Preliminary experimental study of a supercriticalCO₂ power cycle test loop with a high-speedturbo-generator using R134a under similarity conditions

Abstract

Research on applying a supercritical carbon dioxide power cycle(S-CO₂) to concentrating solar power (CSP)instead of a steam Rankine cycle or an air Brayton cycle has beenrecently conducted. An S-CO₂ system is suitablefor CSP owing to its compactness, higher efficiency, and dry-coolingcapability. At the Korea Institute of Energy Research (KIER), to implementan S-CO₂ system, a 10 kWe class test loop witha turbine-alternator-compressor (TAC) using gas foil bearings wasdeveloped. A basic sub-kWe class test loop with a high-speed radialtype turbo-generator and a test loop with a capability of tens ofkWe with an axial type turbo-generator were then developed. To solvethe technical bottleneck of S-CO₂ turbomachinery,a partial admission nozzle and oil-lubrication bearings were usedin the turbo-generators. To experience the closed-power cycle anddevelop an operational strategy of S-CO₂ operatedat high pressure, an organic Rankine cycle (ORC) operating test usinga refrigerant as the working fluid was conducted owing to its operationalcapability at relatively low-pressure conditions of approximately30 to 40 bar. By operating the sub-kWe class test loop using R134aas the working fluid instead of CO₂, an averageturbine power of 400 W was obtained.

Keywords

supercritical CO₂ / power cycle / turbomachinery / compressor / turbine

Introduction

A supercritical carbon dioxide powercycle (S-CO₂) can be applied to diverse energysources such as solar thermal, high-temperature geothermal, wasteheat, nuclear, and coal-fired, and is suitable for distributed powersources owing to its compactness [1].

The USA, the leading country in termsof such technology, has begun developing supercritical CO₂ power generation cycles. Sandia National Laboratoryhas implemented a 250 kWe re-compression closed Brayton cycle usingtwo turbine-alternator-compressor (TAC) units consisting of gas foilbearings [2–5]. In addition, Bechtel Marine PropulsionCo. developed a 100 kWe supercritical CO₂ powergeneration cycle test loop using one turbo-generator and one turbo-compressor[6,7]. South-west Research Institute andGE Global Research have been designing a 10 MWe supercritical CO₂ turbo-expander for high-temperature solar power generationand are building a 1 MWe test facility capable of miniaturizing andtesting the device [8,9]. Echogen has also developed an 8MWe supercritical CO₂ power generation systemfor waste heat recovery, and a 2.4 MWe power was generated [10].

The S-CO₂ isdifficult to implement and operate in an actual cycle owing to thedifficulty in applying high-speed turbomachinery technology, includinga turbine wheel, compressor impeller, bearings, seals, axial forcebalancing and rotordynamics. In fact, Sandia National Laboratory andBechtel Marine Propulsion Company cannot operate their test loopsat their respective design point because both TAC units cannot reachthe designed rotational speed of 75000 RPM [2–7].

In terms of CO₂ cycle, some kinds of CO₂ expanders have beendeveloped to use in the CO₂ refrigeration cycleor CO₂ heat pump cycle. A piston expander,a rolling piston expander, a vane expander, a scroll expander, a screwexpander and a turbo expander were representative types of CO₂ expander [11]. However, these conventional expanders can be used for small andlow temperature/pressure thermodynamic cycles. A radial or an axialtype turbo-generator are suitable for large-scale power plant applicationbecause these turbo-generators are high efficient and can be extendedto hundreds MW-class by multi-staging.

In case of hundreds kW laboratory-scaleS-CO₂ turbo-generator, the rotational speedof turbomachinery is extremely high because the mass flow rate islow and the expansion ration across the turbine is low. Because theturbine outlet pressure is still higher than the critical pressureand the maximum turbine inlet pressure is limited by material technologylevel, the expansion ratio is low. Therefore, Sandia and Bechtel usedgas foil bearings suitable for high rotational speed and high pressureconditions of S-CO₂. The gas foil bearingscould be installed inside of a hermetic turbo-generator casing, therefore,additional seal and lubrication system were not needed. However, thegas foil bearings were weak for high axial force and high windageloss conditions.

Therefore, the objectives of thisstudy are to develop an S-CO₂ system with aturbo-generator resolving bearing failure problems reported by priorresearch groups by using conventional oil-lubricated bearing insteadof the gas foil bearing. Design and operation procedure of S-CO₂ system are described. In addition, preliminary powergenerating operation results using R134a as a working fluid, whichare meaningful for developing a full S-CO₂ system,are presented.

Supercritical carbon dioxide power cycle test loops at KIER

The Korea Institute of Energy Research(KIER) has developed three experimental test loops, given in Table1. This table describes the major features of KIER’s test loopsas compared to those in the USA. The following paragraphs introducethe three supercritical CO₂ test loops (10kWe class, 100 kWe class, and sub-kWe class). Details of the sub-kWeclass are described in later sections of this paper.

The first test loop is a simple 10kWe class un-recuperated Brayton power cycle experiment test loop.The loop was designed, manufactured, and operated to test the feasibilityof the S-CO₂. With this test loop, KIER wasthe first organization in Korea to develop a hermetic-type turbine-alternator-compressor(TAC) unit. The TAC unit was designed to have a capacity of 12.6 kWe,and is composed of a shrouded centrifugal compressor, permanent magnetmotor/alternator, and a shrouded radial turbine using gas foil journal/thrustbearings. A shrouded type of compressor impeller and a turbine wheelwith labyrinth seals were designed to overcome the thrust-balancingproblems of a high-pressure fluid. The designed temperature and pressureof the turbine inlet are 180°C and 130 bar, respectively. Therotational speed of the rotor is 70000 RPM. To minimize the windageloss owing to the high-speed rotation of the supercritical CO₂ at high pressures in a hermetic structure, the pressurebetween the compressor and turbine has to be lowered. Therefore, asystem for recompressing the secondary CO₂ leakageflow path was designed. A heating medium oil boiler capable of heatingto 300°C was used as a heat source. Printed circuit heat exchangers(PCHEs) were used on both sides of the heat source and heat sink.For the system operation, the heat source supply was controlled throughthe temperature and flow rate of the heating oil, and the temperatureof the CO₂ supplied to the turbine was controlledusing the heat source bypass flow path on the CO₂ side. For cooling, a 200 RT refrigerator and a PCHE type heat exchangerwere used. For the initial operation, an external inverter was usedto start the TAC unit at 15000 RPM, which is the lift-off speed ofthe gas foil journal bearing. Preliminary testing was successfullyconducted at 30000 RPM, when all states of the cycle exist in thesupercritical region. Because of the low rotational speed of the rotorcompared to the design speed, and because no recuperator was used,electric power was not produced. Details of this are shown in Ref.[12].

Based on this development experience,KIER is currently manufacturing a 100 kWe class test loop. The cycleconsists of two turbines, one compressor, two recuperators, an LNG-firedheater, and a cooler. The design turbine inlet temperature and pressureare 500°C and 130 bar. To reduce the cost and accelerate the developmentschedule, a 60 kWe axial type turbo-generator, which is one of twoturbines in the full cycle, was first manufactured. To drive thisturbo-generator, the heat source, heat sink, heat exchangers, andbalance of the plants of the 10 kWe class test loop were re-used,and some parts are currently being upgraded. This will allow KIERto acquire experience and test results while the 100 kWe class fullcycle test loop is being procured and manufactured.

During the development of the 10–100kWe class test loops, a smaller test loop was necessary to investigatethe diverse cycle configurations such as a transcritical, a Brayton,a recuperated, a recompression and a cascade cycle which are suitablefor diverse applications of S-CO₂ system byminor modifications. In addition, characteristics of the key componentsand operational strategies have to be investigated. For these reasons,a small-scale, minimum-sized experimental test loop was developedthat can be applied to a future scale-up. Although the compactnessof the loop is advantageous for the S-CO₂ ata commercial plant level of several hundred MW, constructing a systemcapable of achieving several to several tens of kW at the laboratoryscale, where the turbomachinery is extremely small, is a significanttechnically challenge. In addition, the number of revolutions increases,which causes problems such as the proper selection of bearings andseals suitable for high-temperature, high-pressure, and/or high-speedconditions, as well as windage loss. To overcome these technical issuesin a small-scale supercritical CO₂ turbomachinery,a design strategy to reduce the rotational speed of the rotor wasdeveloped.

In this paper, the designs of thetest loop and turbo-generator are described. In addition, vibrationand power generating test results of the test loop using a refrigerantof R134a as a working fluid before operating the test loop with CO₂ are described. This preliminary test was essentialbecause it could be operated at relatively low-pressure conditionsof approximately 30 to 40 bar compared to over 100 bar of a supercriticalCO₂ operation. Under moderate conditions, sufficientexercise operations of the closed thermodynamic power cycle with ahigh-speed turbo-generator can be conducted.

Design of sub-kWe class test loop

A transcritical cycle was designedand fabricated using two plunger-type reciprocating supercriticalCO₂ pumps (Catpumps, USA), as shown in Fig.1. To design test loop conditions, the turbine inlet pressure andtemperature were determined. The maximum pressure of the cycle wasdetermined as a 130 bar which is similar level of world’s bestoperating condition reported by Sandia National laboratory. The maximumtemperature was 200°C as the first step target which is relativelymild condition for the piping, valves and pressure vessels of thesystem. The pump inlet conditions were determined considering abilityof the chiller and the effectiveness of the heat exchangers. The temperatureof the coolant by the chiller was 7°C, so considering mild approachtemperature value, a 20°C of pump inlet temperature and its saturationpressure of 57 bar were determined.

Because it is the first step of developingan S-CO₂ system, the main objective of thistest loop is to operate cycle and to drive a turbo-generator producingan electricity using a supercritical CO₂. Therefore,an un-recuperated simple transcritical cycle configuration was determinedas a first step neglecting the cycle efficiency. In addition, theefficiency of the turbine and the pump were assumed as a 0.35 and0.5, respectively. These were extremely conservative values becauseof the losses which are difficult to be calculated. The net powerwas calculated as a sub-kWe-class considering estimated efficiencyrange of the turbine and the pump.

The liquid CO₂ at 20°C, 57 bar, and 0.07 kg/s was pressurized into a supercriticalstate at 130 bar using two positive displacement pumps with differentflow rates by driving inverter-controlled electric motors to testthe various flow conditions. An immersion electric heater heated theCO₂ up to 200°C, and after driving theturbine, the CO₂ was cooled through a weldedplate heat exchanger. In this scheme, two of the PCHEs were not usedfor the transcritical cycle operation. The valves were used to allowone of the PCHEs to operate as a recuperator under the recuperatedcycle configurations, and the other PCHE as a CO₂ gas pre-cooler under other configurations.

To analyze the characteristics ofthe cycle and develop the operation technology, an electric heatersystem, which is easy to operate and control, was used as a heat source.A heat source capable of supplying a total heat of 30 kW was designedand manufactured by winding a 20 kW band heater on the outside ofa 10 kW immersion heater. A refrigerator used for producing coolingwater was manufactured, allowing the heated supercritical CO₂ to be cooled and compressed again. To supply the coolingwater according to the various capacities of the experiments, a refrigeratorof 5 RT, 10 RT, and a 10 kW heater were combined to precisely controlthe temperature of the cooling water. In addition, other componentsin the system include a liquid CO₂ injectionpump, an expansion tank, a syringe pump for regulating the internalpressure and flow of the system, a system for supplementing the CO₂ turbine leakage, various valves and safety devices,and a measuring device [13].

Design of turbo-generator

Figure 2 shows a comparison of thegeneral specific velocity of gas/steam turbines, and the specificvelocity area derived from the design requirements. The design pointof our supercritical CO₂ turbine is describedas a gray area. The specific speed of the target turbine, which isa function of the mass flow rate and the isentropic work, is aroundon the order of 10⁻². This value is not suitable for a radial turbine, which is on theorder of 10⁻¹. To make a small test loop, small mass flow rate was determinedaccording to capacity of pumps, heat exchangers, heater and chillerin the system. However, this was extremely small mass flow rate conditionto design a suitable turbine. Designed rotational speed of the turbinewas 800000 RPM assuming a full-admission nozzle under 0.07 kg/s condition.This was impossible design to manufacture owing to limits of bearingsand seals. However, even if the efficiency of the turbine is reduced,the radial turbine can be operated within the specific speed rangeby applying a partial admission technique using only a part of thenozzle. This design strategy used to reduce the specific speed allowsthe rotational speed of the turbo-generator to be reduced.

In this study, a partial-admissionradial turbine for supercritical CO₂ powerwas designed for the first time in the world and evaluated throughCFD. To design this type of the turbine, first, virtual full-admissioncondition was assumed. It is because that there are no design toolsor skills to design a partial admission turbine directly. Under full-admissioncondition, turbine blade shape and the number of blades and nozzleswere designed. Using Rital, AxCent, and Pushbutton CFD (Concepts NREC),a 22.6 mm diameter partial admission radial turbine was designed,as shown in Fig. 3. The turbine revolves at 200000 RPM and includes9 nozzles and 13 blades. Through trial-and-error, the turbine wasdesigned to have a mass flow rate that is calculated by multiplyingthe target mass flow rate by the number of nozzles. The mass flowrate was designed to be 0.621 kg/s for full-admission, which is ninetimes the 0.07 kg/s flow rate of the target mass for the test loopcondition. The turbine efficiency was designed as 79.7% at full-admissioncondition as shown in Table 2. Then, this turbine was equivalent withthe turbine which has mass flow rate of 0.07 kg/s with a one nozzle.This means that only one nozzle was used in the experiment, and theremaining eight nozzles were blocked. With this design trick, a one-nozzlepartial admission turbine could be designed. However, it is clearthat the turbine efficiency will decreases. Because there are lossesin real condition that could not be considered at the design stageby using a partial admission nozzle, real turbine efficiency haveto be measured by experiments. Figure 4 shows the manufactured turbinewheel and partial admission nozzle. To realize the partial admissionnozzle, one nozzle shape was designed and fabricated such that theworking fluid could flow into the turbine.

On analyzing the development experienceand data pertaining to existing technology from leading countries,it was determined that the primary technical difficulty of the smallturbomachinery is the breakdown of the gas foil bearing owing to thelarge axial force generated under high-pressure and/or high-temperatureoperating conditions during the S-CO₂ cycle.To resolve the thrust load problem, as shown in Fig. 4, a strategyfor constructing a turbo-generator layout using a commercial oil-lubricatedbearing was used. Angular contact ball bearings (SKF) suitable forcompact high-speed rotation were used to simultaneously support theaxial and radial forces generated by the turbine wheel. The turbinewheel and bearings were separated using several labyrinth seals foroil lubrication. A leakage flow of 10% of the mass flow rate of theworking fluid was predicted by calculation. It is determined by designingthe labyrinth seals as a function of pressure difference, channelwidth and mechanical tolerance. Therefore, a system was made to supplementthis flow. A high-speed generator made by E+ A, suitable for operatingconditions at 200000 RPM, was used, and the optimum layout was designedthrough an analysis of the rotor dynamics. A structural analysis wasalso conducted on each part to confirm the structural safety.

There is a possibility that the rotationalstability of the turbo-generator may deteriorate, owing to the impactof the fluid flowing into the turbine wheel. The turbo-generator developedin this study is an ultra-high-speed rotating body, and it is importantto control the assembly tolerance of each part, as well as the vibrationlevel of the entire shaft. As shown in Fig. 4, a gap sensor was installedat the left end of the shaft system, and the distance between thesensor and shaft was measured to monitor the degree of shaft vibrationin the radial direction. The gap-sensor and converter used were anAEC-55 and a PU-02A, respectively; the data collecting device wasan IOtech 650U; and the dedicated software was from EZ-Tomas. Thedesign value was set to prevent contact between the rotating shaftand the gap sensor, and a test was carried out to confirm the rangeof radial vibration of the shaft, even during the actual driving [13].

Preliminary test loop operation

To analyze the rotational stabilityof the turbo-generator, a no-load rotation test driven by an externalelectric source was conducted. Next, a cold-gas test was conducted,during which the turbo-generator was driven using air and nitrogenunder atmospheric conditions to confirm the effects of the partialadmission nozzle on the stability of the rotor. Figure 5 shows thegap sensor-rotor distance in RPM. Allowable vibration limits of theturbo-generator was 40 mm. First, the vibration level when the turbo-generatorwas driven by an inverter was obtained as a reference case. Bold solidline describes a reference value. Then, the pressurized gas was suppliedto the turbo-generator. In case of no load condition, described asa dashed-line, a little increment of vibration was observed. Thisis due to flow impact on the rotor induced by a partial admissionnozzle. After that, electric load was imposed to the turbo-generator,and then the vibration increased owing to change of force conditionon the rotor described as a solid line. Although there were a littleincrement of vibration level due to a partial admission nozzle, allvalues existed under the allowable vibration limit. It was confirmedthat there were no stability problems during rotation, and the testingproceeded to the turbine under actual cycle operation.

As a preliminary operation test,before using supercritical CO₂, R134a refrigerantwas used as a working fluid to test each component, including thepumps, separator, valves, heater, and safety, and to experience theclosed Rankine cycle operation and determine various troubleshootingprocedures. This organic Rankine cycle (ORC) operating test usinga refrigerant as a working fluid was necessary because it can be operatedunder relatively low-pressure conditions of approximately 30 to 40bar. R134a was selected owing to its good reliability at high temperaturesof approximately 300°C, which is a relatively high temperaturefor a refrigerant. Owing to the material compatibility, an O-ring(Kalrez FFKM) in the turbo-generator was changed to the Perfluoroelastomersuitable for R134a.

At the first stage, a pump-coolingtest was conducted by controlling the valves in the loop. Here, therelationship between the amount of working fluid in the test loopand the pressure formation in the loop was tested, which is an importantoperation parameter in a closed power cycle. At the second stage,a cycle test was conducted using a heater and expansion needle valveinstalled in the turbine bypass line. The turbine inlet and outletvalves were closed, and the expansion needle valves were controlled.The temperature and pressure conditions of the cycle were increasedthrough control of the heater and expansion needle valve. The amountof working fluid in the loop was controlled using a filling pump anddrain valve. At the third stage, a turbine operation was conducted.By controlling the inlet and outlet valves of the turbo-generator,the working fluid can flow into and drive the turbine, thereby generatingelectric power.

By conducting these R134a operationaltests, several minor bugs in the test loop were corrected. After correctingany existing problems, an operational strategy was developed and executed,as shown in Table 3.

Figure 6 shows preliminary test resultsof the experimental loop using R134a as time laps. A pressure of 29.5bar and a temperature of 110°C were obtained for the inlet turbineconditions by controlling the closed Rankine cycle operation strategy.The expansion ratio was 3.0, and the rotational speed of the turbo-generatorwas 90000 RPM under the highest power conditions. Within a relativelystable operation region, 400 W of turbine power was successfully obtained,as shown in Fig. 7.

Although the preliminary test wasnot carried out under fully similar conditions, it is meaningful thatall components of the experimental test loop, including the high-speedturbo-generator, could be tested and applied at more moderate operatingconditions using a working fluid that lowers the maximum pressureof the cycle. In addition, the operation and control strategy weredeveloped and tested. After testing the closed power cycle loop operation,a working fluid will be substituted with CO₂ as the next step.

Conclusions

An experimental small-scale, multi-purpose,S-CO₂ test loop was developed using a high-speedradial-type turbo-generator. A partial admission nozzle was designedto obtain the proper turbine size and operational speed for manufacturingunder very small mass flow rate conditions. A commercial oil-lubricatedangular contact ball bearing was used to avoid bearing failure problemsdiscovered by prior research groups. A preliminary experiment testwas conducted using R134a as a working fluid to determine the operationalcharacteristics of the closed Rankine cycle. Turbine power of 400W was successfully obtained under similarity condition.

References

[1] Dostal V, Driscoll  M, Hejzlar P. A supercritical carbon dioxidecycle for next generation nuclear reactors. Massachusetts Institute of Technology,MA 2004, MIT-ANP-TR-100

[2] Wright S A, Radel  R F, Vernon  M E, Rochau  G E, Pickard  P S. Operation and analysis of a supercritical CO₂ Brayton cycle. SANDIA REPORT, 2010: SAND2010–0171

[3] Pasch J J, Conboy  T M, Fleming  D D, Rochau  G E. Supercritical CO₂ recompression Brayton cycle: completed assemblydescription. SANDIA REPORT, 2012:SAND 2012–9546

[4] Conboy T, Wright  S A, Pasch  J, Fleming D,  Rochau G,  Fuller R. Performance characteristics of an operating supercriticalCO₂ Brayton cycle.  ASME Turbo Expo: Turbine Technical Conference & Exposition, 2012, 134(11): GT2012–68415

[5] Conboy T, Pasch  J, Fleming D. Control of a supercritical CO₂ recompression Brayton cycle demonstration loop. Asme Turbo Expo: Turbine Technical Conference & Exposition, 2013, 135(11): GTP–13–1198

[6] Clementoni E M,  Cox T L. Steady-state power operation of a supercritical carbon dioxide Brayton cycle. ASME Turbo Expo: Turbine Technical Conference & Exposition, 2014: GT2014–25336

[7] Clementoni E M,  Cox T L. Practical aspects of supercritical carbon dioxide Brayton system testing. Proceedings of the 4th International Symposium-SupercriticalCO₂ Power Cycles, Pittsburgh, Pennsylvania, 2014

[8] Kalra C J. Development of high efficiency hot gas turbo-expanderfor optimized CSP supercritical CO₂ power block operation. Proceedings of the 4th InternationalSymposium-Supercritical CO₂ Power Cycles, Pittsburgh, Pennsylvania, 2014

[9] Moore J, Brun  K, Evans N,  Bueno P,  Kalra C J. Development of a 1 MWe supercriticalCO₂ Brayton cycle test loop.  Proceedings of the 4th International Symposium-Supercritical CO₂ Power Cycles, Pittsburgh, Pennsylvania, 2014

[10] Held T J. Initial test results of a megawatt-class supercriticalCO₂ heat engine. Proceedings of the 4th International Symposium-Supercritical CO₂ Power Cycles, Pittsburgh, Pennsylvania, 2014

[11] Zhang Z, Ma  Y, Li M,  Zhao L. Recent advances of energy recovery expanders in the Transcritical CO₂ refrigeration cycle. HVAC & R Research, 2013, 19(4): 376–384

[12] Cho J, Choi  M, Baik Y J,  Lee G, Ra  H S, Kim  B, Kim M. Development of the turbomachinery for the supercritical CO₂ power cycle. International Journal ofEnergy Research, 2016, 40(5): 587–599

[13] Cho J, Shin  H, Cho J,  Ra H S,  Roh C, Lee  B, Lee G,  Baik Y J. Development of the supercritical carbondioxide power cycle experimental loop with a turbo-generator. Asme Turbo Expo: Turbomachinery Technical Conference & Exposition, 2016: GT2017–64287

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag GmbHGermany