WO2006061194A1

WO2006061194A1 - Purge management in fuel cell power plants

Info

Publication number: WO2006061194A1
Application number: PCT/EP2005/013071
Authority: WO
Inventors: Luc Rouveyre
Original assignee: Renault S.A.S.
Priority date: 2004-12-08
Filing date: 2005-12-06
Publication date: 2006-06-15
Also published as: EP1846970A1

Abstract

The present invention provides a fuel cell power plant 120 comprising a fuel cell stack 180 configured to be fed with a regulated supply of pure or quasi-pure hydrogen 140 through a pressure regulator device 260 and having a recirculation loop 200 that is configured for hydrogen recovery from downstream of the fuel cell stack 180. The recovered hydrogen Is fed into the hydrogen feed supply 220. The power plant 120 further comprising a purge device 260 that is located downstream of the fuel cell stack 180, the purge device 260 being under the control of a controller 300 and being configured to purge nitrogen gas from the power plant 120. The controller 160 includes an input configured to receive a signal indicative of an electrical load demanded of the fuel cell power plant and is configured to control the purge device in response to the load signal. The controller 300 is thus able to regulate the level of nitrogen at the entrance to, and/or exit of, an anode of the fuel cell stack and thereby substantially continuously maintain a concentration of nitrogen at said anode which is determined as a function of the electrical load demanded of said fuel cell power plant.

Description

PURGE MANAGEMENT IN FUEL CELL POWER PLANTS

Field of the Invention

The present invention relates to fuel cell power plants and in particular to purge management for fuel cell power plants.

BACKGROUND TO THE INVENTION

Fuel cells are often constructed from a stack of elementary cells, a schematic representation of such an elementary cell being illustrated in Figure 1. A typical elementary cell 10 comprises a pair of electrodes 4 that are separated from each other by an electrolytic membrane 5 and are separated from collectors of electric current 2 by a pair of joints 3. The cell 10 is then encased between a pair of bi-polar plates 1 , which give the cell 10 mechanical stability and permit the distribution of gas in the anodic chambers. In addition, the bi-polar plates assure the transfer of electrons from one cell to another, which gives them their bipolar denomination. The positive pole of one cell is electrically connected to the negative pole of the next cell. The reaction is exothermic and the heat given out may typically be evacuated by an internal serpentine cooling system that runs around some of these bi-polar plates. The plates may be made in treated aluminum or in graphite, which assures good electrical conduction between the cells.

The membrane 5 may be a proton exchange membrane (PEM) and substantially solid. In many fuel cells, the PEM is kept damp with water so as to improve its efficiency. A known way of keeping the PEM humid is by adding water to the reactant gases. In some installations, notably a vehicle, it can sometimes prove difficult to provide sufficient water for this humidification.

To alleviate problems of water supply, systems have been developed in which water is recovered within the fuel cell system. Examples of previously published proposals that deal with the issue of water recovery for humidification may be found in WO-03/043114, US-6541141 and US-5441821.

A general arrangement for water recovery in a fuel cell system 12 is now discussed with particular reference also to Figure 2. The fuel cell stack 18 delivers an electrical current which induces the consumption of hydrogen. The recirculation mode enables the supply of hydrogen to the anode of the fuel cell at a level above stoichiometric, even though for global balance the theoretical rate of hydrogen usage is in the order of 100%.

The basic system 12 comprises a hydrogen supply 14 that is metered 16 into the fuel cell stack 18. The hydrogen is available under a pressure that is higher than the pressure of the rest of the system 12 and is supplied into the fuel cell stack 18 by a regulating valve 16, usually involving pressure reduction. In the absence of such a regulating valve 16, pressure of the hydrogen at the anode would drop. The regulating valve 16 regulates the inlet pressure of the hydrogen automatically and rapidly so as to compensate for the consumption induced by the electrochemical reaction, typical response time being less than 0.1 to 0.5 seconds.

The fluids being released at the anode are redirected through a recirculation circuit 20 to the inlet of the fuel cell stack 18, entering the stream of inlet hydrogen at a point 22 that is downstream of the regulating valve 16. In the recirculation circuit 20, a recirculation device 24 is included that is configured to prevent stratification and to enable a homogenous mix of hydrogen, water and nitrogen, the precise composition of which may vary. This recirculation mode may be found usable up to a mixture containing about 20% hydrogen.

An electrically controlled purge valve 26 is preferably included so as to guard against the effects of entry by contaminants (e.g. water and nitrogen or natural gas) into the anodic chamber. The entry of these different species can introduce pollution and degrade the performance of the fuel cell stack 18. This is because, due to the chemical reaction, part of the water produced migrates from the cathode to the anode and stagnates there (so-called retrodiffusion). In addition, a phenomenon of gas permeation across the membrane enables the migration of a small quantity of nitrogen from the cathode region towards the anode. The addition of these phenomena provokes the generation of a fuel gas, agitated by the recirculation 20, which is admitted into the fuel cell stack 18 and reduces the hydrogen content as a function of the time passed since the last anodic purge. The reduction of the hydrogen content effectively reduces its partial pressure, which results directly in a reduction of the voltage produced by the fuel cell stack 18. In order to maintain the fuel cell stack 18 in good condition, it is therefore important to periodically renew the anodic gas by purging the circuit. Opening the purge valve 26 allows a reaction at the reduction valve 16 that substitutes hydrogen for the mixture of hydrogen, water and nitrogen.

To ensure acceptable durability for the membrane of the fuel cell stack 18, it is necessary to humidify the reactants, air and hydrogen, before their entry into the fuel cell stack 18. The use of dry gas implies the need to charge these reactants with water vapor during their passage into or through the fuel cell stack 18. If this is not done, there is a danger that the membrane will become dehydrated and deteriorate, which could even lead to its destruction. To avoid this deterioration, pre-humidification of the gases is realized upstream of the fuel cell stack 18 using the recirculation circuit 20 internal to the system 12 and using the production of water by the fuel cell stack 18 as a source of the water which can be used for pre-humidification.

Usually, there are two problems with using internal recirculation as a source of the water to be used for pre-humidification. One of these is a potential link between hydrogen and air lines and a second is parasitic consumption of energy by the recirculation devices 24.

It is necessary to provide a humidification unit, which may be in common with the air loop and supplied by recovery of water produced in the fuel cell stack 18 by means of condensers and water traps positioned downstream of the fuel cell stack 18. This air loop, common to both air and hydrogen, represents a possible connection between the air and hydrogen lines and may prove less than totally satisfactory under some fault conditions.

During the passage of hydrogen through the fuel cell stack 18 and its consumption in there, water condenses downstream of the fuel cell stack 18 and can be recovered for subsequent use in humidification of the entry gases. The recirculation of the mixture of gas and water is realized by hydrogen pumps or ejectors. Generally, this type of equipment does not allow for the entry of a mixture containing water or, in the case of ejectors, steam. The instrument that is perhaps most interesting for recirculation is the ejector, because its uses the energy in the pressure reduction of the hydrogen to supply its power module so as to pump the hydrogen exiting the fuel cell stack 18. In that way, it does not really introduce a parasitic consumption of energy into the power supply circuit.

Construction of a typical ejector 30 is illustrated with reference for the moment in particular to Figure 3. The ejector 30 comprises a convergent fluid motor 32 connected in-line to a diffuser nozzle 34. In similar fashion to a venturi device, an inlet 36 for a fluid that is to be pumped leads into the ejector 30 at a point downstream of the fluid motor 32, i.e. a point where a depression is created in use. The working fluid of the fluid motor 32 draws in the fluid to be pumped and they travel into the diffuser nozzle 34 through a mixer 38 that is configured to ensure adequate mixing.

Such ejectors 30 help keep parasitic electricity consumption to a minimum and may therefore be the main recirculation organ of preference. They are sometimes referred to as a trumpet device or jet and as can be seen are static devices destined to suck in and compress or mix gases, vapors, liquids and sometimes solids owing to depression of a primary driving fluid. The latter can be gaseous, in subsonic or supersonic regime, or liquid in incompressible regime and provides the energy necessary for drawing the secondary fluid.

However, this general type of device happens to be very sensitive to variations in the differential pressure and in the volumetric flow of the pumped fluid. Figure 4 is taken from the book "Ejecteurs: Techniques de I'ingenieur, Vol. B 4 250" by Jaques Paulon and shows the characteristic curves for a typical gas injector for the different regimes.

For a subsonic regime, the effect as shown in Figure 4 can be really disastrous as will be shown hereafter. The nitrogen diffusion increases the nitrogen concentration downstream of the fuel cell stack 18. Therefore the ratio between the inlet and outlet pressures of the ejector is reduced. Thereby, the flow recirculated by the ejector 30 decreases notably when it should have increased in order to maintain constant the flow delivering the hydrogen. For a given regime, this increase can be dramatically important when entering the zone of mixed functioning indicated by ¹¹M" in Figure 4.

Problems with the use of a closed hydrogen loop. For reasons of efficiency, the hydrogen loop is operated in a closed loop configuration, in order to reduce as much as possible the hydrogen losses during stack operation and recirculation. Indeed, the fuel cell stack 18 may require an anodic over-stoichiometric mixture, e.g. for reasons associated with flow and diffusion problems of the reactants in the membrane. The recirculation loop makes it possible to respect this restriction coming from the fuel cell stack 18, theoretically without leakage of hydrogen into the environment. The stoichiometry of the hydrogen consumed by the fuel cell stack 18 at the system level is of the order of one.

Several permeation phenomena can, however, disturb the ideal functioning of the recirculation loop. Indeed, due to the different materials used by the membrane, the latter allows slight permeation of nitrogen, hydrogen and oxygen. Thus, in the course of time, the nitrogen diffuses from the cathodic section to the anodic section. Due to the fact that the hydrogen loop is closed, an accumulation of nitrogen is produced and disturbs the good functioning of the fuel cell stack 18. This is expressed as a loss in efficiency of the fuel cell stack 18, exhibited as a drop in output voltage. Figures 5 to 8 show the evolution of the efficiency a fuel cell of 84kW capacity for a partial load of 5.4kW and a nominal full load of 84kW.

Effect of the permeation of oxygen and hydrogen

This problem may be called "cross-over phenomena" and for oxygen and nitrogen the problem is totally different. Indeed, oxygen and hydrogen diffuse respectively from the cathode to the anode and from the anode to the cathode. The oxygen found at the anode reacts at the catalytic sites with hydrogen. Likewise, the hydrogen found at the cathode reacts directly with the oxygen. These two reactions cause a consumption of hydrogen which does not contribute to the conversion into electric energy and therefore leads to a drop in efficiency. This phenomenon is negligible at full load as the flows are too low to be perceived. However, at weak loads these permeation phenomena can noticeably disturb the efficiency of the fuel cell 18 by a drop which can reach 5%. Figures 9 and 10, show the effects as a function of the concentration of nitrogen accumulated in the anodic section for full and weak loads.

Problem of nitrogen elimination

As nitrogen accumulates in the anodic loop, it provokes a drop in tension and thus in efficiency of the fuel cell. Figures 11 to 13 represent the effects of nitrogen accumulation for a fuel cell of 84kW at full load. In Figure 11 , a graph is provided that shows a typical rise in the percentage of nitrogen concentration against time during the time in which it is allowed to accumulate. The nitrogen builds up to say 60% over about 500 seconds. For the same period of time, Figure 12 shows the evolution of current density needed to maintain full power during the functional cycle concerned and Figure 13 shows the evolution of the cell voltage for that cycle.

If no action is taken, the nitrogen can reach a level such that it becomes impossible for the fuel cell 18 to supply the required power. Thus it is indispensable to eliminate part of the nitrogen. Classically, the hydrogen recirculation loops are provided with an anodic purge valve 26 in order to effect this action. The valve 26 is periodically opened and enables the emptying of the circuit to ensure elimination of the nitrogen, which purge cycle can be seen in each of Figures 11 to 13 taking place by way of example shortly after the 500 second point. However, a mode of intermittent purging presents several major inconveniences of its own. For a given load at a constant voltage, the fuel cell 18 will never reach a stationary regime, i.e. one that does not evolve over time, because of the periodic purging. This is due to the fact that nitrogen diffusion is a phenomenon that evolves over time and the same is the case for the purge. Consequently, for a delivered power at constant voltage, the current density as well as the nitrogen concentration in the anodic loop is never constant.

In some cases, the variation in time of the tension (output voltage) of the fuel cell 18 can cause problems. Indeed, the electric circuit fed by the fuel cell 18 must be conceived for the variation in time. Moreover, due to the variation, the dimensioning of the fuel cell 18 must be based on the lower value of the tension. This implies a relative increase in its size, mass and cost.

On top of the problem of non-stationary functioning comes the problem of elimination of hydrogen in the purge. Indeed, for the fuel cell power modules, generally a burner of the catalytic type is used to ensure the neutralization.

Outside this specific context, the use of periodic purges to eliminate nitrogen presents a double problem of dimensioning and functioning.

For the dimensioning of the burner when looking at the evolution of the instant power of the burner for the same example as considered previously, the curve of Figure 14 is obtained. From this typical graph, it appears that the use of a burner alone requires for that burner an instant power in the order of 1OkW to 12kW, whereas the average power for a purge cycle is about 0.42kW. It would be possible to use a storage buffer for the purge, but that might imply the use of one having a volume of 30 liters for the given example and a pump unit for injecting the stored mixture into the burner.

Moreover, for the purges, the purged mixture is composed of hydrogen, nitrogen and water in liquid or vapor form. The presence of water is a phenomenon which can, in certain cases, introduce parasitic effects on the damping of the burner and thus compromise the neutralization of the hydrogen. The purge method for this type of use makes it possible to maintain an acceptable level of efficiency. For a partial load, however, it requires that a level of the order of 15% to 20% and of 70% of the concentration be reached to obtain optimum efficiency for the respective loads of 84kW and 5.4kW. In those two cases, however, the system is outside the functioning zone of ejector technology, which is limited to about 10% of nitrogen. In order to ensure good recirculation, a pump must be installed both for high and for low load. This will lead to a parasitic electricity consumption, the latter being all the more problematic at low load as the nitrogen level required before the purge is of the order of 70% and results in the increase of load loss and the volumetric recirculation flow.

Proposals have been made in the prior art for fuel cell systems that include recirculation and purge of part of the exhaust gases, such as those in US-6124054 and US-2004/0142200. In WO 2004/051780, a fuel cell power plant is proposed in which unused fuel gas is recirculated back into the supply. A purge valve is provided for purging nitrogen, which is regulated such that the concentration of nitrogen in the recirculated fuel gas is kept constant. Various parameters are measured so as to estimate the flow rate for hydrogen purge, which is done with respect to a target value of nitrogen concentration in the hydrogen that is to be recirculated into the fuel supply. The constant nitrogen purge thereby regulates the amount of hydrogen recirculated to a constant level.

Many prior art systems have one or more of the problems discussed above and there is therefore a continuing need to develop improved arrangements for purge management in fuel cells, and in particular for purge management which keeps down parasitic losses.

Summary of the Invention It is an object of the present invention to provide an improved arrangement for purge control in fuel cell power plants, in particular for a vehicle.

Accordingly, the present invention provides a fuel cell power plant comprising a fuel cell stack configured to be fed with a regulated supply of pure or quasi-pure hydrogen through a pressure regulator device and having a recirculation loop that is configured for hydrogen recovery from downstream of said fuel cell stack and for the provision of said recovered hydrogen into said hydrogen feed supply, said power plant further comprising a purge device that is located downstream of said fuel cell stack, said purge device being under the control of a controller and being configured to implement a substantially continuous purge of nitrogen gas from said power plant, characterized in that said controller includes an input configured to receive a signal indicative of an electrical load demanded of said fuel cell power plant and in that said controller is configured to control said purge device in response to said load signal so as to regulate the level of nitrogen at the entrance to, and/or exit of, an anode of said fuel cell stack, whereby said controller maintains a concentration of nitrogen at said anode which is determined as a function of the electrical load demanded of said fuel cell power plant. It is an advantage of the present invention that the continuous purge is effected in such a manner that the concentration of nitrogen at the anode is regulated in proportion to the electrical load on the fuel cell power plant. This ensures that the amount of hydrogen recirculated can also be regulated in proportion to that load, which allows for improved control over parasitic losses when compared with some prior art arrangements.

Said concentration may be further determined as a function of a parameter relating to the functioning of the fuel cell itself, such as a parameter relating to at least one of fuel cell anodic pressure, cathodic pressure, anodic over-stoichiometry or richness of the fuel and the comburant at the entrance to the fuel cell stack.

Said purge device may be used by said controller to regulate the level of hydrogen that is recirculated, so that the amount of hydrogen recirculated is adapted to the load on the fuel cell power plant. Said fuel cell power plant may further comprise a first sensor downstream of said stack and a second sensor upstream of said stack, each said sensor comprising one of a nitrogen concentration sensor or a hydrogen concentration sensor, and wherein the level of hydrogen recirculated may be regulated on the basis of a comparison between signals output from said first and second sensors. Said comparison may be used by said controller to determine hydrogen over-stoichiometry.

Said purge device may comprise a control valve that passes a gas flow that transports a nitrogen flow that is equivalent to a nitrogen flow crossing from a cathode of said fuel cell stack towards an anode thereof. Said purge device may be configured for injection of an airflow during said continuous purge operation.

Said power plant may further comprise an air compressor device configured to supply compressed air into said fuel cell stack and said injected airflow being bled off the output of said air compressor device. Said purge device may comprise a calibrated orifice at its entry for the introduction of the purge gas flow, a passage defined by said orifice extending through a one-way valve biased against the purge gas flow. Said purge device may include a torus configured for regulation of an escape passage for the purge gases, said torus preferably being positioned downstream of said calibrated orifice. Said torus may be inflatable so as to provide variable regulation of said escape flow of purge gases through said purge device. Said torus may also define a passage for air injection into said purge device, said air injection preferably also being used for inflation of said torus.

The purged gas flow may be transported to a catalytic burner so as to dispose of surplus hydrogen in the course of the purge of nitrogen, said catalytic burner preferably being operably connected directly to an outlet of said purge device and more preferably being integrated onto said purge device. Said catalytic burner may be configured to be fed substantially continuously by said purged gas flow.

Said controller may implement said purge under the influence of a signal provided by a nitrogen concentration sensor, said sensor preferably being positioned downstream of said anode in a hydrogen conduit or a by-pass thereof.

Said controller may implement said purge under the influence of a signal provided by a hydrogen concentration sensor. Said fuel cell power plant may further comprise an active recirculation device such as a bi-phasic pump and a static recirculation device such as an ejector, wherein said active recirculation device is used to pump recirculated gas mixture at low load with water in the liquid phase preferably being injected upstream thereof, and said static device is used to recirculate nitrogen at high load.

Said fuel cell power plant may further comprising a first nitrogen concentration sensor downstream of said stack and a second nitrogen concentration sensor upstream of said stack, wherein the level of hydrogen recirculated is regulated on the basis of a comparison between signals output from said first and second sensors, for example by using said comparison to determine hydrogen over-stoichiometry.

Said purge device may be located in said recirculation loop, upstream in that loop of a hydrogen recirculation device. The present invention also provides a method of operating a fuel cell power plant, including: a) feeding a fuel cell stack with a regulated supply of pure or quasi-pure hydrogen through a pressure regulator device; b) recovering hydrogen from downstream of said fuel cell stack and recirculating said recovered hydrogen into said hydrogen feed supply; and c) purging nitrogen gas substantially continuously from said fuel cell power plant using a purge device, the method being characterized in determining a load on said fuel cell power plant and regulating the volume of nitrogen purged in proportion to said load, thereby regulating the amount of hydrogen recirculated also in proportion to said load.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention will now be described by way of example only and with reference to the accompanying drawings, in which :

Figure 1 is a schematic view in section of a fuel cell;

Figure 2 is a basic fuel cell circuit including a recirculation loop; Figure 3 is a schematic view in section through an ejector suitable for use in the present invention;

Figure 4 is a performance graph for the ejector of Figure 3;

Figure 5 is a graph of fuel cell efficiency at partial load (nominally 5.4 kW) for a fuel cell having nominal maximum power rating of 84 kW; Figure 6 is a graph of fuel cell efficiency at full load for a fuel cell having nominal maximum power rating of 84 kW, which may be the same fuel cell as the one considered in Figure 5 when running at partial load;

Figure 7 is a graph of efficiency lost for the fuel cell of Figure 5 when running at partial load (5.4 kW), expressed as the nitrogen concentration (X axis) against the percentage loss of output (Y axis);

Figure 8 is a graph of efficiency lost for the fuel cell of Figure 6 when running at a nominal full load (84 kW); Figure 9 is a graph of the effects of cross-over, represented in the form of hydrogen stoichiometric ratio (SR) gap due to permeation effects at partial load

(5.4 kW) against nitrogen concentration for a fuel cell having nominal maximum power rating of 84 kW; Figure 10 is a graph of the hydrogen stoichiometric ratio (SR) gap due to permeation effects at nominal full load (84 kW);

Figure 1 1 is a graph of the evolution of nitrogen concentration against time during a nitrogen purge cycle, for a fuel cell having nominal maximum power rating of 84 kW; Figure 12 is a graph of the evolution of current density against time during a nitrogen purge cycle, for a fuel cell having nominal maximum power rating of 84 kW;

Figure 13 is a graph of the evolution of cell voltage against time during a nitrogen purge cycle, for a fuel cell having nominal maximum power rating of 84 kW;

Figure 14 is a graph of the evolution of burner power against time during a nitrogen purge cycle, for a fuel cell having nominal maximum power rating of 84 kW;

Figure 15 is a graph of fuel cell efficiency at 5.4 kW gross output from a fuel cell having a nominal full power rating of 84 kW, i.e. while it is running at a partial load of nominally 5.4 kW;

Figure 16 is a graph of fuel cell efficiency at 84 kW gross output from the fuel cell, i.e. while it is running at a full load of nominally 84 kW;

Figure 17 is a graph of fuel cell efficiency at partial load (5.4 kW); Figure 18 is a graph comparing fuel cell efficiency at nominal full load (84 kW) for various conditions;

Figure 19 is a graph of efficiency lost at partial load (5.4 kW) due to permeation effects, illustrated as a function of the concentration of nitrogen at the anode;

Figure 20 is a graph of efficiency lost at nominal full load (84 kW) due to permeation effects, illustrated as a function of the concentration of nitrogen at the anode; Figure 21 is a is a graph of the hydrogen stoichiometric ratio (SR) gap due to permeation effects at partial load (5.4 kW), illustrated as a function of the concentration of nitrogen at the anode;

Figure 22 is a is a graph of the hydrogen stoichiometric ratio (SR) gap due to permeation effects at nominal full load (84 kW), illustrated as a function of the concentration of nitrogen at the anode;

Figure 23 is a graph of nominally optimal evolution of nitrogen concentration on the side of the anode, against fuel cell power (kW);

Figure 24 is a schematic section through a nitrogen purge valve; Figure 25 is a schematic diagram of a fuel cell system according to an embodiment of the present invention including the purge valve of figure 24;

Figure 26 is a modification to the arrangement of figure 24;

Figure 27 is the schematic diagram of figure 25, with the modification of figure

26 included; Figure 28 is a schematic diagram of a fuel cell control system according to the present invention;

Figure 29 is a variation of the system of Figure 28;

Figure 30 is a schematic diagram showing a first position for a sonic sensor;

Figure 31 is a schematic diagram showing a second position for a sonic sensor; Figure 32A shows schematically the form of a typical sonic sensor;

Figure 32B illustrates the derivation of parameters used in the calculations associated with Figure 32A;

Figure 33 shows a fuel cell cooling loop and an associated graph representing the localized pressure within that circuit; Figures 34 and 35 are associated graphs of the evolution of nominally optimal nitrogen concentration on the anode side of the fuel cell, showing the working regions of the ejector with respect to the working region of the bi-phasic pump and their combined effects;

Figure 36 is a schematic diagram of the water recirculation circuit with the combined effects of the biphasic pump and the ejector;

Figure 37 shows three flow regimes for the membrane electrode assembly

(MEA); Figure 38 is a graphical representation of the three regimes illustrated in figure 37, shown positioned relative to speed of the liquid phase (VLS) and the speed of the gas phase (VGS);

Figure 39 is a schematic diagram illustrating control of the recirculation pump; Figure 40 is a perspective view of a recirculation pump and an associated fuel cell illustrating connections disposed on typical such apparatus; and Figure 41 is a schematic diagram of a possible fuel cell system including a purge arrangement according to an embodiment of the present invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

Referring to the drawings, in order to eliminate nitrogen without the disadvantages involved in purging by periodic and temporary opening of an electrically operated valve, the present invention discloses the use of a continuous purge. This is realized by using a control valve which passes a gas flow that transports a nitrogen flow which is equivalent to the nitrogen flow crossing the membrane from the cathode towards the anode. In this way, it is possible to regulate the concentration of nitrogen well enough, as a function of electrical load on the fuel cell, to approach a substantially stationary or steady- state situation for the system and thereby to regulate the recirculation of hydrogen in proportion to the electrical load..

Evolution of the concentration of nitrogen in the anodic section as a function of the electrical load

The continuous purge performs the same function as the temporary or periodic purge, eliminating also hydrogen in the course of eliminating nitrogen. In order to neutralize that hydrogen flow, the purged flux is directly sent to a catalytic burner. By this, the catalytic burner is continuously fed and is not disturbed by the presence of water which can stop the damping of the latter. By way of example, the power required for the burner will typically be at the most in the order of 0.8kW (compared with the figure of 1OkW to 12kW in Figure 14). The amount of hydrogen lost is more important than with a periodic purge but that loss is compensated by the fact that, due to the stabilization of the nitrogen concentration, the fuel cell does not drop in voltage. Taking again the exemplary fuel cell having a nominal full load capacity of 84kW, Figures 15 to 22 show, for a partial load of 5.4kW and the nominal full load of 84kW, the evolution of efficiency and over-stoichiometry in function of the amount of nitrogen present at the anode before purge. As shown in Figures 15 to 22 with Figure 23, there exists an optimum nitrogen concentration which makes it possible to achieve the best efficiency. Indeed, the continuous purge of the present invention makes it possible to obtain a controllable nitrogen concentration at the anode, i.e. a concentration for which it is possible to predetermine the value. This specific nitrogen concentration tends to limit the hydrogen permeation towards the cathode, cross-over phenomenon, and the permeation of hydrogen towards the anode. This positive impact combined with the hydrogen loss due to the continuous purge means that optimum functioning in terms of efficiency exists.

The optimum nitrogen concentration is a function of several parameters. The first parameter is the load of the fuel cell. Less important parameters are those related to the functioning of the fuel cell, such as anodic pressure, cathodic pressure, anodic over-stoichiometry and cathodic over-stoichiometry, richness of the fuel and the comburant at the entrance of the fuel cell and the functioning temperature of the fuel cell. The Figure 23 shows an example of the evolution of the optimum nitrogen concentration in function of the load of the fuel cell with a power of 84kW for a functioning pressure of 17OkPa and temperature of 85 deg. C.

Initiation of continuous purge In the non-limiting embodiment, the continuous purge of the present invention is initiated and controlled using an escape valve coupled to a controller. A typical arrangement 120 may be seen by way of example in Figures 24 to 31.

Referring for the moment in particular to Figures 24 and 25 together and to Figures 26 and 27 together, two examples of a fuel cell system 120 according to the present invention are illustrated and each one comprises a modification to the system previously discussed in relation to Figure 2. In common between the two example, the modified system comprises a fuel cell stack 180 fed with hydrogen gas H2 from a hydrogen supply 140 via a regulator valve 160.

The regulator valve 160 includes an input 220 from a recirculation circuit

200, by which means the hydrogen gas is humidified upstream of the fuel cell stack 180 using water recovered from the outlet of the stack 180 under the influence of a signal supplied by a nitrogen concentration sensor 250 and regulated by a hydrogen recirculation device 240.

The hydrogen (H₂) is available under a pressure that is higher than the pressure of the rest of the system 120 and is supplied into the fuel cell stack 180 by a regulating valve 160, involving pressure reduction. In the absence of such a regulating valve 160, pressure of the hydrogen at the anode would drop. The regulating valve 160 regulates the inlet pressure of the hydrogen automatically and rapidly so as to compensate for the consumption induced by the electrochemical reaction, typical response time being less than 0.1 to 0.5 seconds.

To ensure acceptable durability for the membrane of the fuel cell stack 180, it is necessary to humidify the reactants, air and hydrogen, before their entry into the fuel cell stack 180. The use of dry gas implies the need to charge these reactants with water vapor during their passage into or through the fuel cell stack 180. If this is not done, there is a danger that the membrane will become dehydrated and deteriorate, which could even lead to its destruction. To avoid this deterioration, pre-humidification of the gases is realized upstream of the fuel cell stack 180 using the recirculation circuit 200 internal to the system 120 and using the production of water by the fuel cell stack 180 as a source of the water which can be used for pre-humidification.

During the passage of hydrogen through the fuel cell stack 180 and its consumption in there, water condenses downstream of the fuel cell stack 180 and is recovered for subsequent use in humidification of the entry gases. The recirculation of the mixture of gas and water is realized by hydrogen pumps or ejectors 240. Generally, this type of equipment does not allow for the entry of a mixture containing water or, in the case of ejectors, steam. One arrangement suitable for use as the hydrogen recirculation device 240 of the present invention is the ejector 30 described in relation to Figures 3 and 4. In place of the electrically controlled valve 26 in the prior art arrangement of Figure 2, the present invention provides a nitrogen purge valve 260 of the general type illustrated by way of the first functional example with reference to Figures 24 and 25 or, by way of the second functional example with reference to Figures 26 and 27. Such a valve preferably comprises in either structure (Figure 24 or Figure 26), a calibrated orifice 262 leading through a one way valve 264 to a torus 266, which torus is preferably inflatable.

The one way valve 264 may comprise a ball 268 biased against the direction of flow and, for that, being biased against a frustro-conical seat 270 by a helical spring 272. After passing through the torus 266, the purge gases enter a catalytic burner 274 in which they are burnt off with the help of air injection 276.

In the example of Figures 24 and 25, the air injection is introduced directly into the catalytic burner 274 via an air injection port 276. In the case of an inflatable torus 266, it may be found advantageous to use the same supply 284 to inflate the torus 266. In Figures 26 and 27, the air injection 284 enters the catalytic burner 274 via the torus 266.

In both examples, the fluids being released at the anode are redirected through the recirculation circuit 200 to the inlet of the fuel cell stack 180, entering the stream of inlet hydrogen at an inlet 220 to the regulating valve 160. In the recirculation circuit 200, the recirculation device 240 is configured to prevent stratification and to enable a homogenous mix of hydrogen, water and nitrogen, the precise composition of which may vary. This recirculation mode may be found usable up to a mixture typically containing about 20% hydrogen. The nitrogen purge valve 260 guards against the effects of entry by contaminants (e.g. water and nitrogen or natural gas) into the anodic chamber. The entry of these different species can introduce pollution and degrade the performance of the fuel cell stack 180. This is because, due to the chemical reaction, part of the water produced migrates from the cathode to the anode and stagnates there (so-called retrodiffusion). In addition, a phenomenon of gas permeation across the membrane enables the migration of a small quantity of nitrogen from the cathode region towards the anode. However, the purge valve 260 is also configured or controlled for implementation of the continuous purge of the present invention.

There are two problems usually encountered with using internal recirculation as a source of the water to be used for pre-humidification. One of these is parasitic consumption of energy by the recirculation devices 240 and a second is a potential link between hydrogen and air lines.

The first of these issues may be dealt with by implementing the hydrogen recirculation device 240 of the present invention in the form of the ejector described in relation to Figures 3 and 4 or by an equivalent of it. In the exemplary embodiment, air is compressed by a compressor 280 before entering the fuel cell stack 180. The exhaust of the fuel cell stack 180 is used to drive a turbine 282, but a bleed-off from the exhaust gases is taken downstream of the turbine, regulated by an air pilot valve 284. This bleed is fed into the catalytic burner 274 in the form of air injection 276 via a dedicated air- injection port 276 (Figures 24 and 25), while the same supply may be used to inflate an inflatable torus 266. In the alternative, the bleed may be fed into the catalytic burner 274 in the form of air injection 276 via the inflatable torus 266 (Figures 26 and 27).

The assembly 120 of either example creates a useful degree of freedom. It enables a practical dynamic control of the nitrogen concentration at the anode. Thus, even though the dimensions of the stack 180 are carefully chosen, it becomes possible to control the nitrogen concentration at the anode in an optimum manner for the different loads of the fuel cell. This optimum can be a balance between the functioning of the ejector 240 and the overall efficiency that can be achieved. Furthermore, the degree of freedom makes it possible to develop different modes of recovery for operation of the hydrogen recirculation sub-system 200.

Still further, an improvement may be obtained with respect to the reliability of the system. The flow of the continuous purge is below what is evacuated using the prior art periodic purges. Consequently, if there is a need for a burner to neutralize the lost hydrogen, then in the case of a continuous purge of the present invention the burner will be of a lower power, lower cost and easier integration. Likewise, in case of failure of the purge system 260, the controlled escape enables the integration of a calibrated opening 262 ensuring a minimum escape flow without clearly disturbing the functioning of the power module. Whereas for the purge, a blockage of the valve 260 or a deterioration of the conduit, implies a potentially important escape due to the level of the flow circulating through it. The functioning mode with controlled escape thus presents the advantage of increased reliability and a passive recovery mode, via the calibrated orifice 262, without putting into question the performance of the system. A simple indication on the dashboard can be used for indicating the need for maintenance, e.g. MIL (malfunction indicator light). The schemes of Figures 26 plus 27 provide an idea of a controlled escape system, driven by a simple actuator for an airflow which totally isolates the fuel flux from any mobile assembly which might present the risk of an escape. Furthermore, this system comprises the calibrated orifice 262 and a one way valve 264. The regulation of the escape flow may be realized due to a hollow and deformable torus 266. Air injection is obtained in the cavity of the torus 266 so as to inflate it to a greater or a lesser degree in order to control the passive surface area which controls the escape. The air is thereafter injected upstream of the catalytic burner 274 either by an external conduit 276 (Figures 24 and 25), or via the torus 266 (Figures 26 and 27) which in that case comprises the escape orifices.

Regulation of continuous purge

The control chain of the continuous purge has for a purpose the regulation of the nitrogen level at the entrance to, or at the exit of, the anode. The architecture of the control chain for anodic pressure regulation is represented schematically in Figure 28 and includes a controller 300, that is illustrated for convenience as integrated with the regulator 160 and which includes a map or calculator 302 for optimal nitrogen concentration. A measurement chain is integrated into the control chain in the form of the nitrogen concentration sensor 250, which returns to the regulator 160 the value of the nitrogen concentration at the anode. The regulator 160 responds by sending a task to the air pilot valve 284 in order to obtain the objective value of the nitrogen concentration. The objective value depends on the functioning parameters of the fuel cell, for example; output current, anodic pressure, cathodic pressure, temperature, anodic over-stoichiometry, cathodic over-stoichiometry and output voltage of the fuel cell 180. More or less values may be used and these values may be supplied by appropriate sensors 310.

In particular, the controller 300 is provided with an input signal that is indicative of the electrical load demanded of the fuel cell power plant. This signal may be determined using a current sensor that measures the electrical current drawn by an electrical drive train of a vehicle to which the fuel cell power plant is fitted by way of prime mover.

In a vehicle in which an electrical storage device such as a battery or an ultra-capacitor is fitted, for example for traction purposes, a load signal may be available or developed from a power train control unit that determines the traction power required. In that way, the state-of-charge of such a storage device may directly dictate the target concentration of nitrogen.

Some power devices used in an electrical power distribution network to which the fuel cell power plant is connected, such as electrical motor inverters or DC-DC converters, may include sub-controllers of their own that determine the electrical load. Under such circumstances, the signal indicative of electrical load could be transferred from these devices to the controller 300.

Howsoever the load signal is derived, it is the job of the controller 300 to calculate the target concentration of nitrogen at the entrance to, and/or exit of, the anode of the fuel stack 180. The controller 300 is then in the position to control, in response to that load signal, the substantially continuous purge so as to maintain a target concentration of nitrogen at the anode that is in proportion to the fuel cell working load.

Referring now for the moment to Figure 29, an alternative or modification to the arrangement of Figure 28 is illustrated. If the air pilot valve 284 solution of the arrangement in Figure 28 presents problems, a purge valve with an integrated actuator can be installed (264; 284). For that embodiment, the actuator(264; 284) must be conceived for use in a hydrogen atmosphere and must also not be too sensitive to an intermittent flux of water in liquid phase. Nitrogen concentration measurement chain.

Referring now in particular to Figures 30 and 31 , several technologies can be used for the measurement of nitrogen levels. It is important to note that all these technologies have at an initial stage the need to integrate the measurement of the temperature of the mixture, either for direct use or for the calculation of the saturated vapor pressure. The sensor may be placed directly in series with the hydrogen conduit 304 downstream of the anode A, as illustrated in Figure 30. When placement directly in the hydrogen conduit is problematic, for example because of the presence of liquid water, for the protection of the sensor 250A it may prove preferable to position it in a by-pass to that hydrogen conduit 304. A by-pass arrangement 252 is illustrated schematically in Figure 31.

Figure 32A illustrates how such a sonic sensor 250A is used in an arrangement like that of Figure 30 or Figure 31. The principle of operation of the sensor 250A is the measurement of the flow speed of gas via the Doppler effect. This sensor 250A thus also makes it possible to measure the speed of sound in the hydrogen conduit. This information can be directly useful for determining the nitrogen concentration in the hydrogen conduit 304. The sensor 250A has the configuration shown schematically in Figure 32A and Figure 32B illustrates how the information is obtained on the principle of time differential measurements. A signal is sent across the hydrogen conduit 304 along a fixed distance L and the Doppler shift of its reflection is measured. The transmission time is t1 and the return time of the reflected signal is t2.

Formulae for Figures 30 to 32B

The two differential time values follow the following formulae (1 & 2)

— 1 C

= — ^■ + — V

[1]

L L

— 1 _ C — V

[2]

U ^~ L L For the flow measurement, the flow speed is obtained from Formula 1 Formula 2, giving us the following relationship; (3)

1-1 = 2-^ f, t₂ L

If (1 ) and (2) can be added up, the speed of sound "c" is obtained directly.

/, t₂ L

The value of the speed of sound in the hydrogen conduit is directly related to the composition of the gas by the following relationship:

where "M" is the molecular mass of the mixture and y represents the relationship of the specific temperatures at constant pressure and volume of the mixture.

If the sensor 250A is positioned in a well chosen position where the gas is always quasi-saturated with steam, for example downstream of the anode, it will be possible to directly measure the value of "M" by measuring "C" and thereby to determine the corresponding concentration of nitrogen.

Using a hydrogen sensor to effect the measure. As an alternative to developing a special sensor 250A, for example if it proves unacceptably expensive, it is also possible to adopt a sensor configured to determine the level of hydrogen and this type of sensor is under development at several suppliers. This allows an indirect measure of the quantity of nitrogen, the link between the two options being shown in the following relationship.

C — \ — C — vaμsaturated

'N₁ Pressure

Hydrogen recirculation loop.

Reference is now made to Figures 33A and 33B. Apart from the problems caused by the cross-over phenomenon and the accumulation of hydrogen in the anodic recirculation loop, a problem arises regarding the durability of the membrane. Indeed, only the hydrogen recirculating in the recirculation loop 220 is humidified. The hydrogen that is injected into the loop in order to compensate for hydrogen consumption is not humidified. Furthermore, it has been shown that for a vapor-saturated fluid at the entry of the anode A (Figures 33A and 33B; position 4), part of the vapor that it transports will condense during the crossing of the fuel cell 180 due to the disappearance of hydrogen. This water in liquid form is what the fuel gas entering the system requires.

In order to alleviate this problem, the anodic recirculation loop 220 comprises in series a biphasic recirculation organ 330 and an exchanger 332 serving as an evaporator/vaporizer, through which passes the fuel cell cooling loop 338. This arrangement is illustrated in Figure 33A, along with a graphical representation in Figure 33 B of the pressure effects against the location in that loop. The loop starts at the exit of the anode A (position 5) and enters a recirculation organ 330 via a restriction (position 6 to 6_bi_S). The active organ of the biphasic recirculation loop 220 is of the pump type 330 and ensures the recirculation of the mixture of hydrogen, nitrogen, water vapor and liquid water. Downstream of this recirculation organ 330, the feed-in 336 of the hydrogen loop 220 is realized upstream of a mixer 344 (position 1 ). This feed-in 336 is where hydrogen is added to the loop 220. Before being directed into the fuel cell 180, the final mixture passes from the mixer 344 through the evaporator 332 using the heat bearing fluid of the fuel cell cooling loop 338 as its heat source and ensuring the cooling of the fuel cell 180.

The evaporator 332 may comprise a simple heat exchanger, for example a planar matrix. Downstream of the evaporator 332, the remaining water in liquid phase is separated in a water-trap 342 and directed via a water tank including a purge arrangement 346 into a water loop 340. The recovered water is then conducted to the entrance of the biphasic pump 330 along the water loop 340.

This continuously functioning liquid loop 220 makes it possible to obtain a continuous flow of water in order to ensure the humidification of the hydrogen feed to the hydrogen recirculation loop 220. In order to realize this water recirculation, the restriction running from positions 6 to 6_bi_s is used to draw in water recovered via the water loop 340 and is preferably placed upstream of the entrance of the recirculation pump 330 in order to perfect the natural circulation of the water. However, the water loop 340 is not always necessary as it is the restriction of the membrane with regard to the anodic humidification level that determines the necessity or the non-necessity of the water recirculation loop 340. From the downstream side of the water trap 346 (position 3), the humidified hydrogen is fed up to the inlet of the anode A (position 4).

Coupling of the biphasic pump with an injector-optimization of the loop.

Referring now also to Figures 34 to 36, the recirculation loop 220 of the present invention uses the biphasic pump 330 in order to transport a mixture of steam, nitrogen, hydrogen and liquid water. This pump 330 solves the problem of the ejector technology in the context of this application. Indeed, the driving fluid being pure hydrogen coming from storage 140 at high pressure, it would be difficult to use an ejector to pump a mixture of hydrogen, steam, water and nitrogen with a nitrogen concentration above 10%. Moreover, for liquid water the operation becomes impossible. With respect to the nitrogen concentration limit, with a functioning mode with continuous and uncontrolled purge, the ejector will only be useable at high loads. Therefore the present invention proposes to realize a combination: (a) a

Diphasic pump 330 pumping the gas mixture in the case of low loads and injecting the water in liquid phase upstream of the mixer 334 of the recirculator; and an injector recirculating nitrogen in case of high loads, areas in which the optimum nitrogen is low.

Figures 34 and 35 illustrate graphically the operating regions of the two modes and show that a cross-over can be defined in terms of fuel cell load or power being output.

Figure 36 shows the recirculation loop 220 in its total configuration, i.e. with the pump-ejector 330; 30 combination. This can be regarded as a modification to the hydrogen feed arrangement of Figure 33A. The modified hydrogen inlet circuit comprises a valve 50 upstream of an ejector 30. The output of the ejector 30 feeds into the mixer 344 (position 1 ). Within the ejector part of the operating cycle, the ejector 30 draws in the recirculation gases from a point upstream of the restriction (position 6). This means that the recirculation gases pass up into and through the ejector 30, rather than going through the pump 330.

The advantage of using the ejector 30 is that at high loads the latter will unburden the pump 330 for the gas phase. As the ejector 30 is a passive apparatus, using the driving hydrogen pressure as a driving force for recirculation, in terms of electricity consumption, the savings achieved may reach 1 kW for a fuel cell of 84kW.

The ejector 30 may be by-passed by a discharge conduit 50 provided with a pressure relief valve 52. This conduit 50 serves to prevent a possible blocking of the ejector 30, which might otherwise result in stopping the hydrogen feed of the hydrogen feed loop.

Control of the recirculation loop

The function of the recirculation loop is to ensure an optimum use of hydrogen while respecting the functional restrictions of the fuel cell.

The functioning restrictions can originate out of several phenomenon, for example: • the search for optimum operation in terms of energy efficiency; • the search for optimum balance between the consumption of the recirculation loop and the polarization of the recirculation loop; and

• the flow of cooling fluid in the anodic sections.

The internal flow paths, implemented by the heat-exchange fluid so as to ensure is cooling, are either mono-phase (gaseous) or bi-phase (liquid-gas). A heat-exchange flow that is solely gaseous does not present any functional problems. It is limited mainly by its loss of energy while crossing the fuel cell. On the other hand, the bi-phase cooling flow can take several different forms now discussed with reference to Figures 37A to 38. The bi-phase cooling flow is considered now in three principal forms; stratified regime (Figure 37A), intermittent regime (Figure 37B) and annular regime (Figure 37C).

In each of the Figures 37A to 37C, two views are provided of a cooling channel 60. The left-hand view is a cross section across the channel and the right-hand view is a section along the cooling channel 60 of the membrane. The views each show the passage of the gas phase 62 and the liquid phase 64 for their respective conditions.

The stratified regime (Figure 37A) is the target regime for the cooling circuit of the present invention. In effect, while the bi-phase mix 62; 64 circulates in the channels 60 between the bi-polar plates, the two phases separate out into strata and permit a direct contact between the membrane 5 and the reactants. The liquid phase 64 runs along two of the walls of the channel 60 (e.g. top and bottom) and the gas phase 62 can pass between the side walls and along through the sandwich channel defined by the liquid phase 64. The intermittent regime corresponds to an increase in the level of the liquid phase 64. If this rise passes a certain limit, waves 66 will form and provoke blockages across the channels 60 that interfere with the passage of the gas phase 62. The result is an unstable coolant supply 62, 64 to the membrane 5, which can have a generally destabilizing effect on overall operation of the fuel cell 10.

The annular regime (Figure 37C) corresponds to too high a speed for the gas phase 62. This spreads the liquid phase over all the inner walls of the cooling channel 60 and compromises by obstruction the rapid passage of the reactant to the membrane 5.

It is possible to represent the different regimes of Figures 37A to 37C graphically, as shown in Figure 38 in which they are shown with respect to the speed of the gas phase (X-axis: VGS) and speed of the liquid phase (Y-axis: VLS).

Control of recirculation

The purpose of controlling the recirculation circuit is to control the flux of the recirculation in order to respect the constraints of fuel cell running. These constraints can be translated into a demand for over-stoichiometry of hydrogen upstream of the fuel cell 180. The overall command structure is represented schematically in Figure 39.

Because of restraints in the ejector technology, the recirculation starts in a region in which the value of the nitrogen controlled is at a maximum of about 10%. The ejector does not need its own control and is continuously supplied with drive fluid by the hydrogen supply valve via the pressure regulator. In the case of over-supply by drive fluid, the by-pass 50, 52 is there to provide relief. In this manner, the ejector 30 pumps as much as it can. It can therefore be seen that the actuator that is regulated for the recirculation loop is the bi-phase recirculation pump 330. Control can therefore be based on the frequency of its operation, for example its speed of rotation.

Control in the case where the recirculation circuit is a closed loop requires a measure allowing it to return to the nominal hydrogen stoichiometry. One solution is to install a hydrogen bleed at the entry to the fuel cell 180, but this solution is often difficult to obtain and needs to be coupled to sensors monitoring the evolution of the gas composition over time. However, for the regulation of the level of hydrogen by purge control, a measurement chain is already being used giving the level of the concentration of nitrogen downstream of the fuel cell 180. The installation of a second measurement chain upstream of the fuel cell 180 and a comparison between the measurements enables the determination of over-stoichiometry of hydrogen. The second measurement of the concentration of nitrogen should best be realized upstream of the evaporator. In this case, the debit can be approximated in the following manner:

where:

C is the measured concentration of nitrogen; In, out s the position with respect to the fuel cell 180 QH2 is the hydrogen consumed by the fuel cell 180, obtained indirectly from the output current; and Pvap is the pressure of the saturated vapor.

Knowledge of the discharge of nitrogen allows the determination of the over-stoichiometry of hydrogen at the inlet to the fuel cell. This measurement is treated by the regulator 160, which sends a command to the bi-phase pump 330 giving its speed of rotation.

The present invention is for use in power modules comprising a fuel cell benefiting from a feed of pure hydrogen or quasi-pure hydrogen. Even though the examples are for a fuel cell of 84kW nominal power, the present invention is suitable for many embodiments of power module. These may include mobile applications for transport (including traction) and stationary uses, the power of which is for example situated in the range of less than one to say several hundreds of kW.

Embodiments of the innovation may comprise the following features:

• A unit capable of ensuring a continuous nitrogen purge in order to avoid the phenomenon of nitrogen accumulation in the anodic loop. Thus, the aim is to realize a balance between the nitrogen that has permeated and the nitrogen ejected during the purge. Figure 40 illustrates what a practical arrangement might look like, in combination with the schematic diagram of

Figure 41 of the relevant portion of the fuel cell circuit. This embodiment may be regarded as a modification to the system illustrated with respect to Figure 36. The modification comprises the addition of the valve for continuous purge, in the form in which air is supplied from an air supply via a pilot valve. The purge valve ejects a mixture of water, steam and/or liquid, hydrogen and nitrogen. The mixture downstream of the purge valve is directed towards the burner serving to neutralize the hydrogen before its ejection into the environment. One of the advantages of the system is that through the continuous purge one obtains a continuously functioning burner. Thereby the burner functions continuously and has no damping and starting problems due to the presence of water in the mixture in the case of the traditionally used periodic purges. Moreover, in the case of punctual/periodic purges, the peak power of the burner is of the order of 10kw, for an average power of 400W. In the case of the present invention, the power for the dimensioning of the burner will only be the 800 W for normal functioning.

• The purge valve can be a valve with classical control with an electrical activator. However, the invention can also use a valve with variable opening and/or a deformable torus, controlled by an air injection. This air injection is piloted via a controlled air valve fed by the air compression unit feeding the hydrogen power unit. Thereby it is possible in that case to disconnect the activator (air piloting valve) of the purge valve and thereby to avoid the intrinsic problems of the presence of hydrogen. Moreover, the air used in the purge valve can be exploited as comburant feed for the nitrogen neutralization burner.

• The piloting rules of the unit consist of achieving the optimum nitrogen composition enabling to obtain the energetic optimum taking into account the ejected hydrogen, the drop in the polarization of the fuel cell due to the presence of nitrogen and finally the phenomena of cross-over of hydrogen and of oxygen. The search for the objective point can be realized by a calculator comprising either a cartography or an optimum search routine.

The control is of the closed loop type using a nitrogen concentration measurement chain downstream of the fuel cell. • The hydrogen recirculation system may comprise an assembly in series of a bi-phasic pump for the liquid water and hydrogen mixture at the exit of the fuel cell, an evaporator and finally a phase separator. The pump enables the recirculation of the mixture downstream of the fuel cell towards the evaporator. This mixture consists of hydrogen, nitrogen, water vapor and liquid water which may have condensed in the fuel cell due to the hydrogen consumption. Downstream of the pump is added to this mixture the hydrogen feed of the system, unhumidified hydrogen. In order to obtain an acceptable availability of the membrane of the fuel cell, this final mixture is injected into the evaporator so as to reach its water vapor saturation point. This in order to ensure an acceptable life span of the membrane. The evaporator uses as heat source the heat bearing liquid ensuring the cooling of the fuel cell. Of course, if the cooling flow of the fuel cell is too high for the evaporator, a by-pass 252 can be installed in order to avoid too significant losses. Downstream of the evaporator, a phase separator is installed in order to recover the excess of water in liquid phase and not to disturb the functioning of the fuel cell due to the presence of water in the liquid phase upstream of the anode. The separator also serves as a water buffer and comprises and automatic purge in order to avoid all blocking of the separator by the liquid phase. Moreover it is possible, during certain stages, that the water having condensed downstream of the anode doesn't suffice for the humidification of the entire mixture upstream of the anode. In order to solve this problem, the water recirculation loop is installed between the water buffer, water trap or phase separator and upstream of the biphasic recirculation pump. This loop enables the continuous water circulation so as to be able to humidify the mixture upstream of the anode. The recirculation flow is achieved naturally by the pressure differential and is calibrated by the geometry of the water recirculation line.

• The control of the biphasic pump which ensures the recirculation of the mixture leaving the fuel cell and the liquid water injection in the evaporator is controlled in order to respect certain functioning restrictions of the fuel cell. These restrictions can be expressed as an over-stoichiometry or richness in hydrogen upstream of the fuel cell. The value of the over-stoichiometry results form the determination of the energetic optimum, balance between the polarity of the fuel cell and the recirculation system, or the optimum for the internal flow of the anode. The control is of the closed cycle/loop type, using a differential measurement of the nitrogen level in order to calculate the richness in hydrogen upstream of the anode. The control loop acts on the speed regulator of the pump in order to reach the calculated objective.

• Moreover, a combined biphasic pump-ejector can be installed. The biphasic pump ensures part of the recirculation of the mixture leaving the fuel cell and the injection of water into the evaporator. The injector for its part ensure the recirculation of the other part of the mixture leaving the anode of the fuel cell when this is possible. The ejector uses in that case the hydrogen flux feeding the fuel cell in order to obtain the pumping of part of the mixture from downstream the anode. However, the latter is only really active when the pumped mixture contains not more than 10% nitrogen. According to the calculations made, the ejector can only ensure its operation during high load functioning. Indeed, for low charges, in order to reach the energetic optimum it is necessary to have a high level of nitrogen, e.g. up to 30-40% in order to counterbalance the phenomena of cross-over and nitrogen permeation. Thus, the ejector makes it possible to relieve the pump at high load. This relief can enable a reduction in the electric consumption necessary for the recirculation which can reach 1 kW for a fuel cell of 84kW, and thereby to improve the efficiency of the power module and to contribute to a reduction of the power and the cost of the fuel cell for a same net power of the fuel cell system. For the low load area, the recirculation is principally ensured by the pump.

• The recirculator has no active control and only responds when called upon by the environment, pumped fluid and working fluid. However, it can be equipped with a by-pass 252 provided with a calibrated pressure release valve in order to achieve regulation and to avoid a blockage of the ejector. The recirculation is thus regulated in an active manner only by the regulation of the rotation frequency of the biphasic pump.

Thus, the innovations of the present invention make it possible to obtain a compact recirculation loop due to the incorporation in series of the pump, the mixer, the evaporator, the water trap. This avoids the need for a membrane evaporator which represents a risk of hydrogen leaks. Moreover, an optimization in functioning is doubly assured by the continuous purge of part of the mixture at the exit of the anode and by the combined pump- ejector.

Claims

1 ) A fuel cell power plant comprising a fuel cell stack configured to be fed with a regulated supply of pure or quasi-pure hydrogen through a pressure regulator device and having a recirculation loop that is configured for hydrogen recovery from downstream of said fuel cell stack and for the provision of said recovered hydrogen into said hydrogen feed supply, said power plant further comprising a purge device that is located downstream of said fuel cell stack, said purge device being under the control of a controller and being configured to implement a substantially continuous purge of nitrogen gas from said power plant, characterized in that said controller includes an input configured to receive a signal indicative of an electrical load demanded of said fuel cell power plant and in that said controller is configured to control said purge device in response to said load signal so as to regulate the level of nitrogen at the entrance to, and/or exit of, an anode of said fuel cell stack, whereby said controller maintains substantially continuously a concentration of nitrogen at said anode which is determined as a function of the electrical load demanded of said fuel cell power plant.

2) A fuel cell power plant according to claim 1 , wherein said concentration is further determined as a function of a parameter relating to the functioning of the fuel cell itself, such as a parameter relating to at least one of fuel cell anodic pressure, cathodic pressure, anodic over-stoichiometry or richness of the fuel and the comburant at the entrance to the fuel cell stack.

3) A fuel cell power plant according to claim 1 or claim 2, wherein said purge device is used by said controller to regulate the level of hydrogen that is recirculated, so that the amount of hydrogen recirculated is adapted to the load on the fuel cell power plant.

4) A fuel cell power plant according to claim 3, further comprising a first sensor downstream of said stack and a second sensor upstream of said stack, each said sensor comprising one of a nitrogen concentration sensor or a hydrogen concentration sensor, and wherein the level of hydrogen recirculated is regulated on the basis of a comparison between signals output from said first and second sensors.

5) A fuel cell power plant according to claim 4, wherein said comparison is used by said controller to determine hydrogen over-stoichiometry.

6) A fuel cell power plant according to any preceding claim, wherein said purge device comprises a control valve that passes a gas flow that transports a nitrogen flow that is equivalent to a nitrogen flow crossing from a cathode of said fuel cell stack towards an anode thereof.

7) A fuel cell power plant according to any preceding claim, wherein said purge device is configured for injection of an airflow during said continuous purge operation.

8) A fuel cell power plant according to claim 7, said power plant further comprising an air compressor device configured to supply compressed air into said fuel cell stack and said injected airflow being bled off the output of said air compressor device.

9) A fuel cell power plant according to any preceding claim, wherein said purge device comprises a calibrated orifice at its entry for the introduction of the purge gas flow, a passage defined by said orifice extending through a one-way valve biased against the purge gas flow.

10) A fuel cell device according to any preceding claim, said purge device including a torus configured for regulation of an escape passage for the purge gases.

11 ) A fuel cell according to claim 10, wherein said torus is positioned downstream of said calibrated orifice. 12) A fuel cell according to claim 10 or claim 11 , wherein said torus is inflatable so as to provide variable regulation of said escape flow of purge gases through said purge device.

13) A fuel cell power plant according to claim 12, wherein said torus also defines a passage for air injection into said purge device.

14) A fuel cell power plant according to claim 13, wherein said air injection is also used for inflation of said torus.

15) A fuel cell power plant according to any preceding claim, wherein the purged gas flow is transported to a catalytic burner so as to dispose of surplus hydrogen in the course of the purge of nitrogen.

16) A fuel cell power plant according to claim 15, wherein said catalytic burner is operably connected directly to an outlet of said purge device.

17) A fuel cell power plant according to claim 15 or claim 16, wherein said catalytic burner is integrated onto said purge device.

18) A fuel cell power plant according to any one of claims 15 to 17, wherein said catalytic burner is configured to be fed substantially continuously by said purged gas flow.

19) A fuel cell power plant according to any preceding claim, wherein said controller implements said purge directly under the influence of a signal provided by a nitrogen or hydrogen concentration sensor.

20) A fuel cell power plant according to claim 19, wherein said nitrogen or hydrogen concentration sensor is positioned downstream of said anode in a hydrogen conduit or a by-pass thereof. 21 ) A fuel cell power plant according to any preceding claim, wherein said controller implements said purge under the influence of a signal provided by a hydrogen concentration sensor.

22) A fuel cell power plant according to any preceding claim, further comprising an active recirculation device such as a bi-phasic pump and a static recirculation device such as an ejector, wherein said active recirculation device is used to pump recirculated gas mixture at low load with water in the liquid phase preferably being injected upstream thereof, and said static device is used to recirculate nitrogen at high load.

23) A fuel cell power plant according to claim 22, wherein both said active and said static recirculation devices are regulated with respect to a signal from at least one sensor of hydrogen or nitrogen concentration in the region of said stack.

24) A fuel cell power plant according to any preceding claim, wherein said purge device is located in said recirculation loop, upstream in that loop of a hydrogen recirculation device.

25) A vehicle including a fuel cell power plant comprising a fuel cell stack configured to be fed with a regulated supply of pure or quasi-pure hydrogen through a pressure regulator device and having a recirculation loop that is configured for hydrogen recovery from downstream of said fuel cell stack and for the provision of said recovered hydrogen into said hydrogen feed supply, said power plant further comprising a purge device that is located downstream of said fuel cell stack, said purge device being under the control of a controller and being configured to implement a substantially continuous purge of nitrogen gas from said power plant, characterized in that said controller is configured to control said purge device to implement said substantially continuous purge operation in such a manner as to regulate the level of nitrogen at the entrance to, and/or exit of, said anode so as maintain a predetermined concentration of nitrogen at said anode, and in that said nitrogen concentration is determined as a function of the electrical load demanded of said fuel cell power plant.

26) A method of operating a fuel cell power plant, including: a) feeding a fuel cell stack with a regulated supply of pure or quasi-pure hydrogen through a pressure regulator device; b) recovering hydrogen from downstream of said fuel cell stack and recirculating said recovered hydrogen into said hydrogen feed supply; and c) purging nitrogen gas substantially continuously from said fuel cell power plant using a purge device, the method being characterized in determining a load on said fuel cell power plant and regulating the volume of nitrogen purged in proportion to said load, thereby regulating the amount of hydrogen recirculated also in proportion to said load.