressure Exchanging Ejector andRefrigeration Apparatus and Method”, U. S. Patent5,647,221 (1997).

9. Garris, C. A., Hong, W. J. , “A Radial -Flow Pressure-Exchange Ejector”,Paper No. FEDSM97-3339,International Symposium on Pumping Machinery, 1997ASME Fluids Engineering Division Summer Meeting,Vancouver, CA.(1997).

10. Garris, C.A., Hong, W. J., Mavriplis, C. M.,Shipman, J., “An Experimental and Numer ical Study ofthe Pressure-Exchange Ejector”, Paper No. FEDSM98-5123, Forum on Fluid Machinery, 1998 ASME FluidsEngineering Division Summer Meeting(1998).

11. Herold, K. E., Radermacher, R., Klein, S. A.,“Absorption Chillers and Heat Pumps”, CRC Press(1996).

12. Decher, R., “Direct Energy Conversion:Fundamentals of Electric Power Production”,p.5,OxfordU.Press (1997).

13. Heywood, J. B., “Internal Combustion EngineFundamentals”,p.887,McGraw-Hill (1988).

14. Garris, C.A., Hong, W. J., Mavriplis, C. M.,Shipman, J., “An En vironmentally Friendly Pressure-Exchange Ejector Refrigeration System”, Paper No.FEDSM98-4812, Forum on Industrial andEnvironmental Applications of Fluid Mechanics, 1998ASME Fluids Engineering Division Summer Meeting(1998).

15. Divinis, N., “A Comparative Study of the Effects ofCurvature and Boundary Layers on Two-DimensionalSupersonic Nozzles”, MS Thesis, The GeorgeWashington University (1997).

16. Goldman, L. J., Vanco, M. R., “Computer Programfor Design of Two-Dimensional Sharp-Edged-ThroatSupersonic Nozzle with Boundary Layer Correction”,NASA Technical Memorandum, TM X-2343. NASACOSMIC Program LEW-11636 (1971).