# Development of the adjustable proportion fluid mixture cycle.

One of the few innovations to emerge from the field of thermodynamics in recent years is the adjustable proportion fluid mixture cycle, of which the Kalina cycle is best known. Test results from a plant at the Rocketdyne facility in Canoga Park, Calif., show encouraging results.

There have been few genuinely new ideas in classical thermodynamics put forth since early in this century. One of them is the adjustable proportion fluid mixture (APFM) cycle, of which the Kalina cycle is the best-known example. Its creator, Dr. Alex Kalina, started with a question: since variable composition is an important aspect of a regenerative ammonia refrigeration cycle, might a similar principle be used to improve the efficiency of the Rankine Cycle in the conversion of heat to work?

An analysis of several possible cycles based on this principle shows that the additional degree of freedom provided by variable composition can indeed lead to such an improvement. To show economic feasibility and resolve operational questions inevitably associated with a new technology, a 3 MWe demonstration plant has been built at the Rocketdyne test facility in Canoga Park, Calif., see Figure 1. The facility was first operated in early December 1991 with encouraging results. A smaller, 160 kWe, APFM plant has been tested in Japan.

Two-phase Mixtures

Of Ammonia and Water

A two-phase (liquid and vapor) mixture of ammonia and water in equilibrium will tend to have a greater concentration of ammonia in the vapor than in the liquid because ammonia is more volatile. Figure 2 is a phase diagram for such a mixture at constant pressure. If a mixture of a specified concentration (expressed as mass fraction of ammonia) which is initially all vapor is gradually reduced in temperature, condensation will begin at the dew-point. This temperature is shown on the diagram by the dew point curve. At the other extreme, when an all-liquid mixture is gradually heated, the bubble point curve shows the temperature at which vaporization (boiling) will start.

Point I on Figure 2 represents an equilibrium mixture of liquid and vapor of some overall concentration, X, at a temperature, T, somewhere between the bubble point and dew point temperatures.

The ammonia concentration of the vapor fraction will be XG (point 2 on the dew point curve) and the ammonia concentration of the liquid fraction will be XL (point 3 on the bubble point curve). If T is increased while X is held constant, the compositions, XL and XG, of the liquid and vapor phases will shift. When liquid initially at composition X (point 4) is evaporated, the initial vapor is nearly pure ammonia (point 5). As the vaporization process is completed (point 6) the last drop of liquid is nearly pure water (point 7) and the composition of the vapor is equal to the overall composition.

During vaporization (still at constant pressure) the temperature of the mixture increases because of the changing liquid composition. If the heat source is, say, the exhaust of a gas turbine, the rising vapor temperature of the ammonia-water mixture can be better matched to the cooling combustion gas than pure water boiling at constant temperature. Since a low-concentration mixture is all liquid at a higher temperature than a high concentration mixture, condensation can be accomplished at a pressure above atmospheric pressure, eliminating the problem of air leaking into the system.

The extra degree of freedom provided by variable composition means that properties tables are essentially three-dimensional. Interpolation from tables is difficult and tedious, particularly if the known properties are other than the tabulated independent variables. Thus computer-generated properties are essential, even for preliminary manual cycle calculations.

(The author has generated a set of computerized properties tables based on the 1984-85 work of Dr. Yehia Ei-Sayed and Dr. Myron Tribus. A MS-Dos-based diskette, with documentation, will be furnished on request. Requests should be made to: Department of Mechanical Engineering; Villanova University, Villanova, PA 19085.)

A temperature-entropy diagram for a mixture that is 70 percent ammonia is shown in Figure 3. Note that constant pressure lines in the saturated mixture region are represented by curves rather than the horizontal lines characteristic of pure water substance.

Principles of Cycle Operation

Figure 4 shows a simplified version of the cycle initially developed for a student computer project which later served as the basis for a detailed study. In addition to the flow through the vapor generator (boiler, BLR) and the turbine (TBN), there are several other flow streams. These streams connect an assortment of heat exchangers and pumps - plus a component designated the separator (SEP) which splits an inlet ammonia/water stream into liquid and vapor fractions.

The topology of the system is shown more clearly in Figure 5, in which, except for the condensers, the heat exchangers have been omitted. The system can be divided into two subsystems: heat/power (HPSS) and distillation/condensation (DCSS). More complex versions of the HPSS are used in large systems and there are many possible arrangements of heat exchangers to transfer thermal energy from one part of the cycle to another. However, the flow path for the DCSS in most of the proposed cycles, including the Canoga Park Demonstration Plant, has the basic arrangement shown in Figure 5.

The working fluid in the boiler (BLR) and turbine (TBN) is approximately 70 percent ammonia. Sufficient weak liquid from the separator (SEP) is mixed with the turbine exit flow (MXR2) so that condensation takes place above atmospheric pressure. In practice, vapor is "absorbed" by liquid so that mixing and condensation take place simultaneously. For that reason the condenser (CNDI) is also referred to as an absorber. The nearly pure ammonia vapor from the separator is mixed (MXR1) with some of the condensate to obtain the desired working fluid concentration. The resulting two-phase mixture must be condensed (CND2) prior to entering the main feed pump. Flow rate of the fluid circulating between the separator and the low-pressure condenser is roughly two to five times the flow rate through the turbine, but the additional power required by the condensate pump (CDP) is very small.

The circled numbers on Figure 5 correspond to key state points on Figure 4. Specification of composition and mass flow rate at the turbine inlet (13), plus temperature, pressure, and composition at the separator inlet (1), is sufficient to determine compositions and flow rates throughout the DCSS.

Complete solution of the cycle requires that the First and Second Laws of Thermodynamics be satisfied, and the exact procedure is specific to the configuration. For the cycle shown in Figure 4, separator pressure is set by the saturated liquid exiting CND2. The pressure at CND2 depends on ammonia concentration and on the temperature which can be maintained by the counterflow condenser cooling water. Then, for a specified separator inlet temperature, there is a unique separator inlet composition and consequent flow rate for which the First Law is everywhere satisfied. Satisfaction of the Second Law requires careful attention to the presence of "pinch points" every place in the cycle where there might be a transition between liquid-vapor mixture and pure vapor or pure liquid.

Specifying properties at the separator inlet is convenient for calculation, but in practice, flow rates are the independent variables and the system is monitored by observing property values at selected locations.

The Canoga Park Facility

The Energy Technology and Engineering Center (ETEC) is a U.S. Department of Energy (DOE) facility located near Canoga Park, Calif., near Los Angeles. It is operated on behalf of the DOE by the Rocketdyne Division of Rockwell International and is probably best known as the facility where very large rocket engines have been tested.

The Sodium Component Test Installation (SCTI) at ETEC is, as its name implies, devoted to development of liquid-metal technology. It is gas-fired and the combustion product exhaust is almost ideally suited as the heat source for a prototype Kalina cycle "bottoming" power plant. The "almost" qualification is appropriate because provision of an auxiliary heat source proved to be too expensive and therefore the plant can be tested only when the SCTI is running.

The Kalina cycle demonstration plant project was initiated in October 1987. Four years later in the late fall 1991, construction and pre-startup checks were completed. On December 10, 1991, an initial test was run with the turbine bypassed and turbine exhaust conditions simulated. Results of the test have been summarized by Exergy Inc. (Hayward, Calif.), see Table 1. Further testing could not occur until restart of the SCTI in early June 1992.
```                     Rankine      KCS1D2   KCS6D2   KCS8D3
cycle
ABB 13E              2 pressure
Gas turbine            level
Bottoming cycle        74.00      82.03     88.35    90.62
output (MW)
Bottoming cycle        30.58      33.88     36.51    37.39
thermal efficiency
Bottoming cycle        65.19      73.14     78.73    80.38
2nd Law Efficiency
Improvement ratio       1.00       1.08      1.19     1.23
Kalina/Rankine
Combined cycle        216.17      224.20   230.52   232.79
output (MW)
Combined cycle         50.28       52.15    53.62    54.14
thermal efficiency
GE70001F             3 pressure   KCS1D2   KCS6D2   KCS8D3
Gas turbine            level
Bottoming cycle         78.73     90.19    97.91    99.92
output (MW)
Bottoming cycle         30.70     35.50    38.53    39.08
thermal efficiency
Bottoming cycle         63.96     73.67    79.88    80.68
2nd Law Efficiency
Improvement ratio        1.00      1.15     1.25     1.28
Kalina/Rankine
Combined cycle          237.12    248.17   255.89   257.90
output (MW)
Combined cycle           53.41     55.91    57.65    58.10
thermal efficiency
```

Figure 6 is a schematic diagram of the cycle which has been implemented at Canoga Park. Condensate flow is split to optimize heat recovery from.the turbine exhaust and separator liquid streams. The flow is then recombined so that, with respect to composition and flow rate, the arrangement of the fluid flow circuits is identical to that shown in Figure 5.

The components shown as RC2 and MXR3 on Figure 6 are implemented at State Point 33 inside the separator tower. Similarly, the high pressure and low pressure towers contain combinations of falling streams, coils of tubing, and spray nozzles which are the physical counterparts of many of the boxes in Figure 6. The turbine system can be bypassed, with heat input and pressure drop regulated so that the exit flow matches turbine exhaust conditions.

Practical Issues

Several independent analyses have shown the gains which are possible using an adjustable proportion mixture as the working fluid. Aside from a natural reluctance to forsake proven technologies, what are the potential obstacles to implementation?

Corrosion: Ammonia reacts with copper-based alloys so the components must be fabricated from iron and steel. At high temperatures, stainless steel must be used and the temperature at which one must switch to the more expensive alloys must be confirmed by operating experience.

Nitriding: Ammonia also reacts with steel under special circumstances. The nitriding process, in which steel is exposed to ammonia gas at temperatures of 500 [degrees] C to 540 [degrees] C, is used to give the steel a hard, wear-resistant coating. However, for nitriding to occur, the flow rate of ammonia must be kept very low to allow sufficient residence time for the chemical reaction to take place. The process should be self-limiting. Samples tested at very low flow rates showed some nitriding, but at flow rates far less than those in the system, the nitride disappeared. Nevertheless, a test section, which can be isolated from the rest of the system for periodic removal and examination, has been included in the Canoga Park facility.

Decomposition: At sufficiently high temperature, ammonia will decompose. The rate of decomposition during operation affects the need for makeup and the need for removal of the noncondensible gases which are the products of decomposition. The presence of water vapor tends to suppress the reaction and, as with nitriding, reaction rate is crucial. Samples of vapor taken from the low-pressure condenser at the conclusion of the December test showed only 25 ppm free hydrogen.

Control: As previously mentioned, the system is controlled primarily by controlling flow rates. These are set by matching measured properties at various points in the cycle with calculated properties. Transient response is difficult to estimate analytically but preliminary results are encouraging.

Heat Transfer Surface: At many places in the cycle, heat transfer must take place between two streams which are both two-phase mixtures. There is considerable uncertainty regarding the amount of surface needed to maintain a specified temperature drop between such streams. However the greatest uncertainty is in the smallest heat exchangers, which cost only a small percentage of the installation cost.

Ammonia Safety: Ammonia is toxic but its presence is easily detected, it is easily absorbed in water, and substantial experience with large refrigeration systems exists. With proper precautions this does not appear to be a significant limitation.

Exergetic Efficiency

Let heat duty be defined as thermal energy per unit mass of working fluid transferred from the combustion gas heat source as the gas cools and the working fluid is heated. Let [T.sub.o] be the temperature of the environment. Then, if temperature is replaced by exergetic temperature, defined using absolute temperatures as (T-[T.sub.o])/T, the area between the curves can be interpreted as exergy destruction and thus is a direct indication of effective utilization of available energy.

(At temperature T, the maximum increment of work for a heat interaction dQ is proportional to the Carnot efficiency at that temperature, where Carnot efficiency is the same as the exergetic temperature defined above,

[dW.sub.max] = ((T-[T.sub.o])/T) dQ.

The maximum work for given heat duty is then the integral of [dW.sub.max] over the temperature range of the heat interaction,

[W.sub.max] = ((T-[T.sub.o])/T) dQ.

T

The area between the curves of heat delivered and heat received is thus a direct measure of exergy destruction. See Figure 7, noting the shaded areas of the plots of exergetic temperature versus heat duty.)

A major advantage of adjustable proportion fluid mixture cycles is the flexibility with which energy can be transferred internally to maximize the exergetic (Second Law) efficiency of the cycle. To put it another way, the local temperature gap between the hot gas heat source and the working fluid can be kept small because of the extra design freedom made possible by variable composition and internal fluid circulation.

A series of increasingly sophisticated bottoming cycles were analyzed at Exergy, using two standard gas turbines, to show potential performance compared with multiple pressure steam bottoming plants. Table 1 shows calculated performance. The corresponding plots of temperature and exergetic temperature of heat source and working fluid as a function of heat duty are shown in Figure 7. Capital cost of the S6D2 (intermediate) system was estimated to be \$1020/kW, with the distillation/condensation subsystem accounting for about 20 percent of the total cost of the system.

Gas turbine bottoming is the most obvious application for APFM cycles but other possibilities range from geothermal to direct coal fired.

Key Test Results:

Canoga Park Demo Plant

In addition to the verification of the principles and operability of the Kalina cycles, the following important results were observed, according to Exergy Inc.: * The target difference in the ammonia/water concentration of the working fluid between the low-pressure condenser and the high-pressure condenser (components of the DCSS) was reached in less than one hour. Although the heat supply to the DCSS was only 60 percent of the design capacity, the DCSS distilled the full inventory of the working fluid. In normal operation and "warm" starts, this startup procedure will be even more rapid. This demonstrates that the Kalina cycle plant can be started quickly. * As a corollary to the above, the system is highly responsive. During operation of the DCSS, the composition in the high-pressure condenser was adjusted from 80 percent ammonia to 72 percent ammonia in less than six minutes. This confirms the ability of the Kalina cycle to change composition rapidly to adjust to changes in environmental or load conditions. * There was no perceptible decomposition of ammonia. During operation, the temperature differences between the condensate and cooling water in the CDSS did not increase, indicating no increase in the amount of noncondensible gases in the system. A later chemical analysis did not reveal any presence of hydrogen, which is a product of ammonia decomposition. On the basis of these tests and previous experimentation, it was concluded that if dissociation occurs at a higher temperature than reached during these tests, it will have negligible impact on the operation of the cycle.
COPYRIGHT 1992 American Society of Mechanical Engineers
No portion of this article can be reproduced without the express written permission from the copyright holder.

Title Annotation: Printer friendly Cite/link Email Feedback Advanced Energy Systems; includes related article on fuel efficiency Marston, Charles H.; Moran, Michael J. Mechanical Engineering-CIME Sep 1, 1992 2761 The open approach to FEA integration in the design process. Commercialization of fuel cell technology. Electric power plants Power plants