Working Fluid Control in Non-Aqueous Vapour Power Systems

This invention is concerned with improving the performance of vapour power plant. The invention is particularly, but not exclusively, concerned with closed- circuit vapour power generating systems operating on a Rankine cycle using nonaqueous working fluids, such as organic working fluids in place of steam. Such systems are primarily designed for the recovery of heat from a variety of non- fossil fuel heat sources such as geothermal brines, internal combustion engine exhaust and cooling jacket heat and industrial waste and process heat, where maximum working fluid temperatures are rarely much in excess of 15O ⁰ C, although they also have applications for higher temperatures. Adopting the refrigerant classification system, working fluids commonly used or proposed for these systems are R124 (Chlorotetrafluorethane), R134a (Tetrafluoroethane) or R245fa (1,1,1,3,3-Pentafluoropropane), or light hydrocarbons such as isoButane, n-Butane, isoPentane and n-Pentane.

As discussed in more detail below, the typical Organic Rankine Cycle includes a turbine expander and the working fluid entering the turbine must be dry.

The invention in a first aspect provides a vapour power generating system for generating power by using heat from a source of heat, comprising a closed circuit for a non-aqueous working fluid, the system including a boiler for heating the fluid under pressure with heat from the source to generate vapour, a positive displacement expander for expanding the vapour to generate power, a condenser for condensing the outlet fluid from the expander and feed pump means for returning condensed fluid from the condenser to the boiler wherein the line

connecting the boiler to the expander includes a sensor or sensors for monitoring the dryness fraction of working fluid leaving the boiler, and the circuit includes a controller coupled to the sensor or sensors for controlling the dryness fraction of working fluid leaving the boiler.

Preferably the working fluid leaves the boiler as a wet vapour.

A potential advantage of using a positive displacement expander is that it will operate in an Organic Rankine cycle, admitting wet working fluid, with efficiencies nearly equal to those of a turbine expanding dry vapour, since the presence of liquid seals clearances in the expander and enhances the lubrication of the meshing screw rotors or corresponding rotating parts in other types of positive displacement machine.

Figure 10 shows a typical Rankine Cycle PQRSSTP in which the working fluid is expanded as dry vapour in a turbine with a superheat indicated by SS' which must be removed by heat exchange before condensation can commence.

As can be seen, the effect of increasing the wetness of the working fluid at the expander inlet by means of the cycle PQ'R'STP is to reduce or eliminate the superheat at the expander exit (S') and, if required, even permit the working fluid to leave the expander as wet vapour. Since less heat is required per unit mass flow to only partially evaporate the working fluid, given the same supply of heat, as the wetness at the expander inlet increases, relatively more heat can be supplied to the feed heating process. Thus the working fluid can attain a higher evaporating temperature at Q' and pressure while its mass flow rate is increased.

Also, reducing the need for desuperheat after expansion tends to increase the specific enthalpy drop in the expander (expansion work per unit mass flow). These effects tend to increase the power output. However, in the absence of superheat at the end of expansion, there is a more general trend for the specific enthalpy drop in expansion to decrease as the initial dryness fraction of the vapour decreases. These conflicting trends generally lead to an optimum inlet dryness fraction of the order of 0.75-0.95, depending on the choice of working fluid, the inlet pressure and the condensing temperature. This inlet value corresponds closely to the vapour leaving the expander in the nearly dry saturated condition. The power output and, hence the efficiency of the cycle are then of the order of 3 -10% more than when the fluid enters the expander as dry vapour.

Since desuperheating of the working fluid needs a larger heat transfer surface than condensing and since increased power output leads to reduced heat rejection, the effect of increasing the expander inlet wetness can lead not only to increased plant output but also to reduced heat exchanger surface area and hence lower capital cost per unit power output.

Preferably the liquid phase of the working fluid contains a lubricant for the expander which lubricant is soluble or miscible in the liquid phase.

The admission of wet vapour to the expander can thus also be used to improve the Organic Rankine Cycle (ORC system) by simplifying and reducing the cost of expander lubrication by dissolving or otherwise dispersing up to approximately

5% lubricant by mass in the working fluid inventory. The lubricant is then

transported by the working fluid in the liquid phase through the boiler to enter the expander, where it will concentrate to lubricate the rotors as the liquid working fluid evaporates during the expansion process. Also, some of the lubricant- enriched, pressurised working fluid leaving the feed pump, prior to entering the boiler, can be distributed along a bearing supply path to one or more bearings where frictional heating will evaporate it off to leave sufficient lubricant for lubrication. One possible arrangement for this is shown in Figure 8.

It is common practice, in vapour power plant, to divide the boiler into two sections. In the first, known as the feed heater, the liquid is heated up to its boiling point. In the second, known as the evaporator, further heating causes some liquid to evaporate. At the evaporator exit, the vapour so formed is separated from the liquid and passes to the expander, while the residual liquid is returned to the evaporator inlet. Separation usually takes place in an external vessel, as shown in Fig 1 , but it can also be within the main shell of the boiler casing. However the separation is achieved, measurement of the liquid level in the separator can supply a convenient signal for controlling the flow from the feed pump when the demand for power changes. Thus, if the liquid level rises, the flow from the pump needs to be reduced, while if it falls, the pump flow needs to be increased. This can be achieved by using the signal obtained from the changed liquid level either to open or close a bypass valve that recirculates some of the liquid leaving the feed pump back to the pump inlet, or to vary the feed pump speed.

If the fluid is to enter the expander as a two-phase fluid, separation of the vapour from the liquid leaving the boiler implies that liquid must be drawn from some

other point in the system and injected with the vapour. There are many problems associated with such an arrangement. The two most important are:

i) It is very difficult to control the liquid injection rate to maintain the required dryness fraction over the entire operating range of the system.

ii) In the case where lubricant is added to the working fluid, separating the vapour from the liquid phase means that the fluid entering the expander is likely to contain none, because the lubricant will remain within the liquid component of the working fluid and thus then tend to gather and recirculate within the evaporator section of the boiler.

Since the injected liquid needed to obtain the required dryness fraction at the expander inlet is only a small fraction of the total mass flow, normally being of the order of 5-10% of the total, the lubricant transported in it will be only a tiny fraction of the total mass flow and will be insufficient to maintain the desired rate of its circulation around the system. Thus both the rotors and the bearings will receive insufficient oil for adequate lubrication.

Preferably the boiler is a single pass boiler, heater or system. A single pass system, is one in which the working fluid enters at one end as cold liquid and leaves as vapour with the desired degree of wetness at the other end, without any liquid recirculation. This would then eliminate the need for two separate heat exchangers and a separator. Also, if lubricant is added to the working fluid, the liquid content of the wet fluid leaving the boiler would contain the entire lubricant

content required and ensure its circulation through the expander, condenser and feed pump, round the system. Sufficient lubricant would then be distributed to all parts of the system to permit adequate lubrication.

Other preferred features of the invention in its first aspect are set out in dependent claims 3 to 11. The invention in a second aspect provides a method as set out in claim 14. Preferred features in the second aspect of the invention are set out in dependent claim 15.

As discussed in more detail below, embodiments of this invention are concerned with a means of flow control that enables the feed pump to supply the correct mass flow of working fluid to the boiler so that by its passage through the boiler in a single pass, the working fluid will enter the expander with the required dryness fraction over the entire operating range of the power generating system.

The invention will now be further described by way of example with reference to the attached figures in which:

Figure 1 is a circuit diagram of a typical Organic Rankine cycle vapour power generating system;

Figure 2 is a temperature entropy diagram illustrating a typical Rankine cycle vapour power plant;

Figure 3 is a circuit diagram of a vapour power generating system according to the invention; Figure 4 is a sectional view through the boiler exit sensing arrangement of figure

3;

Figure 5 is a circuit diagram of an alternative embodiment of the invention using an alternative boiler exit sensing arrangement;

Figure 6 is a sectional view through a boiler exit sensing arrangement suitable for use in the vapour power generating system of figure 5; Figure 7 is a sectional view through an alternative boiler exit sensing arrangement suitable for use in the vapour power generating system of figure 5; Figures 8 and 9 are circuit diagrams of further alternative embodiments of the invention; and

Figure 10 is a temperature-entropy diagram illustrating the effect of varying the expander inlet dryness fraction with the systems of figures 3 to 9.

The known circuit of Figure 1 comprises a heat exchanger assembly 1 for heating the working fluid in counterflow heat exchange with a hot liquid such as geothermal brine or waste from an industrial source. Typically the working fluid temperature of an Organic Rankine Cycle is up to about 15O ⁰ C. It is however, possible to have higher working fluid temperatures and Organic Rankine Cycles with working fluid temperatures up to 300 ⁰ C are known.

The heat exchanger assembly 1 defines a path 2 for the hot fluid from the source, the path 2 extending from an inlet 3 to an outlet 4. The assembly also defines a path, extending in counterflow heat exchange with the path 2, through a heater section 5, for heating liquid working fluid, and an evaporator section 6 for evaporating at least some of the working fluid.

A line 7 leads from the outlet of the evaporator 6 to a separator 8, at a higher level than the heater section 5, for separating the vapour component of the evaporator output from the liquid component. Line 9 serves to return the hot

liquid component to the junction 11 between the heater and evaporator sections 5 and 6.

A line 12 connects the vapour output of the separator 8 to the inlet 13 of a turbine 14 for expanding the vapour to a lower pressure and thereby generating power to drive an external load such as an electrical generator G.

A line 15 leads from the exhaust outlet 16 of the expander to a condenser 17 for condensing the expanded vapour in heat exchange with a cooling fluid flowing through a circuit 18.

A line 19 connects the liquid outlet of the condenser to a feed pump F for returning the liquid to the heater under pressure through a line 20.

The liquid level in the separator 8 is measured and the measurement is used to supply a signal to the feed pump F. The liquid level in the separator 8 therefore controls the pump F as demand for power changes. Thus, if the liquid level rises, the flow from the pump needs to be reduced, while if it falls, the pump flow needs to be increased. This can be achieved by using the signal obtained from the changed liquid level to open or close a bypass value that recirculates some of the liquid leaving the feed pump back to the pump, or to vary the feed pump speed.

As can be seen from the temperature-entropy diagram PQRSSTP shown in Fig 2, a common characteristic of the organic working fluids in the typical Organic Rankine Cycle (ORC) of figure 1 is that as expansion proceeds (R to S), working fluid entering the expander, initially as dry vapour, leaves the turbine expander with some degree of superheat.

The need for the working fluid to be free of liquid at the expander inlet 13 is an essential feature of the known ORC system with a turbine expander. The working fluid entering the turbine 14 must be dry, in order to maintain a high turbine efficiency and to prevent blade erosion by any entrained mist of droplets if the fluid entering the expander were wet.

The present invention results from the realisation that if the expander is of the positive displacement type, such as a plural screw machine, it is possible to obtain identical or even improved efficiencies if the working fluid enters the expander with entrained liquid without damaging the expander. In preferred embodiments of the present invention (see, for example, figures 3, 5 and 8), the turbine expander in a typical Organic Rankine Cycle system is replaced with a positive displacement expander (e.g. a plural-screw expander) and the dryness fraction of vapour leaving the boiler and entering the expander is controlled to have a low but significant liquid content.

It has been shown that twin-screw expanders are potentially as efficient as turbines, at lower power outputs (e.g. up to approximately 3MW), and can be manufactured cheaply, by modifying standard, mass produced oil flooded screw compressors and running them in the reverse direction of rotation. A suitable twin-screw expander is described in WO 2005/031123.

The Organic Rankine Cycle circuit of figure 3 includes a single pass boiler 21 for heating the working fluid in counterflow heat exchange to a temperature up to about 150 ⁰ C, with a hot liquid such as a geothermal brine, waste heat from an

industrial source or heat from the cooling system and exhaust of an internal combustion engine.

A line 22 leads from the outlet of the single pass boiler 21 to the inlet of a twin- screw expander 23. A working fluid sensing or monitoring arrangement 24 monitors the working fluid as it exits the boiler. The sensing or monitoring arrangement 24 (see figures 3 and 4) includes a pressure transducer 25 for determining the pressure of the working fluid 22 in the line as it leaves the boiler 21 , and a sampling or tapping off tube 26 for drawing off a small amount of the working fluid leaving the boiler to determine the working fluid's specific enthalpy as it leaves the boiler. The fluid drawn off for this determination is returned via a line 27 to the main circuit in the line 15 connecting the expander outlet 16 to the condenser 17.

As shown in figure 4, a small amount of the working fluid is tapped off from the pipe or line 22 connecting the boiler 21 to the expander inlet, through a tube 28 having its inlet facing upstream to the flow of working fluid in the line 22. The tube 28 contains a restrictor 29 which throttles the tapped off fluid. The throttling process occurs within the tube 28 while it is still within the main expander inlet pipe 22 so that any heat transfer to the throttled fluid will tend to raise the specific enthalpy of the drawn off fluid. However, calculations have shown this heat transfer to be negligible. The throttled fluid then enters a thermally insulated plenum chamber 30, where the pressure and the temperature of the throttled fluid are measured by pressure and temperature transducers 31 and 32 respectively.

The vapour thus obtained is then returned to the condenser inlet pipe 15, where it mixes with and is condensed with the fluid leaving the expander 23. The assumption made is that since the working fluid leaving the boiler 21 should never leave the expander 23 with more than minimal wetness, after work has been extracted from it, the throttled vapour drawn off from the exit of the boiler 21 and bypassing the expander 23, must always be superheated, since it has had no loss of enthalpy as a result of the expansion. The pressure and temperature measured in the plenum chamber 30 will then constitute two independent properties from which the specific enthalpy h _f v of the throttled vapour can be calculated in a microprocessor 33 in communication with the transducers 31, 32 using known thermodynamic property subroutines (see below). Assuming no heat loss, this enthalpy value will then be the same as that of the wet vapour leaving the boiler.

The value of the pressure, measured in the main delivery pipe 22 by the pressure transducer 25 close to the point where the throttled fluid is tapped off, is sufficient an input to enable the values of saturated liquid h _f and saturated vapour h _g specific enthalpy to be calculated by the microprocessor 33 with the aid of known thermodynamic property subroutines (see below).

There are a number of known sets of thermodynamic property subroutines that estimate the thermodynamic properties of a working fluid with high accuracy, given any two independent properties. The subroutines can also derive the thermodynamic properties of saturated liquid and saturated vapour of a fluid, given the pressure only. One such set, which is now the most widely used, has been prepared by the United States National Institute of Standards and is known

by the acronym NIST, the latest version, at the time of writing, being NIST7. This can be used for a large range of fluids, including nearly all those used or considered for power generation and refrigeration systems, and includes both pure fluids and their mixtures.

For any given working fluid, a succession of cycle analyses (see below) must be performed for the range of boiler inlet pressures and condensing pressures over which a designated power plant is required to operate and from each of these, the desired dryness fraction X _d of the fluid leaving the boiler is determined. The results can be summarised in the form of a function that expresses the required boiler exit dryness fraction, X _d , in terms of the boiler inlet pressure and condensing pressure. This function is then input to the boiler control microprocessor 33.

Using a set of subroutines such as those of NIST, described above, the microprocessor 33 derives the values of the specific enthalpy of saturated liquid, h _fi and saturated vapour, h _g , at the boiler exit pressure measured by transducer 25. These are then combined with the value of desired dryness fraction derived from the boiler exit and condenser pressure function to estimate the desired boiler exit specific enthalpy h _d from the well known thermodynamic relationship h _d = (1-X _d )-h _f +X _d h _g . This is the target function for the control system.

Using the same set of subroutines (e.g. the NIST subroutines), the value of specific enthalpy h _t v derived from the pressure and temperature of the throttled vapour is also obtained. This is the measured input.

The difference between the target function h _d and the measured input h _tv then gives a signal to control flow to the boiler and hence the mass flow rate, which in turn alters the dryness fraction of the fluid leaving the boiler. Thus, should the measured specific enthalpy h _tv of the throttled fluid be too high, then the flow to the boiler is increased, either by increasing the pump speed or closing a feed pump bypass valve as is done by a normal boiler float control valve. Similarly, if the specific enthalpy htv of the throttled fluid is too low, then the flow to the boiler is decreased.

As is clear from the control process described above, it is necessary to establish the desired or target dryness fraction x _d (and hence the target or desired enthalpy h _d ). This can be done and a boiler control equation derived by a person skilled in the art as follows:

i) Given the boiler evaporating and condensing temperatures and pressures, he/she would carry out a thermodynamic cycle analysis to determine the cycle efficiency and power output per unit mass flow of working fluid circulating in the cycle. This would be based on the assumption of constant pressure heating in the boiler up to the saturated vapour condition, adiabatic expansion in the expander, constant pressure cooling in the condenser down to the saturated liquid condition and, finally adiabatic compression of the working fluid back to the boiler inlet condition. The NIST subroutines, or their equivalent, can be used to do this within the framework of a computer

program. Such procedures are well known to anyone with any interest in this area. ii) Given the source from which the boiler receives heat, either in the form of the total heat available or the mass flow of the fluid, its specific heat and its allowed temperature drop, this total heat available is divided by the heat transferred to the boiler per unit mass flow of working fluid, as derived in the cycle analysis. This gives the required mass flow of working fluid in the cycle and hence, given the cycle efficiency, as derived from i), the power output. iii) Processes i) and ii) are repeated, assuming that the working fluid leaves the boiler with different degrees of wetness (or to use thermodynamic terminology, varying dryness fraction). Either by comparison, if these analyses are done manually, or by use of a minimisation routine, if using a computer program, the dryness fraction is found that gives the maximum power output. This is the value for the required dryness fraction Xd at the boiler exit at the design point, iv) Procedures i) to iii) are then repeated at different boiler saturation conditions (pressure and temperature), corresponding to the range of part load conditions likely to be encountered. For each of these, the optimum dryness fraction X _d is determined. v) Procedures i) to iv) may have to be repeated for a range of condensing temperatures and pressures, corresponding to the fact that during the year, atmospheric or cooling water temperatures vary seasonally and these will affect the value of the temperature at which condensation takes place. To a lesser extent, condensing temperature also varies with the heat input at part load, because, as

the heat input to the boiler goes down, the required heat to be rejected to the condenser is reduced and hence the same flow of coolant will result in a lower temperature difference between the coolant and the working fluid. This is, however, a second order effect. vi) Using procedures i) to v) inclusive, a function can be obtained, that expresses optimum dryness fraction as a function of boiler and condenser pressure or temperature. vii) For each of a series of selected boiler exit and condensing conditions, covering the range of normal power plant operation, using the NIST routines, or their equivalent, the working fluid temperature is calculated that would result from throttling the working fluid from the defined boiler exit to the condensing pressure. The saturation temperature at the same condensing pressure can be obtained from the NIST subroutines, or their equivalent. If this value is subtracted from calculated temperature of the throttled fluid, then the difference is equal to the degrees of superheat required as a result of throttling, which would correspond to the required boiler exit dryness fraction. Since the range of temperatures and pressures over which the system will operate is relatively small, a simple linear function, such as that given below will estimate this value to an acceptable degree of accuracy and, hence can be used as a control function. To quantify this accuracy, in the case of the working fluid R124, over the range of operation for which this working fluid is likely to be required to operate, an error of 2.5 ⁰ C in the value of the function, thus derived, would only cause approximately a 1% error in the value of dryness fraction required. Hence, this is a very robust function which is fairly

insensitive to either measuring or estimation errors in maintaining the boiler at optimum conditions over its operating range.

There are a few caveats, which are not included in the discussion above of a method for deriving the boiler control equation. One of these is how one estimates how the heat input to the boiler varies as the boiler pressure changes. This depends, to a major extent on how the speed of the expander is controlled. Thus, if the expander operates at a constant speed, corresponding to, say 50Hz electrical generation, then this is determined by assuming that the volume flow rate to the expander inlet, approximately constant. However, if the expander has a variable speed control, then the boiler pressure will vary far less as the load changes. Nonetheless, the control function for dryness fraction still stands for both situations but could be made more accurate for the variable speed case because the range of pressures over which the boiler operates is more restricted. Here, it has been assumed that the expander operates at constant speed. This results in a far cheaper system but which is less efficient when operating at part load. Since, in the main, the heat sources are waste heat, part load power requirements are not so critical.

Another condition is what to do if the fluid leaves the boiler too wet to attain superheat conditions after throttling. In that case, the function will result in negative or zero value to the superheat function. Under these conditions, the feed pump by-pass valve would be fully opened or the pump speed controller reduced to a minimum value, such that the flow through the boiler would be so small that the available heat supply to it would be sufficient to restore the fluid to superheat conditions after throttling the fluid leaving it.

The relationship between saturation pressure and temperature in fluids is a highly non-linear function and, in most cases, is expressed, fairly closely by an inverse logarithmic function (the Cox-Antoine Equation). However, the linear function seems to work well enough here because, over a limited range, almost any two related properties can be expressed, one as a linear function of the other. Any of the following relations may therefore be equally valid.

i) Degrees of Superheat = Constant x Boiler Pressure - Constant x Condensing Temperature. ii) Degrees of Superheat = Constant x Boiler Temperature - Constant x

Condensing Temperature, iii) Degrees of Superheat = Constant x Boiler Pressure - Constant x

Condensing Pressure. iv) Degrees of Superheat = Constant x Boiler Temperature - Constant x

Condensing Pressure.

A second method of controlling the boiler exit (and expander entry) dryness fraction has its essential components shown in figures 5 and 6. Although it is broadly similar to the method of control of figures 3 and 4 in principle, it is simpler in that control can be achieved with the aid of a microchip 34 controller without the need for a microprocessor containing complex thermodynamic property subroutines.

The arrangement shown in figures 5 and 6 is similar to that of figures 3 and 4 in that it includes a pressure transducer 25 for determining the working fluid

pressure as it leaves the boiler, and a tube 26 for drawing or tapping off a small amount of the working fluid. The tube 26 again includes a throttle 29 and supplies the throttled (and now dry superheated) working fluid sample to a plenum 30. However, in this case the pressure transducer in the plenum chamber is not required. The system has a plenum temperature thermocouple or transducer 32 in the plenum chamber 30 and a first condensation temperature thermocouple 35 in the line 15 between the expander and condenser . The degrees of superheat of the throttled fluid leaving the plenum chamber are determined by measuring the difference between the temperature of the fluid entering the plenum chamber and that leaving the expander (and entering the condenser), which is normally just saturated. This is the control signal. The target function is obtained from a simple linear algebraic function that estimates what the degrees of superheat in the drawn off sample of fluid, after throttling should be, expressed in terms of the expander inlet pressure and the condensing temperature. The simple linear algebraic function is of the form

Target Signal = A+B x boiler pressure - C x Condensing Temperature

where A, B and C are constants determined by the working fluid used and the required range of boiler exit and condensing temperatures. The values of the target signal are derived by thermodynamic cycle analysis in a manner similar to that described above for the boiler control equation of figures 3 and 4.

The embodiments of figures 5 and 6 include two separate thermocouples 35, 36 measuring the condensation temperature. The first thermocouple 35 provides the signal to determine the temperature difference used to calculate the level of

superheat. The second thermocouple 36 produces a temperature signal inputted to the target function set out above.

At higher working fluid temperatures, even though the fluid enters the expander as wet vapour, the optimum performance is achieved with the working fluid leaving the expander as superheated vapour. In that case the vapour leaving the expander 23 is not saturated vapour. The degrees of superheat cannot then be directly obtained from the difference in temperature between the plenum chamber and line 15 between the expander 23 and condenser 17 using the arrangement of figures δ and β.

In those cases (see figure 7), a small heat exchanger 37 typically with a heat transfer capacity of less than 200 Watts, can be included in the line or pipe connecting the plenum chamber 30 to the line 15 between the expander and condenser. This heat exchanger 37 partially condenses the vapour leaving the plenum chamber.

An example of the arrangement of figures 5 and 7 applied to a specific case is given as follows.

Heat recovered from the waste heat obtained from the cooling water jacket and exhaust gases of a 250 kW gas engine is used to drive a small Wet vapour Organic Rankine Cycle (WORC) ^" system using a screw expander. The working fluid is R124, which leaves the boiler at 8O ⁰ C and condenses at 3O ⁰ C at the design point. Depending on the local conditions, the condensing temperature can vary between 2O ⁰ C and 35 ⁰ C, while at part load, the temperature and

pressure of the working fluid leaving the boiler are reduced in order to maintain approximately constant volume flow to the expander.

In this case, the expander has a 2" (50.8 mm) bore pipe connecting it to the boiler exit. The probe diameter, through which the fluid to be throttled is drawn off, has a bore of 5 mm and the restrictor within it has a bore of 2 mm.

At the design point, the mass flow of fluid to the expander is approximately 2.4 kg/s while the fluid drawn off through the probe will choke in the restrictor so that its mass flow rate is limited to approximately only 0.01 kg/s. Thus, less than 0.5% of the fluid flow is throttled to obtain the control signal and the effect on the system performance resulting from drawing it off is therefore negligible.

The heat input to raise 0.01 kg/s of wet vapour in the boiler is approximately 1.8 kW. However, assuming that the 5 mm probe is 50 mm in length and entirely contained in the 2" pipe, the heat transfer to the throttled fluid within it from the main hot vapour stream is only of the order of 4 Watts. Hence errors in measurement due to heat transfer are negligible.

The fluid emerges from the 5 mm diameter probe, into the plenum chamber at a velocity of approximately 20 m/s, which gives a good heat transfer coefficient to the thermocouple recording its temperature and hence a fast response to changes. It then leaves the plenum through a 12.5 mm bore pipe. The fluid velocity in the 12.5 mm bore pipe is then only of the order of 3.5 m/s and this leads to a pressure loss along the pipe of only 0.005 bar per metre of pipe length. Hence if this fluid is returned to the condenser inlet, the pressure loss between

the plenum chamber and the condensate can be ignored. In this case, the fluid leaving the expander and entering the condenser is just dry and saturated.

To give an idea of the size of heat exchanger required, if the vapour leaving the expander were superheated, it was estimated that in this case the heat to be extracted is only 220 Watts in order to partially condense vapour leaving the plenum chamber. This means that the heat exchanger is small and therefore cheap.

The target function for control of the boiler exit dryness function may then be obtained from the following relationship.

Degrees of Superheat = 11.6+1.6 x boiler pressure - 0.7 x Condensing Temperature

-Boiler pressure in bar

-Condensing temperature in deg C

If this is subtracted from the measured temperature difference between the fluid entering the plenum chamber and that of the main mass flow entering the condenser, the resulting signal can be used to increase the boiler feed rate when it is greater than zero and to reduce the boiler feed rate when it is negative.

Analyses have shown that the above function will enable the dryness function of the fluid leaving the boiler to be controlled over the entire operating range of the system, allowing for seasonal changes to the condensing temperature, with errors of less than 1% of the required value for optimum performance.

Co-pending patent applications GB 0526413.0 and GB 0511864.1 describe closed-circuit vapour power generating systems in which lubricant is soluble in, or miscible with, the working fluid. Such arrangements can be combined with the cycles of figures 3 to 7 described above. Figures 8 and 9 illustrate such systems in which the dryness fraction control is carried out using the microprocessor control arrangement of figures 3 and 4, or 5 to 7 respectively.

The systems shown figures 8 and 9 are similar to those of figures 3 and 4, and figures 5 to 7 respectively and reference is therefore made to the discussion above of figures 3 and 4, and figures 5 to 7. The additional features of the arrangements of figures 8 and 9 including expander bearing lubrication are that a line 38 runs from the line 20 to bearing housings 39, 40 containing bearings for the rotating elements of the expander 23. As discussed in co-pending application GB 0526413.0, the bearing housings 39, 40 provide sufficient space around the bearings for the oil content to be concentrated as the working liquid evaporates into the expander as a result of heat generated in the bearings. As oil leaves the bearings and flows into the expander, it is constantly replaced by further oil from the line 38. The oil leaves the expander outlet 16 with the vapour and dissolves into the liquid condensed in the condenser 17.