A breakthrough in regenerators

Turbomachinery International, May/Jun 2006 by Wilson, David

NEW DEVICE CLAIMS TO INCREASE MICROTURBINE EFFICIENCY TO 50%

Recuperative (steady flow) heat exchangers that have been developed for waste-heat recovery in gas turbines have fundamental limitations. No one has been able to make a successful ceramic recuperator, so the temperatures have been limited to those that can be handled by metals - up to 650°C for stainless steels and perhaps 950°C for jet engine superalloys.

To develop a solution, researchers at the Massachussetts Institute of Technology (MIT) have instead worked on regenerative (flow switching) heat exchangers. Wilson Turbo Power, Inc., (WTPI; Woburn, MA, www.wilsonturbopower.com) has commercialized a ceramic regenerator (Photo) - based on a MIT patent - with long-lasting seals and having a potential effectiveness of over 99%. The company is developing a version of the regenerator for a 300 KW microturbine so that it can achieve 50% electrical efficiency.

Satisfying current needs

To achieve the highest efficiencies, today's turbines and solid-oxide fuel cells require regenerators that take gases at 950°C or above. This temperature excludes metallic recuperators.

It is generally uneconomical to make recuperators of high effectiveness because they become large. To go from 90% to 95% effectiveness requires an approximate doubling of recuperator volume, and another doubling to go to 97.5%. Using smaller air and gas passages would reduce the size, because the volume of any heat exchanger is approximately proportional to the square of the passage diameter. But the cost of producing smaller passages in superalloys is high, and the passages begin to act as air filters, getting clogged by dirt. For example, the cost of an Inconel 625 recuperator for a 100-kW microturbine is over four times that for an equivalent ceramic regenerator.

Regenerators start at a lower volume than their recuperator equivalents, and the doubling in volume occurs mainly by increasing the thickness of the thin rotating disks that transfer the heat. The increase in disk size does not cause a corresponding increase in the size of the headers and casings. Regenerators can use small passages because the flow reverses every few seconds giving a degree of self-cleaning, and the cost of the extruded ceramic honeycomb, while not trivial, is much lower than for the metal equivalents. Lastly, heat-resisting metals have higher thermal expansion coefficients than ceramics, making large recuperators liable to thermal-fatigue cracking. For instance, the expansion coefficient of 309 stainless steel is ten times that of corderite ceramic used in the MIT regenerator.

Ceramic rotary regenerators have had their own problems, principally caused by the friction of the rubbing seals used to reduce leakage of the compressor delivery flow as the regenerator disk rotates. On some occasions the regenerators lasted just a few hours before the compressed-air leakage rose to a point where the engine efficiency fell disastrously.

MIT has been awarded a patent on a discontinuous-rotation regenerator designed to overcome this problem (see www.turbomachinerymag.com archives for more illustrations). Instead of rotating steadily against rubbing seals, the regenerator disk is stationary for about ten seconds with seals clamped against the surface. Then the seals are lifted by 25 - 50 microns (0.001 - 0.002 inches) and the disk is rotated rapidly through an increment of 30 to 180 degrees in a fraction of a second, after which the seals are again clamped on to the disk. Thereby leakage is greatly reduced and wear virtually eliminated.

The regenerator was first tested in January 2006 and achieved most objectives almost immediately, having an effectiveness of over 99% with an inlet temperature of 925°C (1,700°F) [1].

The basic principle of the seals, which surrounds circular ducts in this demonstrator, is that they are formed of several sliding elements, each of which is pushed down on to the regenerator face by springs. Separate elements are used because the initial flat face of the regenerator disk may undergo some eventual distortion. The number of sliding elements could be increased to ensure that leakage would be small, however large the distortion. The first build had eight elements.

After the sealing elements are pushed down on the sealing face of the disk, a surrounding squeeze band is tightened, converting the assembly of individual elements into a rigid cylinder and preventing compressed-air leakage between them. The squeeze band is used to lift the seal assembly slightly for the next incremental rotation, after which the band is loosened. All the seal pieces snap back on the sealing face, and the squeeze band is tightened. Thus the seal assembly conforms to the new shape of the face, should it deviate from being plane.

Other features of the device include a low relative rotation rate (4 rpm), a lowporosity honeycomb (60% open area), and a limited temperature gradient to avoid large thermal stresses (7°C/mm). Conductance ratio, indicated by the facearea split between the two fluids, is chosen so as to reduce the pressure drop on the low-pressure gas and increase it on the high-pressure flow.