Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

"Practical Considerations in Scaling Supercritical Carbon Dioxide Closed Brayton Cycle Power Systems" Robert L. Fuller, William Batton BARBER-NICHOLS INC. 6325 West 55th, Arvada, Colorado, 80002 Phone: 303-421-8111, Fax: 303-420-4679, Email:[email protected];[email protected] Abstract - The supercritical CO2 closed Brayton cycle offers a high efficiency power generation system from relatively low temperature sources. The thermodynamic properties of CO2 including the high density make the turbomachinery and heat exchangers very compact. Unique challenges exist in the design of power systems using super critical CO2. The challenges to be addressed depend on the power level and the thermodynamic cycle. Several power levels and thermodynamic cycles are presented. Turbine and compressor design, power conversion technology, seals, bearings, and rotor dynamics are some of the items discussed. I. INTRODUCTION Supercritical carbon dioxide power cycles are being investigated as a way of making electric power with good efficiency at relatively low temperatures. Carbon Dioxide is unique in that the critical temperature and pressure are 87F and 1100 psi. Operating a gas compressor near the critical point reduces the compressor power requirement, increasing cycle efficiency. Operating near the critical point also means high fluid pressure and density which lead to engineering design problems that are unique. To prove out issues regarding supercritical CO2 power systems, several scaled systems are being designed. Many design problems have to be contemplated for scaled systems that are different than that for larger systems. These design problems and options are identified for scaling of 200kW, 3 MW, and 300 MW systems. A wide range of engineering skills are necessary to analyze the options when scaling these systems. These skill sets range from thermodynamic analysis to aerodynamics, to rotor dynamics, to stress analysis. Each skill is interdependent as seen in Figure 1 and leads, in many instances to an iterative design process. The machinery design solution is greatly accelerated by having a good experience base to draw upon. While the engineering design process has been done many times for low pressure Brayton cycles, the unique nature of CO2 leads to new design challenges that must be overcome for a successful design.

Figure 1. Turbomachinery Scaling Process Flow

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

II. Cycle Analysis Cycle analysis is an important first step in determining the suitability of supercritical CO2 for an application. Consideration has to be given to the unique properties of CO2. This includes a compressor inlet temperature of 90F to 100F and compressor inlet pressure of 1100 psi to 1500 psi. Compressor inlet temperatures of up to 140F have been contemplated, but with a reduction of cycle efficiency below that of competing technologies. Also of primary importance is the turbine inlet temperature. Supercritical CO2 cycles have shown to be advantageous for applications for turbine inlet temperatures as low as 500 deg F. Higher turbine inlet temperatures provide higher efficiencies but depending on the application, structural limitations of the high pressure turbine containment structure should be considered from a materials and safety perspective. To understand a method of scaling the supercritical CO2 cycle, consider the recompression cycle as shown in figure 2. This cycle was developed at Massachusetts Institute of Technology to overcome efficiency limitations in a standard recuperated Brayton cycle. This cycle was developed for a 300 MWe nuclear application and has the ability to produce over 45% efficiency with a 1022F turbine inlet temperature (1). Other cycles have been contemplated, primarily for nuclear applications (2).

Figure 2. Recompression Cycle Simple Diagram The cycle state points are listed in table 1. To scale the cycle from 300 MWe to 3 MWe and then to 200 kWe, the mass flow is reduced proportionally while keeping the pressures and temperatures equivalent. If the head remains constant across a turbine or compressor, and the mass flow changes, the turbo-machinery must adapt. In some instances the type of generator, compressor, or turbine technology must change. Also seals, bearings, and even rotor configurations must change as well to compensate for changing mass flow. If acceptable, the cycle design may have to be reconsidered in light of these technology choices with the goal of the scaling system within available technology choices.

Turbine

Re-compressor

Main Compressor

Inlet Pressure (psia)

2876

1116

1115

Exit Pressure (psia)

1145.8

3000

3000

Inlet Temperature (F)

1022

157

89.6

Table 1. Recompression Cycle State Point Information III. Turbomachinery Considerations: III (a). Generator Technology It is desirable for many applications to create either 50 or 60 Hz power. For large power plants this is a synchronous generator

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

operating at 3000 or 3600 rpm and operating at high voltage, 13.8 kV or above. This can typically be accommodated by multistage compressors and turbines operating on a single shaft. As the power level decreases it can be more difficult to provide efficient turbine and compressor machinery that can operate at synchronous operating speed. At this point several options exist to provide power at higher shaft speeds. The first option is to utilize a speed decreasing gearbox. This allows the turbine and compressor shaft to operate at higher speeds while utilizing a standard 1800 or 3600 rpm generator technology. Gearboxes are reasonably efficient, but require lubrication, extra seals, bearings, increased space, higher noise, and decreased reliability. Operating the gearbox in a hermetic type closed Brayton cycle would make it difficult to lubricate the bearings and gears with oil. The oil could tend to migrate or breakdown at supercritical CO2 conditions. The migration could introduce oil in areas such as the heater providing an opportunity to coke or otherwise inhibit proper heat exchanger performance. Another solution is to utilize 400 Hz power if it is suitable for the application. This would typically be for shipboard power applications where the distribution losses inherent in 400 Hz power are mitigated by the island nature of the load. This would allow consideration of synchronous machines operating at 12 to 24 krpm. While these higher speed machines are prevalent on aircraft at lower power levels, they would need to be developed for applications in the .5 to 20 MWe range where they may be desirable.

Still another solution is to utilize power conversion technology to take the high frequency generator output, convert this to DC voltage and then invert to 50 or 60 Hz. This technology has primarily been used for low voltage (<600 VAC) systems up to 2 MWe for conventional commercial applications such as uninterruptible power supplies and micro-turbines. They typically utilized permanent magnet designs that generate voltage proportional to speed. This high frequency voltage is rectified to DC and then reconverted to lower frequency alternating current. This technology has been extended to medium voltage systems, primarily for motors driven compressors for offshore oil platforms where space is at a premium. It has the advantage of allowing the turbo-machinery to operate more optimally at off design conditions by running at speeds not grid synchronous. The additional EMI/EMC can also be challenging for some applications. Typical efficiencies are less than a standard large synchronous generator and the system requires large control circuits that must be housed and cooled. The cycle designer should know the limitations of the medium voltage drive technology and be prepared for the associated development activity if an off-the-shelf application can not be found. At power levels below 350 kWe, technology from the aerospace industry can be used. Single magnet, two pole machines operating on air bearings have been sufficiently developed and can be applied to supercritical CO2 systems. These machines are constructed in a manner to allow high speeds of up to 650 ft/sec surface speeds for the rotor with metallic magnet retention, and up to 900 ft/sec for carbon graphite magnet retention. The cylindrical boundary between the rotating magnetic assembly and the stator can be thought of as carrying a shear stress that produces the torque for motoring or generating. This air gap shear stress value is a useful calculation tool to size a motor or generator. Since a maximum surface speed value is dictated by the stress of available materials, the length of the magnetic machine can be calculated if an appropriate value for air gap shear stress can be determined. For most high speed aircraft style generators this value is 3 psi, for more common electric motors this may be 1 psi, and for large power plant generators, this value may be 10 psi (2). The motor/generator drive controls and foil bearing technology that has been utilized in aircraft style systems has been developed sufficiently to apply to the CO2 application. III (b) Turbine and Compressor Design The turbine and compressor designs are dictated by the operating regime, specified primarily by pressures, temperature, and flow rates. As the output power is scaled to lower power levels by reducing mass flow, changes in the turbo-machinery must follow to maintain optimum efficiency. The optimum shaft speed for high compressor and turbine efficiency will increase as the mass flow is decreased. This can be seen by the concept of specific speed of an expander in equation 1, and for a compressor in equation 2. The specific speed is utilized relative to the turbine or compressor type to optimize the performance. The specific speed can be used to determine the approximate efficiency of a compressor or a turbine based on empirical information gathered in Balje (2).

Ns

rpm * V3 H ad

3/ 4

(1) Turbine specific speed calculation

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

Ns Where

rpm * V1 H ad

3/ 4

(2)

Compressor specific speed calculation

rpmdenotes the shaft speed, V3 is the expander exit volumetric flow, V1 is the compressor inlet volumetric flow, and

H ad is the adiabatic head across the device. Optimum specific speed is related to the type of turbine or compressor technology utilized. It can be seen from the above equations that for a desired adiabatic head and volumetric flow, only the shaft speed can be changed to realize the optimum specific speed and therefore efficiency. A radial turbine, for instance, shows an optimum specific speed of 60 to 80 (3) as seen in figure 3 (4). The same type of optimum efficiency vs. specific speed calculation can be done for compressors. If the compressor(s) and turbine(s) are on the same shaft, then the overall optimum efficiency may force a balanced approach to compromise the efficiency of the turbine and compressor for the benefit of the overall cycle efficiency. Since the adiabatic head can be divided by multiple stages; the radial and axial machinery can be applied as multiple stages within the aerodynamic, structural, rotor dynamic, bearing, seal, and performance range to meet the desired result. As the cycle is scaled to lower power levels, it becomes increasingly more difficult to apply multiple stages to the scaled solution and maintain equivalent cycle efficiencies. On scaled systems of very low power level, the mass flow rate can result in very small turbomachinery blade heights that are detrimental to component efficiencies and require very high speeds. An example set of calculations for three system power levels can be done for the recompression cycle cited previously. For 300 MWe it is anticipated that a 3600 rpm synchronous generator technology can be utilized. For this analysis it was assumed that all turbomachinery is operating on a single shaft driving a generator. Referring to table 2, it is plausible to design adequate turbines and compressors, but using single stage is not a viable solution. Given the shaft speed of 3600 rpm, and high volumetric flow rates, it is plausible that multistage radial or axial turbines and compressors could yield a satisfactory solution.

300 MW Inlet Pressure (psia) Exit Pressure (psia) Inlet Temperature (F) Efficiency Head (lbf*ft/lbm) Mass Flow (lb/sec) Volume Flow (ft^3/sec)* Desired Shaft Speed (rpm) Specific Speed (1 stage) Optimum Specific Speed** Approximate Wheel Diameter (in) *Turbine Exit, Comp Inlet **Radial Type

Turbine 2876.0 1145.8 1022.0 0.9 44937.3 7667.0 2099.6 3600.0 53.4 80.0 60.0

Re-compressor 1116 3000 157 0.89 17855.1 3066.8 298.1528291 3600.0 40.2 100 to 150 Not provided

Main Compressor 1115 3000 89.6 0.89 6457 4600 123.1988859 3600 55.5 100 to 150 Not provided

Table 2. 300 MW Turbomachinery Example To scale the 300 MW to 3 MW, the mass flow was reduced by a factor of 100, table 3. A shaft speed increase to 50,000 rpm is required to yield a turbine near the goal of 80. Even in this case a single stage radial turbine would probably not yield 90% efficiency. A generator operating at 50,000 rpm with an output of 3 MW does not exist and would be extremely difficult, if not impossible to design. It is clear that the machinery should be broken into multiple stages to slow the shaft speed to a value that is reasonable for a generator technology to be found. In this case, the number of stages would be quite high to reduce the speed to 3600 rpm and the number of stages required to do so would not yield high efficiency turbines or compressors. Therefore, a high speed gearbox, high speed permanent magnet or synchronous machine would be necessary. Another alternative is to change the cycle for a lower pressure ratio, which would lower the head requirement on the turbomachinery, which would also lower the overall converted efficiency.

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

3 MW Inlet Pressure (psia) Exit Pressure (psia) Inlet Temperature (F) Efficiency Head (lbf*ft/lbm) Mass Flow (lb/sec) Volume Flow (ft^3/sec)* Desired Shaft Speed (rpm) Specific Speed (1 stage) Optimum Specific Speed** Approximate Wheel Diameter (in) *Turbine Exit, Comp Inlet **Radial Type

Turbine 2876.0 1145.8 1022.0 0.9 44937.3 76.7 21.0 50000.0 74.2 80.0 6.0

Recompressor 1116 3000 157 0.89 17855.1 30.66 2.98 50000 55.9 100 to 150 Not provided

Main Compressor 1115 3000 89.6 0.89 6457 46 1.23 50000 77.0 100 to 150 Not provided

Table 3. 3 MW Turbomachinery Example Extending the scaling of the 300 MW to 300 kW reduces the mass flow further. Utilizing equations 1 and 2, the specific speed of the turbine and compressors at 125,000 rpm is still insufficient to support high efficiencies. A generator operating at a shaft speed of 125,000 rpm and yielding 300 kW is not technically feasible. In this case it is absolutely necessary to split the cycle into two rotating shafts and also lower the head across each turbine or compressor. Adding multiple stages to the turbines and compressors is not feasible due to the very small wheel diameters and poor rotordynamic outcome. Sealing solutions for shaft speeds for this equipment are limited. Also blade heights are small and the turbine and compressor efficiencies will suffer. It would be much easier to start with a reasonable 300 kW system and scale upwards to 300 MW from a turbomachinery perspective. 300 kW Inlet Pressure (psia) Exit Pressure (psia) Inlet Temperature (F) Efficiency Head (lbf*ft/lbm) Mass Flow (lb/sec) Volume Flow (ft^3/sec)* Desired Shaft Speed (rpm) Specific Speed (1 stage) Optimum Specific Speed Approximate Wheel Diameter (in) *Turbine Exit, Comp Inlet **Radial Type

Turbine 2876.00 1145.80 1022.00 0.90 44937.28 7.70 2.10 125000.00 58.69 80.00 1.79

Table 4. 300 kW Turbomachinery Example

Recompressor 1116 3000 157 0.89 17855.1 3.07 0.298 125000 44.2 100 to 150 Not provided

Main Compressor 1115 3000 89.6 0.89 6457 4.6 0.123 125000 60.9 100 to 150 Not provided

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

Figure 3. Radial Inflow Turbine Maximum Efficiency from Reference (4) III (c). System design including bearings, seals, and Windage. Due to the high operating pressures of the overall supercritical CO2 system, it is necessary to remove as many rotating shafts from the high pressure gas as possible to reduce windage losses. This is primarily the bearings and the generator components. Shaft seals need to be employed to either allow a low pressure CO2 cavity for generator and bearing operation, or to relocate the generator and bearings to an ambient environment. Proper seal technology is necessary to provide a barrier to leakage across the high pressure boundary. In the case of a hermetically sealed system, the boundary separates high pressure CO2 to a low pressure CO2 cavity that has been evacuated to a lower pressure level. In most applications a pump is used for cavity evacuation and the CO2 is reintroduced to the system at a high pressure point in the system. The pump power parasitic loss can subtract from the overall system efficiency and is proportional to the differential cavity pressure and seal leakage flow rate. From experience, a cavity pressure in CO2 for generators and bearings would be in the 100 to 200 psia range for low windage losses. This pressure is low enough for the high pressure CO2 to turn partially liquid as it flows to the low pressure cavity, depending on the temperature of the gas. The closed loop concept also has other implications. When the system is shut down, the components in the system may be at supercritical CO2 conditions. Since CO2 can permeate materials and is a good solvent, the generator and bearing materials and lubrication methods must be chosen accordingly. Also, upon startup, the cavity pump down rate must be monitored to keep any materials from deterioration due to CO2 outgassing. In the case of a seal buffering to ambient, air, the CO2 leakage may turn solid (as dry ice) and must be replaced in the system as it leaks out to keep the closed Brayton cycle operating properly. Several seal types have been identified depending on the shaft speed, diameter, pressure, and cavity pump availability. Labyrinth seals are the most economical and compact. They can be used when pumping power consequences are not counted against the overall cycle efficiency, i.e. for laboratory demonstration equipment. Brush seals have been investigated, but are typically available as custom designs for volume applications. Liftoff gas seals have been utilized in commercial equipment on CO2 at high pressures and reasonable shaft speeds. They have the lowest leakage rate and should be investigated further relative to the various power system configurations as well as other types of seals that may be suitable. The bearings should operate on the low pressure side of the CO2 or in ambient air environment. This could allow the use of oil lubricated bearings. In the case of low pressure CO2 a gas oil separation system would be necessary to remove the possibility of the oil migrating in the system. This may be difficult to do to the degree necessary in some applications. Foil bearings have

Proceedings of SCCO2 Power Cycle Symposium 2009 RPI, Troy, NY, April 29-30, 2009

been utilized in low pressure CO2, but have been so far limited to light rotors. Foil bearings also have a limit on the minimum shaft speed and must be aggressively started to the minimum speed to prevent foil coating damage. A third alternative is magnetic bearings. Magnetic bearings have been utilized in larger machinery. They are more expensive and complicated than other types of bearings, but may be a good choice considering the operating conditions for some systems. IV. CONCLUSIONS Supercritical carbon dioxide Brayton systems are difficult to scale over a wide output power range. The technologies utilized vary widely over a scaling range of 1 :1000. Small scale systems will have lower efficiencies and therefore may be adequate for laboratory or demonstration use, but may not be competitve against other power cycles. At low power levels, technologies have to be developed that are equally as challenging as engineering development than large scale power systems, but may have no practical application on the larger system design. ACKNOWLEDGMENTS We wish to thank Sandia National Laboratories and Knolls Atomic Power Laboratory for the opportunity to work on supercritical carbon dioxide power system development. REFERENCES 1.

V. Dostal, M.J. Driscoll, P. Hejzlar, “A Supercritical Carbon Dioxide Cycle for Next Generation Nuclear Reactors” MITANP-TR-100 March 10, 2004.

2.

C.H. OH et.al, “Development of a Supercritical Carbon Dioxide Brayton Cycle: Improving VHTR Efficiency and Testing Material Compatibility” Final Report, Idaho National Laboratory, INL/EXT-06-01271, 2006

3.

O.E. Balje, “Turbo machines: A Guide to Design Selection and Theory”, ISBN 0471060364

4.

H.Rohlik, “Analytical Determination of Radial Inflow Turbine Design Geometry for Maximum Efficiency,” NASA Technical Note TN D-4384 , 1968.