OVERVIEW OF LIQUID METAL MHD POWER SYSTEMS Edward S.

Pierson Department of Engineering, Purdue University Calumet, Hammond, IN 46323 USA [email protected] Abstract: The focus is on descriptions of the most-promising liquid metal MHD cycle concepts along with a summary of the relative advantages and disadvantages of liquid-metal MHD and of each concept. The objective is to provide the reader with an overview of the concepts and the background to choose a cycle for a given application and temperature range. 1. Introduction Magnetohydrodynamic (MHD) power systems combine the turbine and electrical generator of a conventional (steam or gas turbine) power plant into a single unit, thus eliminating a significant component and its associated temperature limitations. Liquid-metal MHD (LMMHD) power systems were developed to combine the advantages of a liquid metal (high electrical conductivity independent of temperature) as the electrically-conducting fluid with a gas or volatile liquid (different or the same as the electrically-conducting fluid) as the thermodynamic working fluid. This avoids the high temperatures and special processes used with plasma MHD to ensure adequate electrical conductivity. Basically, all LMMHD cycles consist of a heat source, a method of accelerating or pumping the electrically-conducting liquid by the thermodynamic working fluid, a (dc conduction or ac induction) MHD generator (with a pure liquid or a two-phase working fluid), a separator for the two fluid streams, methods (such as pumps) to close the two fluid loops, and a heat sink, see Fig. 1. The approach followed here is first to describe the initial LMMHD cycle, and then to use that as the basis for describing the advantages and limitations of the LMMHD concepts. This is followed by a summary of the two cycle variations that, to the author, appear most attractive. Other LMMHD concepts have been developed in France [1,2], Germany [3], the USSR [4], etc. The condensing cycles [5] are expected to have a lower efficiency [6,7] because of the losses associated with the condensing injector (where the liquid and vapor are mixed, and the vapor condensed to produce the high pressure necessary to push the liquid through the generator). For a more-detailed review of the LMMHD cycles, see Petrick and Branover[7]. 2. The first LMMHD cycle The first LMMHD system, proposed by Elliott in 1962 [8], see Fig. 1, uses lithium as the liquid metal and cesium as the vaporizable fluid in a high-temperature Rankine cycle chosen for space applications. In this system, the vapor accelerates the liquid to a high velocity. The liquid is then separated from the vapor, and the high-velocity liquid enters the generator where most of its kinetic energy is converted to electrical energy. The liquid uses the remaining kinetic energy to return (through the diffuser and heat source) to the mixer, the vapor is condensed and then pumped to the mixer. This is commonly referred to as the separator cycle because (ideally) only the liquid flows through the generator. This cycle has been extensively studied so that performance predictions should be accurate [7,9]. As part of the study, a 300 kW integrated system using NaK and nitrogen to simulate the anticipated high-temperature system in the parameter ranges of interest was built and tested. The ratio actual system efficiency to Carnot efficiency that appeared to be achievable was ~0.20 to 0.30 [7]. Limitations are the high velocity required from the nozzle (to provide the pressure

for the rest of the cycle), and the inability to obtain complete separation of the phases. Although the vapor carryover with the liquid is small by weight (a few percent), by volume it can be significant and impacts the performance of diffusers and the generator. Historically, Elliott proposed this cycle with a dc conduction generator. At the same time Jackson and Pierson [10] were investigating Figure 1. Separator cycle the pure-liquid induction generator without an application in mind. The two quickly teamed up. Unfortunately, the induction generator does not work well with a two-phase fluid as the conductivity is too low. 3. Features and liabilities Using this initial cycle as a model, the advantages of LMMHD include: The ability to tailor a system to any temperature range through the choice of the working fluids, thus offering the potential to serve many special needs at higher efficiencies than alternatives. Thus the concept was originally proposed for space where a high radiator temperature is essential for compact systems, and later modified for ground-based applications. For example, both gas cycle and vapor cycle versions are possible. The almost-constant-temperature expansion of the working fluid in accelerating or pumping the liquid, leading thermodynamically to a higher cycle efficiency. This corresponds to infinite reheat in a conventional Rankine cycle. The high electrical conductivity of the liquid metal (as compared to the plasmas used in other MHD systems). The disadvantage is the losses associated with pumping or accelerating the liquid metal. (The pumping process, which can occur inside the generator or in a separate nozzle, effectively replaces the turbine in a conventional power plant.) These losses may show up as pump power, increased pressure drop of the working fluid, etc. There are several types of losses unique to these two-phase cycles, as can be seen from examining Fig. 1. These are: Mixer losses, pressure drops and poor mixing. In this cycle vaporization of the liquid occurs partly or completely in the mixer. Slip losses in the nozzle or generator, where the vapor moves faster than the liquid. Poor heat transfer between the two phases, impacting the desired almost-constanttemperature expansion. Separation losses of two types, friction on the impact surface, and incomplete separation (some liquid goes with the vapor, some vapor stays as bubbles in the liquid). These are significant because of the impact on other components, such as extra heat loss in the reject heat exchanger and the next item. Diffuser losses in the liquid loop because of the vapor present. The separation process can be particularly difficult because the desired good mixing with small bubbles in a continuous matrix of liquid is particularly hard to separate. We like to think of generating a champagne foam that breaks down easily, rather than a beer foam.

The goal is to build on the advantages while minimizing the disadvantages or losses. Multiple cycles or concepts have been proposed in response to the many needs and in attempts to minimize the losses, as mentioned above. They differ in the choices of mixing, acceleration or pumping, and separation processes for the two fluids. The first were single-fluid systems, but their efficiency tends to be lower [6]. Later cycles made improvements based on careful study of the significant loss mechanisms in the earlier concepts [7]. Among these, The two-phase generator cycle, where the vapor/gas-liquid mixture is the working fluid in the generator, and The OMACON (optimized MHD conversion system) cycle, which uses natural circulation to minimize the losses associated with pumping the liquid metal appear to be the most attractive. With limited space, only these two concepts are described. 4. The two-phase generator cycle The two-phase generator cycle, Fig. 2, was developed by Petrick and Lee[11] after an extensive study of the previously-proposed cycle concepts to minimize the loss mechanisms. By mixing the two fluids together and then expanding the mixture through the generator, the necessity for high velocities, and in particular separating the two phases at high velocity, was avoided. (The down side is the two-phase mixture in the generator, but the separator cycle because of vapor carry-over also had two phases in the generator.) After the generator, the liquid can be returned to the mixer (through the heat source) by either An additional expansion through a two-phase nozzle so that the liquid has enough kinetic energy after the separator to use a diffuser to return to the mixer (solid lines in Fig. 2), or Separating the phases and using some of the electrical output from the generator to drive an EM Figure 2. Two-phase (electromagnetic) pump (dashed lines in Fig. 2). generator cycle Because the top and bottom pressures for the cycle are normally constrained, the former has more pressure across the generator, but uses some of the generators output to drive the EM pump; the latter has less output from the generator but no EM pump. The optimum version depends on the operating conditions and component efficiencies. Heat may be added to the gas or vaporizable fluid, or the liquid metal; generally better performance is obtained heating the liquid metal because of its excellent heat transfer properties. Also, the top temperature for the expansion will generally be higher (higher Carnot efficiency) than if just the thermodynamic fluid is heated. Following the separator, a gas or vapor turbine can be added; the impact on efficiency depends again on parameters and component efficiencies. A regenerator is generally necessary, even with a turbine, to minimize the rejected energy; this is a consequence of the almost-constant-temperature expansion. The regenerator has a significant impact on efficiency. This cycle exists in three versions: A gas or Brayton version using helium or nitrogen as the noncondensible gas and a liquid metal chosen by the operating temperature, as in Fig. 2. Actually this is an Ericsson cycle because of the almost-constant-temperature expansion of the gas, resulting in a significantly higher ideal cycle efficiency.

 A vapor or Rankine version using, for example, water and a compatible liquid metal such as tin. Thermodynamically this is an infinite reheat cycle. This differs from the cycle in Fig. 2 only in replacing the reject heat exchanges with a condenser, and ensuring that boiling occurs at some point. The almost-constant-temperature expansion means that the vapor leaving the generator or nozzle is highly superheated, excellent conditions for the entrance to a turbine. An open-cycle version using combustion gas as the thermodynamic fluid and liquid copper, discussed below. The versions have been studied for many applications, as summarized below for two solar and coal. Unfortunately, a complete system has never been built. 4.1 Solar Applications [12] LMMHD can be used at low temperatures (350-700 K) with solar collectors, as well as geothermal and waste heat sources, probably in small, dispersed applications. For example, with source and sink temperatures of 353 and 298 K (flat plate collectors), the Carnot efficiency is 0.156, and the calculated system efficiency including a solar collector efficiency of 0.6 is 0.043. If the liquid metal is used in the collectors, the top temperature would be higher (378 K) and the efficiency goes to 0.058. With low-concentration collectors, 400-700 K, significant improvements are possible because using the same liquid metal in the collectors with a directcontact boiler can result in a 25 % increase in the top temperature of the thermodynamic fluid with essentially the same collector efficiency. For temperatures above ~810 K, the Brayton version becomes attractive because of its efficiency, inherent simplicity, and low sensitivity to design power level (MHD is a volume effect). It also couples well to sodium-cooled solar collectors, which have better properties. At 1089 K the efficiency is calculated at ~0.5 for both the pure LMMHD (no gas or steam turbine) and the LMMHD-gas turbine cycles with helium and lithium as the fluids. As the temperature is increased, the efficiency increases and then starts to decrease. This decrease is caused by the carry-over of liquid metal vapor with the gas from the separator, causing the heat rejected to increase. At lower temperatures sodium is used, but above ~867 K lithium is required to minimize vapor carry-over. 4.2 Open-Cycle LMMHD [13] This novel concept, Fig. 3, uses combustion gas as the thermodynamic fluid and liquid copper. The fuel (coal) is burned with compressed air, and the hot gas is mixed with liquid copper. After the LMMHD generator (and nozzle), the combustion products are separated out (in a rotating separator to minimize loses and improve separation effectiveness). The (cleaned by the copper) gas then enters a conventional steam plant for removal of the remaining energy before exhausting to the atmosphere. To explore its technical and economic feasibility, a conceptual design study of the retrofit of an existing oil-fired steam plant with a new coal-fired LMMHD topping cycle was performed, and extensive parametric studies carried out to establish optimum parameter ranges for the retrofit cycle. A conceptual design was developed for the plant and components in sufficient detail that a cost estimate could be made by an architect/engineer firm. A significant feature of this concept is the ability to burn coal in an environmentally-acceptable fashion. If the combustion gas is maintained sub-stoichiometric until after separation from the copper, test-tube size experiments have demonstrated that the copper

will remove virtually all of the sulfur and nitrogen oxides from the combustion gas, and probably the slag (ash). The sulfur can then be easily and economically removed from the copper in a slag/sulfur removal system (with a flow rate ~0.005 of the total copper flow) as concentrated SO2 by simply bubbling air or oxygen through the copper. This results in significant cost savings; and in fact the copper clean-up is potentially economically viable on its own.

Figure 3. Open-cycle LMMHD system

Features of the retrofit, besides cleaning the combustion gas, are: Because the combustion gas leaving the separator is clean, the existing boiler can be used with, at most, minor modifications. With the copper, no stack gas cleanup or slag/ash removal from the boiler is needed. The LMMHD components are small , and can be fabricated and assembled in a factory with minimal on-site assembly. The combustion gas is used for preheating, eliminating trace heating. Start-up is simple. The original oil-fired plant produces a net plant output of 65.1 MWe at an efficiency of 0.289. Converting to coal or a coal gasifier by conventional means was shown in other studies to decrease the output power and efficiency. The LMMHD system at stoichiometric ratios of 0.8 before the copper mixer and 1.05 in the boiler, and a copper peak temperature of 1501 K, produces 119 MWe at an efficiency of 0.341 (increases of 83% and 18%, respectively), with higher efficiencies possible at higher stoichiometric ratios and copper temperatures. The total retrofit cost was calculated in 1980 as ~$104 million. For this cost, the net output power is approximately doubled, the efficiency is higher, and a low-operating-cost slag/sulfur removal system is provided. In all the cases studied, the LMMHD conversion option appeared economically advantageous relative to the other coal options or the existing system. 5. The OMACON Cycle OMACON stands for optimized MHD conversion system, developed by Petrick and Branover [7, 14]. The system uses natural circulation to eliminate the need for pumping the liquid metal. The basic OMACON system, Fig. 4, consists of two pipes (riser and downcomer) connected at the bottom by a crossover pipe and at the top by a gas (vapor)-liquid separator. Gas (or vapor) at pressure is injected into the base of the riser producing a two-phase media; the gas then expands nearly isothermally up the riser, providing the driving force (density differential) to push the compatible liquid metal around the circuit. The gas is separated at the top of the flow loop, and the liquid metal passes down through the MHD generator. The benefits of the OMACON system are: The high loss two-phase-flow generator liquid metal MHD system components -- twophase nozzle, separator, diffuser, are eliminated. The two-phase flow limitations on MHD generator performance are eliminated. The single-phase-flow MHD generator can achieve efficiencies of 80 to 85%, and can be designed from the data base that exists today.

 The single-phase MHD generator can use low-flux-density magnets. The system can be designed for low frictional losses - thus maximizing performance. The performance is independent of void fraction. The system can be operated as a Rankine or Brayton cycle, and readily adapted to various temperature ranges by the selection of the fluids. A limitation of the OMACON concept is the physical size required for practical pressure differences, and the liquid metal inventory. These can be minimized by using a heavy liquid, multiple stages, and/or a hybrid concept Figure 4. OMACON cycle that combines features of the OMACON system with those of other LMMHD systems, such as the nozzle-separator-diffuser [15]. Several models of the OMACON system have been built and tested [14], and extensive performance studies completed. Given the simplicity of the system and the available data base, performance predictions should be fairly accurate. 6. References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] Bidard, R.: Sterlini, M. J.: Etude des Milieux Biphases en Vue de Leur Utilisation dans des Dispositifs Magnetohydrodynamiques. Electricity from MHD, Proc. of a Symp., Salzburg, paper SM-107/75j (1966) Thibault, J. P.; Joussellin, F.; Alemany, A.; Dupas, A.: The Possible Use of a Metal-Gas MHD Energy Converter in Space. Acta Astronautica 10 (1983) 587-589. Radebold, R. D.; et al: Energy Conversion with Liquid Metal Working Fluids in the MHD-Staustrahl Rohr. Proc. Sixth Intl. Conference on Magnetohydrodynamic Electric Power Generation, Washington (1971) Aladiev, I. T.; Teplov, S. V.; Telmach, I. M.: Experimental Potassium MHD Installation of 1 kW. Proc. Fifth International Conference on Magnetohydrodynamic Electric Power Generation, Munich III (1975) Brown, G. A.; Lee, K. S.: A Liquid Metal MHD Power Generation Cycle Using a Condensing Ejector. Proc. Intl. Symposium on Magnetohydrodynamic Electrical Power Generation, Paris, paper 60 (1964). Grolmes, M. A.; Petrick, M.; Jerger, E. W.: Condensing Injector Experiments and Analysis of Performance with Supersonic Inlet Vapor. Proc. Intl .Symposium on Electricity from MHD, 2 Salzburg (1966). Petrick, M.; Branover, H.: Liquid Metal MHD Power Generation - Its Evolution and Status. Single-and Multi-Phase Flows in an Electromagnetic Field, Prog. in Astronautics and Aeronautics. 100, AIAA, NY, (1985) 37l-400. Elliott, D. G.: Two-Fluid Magnetohydrodynamic Cycle for Nuclear-Electric Power Conversion. ARS Journal. (1962) 924-928. Liquid Metal Magnetohydrodynamics Technology Transfer Study, Vol. II. Rept. 1200-59, Jet Propulsion Laboratory, Pasadena (1973). Jackson, W. D.; Pierson, E. S.: Operating Characteristics of MHD Induction Generators. I.E.E. Conference Report, Series No. 4. , (1962) 38-42. Petrick, M.; Lee, K. Y.: Performance Characteristics of a Liquid Metal MHD Generator. Proc. Intl. Symposium on Magnetohydrodynamic Electrical Power Generation, Paris, Vol. 4 (1964). Pierson, E. S.; Branover, H.; Fabris, G.; Reed, C. B.: Solar Powered Liquid-Metal MHD Power Systems. Mechanical Engineering. 102 (October 1980) 32-37. Pierson, E. S.; Herman, H.; Petrick, M.; Grammel, S. J.; Dubey, G.: Retrofit of Coal-Fired Open-Cycle Liquid-Metal MHD to Steam Power Plants. Proc. of the 16th Intersociety Energy Conversion Eng. Conf., Atlanta (1981) 1525-1530. Branover, H.; El-Boher, A.; Lesin, S.; Unger, Y.; Petrick, M.: Testing of OMACON-Type MHD Power Systems. Liquid Metal Flows: Magnetohydrodynamics and Applications, Prog. in Astronautics and Aeronautics. 111, AIAA, NY, (1988) 209-229. Petrick, M.; Pierson, E. S.; Berry, G. F.: Hybrid OMACON: An Optimum Liquid-Metal MHD System Configuration? Liquid Metal Flows: Magnetohydrodynamics and Applications, Prog. in Astronautics and Aeronautics. 111, AIAA, NY, (1988) 158-188.