You are on page 1of 6

MST TOC

Proceedings of ASME Turbo Expo 2003 Power for Land, Sea and Air June 16-19. 2003, Atlanta, Georgia, USA

GT2003-38871

REGENERATIVE HEAT EXCHANGERS FOR MICROTURBINES, AND AN IMPROVED TYPE


David Gordon Wilson Wilson TurboPower Inc. 21 Winthrop St. Winchester MA 01890-2851, USA

ABSTRACT The principle of operation of regenerative heat exchangers (regenerators) is explained, and the characteristics are compared with those of recuperators. The design rules for heat exchangers in general are stated, and the particular design rules for regenerators are discussed in more detail. Problems in past regenerators have led to an improved type, which is described. Design studies of a regenerator for a typical microturbine at three different effectivenesses are given, and leakage rates estimated. At the highest regenerator effectiveness, and using the improved compressor and turbine design permitted by this effectiveness, it is estimated that a 300-kW microturbine should achieve 50% electrical efficiency. INTRODUCTION Primitive forms of heating systems illustrate well the different operation of recuperators and regenerators. A kettle on a gas burner is a heat exchanger, with water on one side and hot gas on the other. This can be designated to be a recuperator, although a better example is that of the heater of a building heating system, with the products of gas or oil combustion blown steadily through on one side and ducted air, water or steam flowing on the other side (figure 1).
In a Roman villa (circa first century AD) near where the author was born in Britain another method was used for heating the water for the baths. A fireplace was built into the base of a small hill or mound, with the chimney set in the hill. Favored British slaves would carry cold boulders up the hill and drop them into the chimney. Unlucky slaves would remove the now-hot boulders from above the fireplace and carry or roll them into the baths. This is a regenerative heat-exchanging system, and it undoubtedly predated the recuperator. It has been used to generate hot gases for various metallurgical duties for centuries, although, in contrast to the moving stones of the Roman baths, the ceramic brickwork or stones have usually Figure 1: Recuperator (home heating system). been stationary and the two streams have been switched so that they pass alternately through the ceramic. This is the case for the Cowper or hot stoves used for heating blast-furnace air, and for the reverbatory or regenerative heat exchangers in open-hearth furnaces. Figure 2 is a diagrammatic representation of a pair of switching regenerators. From these simple examples a generalization can be made: that recuperators have steady flow, while regenerators of all types have discontinuous flow through their heat-exchanging passages In 1920 Fredrik Ljungstrom invented the rotary regenerator, in which the ceramic or metallic thermal-storage elements (forming what has come to be called a matrix or core) are mounted in a circular carrier arranged to intercept the hot-fluid and cold-fluid ducts (figure 3). When this is rotated slowly with the hot-gas stream passing through the matrix in one direction and the cold-gas stream passing through another part of the matrix in the other direction it automates the stone-carriers of the Roman baths, and produces efficient

Copyright 2003 by ASME Copyright #### by ASME

counterflow heat exchange in each stream. Air preheaters built on this principle were and are used to heat the combustion air in steam generators by heat exchange with the exhaust air. The largest are over 10 m diameter. There is little difference of pressure between the two flows, so that the substitution of the sliding seals on the rotating matrix for the valves in a hot stove (where there is a large pressure difference between the two flows) gives a saving in cost and complexity. Whereas the first use of heat exchangers in gas-turbine cycles was by Aegidius Elling in 1903-4 via a recuperator (Wilson & Korakianitis, 1998[1], gas-turbine regenerators were first studied by L. Ritz in Gttingen (Cox and Stevens, 1950)[2] and first developed as hardware by the UK National Gas-Turbine Establishment (NGTE) in the 1945-49 period when three successive full-size versions were made. Although ceramics were contemplated, the matrices were made of stainless-steel strips, plain and crimped, wound around a mandrel. This was the first known use of rotary regenerators for two flows at quite different pressures. The NGTE devoted considerable effort to the resulting sealing problem, and developed smart seals (figure 4) that could follow a matrix face that was obviously not going to remain flat, because of the large coefficient of thermal expansion of stainless-steels. Unfortunately the NGTE closed off work on regenerators shortly thereafter: its mission was principally to develop better aircraft engines, and the author remembers the NGTE director,

Figure 2: Diagrammatic representation of switching regenerators using a ceramic matrix, and Corning developed aluminum-silicate honeycomb matrices for Rover in the UK. A Rover-BRM car with twin ceramic regenerators entered in the Le Mans 24-hour race in 1965 and achieved a performance far ahead of forecasts (Wilson, 1995). This helped to engender a wave of enthusiasm for the ceramic regenerator, and it has been specified for most automotive gas turbines since that time. Ceramic regenerators have not, however, met with total success. Ford produced a truck engine with lithium-aluminum-silicate regenerators that turned out to be susceptible to attack from an unexpected combination of two common gas-borne contaminants (sodium chloride and fuel sulphur). Corning switched to magnesium-aluminum silicate, but Ford, along with other companies in the automotive industry, was reluctant to trust ceramics subsequently. Other companies continued development, mainly in Japanese- and US-government programs, but most encountered rapid wear in either the ceramic matrices or the seals, leading to penalizing leakage of compressor air. This became the Achilles heel of gas-turbine regenerators. Gas-turbine heat exchangers in general have had a mixed history. The most successful recuperators appear to be the small annular units produced by Solar for the Capstone 30-kW microturbines, and those designed and manufactured by Ingersoll-Rand-NREC for its own microturbine. The highest production of recuperators has been for the Avco AGT-1500

Figure 3: Ljungstrom-type rotary regenerators. (From Wilson & Korakianitis, 1998) Hayne Constant, stating at a student talk that his colleagues had found how to produce higher pressure ratios in axial-flow compressors so that heat exchangers would not be required. Similar regenerators with stainless-steel matrices (using compliant pressure-loaded seals) were introduced in 1963 by George Huebner, chief engineer at Chrysler, in an audacious and amazingly successful demonstration of fifty Chrysler gas-turbine automobiles that were put in the hands of over 200 selected members of the public for up to three months (Wilson, 1995[3]). The engines and the regenerators performed very well, but the cost of the high-temperature materials alone ruled the engine out of the running for a mass-produced vehicle. Huebner had worked with Corning Corp. on the possibility of

Copyright 2003 by ASME Copyright #### by ASME

an engine thermal efficiency of about 0.54 is predicted, leading to the possibility of an electrical efficiency of 0.50. A highly beneficial result of this low-pressure-ratio cycle for automotive engines is that they are given a wide area of high efficiency, even at quite low loads. This comes from the wide area of efficient operation of low-pressure-ratio compressors, and from the fortunate characteristic of heat exchangers in actually improving their performance at low loads. (The same

Figure4: NGTE self-adjusting smart seals. (Cox and Stevens, 1950) engine for the US Abrams tank. Thermal stresses increase rapidly with unit size, particularly in devices made of stainless-steel, and the AGT-1500 annular heat exchangers for these 1-MW-class engines had significant cracking problems. Most of the failures were ascribed to low-cycle thermal fatigue from the rapid load changes that a tank engine must undergo, but some blame was placed on the flexing of the matrix from travel over rough terrain. The most ambitious recuperators yet commissioned for gas turbines have been those for the WR-21 intercooled-recuperative engine (about 20 MW) for future propulsion units in the US, UK, French and other navies. In the development of these recuperators, Sanders and Louie (1999)[4] reported serious cracking problems, and other recuperators are being considered as replacements. Past ceramic rotary regenerators have had the wear and consequent leakage problems mentioned above, coupled with occasional material susceptibility.

Figure 5: Influence of regenerator effectiveness on cycle thermal efficiency and optimum pressure ratio. heat-transfer area is available for reduced mass flows.) On the other hand, a problem introduced by the use of heat exchangers is that the reduction in the compressor pressure ratio also reduces the turbine expansion ratio, and with it the temperature drop through the turbine (figure 6). Uncooled metal turbines can be run at a turbine-inlet temperature of 950 C. However, the maximum inlet temperature of a stainless-steel recuperator is about 700 C, which is not a good match for a turbine of low pressure ratio, and either the pressure ratio or the turbine-inlet temperature or the recuperator effectiveness is usually compromised.

THE NEED FOR HEAT EXCHANGERS This paper is concerned with small engines, for the following reason. Whereas large engines can attain high efficiencies by using high turbine-inlet temperatures coupled with high pressure ratios, small engines cannot. The turbine blades of small engines are too small to be effectively cooled, so that inlet temperatures must be kept relatively low. Nor can small engines be designed with high pressure ratios, because the blade length at the outlet of the centrifugal compressors used in small gas turbines becomes very small, similar to the required blade-to-casing clearance. This greatly decreases the compressor efficiency. Fortunately, the heat-exchanger cycle requires a low pressure ratio. A heat exchanger is, therefore, the ideal component of a small-gas-turbine cycle.
The design-point efficiency of a heat-exchanger cycle increases as the effectiveness (or efficiency) of the heat exchanger increases (figure 5), and can exceed the thermal efficiency of high-pressure-ratio high-temperature cycles. As the effectiveness increases, the design-point pressure ratio for maximum efficiency decreases. Figure 5 is calculated for a 300-kW microturbine with a three-stage centrifugal compressor and a three-stage axial turbine, giving high component efficiencies because of the low loading and the low kinetic energy lost at outlet. With a regenerator of 0.975 effectiveness,

Figure 6: Influence of turbine expansion ratio on turbine outlet temperature.

Copyright 2003 by ASME Copyright #### by ASME

There are two ways out of this dilemma. One is to use an intercooled cycle coupled to a heat exchanger (as in the WR-21 engine above). The intercooler produces an optimum pressure ratio much higher than for a non-intercooled cycle, and the higher expansion ratio through the turbine then allows the use of a stainless-steel recuperator even with a high turbine-inlet temperature. The other way is to use a ceramic heat exchanger, which can withstand very high temperatures. There have been many attempts at producing ceramic recuperators, none wholly successful, at least because of the size effects mentioned below. Ceramic regenerators, however, seem to be a natural fit for a small gas turbine. They can also be made at higher effectivenesses than can recuperators (e.g., 0.975 vs. 0.90), and can easily withstand the higher turbine-exit temperatures that result from using the reduced pressure ratio associated with this capability. Their use in gas turbines has been highly restricted because of the wear, leakage, and occasional susceptibility to corrosion that past regenerators have exhibited. A solution to these problems is discussed below.

likely to be small, 50 - 150 mm, so that the regenerator rotor may turn out to be a relatively thin flat disk. It is much more difficult to combine these two goals in a recuperator than in a regenerator.

Pressure-drop balance It is so easy to produce a low pumping-power requirement with regenerators that one can reach a level where there isnt enough pressure drop to spread the flow out across the core. The authors graduate class performed an analysis of the design of a fairly typical twin-regenerator engine and showed that the air-side pressure drop was 0.1%, not enough to cause the flow to spread across the core face. On the exhaust-gas side, however, it was over 15%, highly penalizing to the engine performance. The author, in his designs, generally specifies core pressure drops of about 2.5% on the gas side and a minimum of 1% on the air side. This can be achieved by manipulating another degree of freedom, the conductance ratio, which can be translated (for regenerators) as approximately the gas-to-air face-area ratio for the two streams. It should increase with design-point pressure ratio (Wilson & Korakianitis, 1998).
The conductance on one side is the product of the mean heat-transfer coefficient, h, and the heat-transfer area, Ah. In the laminar flow found in small regenerators, the non-dimensional heat-transfer coefficient or Nusselt number, Nu, is constant: for instance, for passages with square cross-sections it is about 3.6, regardless of flow velocity. The Nusselt number is defined as (h.dh/k), where dh is the hydraulic diameter and k is the fluid thermal conductivity. The heat-transfer coefficient, h, is therefore inversely proportional to the hydraulic diameter. A small value of dh produces both a high heat-transfer coefficient and a high heat-transfer area per unit volume, and thus a high value of the thermal conductance.

ASPECTS OF HEAT-EXCHANGER DESIGN Most turbine designers are unaware of the multidimensional freedom allowed to heat-exchanger designers. One can start with tubes of various sizes and cross sections or plates with and without fins or containers full of sand or gravel, and with various flow arrangements, and almost any of these can be contrived to produce an effectiveness of almost any value. The freedom is so great as to be unsettling to many. The author has therefore published guidelines (Wilson & Korakianitis, 1998) to the methods of Kays and London, (1985)[5]. One is that, to produce a heat exchanger of small volume, one should use small passages, because the amount of heat transferred per unit volume is inversely proportional to the hydraulic diameter of the passages for turbulent flow, and to the square of the hydraulic diameter for laminar flow.
Gas-turbine rotary regenerators operate deeply in the laminar region. It is possible to use much smaller passages in regenerators than in recuperators for two reasons. One is that the flow reverses frequently, so that if the passages are fine enough for the core face to act as a filter, the dirt, etc., that is collected is blown off a second or two later. The second reason is that the fine passages do not have to be brazed or otherwise joined to a header, as in a recuperator. The construction cost of a regenerator core is therefore much less than for a recuperator. Another guideline, surprisingly completely independent of the first guideline, is that to reduce the pumping power required to push the flows through the passages, one should choose low Mach numbers or flow velocities, because the pumping power is proportional to the square of the Mach number. Combining these two guidelines, one can specify a heat exchanger that simultaneously has low pressure drops and a small core volume. The core will have a large cross-sectional, or face, area to produce low flow velocities, and small passages (hydraulic diameters of 0.5 - 1.0 mm are frequently used in regenerator cores.) The passage length is

REGENERATOR IMPROVEMENTS These advantages of regenerators for small gas turbines seemed substantial enough for some effort to be devoted to overcoming the problems discussed earlier. The results are as follows. Discontinuously moving core In traditional regenerators, the core rotates at a steady (low) speed, with the seals pressed against the core faces and rims. The resulting scraping causes rapid wear. During a research project at MIT it was decided to investigate the effect on thermal performance of rotating the core in increments, so that the seals could be clamped for considerable periods. It was found that there were no adverse effects. MIT patented the concept (Wilson, 1993 & 2001[6]) and Wilson TurboPower has licensed the technology. In this regenerator the core is first held stationary, with the seals held on to it with just enough pressure to reduce leakage to a negligible amount. After a period of a second or two (this depends on the regenerator size), the seals are lifted just enough to allow the core to be rotated rapidly through an increment (30 degrees in our case), after which the seals are again pressed down. Thus all wear is eliminated, as is most of the seal leakage (because the seals are clamped for 95% of the time). During this rapid movement the flow areas do not change, so that there should be no

Copyright 2003 by ASME Copyright #### by ASME

propagation of pressure peaks. There will be a very small amplitude of outlet-temperature fluctuation for which the controls on firing rate will compensate. A diagrammatic representation is shown in figure 7. There is virtually no difference in engine installation from that of steadily rotating regenerators except for the incremental drive and seal-lifting mechanisms.

A maximum axial core temperature gradient is specified An unthinking designer of a rotary regenerator can specify an extremely small hydraulic diameter that results in a very short passage length or disk thickness. This produces a high temperature gradient, because one face of the disk becomes, within a short time of start-up, close to turbine-outlet temperature, while the other face approaches compressor-outlet temperature. We have set a limit for the air-flow temperature gradient (as a surrogate for the matrix-wall temperature gradient) of 7.5 degrees K per mm. This is used to set the minimum hydraulic diameter of the passages. The conductance ratio is used to adjust the percentage pressure drops As mentioned earlier, the hot/cold conductance ratio is used to arrive at hot and cold pressure drops that do not unduly penalize the engine performance, that provide for the flows to spread across the inlet faces, and that do not result in an overly large core diameter. EXAMPLES OF REGENERATOR DESIGNS Some designs that result from the above technology and design approach are illustrated in the following table for three regenerators applied to a small engine (of 40 kW electrical output). The first is that already fitted, with an effectiveness of 0.9 and a leakage rate of 4% (of the compressor flow). The second and third are designs using the present technology, having a maximum leakage of 2% and effectivenesses of 0.95 and 0.975. The engine has an air mass flow of 0.35 kg/s, a compressor pressure ratio of 2.73, a compressor-inlet temperature of 288 K, and a turbine-inlet temperature of 1228 K. The total-pressure loss for the cycle is 12%, and the mechanical efficiency is 98%. The polytropic efficiencies of compressor and turbine are 0.84 and 0.87. The calculations were made for LAS ceramic honeycombs having square-section passages and a porosity of 70% (i.e. the solids content was 30%).
The predicted engine thermal efficiencies are well below those of the 300-kW engine for which figure 5 was produced because the component efficiencies for a small engine are below those for a larger engine; because a single-stage compressor and turbine are used, for which the kinetic-energy leaving losses are high; because the pressure ratio is not optimized for each value of the regenerator effectiveness; and because the turbine-inlet temperature is below that used for the 300-kW engine. In going from an effectiveness of 0.9 to 0.95, and from 0.95 to 0.975, in each step the thermal losses are halved, and the size of the regenerator active disk more than doubles. The actual size includes solid rims, a central hole of one-quarter the outer diameter, and allowances for the face area taken by seals. The calculations, the results of which are given in the table, show how the diameter and thickness of a regenerator disk change when the pressure drops are specified to be constant. The compressed air trapped in the pores of the matrix as it passes into the exhaust stream amounts typically to 0.5% of the compressor mass flow. Predictions from models of the

Figure 7: Diagrammatic representation of a regenerator with a discontinuously moving core.

Core is produced in sectors The core is produced in twelve sectors of 30 degrees each, having solid-ceramic rims around the non-flow directions. Doing so allows the seals to press against a flat, solid, surface. The thermal stresses are greatly reduced compared with those in a monolithic disk. The production cost of manufacturing a disk of twelve small identical components is considerably less than that for a single large disk. In addition, it is possible to restore a disk that has damage in only one or two sectors. Foamed ceramic material may be used Traditional regenerator cores, metal or ceramic, have been produced with honeycombs of hexagonal, square or triangular passage sections. The first cores were wrapped from stainless-steel foil, alternately plain and crimped, and a similar method, using ceramic-slurry-coated tea-bag paper, was used for Cornings glass-ceramic cores. Later, extrusion methods were developed to produce honeycomb catalyst beds for automobile exhausts, and were applied to regenerator cores. The extrusion nozzles are expensive, and the extrusion-pressure loads on the nozzles are such that the maximum diameter of the cores is about 300 mm. This is sufficient for core sectors to be produced from extruded honeycomb sections. It is also possible for the core sectors to be produced in foamed ceramic, giving greater durability at considerably lower cost.

Copyright 2003 by ASME Copyright #### by ASME

discontinuous-rotation regenerator are for the seal leakage to be a further 0.5%, so that the total leakage would be 1%, although 2% has been specified to be conservative. The 4% leakage of the existing regenerator is considerably lower than that from much recent experience (Shimada et al, 1993[7]) Other provisions in the design approach (such as annual inspections and regenerator-face and seal-face grinding if necessary) are aimed at giving a life of 40,000 hours minimum.

TABLE OF RESULTS OF REGENERATOR PRELIMINARY DESIGNS FOR A 40-kWe ENGINE


EFFECTIVENESS Turbine-exhaust temperature Air outlet temperature
No. of transfer units, Ntu Rotation rate, non-dimensional Free face area, hot side Pressure drop, hot side Outer diameter of active matrix Channel length (disk thickness) Mass of active matrix Mean flow velocity, cold side Pressure drop, cold side Time for one rotation Thermal efficiency of engine Fuel-air ratio x 10,000 From cycle calculations From cycle calculations From Kays & London From Kays & London Calculated Specified Calculated Calculated Calculated Calculated Calculated Calculated Calculated Calculated

K
K m2 % mm mm kg m/s % s % -

0.9 1,153 1,085 10 3 0.06 2.5 446 70 7.7 11.6 0.8 9.3 35.3 80

0.95 1,153 1,119 23 3 0.09 2.5 545 108 18 8.1 0.9 20.9 40.3 71

0.975 1,153 1,136 50 3 0.14 2.5 659 162 39 5.7 0.93 45 42.8 67

exchanger for gas-turbine power plant. Proc. Inst. Mech.Engrs. 150, vol. 163, W.E.P. 60, London, UK. [3] Wilson, D.G. (1995). Automotive gas turbines: the pioneers. Global Gas Turbine News, May/June, I.G.T.I., A.S.M.E., N.Y, NY. [4] Sanders, R.C., & G.C. Louie (1999). Development of the WR-21 gas-turbine recuperator. ASME paper 99-GT-314, N.Y., N.Y.. [5] Kays, W.M. & A.L. London (1985). Compact heat exchangers. Third edition, McGraw-Hill, N.Y., N.Y. [6] Wilson, D.G. (1993 & 2001). Heat exchanger containing a component capable of discontinuous movement. U.S. patent no. 5,259,444 (1993), reissued as RE37,134 E (2001). [7] Shimada, Kazuaki, et al (1993). Advanced ceramic technology developed for industrial 300-kW CGT (ceramic gas turbine) research & development project in Japan. Paper no. 93-GT-188, ASME, NY, NY

CONCLUSION In most current microturbines, the use of heat exchangers with high effectiveness, low leakage and low pressure drops, coupled with the ability to withstand high inlet temperatures, would produce a sharp rise in design-point thermal efficiency (typically five percentage points, a gain of around fifteen percent). The off-design performance would also improve. If a turbine were designed to use the full characteristics of the regenerator, it would have a low design-point pressure ratio, a high turbine-inlet temperature, and high component efficiencies, and be capable of 50-percent electrical efficiency for microturbines of 300kW and over. ACKNOWLEDGMENTS Andreas Pfahnl was prominent among the many MIT students who have contributed to the development of the regenerator concept. Rosalind Takata, among colleagues in WTPI, helped substantially in preparing this paper. The author expresses his sincere appreciation for their help. REFERENCES [1] Wilson, D. G. and T.P. Korakianitis (1998). The design of high-efficiency turbomachinery and gas turbines. Second edition, Prentice-Hall, Upper Saddle River, NJ. [2] Cox, M, and R.K.P.Stevens (1950). The regenerative heat

Copyright 2003 by ASME Copyright #### by ASME

You might also like