Analysis of a Stirling-Cycle Power Convertor for Domestic Combined Heat and Power D. J. Buckmaster and W. S. Newman, Senior Member, IEEE Abstract—Use of a Stirling-cycle convertor is analyzed with respect to domestic application to micro Combined Heat and Power. An engine that can convert heat into electric power has the potential to extract more availability from fuel consumption than combustion for space heating alone. A specific conceptual design is presented, illustrating that the technical aspects are manageable. A dynamic analysis shows that such a system could be integrated with line power without expensive tie-in electronics. A cost analysis indicates an installed-system target price for practical consumer adoption. Index Terms—Stirling engine; linear alternator; combined heat and power; co-generation I. BACKGROUND AND MOTIVATION Typically, homes are heated through combustion of hydrocarbon fuels (with a majority of U.S. homes using natural gas). Thermodynamically, this process is wasteful, since the fuel is combusted at a relatively high temperature and the heat produced is mixed with air at room temperature, thus generating significant entropy and corresponding loss of thermodynamic availability. A more efficient use of the heat would be to extract as much electric energy as possible and use the balance for space heating—a concept known as “Combined Heat and Power” (CHP). Assuming a typical heat-exchanger temperature of TH =400C (673K) in a gas- fired furnace and a room temperature of energy TH =20C (239K), for QH of heat released by combustion, a theoretically perfect Carnot engine would be able to extract an amount of work W = QH (TH − TL ) / TH ≈ 0.56QH . As an example, the home of one of the authors consumed 64MCF of natural gas in January, 2011 at a cost of approximately $700. Combustion of this gas corresponds to 67GJ of heat at a cost of $10.54/GJ. For the same region, the cost of electricity to the consumer (fully loaded) was $0.1385/kWh. As an upper bound, if 67GJ of waste heat came from a theoretically perfect heat engine operating between 400C and 20C, gas consumption equivalent to 120GJ of heat would produce 53GJ of electric energy and still provide the desired 67GJ of waste heat. At the fully-loaded consumer price of $10.54/GJ of heat energy from natural gas, 120GJ would have cost $1,265. However, if the electricity produced could be sold back at the consumer’s rate of $0.1385/kWh ($38.47/GJ), the value of the electricity produced would be $2,039. Thus, for this month, instead of paying $700 for heat, 978-1-4577-0776-6/11/$26.00 ©2011 IEEE the consumer would have received space heating for free plus a payment of $774 for electric energy sold. The above limiting analysis assumes an unrealistic efficiency, as well as ignores a barrier in selling net electric power at the consumer’s rate. While all but four states currently have some form of net-metering regulation [1], details vary as to how consumers are credited for net production. Importantly, in many states consumers are reimbursed for excess production at a “wholesale” rate that is significantly lower than the retail cost [2]. Thus, more pragmatically, the optimal power to produce over a month would be no more than the homeowner consumes, resulting in zero net electric energy consumption. Continuing from the same anecdotal example, a representative electric consumption was 850kWh/mo (3.06GJ). Producing this much electric energy per month would have a consumer value of $118/mo, resulting in zero electric bills. If the system were run continuously to produce the desired 3GJ of energy per month, the generator should be sized to approximately 1kW. To further improve the realism, assume the energy convertor operates at 50% of Carnot efficiency (28% conversion of heat to electric energy for temperature reservoirs of 400C and 20C). To produce 850kWh of electric power at this efficiency would require approximately 10.9GJ of gas energy consumption and would exhaust 7.9GJ of waste heat. Correspondingly, an extra 3 GJ of heat energy, purchased for $31.62, would be required to produce the desired 3GJ of electric energy, worth $118, for a savings of approximately $86. This savings could be realized for each month during which at least 7.9GJ of space heating was needed (which was 9 out of 12 months for this anecdotal example). For this example, the annual value of a CHP system operating at 50% of Carnot efficiency would be approximately $775/yr. Such savings would fluctuate with the cost of gas and the cost of electricity. For a simple payback of 5 years, the installed system cost would have to be less than $4,000 to be justified. This rough analysis establishes an estimate for the size and installed cost of a practical domestic CHP system. A significant body of work going back decades exists with regard to evaluating the utility of micro-CHP systems. In [8], the merits of a variety of CHP systems (Stirling-engines, fuel cells, etc.) are evaluated, and the authors concluded that up to 45% of installed central-heating boilers in the Netherlands could be replaced by micro-CHP systems per year based on the units’ average life-cycle. The authors of [9] focus on larger, “mini-CHP” systems and conclude that there is a large potential for CHP in high-load, non-industrial applications such as office buildings and hospitals. Reference [10] evaluates potential of different micro-CHP solutions for use in a particular Belgian city, including an analysis of thencurrently (2009) available commercial products. This included three crank-driven and two “free-piston” Stirling engines, for which the primary barrier to feasibility was a cost 2-5x higher than what the authors considered viable, despite being the technologies nearest to commercial availability. In [11], the authors went so far as to create mathematical models based on appliance usage, home location, and insulation properties to predict end-user cost savings from CHP systems. Publications of this sort have been frequent over more than two decades, especially because of the benefits pointed out in our anecdotal analysis and in [12]: micro-CHP systems can achieve upwards of 90% conversion from chemical energy to “useful” energy in the form of electricity and heat, while centralized electricity production and distribution can squander up to 60% at the point of production, and another 5% in transmission. In addition to cost-benefits provided to the end user, micro-CHP quite simply allows for more efficient use of scarce resources. We next consider a specific technology to serve this domain. compact, result in higher alternator efficiencies and have attractive start-up transient properties. Another development promoting Stirling-cycle designs is the emergence of commercial CAD packages for the design and analysis of Stirling systems. A notable example is “Sage” [5], a software system used in the present analysis for design of Stirling engines. The Sage program assumes periodicity of thermodynamic solutions, from which it can derive nonlinear pressure waveforms, time-varying flow rates and heat transfer, and power fluxes. Figure 1 shows a schematic example of a Sage model for a Stirling engine. The primary modules consist of: piston, displacer, regenerator, temperatures of thermal reservoirs and respective heat exchangers, expansion and compression volumes, and specification of working fluid and charge pressure. The schematic description allows specification of myriad details for each of the modules, as well as operating frequency, piston and displacer amplitude and phase, charge pressure, piston and displacer geometry, dead volumes, heat exchangers and regenerator design details. An illustrative design example is detailed in Section III. II. STIRLING-CYCLE CONVERTOR TECHNOLOGY Among the variety of heat-engine alternatives, Stirling engines promise attractive efficiency, simplicity and reliability. Stirling convertors have been under development and testing at NASA’s Glenn Research Lab since 1983 for intended use in space power systems as described in [6]. An overview of ongoing work is given in [7]. Using radioisotope-based General Purpose Heat Sources, high-temperature heat can be converted into AC electric power. A Stirling converter can have only two moving parts—a displacer and a piston (where an alternator’s armature is coupled to the piston). A relatively modern variation is the thermo-acoustic Stirling convertor, in which a moving displacer is eliminated, reducing the system to a single moving part. An analysis of a high-efficiency, longlife (>100,000 hour) design for NASA space-missions is described in detail in [3] and [4]. Another variation is to add a second piston, which oscillates in opposition to the first piston, sharing a common compression space and resulting in near net cancellation of imbalance forces from the working piston oscillations. Another feature explored at NASA has been the use of gas bearings with carefully tapered features on the piston and displacer. This technology has been demonstrated and is subject to ongoing testing for space-mission applications with life-requirements in excess of 60,000 hours [15]. The analysis of [15] further concludes that all of the technologies necessary for long-lived Stirling conversion-systems exist today, including bearings, materials resistant to temperature-variation induced fatigue, and linear alternators. By carefully choosing stiffnesses of mechanical springs on the piston(s) and displacer, the moving parts can be tuned to oscillate near a natural resonance. Such systems can be more Fig 1: Sage model of Stirling engine. Upper fig: top-level view, consisting of pressure source, piston, displacer and lumped thermodynamics of the heat engine. Lower fig: icons that represent and contain details within the heat engine. III. ILLUSTRATIVE DESIGN A specific Stirling-engine design for residential combined heat and power is described here. Due to net-metering regulations, the optimal electric-power production capability of such an engine would be approximately 1kW (although this value that may vary as consequence of numerous factors). A reasonable assumption for the temperature of the heat source is approximately 400C. Integration with a home heating system requires a relatively high-temperature for the convertor’s “cold” reservoir in order to appropriately heat water for storage or space-heating. Correspondingly, the present design assumes an exhaust temperature of 66C. Tables 1-4 summarize design details for a Stirling engine capable of producing 1kW of electric power from 3.4kW of heat input and providing 2.4kW of waste heat at 66C. Per Sage analysis, the system efficiency is 59.3% of the Carnot efficiency, resulting in converting 29.4% of the heat energy into electrical energy. This design assumes use of dual, opposed pistons that cyclically expand and compress a working fluid of Helium with a system charge of 0.7MPa (100 psi). Together, these pistons displace a volume of +/- 161cm^3. Accounting for gas volume within the system (including compression and expansion spaces, regenerator porosity, heat-exchanger passages and dead volume), the system pressure varies (nearly) sinusoidally from 0.61MPa to 0.79MPa at 60Hz. Length Heater Rejecter 25.000 cm Temperature 672.0 K Tube Inner Diameter 0.333 cm Number of Tubes 156 Length 10.0 cm Temperature 338.7 K Channel Width 0.106 cm Channel Height 0.500 cm Number of Channels 396 Fin Thickness 0.050 cm Table 4: Heat exchanger parameters Fig 2 shows the resulting predicted internal temperature variations at the two heat exchangers. Working fluid He Charge pressure 0.7 MPa (100 psi) Operating frequency 60Hz Compression space mean vol 265 cm3 Pmax/Pmin pressure ratio 1.288 Expansion space mean vol 274 cm3 Heat input at 400C (750F) 3.4kW Heat rejected at 65C (150F) 2.4kW Mechanical work extracted 1.0kW Table 1: System operating conditions Amplitude 1.000 cm Phase 0.000 rad Piston Face Area (each) 80.4 cm2 Spring Constant (each) 10.0 kN/m Amplitude 1.215 cm Phase 1.071 rad Outer Diameter 13.4 cm Displacer Drive Rod Diameter 2.77 cm Hot-facing Area 141.0 cm2 Cold-facing Area 135.0 cm2 Spring Constant 42.6 kN/m Table 2: Piston and displacer parameters Length 4.141 cm Inner Flow Diameter 13.500 cm Outer Flow Diameter 20.786 cm 196.213 cm2 Cross-Sectional Flow Area Porosity Fig2: Sage-predicted temperature variations in the Stirling-engine variable volume spaces. 0.900 Mesh Fiber Diameter 58.708 μm Table 3: Regenerator parameters Fig 3: Positions and velocities of displacer and pistons and mechanical power output from pistons at nominal operating condition. Fig 3 shows the nominal displacements and velocities of the pistons and displacer, as well as the net power required to maintain these profiles. The displacer requires no control force to produce its desired amplitude and phase. By choosing an appropriate mechanical spring constant and displacer dimensions (with respect to computed fluid-flow pressure drops through the regenerator), the displacer resonates at the desired amplitude, phase and frequency due to the timevarying internal pressures acting on opposite faces of the displacer. The pistons also have tuned mechanical springs, but substantial external forces are required to achieve the desired amplitudes and phases at 60Hz. These external forces are braking forces that extract power from the heat engine. These forces are to be produced by rectilinear, oscillating alternators integrated with the pistons (thus moving as a solid body). The alternators are to be designed to provide nominally 110VAC at 60Hz at the design operating conditions of the heat engine. Fig. 3 illustrates that the power extraction profile has the waveform of a sinusoid squared, corresponding to a sinusoidal braking force 180-deg out of phase with a sinusoidal velocity. The power extracted peaks at 1kW in each of the 2 alternators, which nets 1kW average power out, as desired. Details of the linear alternator design are not provided here. Ongoing work at the NASA GRC includes magnetic analysis [16], characterization of magnet deterioration and alternator efficiency [17], reliability [18], and evaluation of constructed systems with efficiencies of 85% [19]. Assuming a similar design for the present context, a linear alternator would exert a force proportional to current and a back-EMF proportional to velocity, where the constant of proportionality, K m , is identical for both effects. A prospective barrier for a practical, low-cost system is that it may be complex and expensive to construct the power electronics required to achieve the design operating conditions and to tie in the power production with line power. However, an advantage of designing a Stirling engine to operate at line frequency and voltage is that the system might require no tiein or control electronics at all (other than a relay to engage or disengage the system). This will be analyzed in the next section. IV. DYNAMIC ANALYSIS One of the challenges of home power generation is how to tie in to the grid. Rectifying power from AC to DC then inverting the DC to AC and synchronizing with line power requires expensive components. A prospect for Stirling convertors, however, is that the alternator can be designed to provide AC power at the nominal line voltage and frequency. This alone, however, would be inadequate for grid tie-in. The convertor’s alternator must be synchronized precisely with the line phasing and the alternator’s voltage must be only very slightly higher than the line voltage, which can vary by 10%. Surprisingly, the Stirling convertor is remarkably simple to interface to line voltage. By connecting the alternator to line voltage, the Stirling convertor will pull in to the proper amplitude and phase, and the desired power production will be delivered to the line. The line voltage will enforce that the alternator back-EMF be close to line voltage, since a significant difference between these voltages would induce large currents in the alternator, thus producing corrective forces that would cause the pistons to oscillate to nearly match the back-EMFs to line voltage. If no heat were provided to the engine, the alternators would behave as motors, consuming power and cooling the tip of the unit (thus acting as a cooler instead of a heater). Even under these dramatically offnominal operating conditions, the piston back-EMFs would be close to line voltage. Under nominal operating conditions, the pistons would tend to overstroke due to power input from the heat source, unless active braking were provided by the alternators. However, any tendency to overstroke (at the design frequency of 60Hz) would result in back-EMF waveforms with voltages in excess of line voltage. As a result, potentially significant current would flow in the alternators, providing power to the line and providing substantial braking forces to the pistons. The system would equilibrate with the back-EMF voltages being near to line voltage and in synchrony. At this operating point, the pistons would necessarily be oscillating at their design amplitude. Proving these assertions requires constructing a dynamic model from the steady-state thermodynamic analysis. To perform this conversion, an intermediate step is to find the sensitivity coefficients of piston and displacer positions and velocities on expansion-space and compression-space pressures. This can be done by analyzing steady-state operations at conditions that are perturbations from nominal, from which sensitivity coefficients for a pressure-waveform model can be derived. From the pressure-waveform dependencies, along with parameters of masses, piston face areas, mechanical spring constants, and alternator parameters (back-EMF constant, resistance and inductance), one can construct a dynamic model that relates piston and displacer accelerations to velocities, positions, line voltage and alternator currents. The process to derive such a model is not detailed here, but techniques to do so are presented in [13] and [14]. For a dynamic analysis, a state vector consists of: the positions of both pistons and the displacer, the velocities of all three of these components, and the currents in the two alternators. This 8-D state vector evolves according to a linearized state-space model of the system dynamics (related through the pressure-wave influence coefficients) as excited by a sinusoidal (line) voltage. In the present analysis, the two alternators are excited in parallel, and each alternator is designed for 110V at 60Hz under nominal operating conditions of the convertor. Fig. 4 shows a start-up transient for the dynamic model of the example Stirling-convertor design. Initially, all three moving parts (dual opposed pistons and displacer) are at their rest positions (center of stroke, enforced by mechanical springs). Sinusoidal line voltage is suddenly applied to the two alternators (emulating enabling the system with a relay). The two pistons pull into synchronism, reaching their design amplitudes and phasing almost immediately. The displacer takes several cycles (about 70ms) to reach its design amplitude. During the transient, none of the components overstroke or behave chaotically. If the line voltage varies, the pistons will vary their amplitudes proportionally. Thus a 10% variation in line voltage will have no disproportionate influence on the system operating conditions. integrated with line power without the need for tie-in electronics. This feature contributes to the economic viability of a domestic CHP system. To make Stirling convertors practical for residential use, the installed system cost must be low. Variables such as energy tax credits, availability of low-interest home-equity loans and electricity and natural gas rates all affect an economic decision, but it is nonetheless clear that a viable unit must be manufacturable at relatively low cost. A prospect to be explored is a thermo-acoustic design. This technology allows for elimination of one of the moving parts (the displacer), replacing it with a tuned cavity and inertance tubes. While the analysis and design of such systems can be complex, the result can be potentially cheap to manufacture. In continuing work, we plan to fabricate and test a Stirlingconvertor design based on the analysis presented here. VI. REFERENCES [1] [2] [3] Fig 4: Simulation of start-up transient. Pistons and displacer start from rest at zero displacement. Line power is applied, and all three components pull into synchronism with the desired phasing and amplitudes. [4] It is also irrelevant what the line-voltage phase is upon start-up (engagement of the alternator coils). The Stirling convertor is surprisingly robust. It would behave like the equivalent of a negative appliance—plugging it in would consume negative power. To halt the convertor, it is only necessary to short the alternator coils. This would result in dynamic braking, bringing the pistons (and, consequently) the displacer to rest. [5] V. CONCLUSIONS AND FUTURE WORK [9] Integrating a heat engine with a residential furnace can accomplish more efficient utilization of finite fuel resources. This is apparent from cost savings of electricity that can be produced from a combined heat and power system. Due to the nonlinear constraint of net-metering regulations, however, it is not always economically advantageous for the consumer to produce any more electric energy than they would consume each month. A consequence of this constraint is that the efficiency of a heat engine that produces electricity is almost irrelevant, provided the waste heat is desired for space heating. In testing at NASA, it has been demonstrated that Stirling converters can be built to be extremely reliable, requiring no maintenance for decades of operation. Another attractive feature, evaluated in simulations presented here, is that a Stirling convertor designed to operate at line voltage and frequency requires no automatic control system and can be [6] [7] [8] [10] [11] [12] [13] [14] Database of State Incentives for Renewables and Efficiency, “Rules, Regulations & Policies for Renewable Energy,” Accessed Feb. 2011. http://www.dsireusa.org/summarytables/rrpre.cfm. J. Barnes, and L. Varnado, “The Intersection of Net-Metering and Retail Choice: An Overview of Policy, Practice, and Issues,” Interstate Renewable Energy Council, Latham, NY, Dec. 2010. http://irecusa.org/2010/12/irec-releases-comprehensive-report-on-netmetering-in-competitive-electricity-markets/ S. Backhaus and G. W. Swift, “A thermoacoustic-Stirling heat engine: Detailed study,” The Journal of the Acoustical Society of America, vol. 107, issue 6, pp. 3148-3166, 2000. M. Petach, E. Tward, and S. Backhaus, “Design of a High Efficiency Power Source (HEPS) Based on Thermoacoustic Technology,” Final report, NASA contract no. NAS3-01103, CDRL 3f, 2004. http://www.lanl.gov/projects/thermoacoustics/Pubs/HEPSFinalDraftU.p df D. Gedeon, “Sage User’s Guide,” 7th Edition, May 2010. http://sageofathens.com/Documents/documents.php. J. Slaby, “Overview of free-piston Stirling technology at the NASA Lewis Research Center,” prepared for Twenty-third Automotive Technology Development Contractors Coordination Meeting, NASA/TM-87156, Oct. 1985. J.G. Schreiber and L.G. Thieme, “Overview of NASA GRC Stirling Technology Development,” in Proceedings of the 1st International Energy Conversion Engineering Conference, Portsmouth, VA, 2003. J. Schulte, “Feasibility study into micro combined heat and power,” in Proceedings of the 31st Intersociety Energy Conversion Engineering Conference and Exhibit, pp. 1314-1319, 1996. W. Dong, M. Lucentini, and V. Nasp, “The potential market analysis of a small cogeneration system based on Stirling cycle,” in Proceedings of the 35th Intersociety Energy Conversion Engineering Conference and Exhibit, pp.719-722, 2002. D. Six, G. Vekemans, and A. Dexters, “Market opportunities for microCHP in Flanders (Belgium)” in Proceedings of the 6th International Conference on the European Energy Market, pp.1-6, 2009. J.M. Pearce, B.A.T. Al Zahawi, and R. Shuttleworth, “Electricity generation in the home: modelling of single-house domestic combined heat and power,” in IEE Proceedings- Science, Measurement and Technology, pp.197-203, 2002. “Small scale CHP: Cogeneration of Heat and Electricity,” Nottingham Energy Partnership, Nottingham NG1 6BJ, UK, Accessed Feb. 2011. http://www.nottenergy.com/renewable-energy/small-scale-combinedheat-and-power-chp. J.S. Rauch, M.D. Kankam, W. Santiago, and F.J. Madi, “A Free-Piston Stirling Engine/Linear Alternator Controls and Load Interaction Test Facility,” in Proceedings of the 27th Intersociety Energy Conversion Engineering Conference, NASA/TM-105825, 1992. T.F. Regan and E.J. Lewandowski, “Stirling System Modeling for Linear Dynamics Analysis,” Proceedings of the Third International Energy Conversion Engineering Conference, San Francisco, CA, 2005. [15] J.G. Schreiber, and NASA Glenn Research Center, “Assessment of the free-piston Stirling convertor as a long life power convertor for space,” in 35th Intersociety Energy Conversion Engineering Conference and Exhibit, pp.1239-1247, 2000. [16] S.M. Geng, G.E. Schwarze, J.M. Niedra, and T.F. Regan, “A 3-D Magnetic Analysis of Stirling Convertor Linear Alternator Under Load,” in Proceedings of the 36th Intersociety Energy Conversion Engineering Conference, pp.441-445, 2001. [17] S.M. Geng, J.M. Niedra, and G.E. Schwarze, “Overview of NASA Magnet and Linear Alternator Research Efforts,” in Proceedings of Space Technology and Applications International Forum, pp.666-673, 2005. [18] A. Shah, I. Korovaichuk, S.M. Geng, and J.G. Schreiber, “Reliability of Radioisotope Stirling Convertor Linear Alternator,” in Proceedings of Space Technology and Applications International Forum, pp.641-647, 2005. [19] J.G. Schreiber, and L.G. Thieme, “Final Results for the GRC Supporting Technology Development Project for the 110-Watt Stirling Radioisotope Generator (SRG110),” in Space Technology and Applications International Forum, pp.325-338, 2007. VII. BIOGRAPHIES Wyatt Newman, Ph.D., P.E., is a professor in the EECS Dept at CWRU with research in mechatronic systems. He has degrees in engineering science (Harvard College), electrical engineering (Columbia U.) and mechanical engineering (M.I.T.). David Buckmaster is a graduate student in the EMAE Dept at CWRU. He has degrees in Computer, Systems/Control, and Electrical Engineering.