US 20020009623 A1
Apparatus and methods of ceasing operation of an electric power generating system improve the cold starting capability of the system. The system comprises a fuel cell stack connectable to an external circuit for supplying electric current to the external circuit. The stack comprises at least one solid polymer fuel cell, and the system further comprises a fuel passage for directing a fuel stream through the stack and an oxidant passage for directing an oxidant stream through the stack, a sensor assembly connected to the stack for monitoring a parameter indicative of stack performance, a controller for controlling at least one operating parameter of the stack, and a control system communicative with the sensor assembly and operating parameter controller. The method comprises adjusting at least one fuel cell stack operating parameter to cause the stack to operate under a drying condition that causes a net outflux of water from the stack, operating the stack under the drying condition until the water content in the stack has been reduced, and interrupting supply of electric current from the stack to the external circuit.
1. A method of ceasing operation of an electric power generating system comprising a fuel cell stack connectable to an external circuit for supplying electric current to said external circuit, said stack comprising at least one solid polymer fuel cell, said system further comprising a fuel passage for directing a fuel stream through said stack and an oxidant passage for directing an oxidant stream through said stack, said method comprising in sequential order:
(a) adjusting at least one fuel cell stack operating parameter to cause said stack to operate under a drying condition that causes a net outflux of water from said stack;
(b) operating said stack under said drying condition until the water content in said stack has been reduced; and
(c) interrupting supply of electric current from said stack to said external circuit.
2. The method of
3. The method of
4. The method of
5. The method of
6. The method of
7. The method of
8. The method of
9. The method of
10. The method of
11. The method of
12. The method of
13. The method of
14. A method of ceasing operation of an electric power generating system comprising a fuel cell stack connectable to an external circuit for supplying electric current to said external circuit, and said stack comprising at least one solid polymer fuel cell, said system further comprising a fuel passage for directing a fuel stream through said stack and an oxidant passage for directing an oxidant stream through said stack, said method comprising in sequential order:
(a) interrupting the supply of electric current from said fuel cell stack to said external circuit;
(b) adjusting at least one of stack temperature, oxidant or fuel stream flow rate, or oxidant or fuel stream pressure to cause a drying condition with a net outflux of water from said stack; and
(c) flowing at least one of fuel or oxidant streams through said stack under said drying condition until the water content in said stack has been reduced.
15. The method of
16. The method of
17. The method of
18. The method of
19. The method of
20. The method of
21. The method of
22. The method of
23. An electric power generation system comprising:
(a) a fuel cell stack connectable to an external circuit for supplying electric current to said external circuit, said stack comprising at least one solid polymer fuel cell and fluid flow passages through said stack;
(b) a sensor assembly connected to said stack for monitoring at least one parameter indicative of stack performance;
(c) a controller for controlling at least one operating parameter of said stack; and
(d) a control system communicative with said sensor assembly and said operating parameter controller, such that upon receipt of a shut down instruction by said control system, said operating parameter controller is operable to adjust at least one stack operating parameter such that said stack operates in a drying condition that causes a net outflux of water from said stack, until the water content in said stack has been reduced.
24. The electric power generation system of
25. The electric power generation system of
26. The electric power generation system of
27. The electric power generation system of
28. The electric power generation system of
29. The electric power generation system of
30. The electric power generation system of
31. The electric power generation system of
32. The electric power generation system of
33. The electric power generation system of
34. The electric power generation system of
35. The electric power generation system of
36. The electric power generation system of
37. The electric power generation system of
38. The electric power generation system of
39. The electric power generation system of
40. The electric power generation system of
41. The electric power generation system of
42. The electric power generation system of
43. The electric power generation system of
 This application a continuation-in-part of U.S. patent application Ser. No. 09/406,318, entitled “Methods for Improving the Cold Starting Capability of an Electrochemical Fuel Cell”. The '318 application is, in turn, a continuation-in-part of U.S. patent application Ser. No. 09/138,625 filed Aug. 24, 1998, entitled “Method and Apparatus for Commencing Operation of a Fuel Cell Electric Power Generation System Below the Freezing Temperature of Water”. The '625 application is, in turn, a continuation of U.S. patent application Ser. No. 08/659,921 filed Jun. 7, 1996, now U.S. Pat. No. 5,798,186 issued Aug. 25, 1998, also entitled “Method and Apparatus for Commencing Operation of a Fuel Cell Electric Power Generation System Below the Freezing Temperature of Water”. The '318, '625 and '921 applications are each hereby incorporated by reference in their entirety.
 The present invention relates to techniques to improve the cold starting capabilities of an electric power generating system comprising a solid polymer fuel cell, and in particular relates to methods and apparatus for reducing water content in the fuel cell when the stack is shut down.
 Electrochemical fuel cells convert fuel and oxidant to electricity and reaction product. Solid polymer electrochemical fuel cells generally employ a membrane electrode assembly (“MEA”) which comprises an ion exchange membrane or solid polymer electrolyte disposed between two electrodes typically comprising a layer of porous, electrically conductive sheet material, such as carbon fiber paper or carbon cloth. The MEA contains a layer of catalyst, typically in the form of finely comminuted platinum, at each membrane/electrode interface to induce the desired electrochemical reaction. In operation the electrodes are electrically coupled to provide a circuit for conducting electrons between the electrodes through an external circuit.
 At the anode, the fuel stream moves through the porous anode substrate and is oxidized at the anode electrocatalyst layer. At the cathode, the oxidant stream moves through the porous cathode substrate and is reduced at the cathode electrocatalyst layer to form a reaction product.
 In fuel cells employing hydrogen as the fuel and oxygen-containing air (or substantially pure oxygen) as the oxidant, the catalyzed reaction at the anode produces hydrogen cations (protons) from the fuel supply. The ion exchange membrane facilitates the migration of protons from the anode to the cathode. In addition to conducting protons, the membrane isolates the hydrogen-containing fuel stream from the oxygen-containing oxidant stream. At the cathode electrocatalyst layer, oxygen reacts with the protons that have crossed the membrane to form water as the reaction product. The anode and cathode reactions in hydrogen/oxygen fuel cells are shown in the following equations:
Anode reaction: H2→2H++2e −
Cathode reaction: ½O2+2H++2e −→H2O
 In typical fuel cells, the MEA is disposed between two electrically conductive fluid flow field plates or separator plates. Fluid flow field plates have at least one flow passage formed in at least one of the major planar surfaces thereof. The flow passages direct the fuel and oxidant to the respective electrodes, namely, the anode on the fuel side and the cathode on the oxidant side. The fluid flow field plates act as current collectors, provide support for the electrodes, provide access channels for the fuel and oxidant to the respective anode and cathode surfaces, and provide channels for the removal of reaction products, such as water, formed during operation of the cell. Separator plates typically do not have flow passages formed in the surfaces thereof, but are used in combination with an adjacent layer of material which provides access passages for the fuel and oxidant to the respective anode and cathode electrocatalyst, and provides passages for the removal of reaction products. The preferred operating temperature range for solid polymer fuel cells is typically 50° C. to 120° C., most typically about 75° C. to 85° C.
 Two or more fuel cells can be electrically connected together in series to increase the overall power output of the assembly. In series arrangements, one side of a given fluid flow field or separator plate can serve as an anode plate for one cell and the other side of the fluid flow field or separator plate can serve as the cathode plate for the adjacent cell. Such a multiple fuel cell arrangement is referred to as a fuel cell stack, and is usually held together in its assembled state by tie rods and end plates. The stack typically includes inlet ports and manifolds for directing the fluid fuel stream (such as substantially pure hydrogen, methanol reformate or natural gas reformate, or a methanol-containing stream in a direct methanol fuel cell) and the fluid oxidant stream (such as substantially pure oxygen, oxygen-containing air or oxygen in a carrier gas such as nitrogen) to the individual fuel cell reactant flow passages. The stack also commonly includes an inlet port and manifold for directing a coolant fluid stream, typically water, to interior passages within the stack to absorb heat generated by the fuel cell during operation. The stack also generally includes exhaust manifolds and outlet ports for expelling the depleted reactant streams and the reaction products such as water, as well as an exhaust manifold and outlet port for the coolant stream exiting the stack. In a power generation system various fuel, oxidant and coolant conduits carry these fluid streams to and from the fuel cell stack.
 When an electrical load (comprising one or more load elements) is placed in an electrical circuit connecting the stack terminals, fuel and oxidant are consumed in direct proportion to the electrical current drawn by the load, which will vary with the ohmic resistance of the load.
 Solid polymer fuel cells generally employ perfluorosulfonic ion exchange membranes, such as those sold by DuPont under its NAFION® trade designation. When employing such membranes, the fuel and oxidant reactant streams are typically humidified before they are introduced to solid polymer fuel cells so as to facilitate proton transport through the ion exchange membrane and to avoid drying (and damaging) the membrane separating the anode and cathode of each cell.
 Each reactant stream exiting the fuel cell stack generally contains water. The outlet fuel stream from the anodes generally contains water from the incoming fuel stream plus any product water drawn across the membrane from the cathode. The outlet oxidant stream from the cathodes generally contains water added to humidify the incoming oxidant stream plus product water formed at the cathode.
 In some fuel cell applications, such as, for example, motive applications, it may be necessary or desirable to commence operation of a solid polymer electrolyte fuel cell stack when the stack core temperature is below the freezing temperature of water. As used herein, the freezing temperature of water means the freezing temperature of free water, that is, 0° C. at 1 atmosphere. It may also be necessary or desirable when ceasing operation of the solid polymer fuel cell stack to improve the cold start capability and freeze tolerance of the stack by reducing the amount of water remaining within the fuel, oxidant and coolant passages of the stack. Upon freezing, water remaining within stack passages will expand and potentially damage structures within the stack such as, for example, the membrane/electrocatalyst interface, the reactant passageways, conduits and seals, as well as the porous electrode substrate material.
 If there is an expectation that a solid polymer fuel cell stack will be subjected to cold temperatures, especially temperatures below the freezing temperature of water, one or more special start-up and shutdown techniques and associated apparatus may be used. These techniques may improve the cold start capability and freeze tolerance of the stack, and improve the subsequent fuel cell performance. A measure of electrochemical fuel cell performance is the voltage output from the cell for a given current density. Higher performance is associated with a higher voltage output for a given current density or higher current density for a given voltage output.
 Water may be introduced into a fuel cell through either or both of the oxidant and fuel supply streams to the fuel cell. Water is produced in fuel cell by the electrochemical reaction at the cathode. Water may escape the fuel cell via one or both of the oxidant and fuel exhaust streams leaving the fuel cell. If the theoretical maximum water flux exiting the fuel cell in vapour form (for example, via the outlet reactant streams) is greater than the water flux introduced and produced, then the fuel cell will operate under a drying condition that causes the fuel cell to dehydrate. In this case there is a “net outflux” of water. Conversely, if the amount of water introduced and produced exceeds the theoretical maximum amount of water exiting the fuel cell in vapour form, a wetting condition exists that causes water to temporarily accumulate in the cell (a net influx of water) until a steady state is achieved. A water balance exists when the net influx and theoretical outflux of water in vapour form is zero.
 According to one aspect of the invention, the freeze tolerance and cold start-up capability of an electric power generating system is improved by removing at least some of the excess water from fuel cells in the system before the system falls below the freezing temperature of water. Water removal is carried out during a system shutdown procedure. The system includes a fuel cell stack having at least one solid polymer fuel cell. Each fuel cell has a membrane electrode assembly (MEA) comprising an anode, a cathode, and an ion exchange membrane interposed between the anode and the cathode. The stack is connectable to supply electric current to an external circuit. The system also includes fuel and oxidant passages that direct respective fuel and oxidant streams through the stack, a controller for controlling certain operating parameters of the stack, a sensor assembly for monitoring at least one parameter indicative of the cell performance of the stack, and a control system that is communicative with the sensor assembly and the operating parameter controller.
 Upon receipt of a shut down instruction by the control system, the operating parameter controller is operable to adjust at least one stack operating parameter such that the stack operates under a drying condition that causes a net outflux of water from the fuel cells. The control system continues the drying operation until the water content in the stack has been reduced.
 The sensor assembly may include sensors for measuring stack voltage, stack resistance, stack temperature, fuel and oxidant pressures, fuel and oxidant relative humidities, and coolant inlet and outlet temperatures. The controller may include apparatus to control fuel and oxidant relative humidities, fuel and oxidant stoichiometries, fuel and oxidant pressures, and stack temperatures. An example of the apparatus to control oxidant stoichiometry is a compressor that is connected to the oxidant inlet passage. An example of the pressure control apparatus are pressure regulators on each of the fuel and oxidant passages. An example of the stack temperature control apparatus is a coolant system having a coolant passage through the stack and a coolant pump that is communicative with the control system. The control system may be a microcontroller or like device.
 If the fuel and/or oxidant supply streams are humidified during normal operation, their relative humidities can be decreased by reducing or stopping the transfer of water to the reactant supply streams. For example, a humidifier bypass system comprises a bypass conduit that allows a reactant supply stream to be directed to the stack in fluid isolation from a humidifier that is used during normal operation to humidify the reactant supply streams. Upon receipt of a shut down instruction by the control system, the humidifier bypass system directs at least some of the oxidant or fuel streams through an associated bypass conduit during the shut down procedure, and discontinues transmission through the bypass conduit when the water content in the stack has been reduced.
 The bypass conduit includes an inlet end that is connected to one of the reactant stream passages at a location upstream of the humidifier, and an outlet end that is connected to the same reactant stream passage at a location downstream of the humidifier. Alternatively, the bypass conduit inlet end can be connected directly to the reactant supply. The humidifier bypass system may include a bypass inlet valve on one of the reactant passages at a location upstream of the humidifier, and a bypass outlet valve on the same reactant passage at a location downstream of the humidifier; the bypass conduit connects the bypass inlet valve to the bypass outlet valve such that reactant fluid can be directed to the stack in fluid isolation from the humidifier.
 According to another aspect of the invention, a method is provided for ceasing operation of an electric power generating system that comprises a fuel cell stack connectable to an external circuit for supplying electric current to said external circuit, and at least one solid polymer fuel cell. The system further comprises a fuel passage for directing a fuel stream through said stack and an oxidant passage for directing an oxidant stream through said stack. The method comprises in sequential order:
 (a) after receiving instructions to shut down the system, at least one fuel cell stack operating parameter is adjusted so that the stack operates under a drying condition that causes a net outflux of water from the stack;
 (b) the stack is then operated under the drying condition until the water content in the stack has been reduced; and
 (c) the supply of electric current from the stack to the external circuit is interrupted.
 The amount of water removed should be enough to remove at least some of the excess water from the fuel cell, but should not be so much as to dry out the membrane. If the membrane water level falls below its critical moisture level (the minimum amount of water needed for the membrane to be adequately ionically conductive), a drop in fuel cell performance will occur, which is observable as a drop in cell output voltage, or an increase in cell resistance. The stack voltage or resistance is monitored during the drying operation by stack voltage and resistance sensors. When the stack performance falls below a selected threshold level (that is, the voltage drops below a threshold value, or the resistance increases to a threshold value), the drying operation is stopped. The threshold values correspond to or above the critical membrane moisture level.
 The water flux of each fuel cell can be controlled by controlling certain operating parameters of the system. These parameters include the fuel and oxidant relative humidities, fuel and oxidant stoichiometries, fuel and oxidant pressures, and the stack temperature. One or more of these parameters can be adjusted so that the stack operates under a drying condition.
 Where one or both of the oxidant and fuel supply streams are humidified during normal operation, the oxidant and fuel relative humidities can be reduced during shutdown by reducing the degree of humidification to the oxidant and/or fuel supply streams. For certain humidifiers, the amount of water transferred to the supply stream passing through the stack can be reduced or stopped altogether. Alternatively, the humidification reduction can be performed by directing at least some of the oxidant and/or fuel streams to the stack to the stack in fluid isolation from the humidifier.
 One or more additional operating parameters may be adjusted to operate the stack under a drying condition. For example, the stack temperature can be increased during the shut down procedure by stopping coolant circulation through the stack. Alternatively or in addition, the fuel and/or oxidant supply pressures can be decreased during the shut down procedure. Alternatively or in addition, the fuel and/or oxidant stoichiometries can be increased during the shut down procedure. Sensors can be provided to measure the fuel pressure, oxidant pressure, fuel relative humidity, oxidant relative humidity, coolant inlet and outlet temperatures, and stack temperature.
 The stack can be disconnected from the external circuit before or after the drying operation. If the latter case, then power to various components in the system (for example, compressor, sensors) are provided by an auxiliary power source such as a battery. As the electrochemical reaction stops after the stack is disconnected, the operating parameters further include air and oxidant flow rates and excludes air and oxidant stoichiometries.
FIG. 1 illustrates a typical fuel cell 10. Fuel cell 10 includes a membrane electrode assembly 12 interposed between anode flow field plate 14 and cathode flow field plate 16. Membrane electrode assembly (MEA) 12 comprises an ion exchange membrane 20 interposed between two electrodes, namely, anode 21 and cathode 22. In conventional fuel cells, anode 21 and cathode 22 comprise a substrate of porous electrically conductive sheet material 23 and 24, respectively, for example, carbon fiber paper or carbon cloth. Each substrate has a thin layer of electrocatalyst 25 and 26, respectively, disposed on one surface thereof at the interface with membrane 20 to render each electrode electrochemically active.
 As further shown in FIG. 1, anode flow field plate 14 has at least one fuel flow channel 14 a formed in its surface facing anode 21. Similarly, cathode separator plate 16 has at least one oxidant flow channel 16 a formed in its surface facing cathode 22. When assembled against the cooperating surfaces of electrodes 21 and 22, channels 14 a and 16 a form the reactant flow field passages for the fuel and oxidant, respectively. The flow field plates 14, 16 are electrically conductive. Coolant channels (not shown) may also be formed on the flow field plate 14, 16 (typically on the other side of the surface having the reactant flow channels) to provide passages for flow of a coolant therethrough.
 Turning now to FIG. 2, a fuel cell stack 100 includes a plurality of fuel cell assemblies, a series of which is designated as 111 in FIG. 2. Each of the fuel cell assemblies includes a membrane electrode assembly 112 interposed between a pair of fluid flow field plates 114, 116. Fuel cell stack 100 also includes a first end plate 130 and a second end plate 140.
 Plate 130 includes fluid inlet ports 132, 134, 136 for introducing fluid fuel, oxidant and coolant streams, respectively, to the stack 100. Plate 140 includes fluid outlet ports 142, 144, 146 for exhausting fluid fuel, oxidant and coolant streams, respectively, from the stack 100. The fluid outlet ports 142, 144, 146 are fluidly connected to the corresponding fluid inlet ports 132, 134, 136 via passages within the stack 100.
 The fuel cell assemblies have a series of openings formed therein, which cooperate with corresponding openings in adjacent assemblies to form fluid manifolds 152, 154, 156, 162, 164, 166 within the stack 100. The fluid manifolds are each circumscribed by a sealant material or gasket. In addition, a peripheral seal at the exterior perimeter of each fuel cell fluidly isolates the interior, electrochemically active portion of the fuel cell from the external environment.
 A fuel stream entering the stack 100 via fuel inlet port 132 is directed to the individual fuel flow field plates via manifold 152. After passing through the fuel flow field plate channels, the fuel stream is collected in manifold 162 and exhausted from the stack via fuel outlet port 142. Similarly, an oxidant stream entering the stack 100 via oxidant inlet port 134 is directed to individual oxidant flow field plates via manifold 154. After passing through the oxidant flow field plate channels, the oxidant stream is collected in manifold 164 and exhausted from the stack via oxidant outlet port 144. A fluid coolant (typically water) introduced via coolant inlet port 136 is directed to coolant channels (not shown) in each flow field plate, or to coolant plate assemblies (not shown) in the stack 100 via manifold 156. The coolant stream is collected in manifold 166 and exhausted from stack 100 via coolant outlet port 146. Coolant manifolds 156, 166 may be fitted with a compliant mechanism (not shown), such as tube cushions or inserts made of closed cell foam, to accommodate the expansion of freezing water. Tie rods 170 extend between end plates 130 and 140 to compress and secure stack 100 in its assembled state with fastening nuts 172 disposed at opposite ends of each tie rod 170, and disc springs 174 interposed between the fastening nuts 172 and end plates 130, 140.
 Each fuel cell 10 in stack 100 can operate satisfactorily only when sufficient water is provided to keep membrane 20 wet and ionically conductive. Water may be introduced in the reactant streams and is produced in the electrochemical reaction at the cathode 22. If the theoretical maximum quantity of water escaping from fuel cell 10 in vapor form via the outlet reactant streams (assuming exhaust gases are saturated with water vapor) is greater than the water quantity introduced and produced, MEA dehydration will tend to occur. During operation, it is important to provide adequate humidification to the MEA, so as to avoid dehydrating the membrane. Mathematically this condition is expressed as:
N w,o,in +N w,f,in +N w,p ≧N w,o,out +N w,f,out (1)
 Nw,o,in is the inlet oxidant molar water flow rate;
 Nw,f,in is the inlet fuel molar water flow rate;
 Nw,p is the produced water molar flow rate;
 Nw,o,out is outlet oxidant molar water flow rate;
 Nw,f,out is the outlet fuel molar water flow rate.
 Equation 1 terms are given by:
 A is the geometric active surface area;
 i is the current density;
 F is the Faraday constant;
 pf,in is the inlet fuel pressure;
 pf,out is the outlet fuel pressure;
 po,in is the inlet oxidant pressure;
 po,out is the outlet oxidant pressure;
 ps,f,in is the inlet fuel water vapor saturation pressure;
 ps,f,out is the outlet fuel water vapor saturation pressure;
 ps,o,in is the inlet oxidant water vapor saturation pressure;
 ps,o,out is the outlet oxidant water vapor saturation pressure;
 RHf,in is the inlet fuel relative humidity;
 RHf,out is the outlet fuel relative humidity;
 RHo,in is the inlet oxidant relative humidity;
 RHo,out is the outlet oxidant relative humidity;
 νf is the fuel stoichiometry;
 νo is the oxidant stoichiometry;
 φf is the hydrogen volume fraction in the dry fuel;
 φo is the oxygen volume fraction in the dry oxidant.
 Equations 5 and 6 can be somewhat simplified when it is realized that for outlet relative humidities lower than 100%, the MEA will be subjected to dehydrating conditions. Therefore, outlet relative humidities of 100% represent a limiting case defining a boundary between drying and wetting conditions (assuming that the exhaust gases are saturated with water vapor). By introducing equations (2) to (6) in equation (1) and simplifying with RHo,out=1 and RHf,out=1, the following equation is obtained:
 The water vapor saturation pressure is computed using a temperature dependent empirical equation:
log p s=−2.1794+0.02953T−9.1837×10−5 T 2+1.4454×10−7 T 3 (8)
 The water vapor saturation pressure at each of the fuel and oxidant inlets and outlets can thus be determined by measuring the temperature at each location Tf,in, Tf,out, To,in, To,out. Generally, these temperatures are closely related to the inlet and outlet coolant temperatures (Tc,in, Tc,out) which in practice are easier to accurately measure due to the larger heat capacity of the coolant.
 Each of the variables in equation (7) represent an operating parameter of fuel cell stack 100. As written, equation (7) defines an operating condition that produces a wetting condition (net influx of water into cell) or water balance (equality in equation (7)). A “water balance” is defined as a balance between water influx and water outflux when the outflux is calculated with the assumption that the exhaust is saturated with water vapor. If the equation was rewritten so that the left side is less than the right side, the equation defines an operating condition that produces a drying condition (net outflux of water).
 While it is generally desirable to operate fuel cell 10 under a wetting condition such that membrane 20 is properly hydrated at all times, excess accumulated water in MEA 12 is not desired if the stack 100 is to be cold started at or below 0° C. after the stack has been exposed to freezing conditions for an extended period of time. “Excess water” is hereby defined as the amount of water exceeding the minimum required to keep the membrane adequately ionically conductive (“critical membrane moisture level”). While some water in MEA 12 is needed to keep membrane 20 moist, excess water in MEA 12 will accumulate in pores of substrates 23, 24 and in flow channels 14 a, 16 a and will eventually freeze when the stack is exposed to temperatures below 0° C. for prolonged periods. If the stack is started before the MEA 12 has a chance to thaw, ice in pores of substrates pores 23, 24 may block or impede the flow of reactant that must pass through substrate 23, 24 and to membrane 20 in order for the electrochemical reaction to proceed. Furthermore, ice accumulation may cause mechanical stresses inside fuel cell 10 that can cause damage to stack 100. It is theorized that reducing the quantity of excess water accumulated in flow channels 14 a, 16 a and in the substrate pores of MEA 12 before stack 100 freezes, will reduce reactant flow blockage caused by ice, and thus reduce the time required for stack 100 to reach a nominal operating state after a cold start-up from below 0° C. or improve cell performance at sub 0° C. temperatures. This can be achieved by operating each fuel cell 10 in stack 100 under a drying condition for a period of time that is sufficient to remove at least some excess water from MEA 12 but not excessively dry out membrane 20.
 As shown in equation (7) a number of operating parameters can be adjusted to change the operating condition of fuel cell 10, including, oxidant and fuel stoichiometries, compositions (that is, volume fraction in reactant stream), relative humidities, pressures, temperatures, and relative flow configurations (for example, concurrent and counter-flow operation). One or more of these parameters can be adjusted so that fuel cell operation is changed from a wetting condition to a drying condition or to a water balance.
 A series of tests were performed to verify the MEA water flux equations (1) through (7) set out above. All tests were performed using a Ballard Mk 513 single cell having a catalyst loading of 0.3 mg Pt/cm2, an N112 Nafion® membrane, and Toray CFP TGP-H-90 electrode substrates, and under the following common operating parameters: 80° C. coolant outlet, a temperature gradient of +10° C. (temperature difference between inlet and outlet coolant temperatures) at a current density of 1 A/cm2, air/methanol reformate (63.5% H2), 4% air bleed, 2.5 bara fuel pressure and 100% fuel inlet relative humidity (RH). The air inlet pressure, oxidant/fuel stoichiometries, and nominal current densities differed between each test. In each test, the fuel cell was first operated under a wetting condition for a period of time sufficient for the fuel cell to produce a steady state voltage. Then, the air inlet relative humidity of the fuel cell was reduced from 100% to 0% and the performance of the fuel cell was monitored by measuring the fuel cell resistance and voltage.
FIG. 3 is a dimensionless representation of equation (7). The x and y axes represent the right hand and left hand sides of equation (7) respectively and the dashed line indicates an equality in equation (7) (water balance). The dashed line therefore separates the graph into a wetting region (y>x), and a drying region (x>y). The operating conditions of the fuel cell in each of the three test runs as theoretically derived from equations (1) through (7) are plotted in FIG. 3 as triangles, squares and circles, respectively. Filled symbols indicate a test being run at 100% air inlet relative humidity and unfilled symbols indicate a test being run at 0% air inlet relative humidity. It can be seen that the change in relative humidity for each of the three test runs shifted the fuel cell operation from the wetting region to either another location in the wetting region closer to water balance (test 1), or into the drying region (tests 2 and 3). The fuel cell was then operated under each of the three test conditions; in each test, the air inlet relative humidity was switched from 100% to 0% at Time=0. The cell responses to the change in relative humidity were recorded by measuring both MEA voltage and resistance and are plotted in FIG. 4. Cell performance is expected to drop as the membrane dries out; an increase in cell resistance and a decrease in cell voltage likely indicates that the fuel cell is being dehydrated.
 The first test (illustrated as triangles in FIGS. 3 and 4) was conducted at 0.542 A/cm2 current density, 2.6 bara oxidant pressure, and 1.5/1.3 oxidant/fuel stoichiometry. The first test was designed to shift the fuel cell from a point in the wetting region to another point in the wetting region that is closer to the water balance (dashed line in FIG. 3) upon the change in inlet RH (at Time=0). A small change in cell voltage and resistance (0.06 mΩ and 15 mV respectively) was found after the oxidant relative humidity was reduced from 100% to 0% at T=0. Since the cell was in theory still operating within the wetting region after the RH was changed, no significant reduction in cell performance was expected due to membrane drying. However, the operating points plotted in FIG. 3 have uncertainties attached to them, and it is possible that the first test run was in fact operating just within the drying region or that the cell performance was affected by mass transfer or other effects.
 The second test (illustrated by squares in FIGS. 3 and 4) was conducted at 0.312 A/cm2 current density, 1.5 bara oxidant pressure, and 1.5/1.3 oxidant/fuel stoichiometry. The second test was designed to shift the fuel cell from the wetting region to just inside the drying region. As expected, the MEA resistance significantly increased and the MEA voltage significantly decreased (0.25 mΩ and 68 mV respectively) then appeared to reach a steady state after about 4 hours. According to the water balance equations, this steady state is predicted to be only apparent, and eventually, the MEA should continue to dehydrate and eventually fail.
 The third test (illustrated by circles in FIGS. 4 and 5) was conducted at 0.021 A/cm2 current density, 1.1 bara oxidant pressure, and 5/2 oxidant/fuel stoichiometry. As shown in FIG. 3, the operating condition of the fuel cell after the relative humidity was reduced to 0% is deeper inside the drying region than the first two test cases, and thus, a greater drying was expected. This expectation was confirmed, as cell performance was found to drop faster and by a greater magnitude than in the first two tests.
 If the fuel cell is operated under a drying condition to remove excess water therein, the drying operation should be stopped before the membrane water level falls below its critical moisture level. As the membrane dries, and especially after the membrane water level falls below its critical moisture level, the internal fuel cell resistance increases and the voltage output decreases significantly. To ensure that the drying operation does not cause the membrane to fall below the critical moisture level, the fuel cell resistance and voltage are preferably monitored during the drying operation. The drying operation is preferably stopped once the resistance has increased to or above a threshold level (or the voltage has decreased below a threshold level).
 This threshold level can be determined empirically as follows. First, a fuel cell (or stack) is operated normally (under a wetting condition) and then under a drying condition and its resistance (and/or voltage)/time curve is determined. The fuel cell or stack is then frozen and restarted at a sub 0° C. temperature under a normal (wetting) operating condition, and the initial performance (before the stack temperature exceeds 0° C.) of the fuel cell or stack is measured. If there is a degradation in initial performance, it can be concluded that the membrane was dried beyond its critical moisture level, and that the drying time has to be shortened (or the rate of drying reduced). Progressively shorter periods of drying times can be tested until a drying time (and corresponding resistance) is found that does not dry out the membrane such that the initial cold start-up performance is degraded. With enough empirical testing, a database can be compiled for appropriate drying times and rates for various operating conditions.
 An example of a resistance/time curve is shown in FIG. 5. The resistance of a Ballard fuel cell stack (10 cell) was monitored during a drying operation. The stack was initially operated at steady state producing 300 A with an air/fuel stoichiometry ratio of 1.8/1.2 and at a stack temperature of 70° C. The inlet oxidant and fuel streams were humidified by passing same through a humidifier upstream of the stack. At time=0, humidification of the oxidant and fuel streams was stopped and the external load was disconnected from the stack. At about 70° C., a drying operation was then carried out in which the unhumidified oxidant and fuel streams continued to flow through the stacks at 89/25 slpm at 0.6 barg for 120 seconds. A relatively linear but small increase in resistance from about 3 mΩ to about 5 mΩ was observed after 60 seconds; a steeper increase in slope was observed at around 90 seconds and continued in a generally linear fashion until the drying operation was stopped; the resistance measured at the end of 120 seconds was 12 mΩ.
 A series of shutdown and cold start tests was also performed on the stack, the resistance of the stack after each drying operation was measured. The stack was initially operated at steady state producing 300 A with an air/fuel stoichiometry ratio of 1.8/1.2 and at a stack temperature of 70° C. The inlet oxidant and fuel streams were humidified by passing same through a humidifier upstream of the stack. At time=0, humidification of the oxidant and fuel streams was stopped and the external load was disconnected from the stack. At about 70° C., a drying operation was then carried out in which the unhumidified oxidant and fuel streams continued to flow through the stacks at a fuel/air rate of 25/89 slpm (for 10 cells) at 0.6 barg. A drying operation was applied for each test run for different time lengths and the corresponding stack resistance was measured at the end of the drying operation, as follows: 12 mΩ (test 1), 7.2 mΩ (test 2), 6.23 mΩ (test 3), 5.2 mΩ (test 4), and 5.99 mΩ (test 5). The stack was then allowed to cool to about 20° C. and was subjected to a second drying operation of unhumidified fuel and air flow at a fuel/air rate of 25/89 slpm (10 cell) and at 0.6 barg for about 1 minute.
 The stack was then cooled to about −10° C. and held at that temperature. Thereafter, the stack was started at about −10° C. and the resistance was measured for each test run as follows: 16 mΩ (test 1), 10 mΩ (test 2), 9.23 mΩ (test 3), 6.2 mΩ (test 4), and 7.89 mΩ (test 5). Current was varied in steps of 5 A between a range of 5 and 50 A for about 10 seconds per step and the cell voltage at each current was measured. It was observed that the higher the measured stack resistance (both at shutdown and at start-up), the lower the measured cell voltage, that is, the worse the initial cold start performance, that is, the performance of the cell below 0° C. It is theorized that the performance losses correlate with the degree of MEA dryness prior to freezing, which is dependent on the parameters of the drying operation during shut down.
FIG. 6 is a schematic diagram of a fuel cell electric power generation system 200 comprising a fuel cell stack 210 according to one embodiment of the invention. Fuel cell stack 210 includes negative and positive bus plates 212, 214, respectively, to which an external circuit comprising a variable load 216 is electrically connectable by closing switch 218. System 200 includes a fuel (hydrogen) circuit, an oxidant (air) circuit, and a coolant water circuit. The reactant and coolant streams are circulated in the system in various conduits illustrated schematically in FIG. 6.
 During normal operation, a hydrogen supply 220 is humidified in humidifier 270 then delivered to stack 210 via hydrogen conduit 261. Flow through conduit 261 is controlled by hydrogen pressure regulator 221. Hydrogen delivery pressure is measured by pressure sensor 271. If humidification of the hydrogen stream is not desired, hydrogen flow may be bypassed around humidifier 270 through three-way valve 272 connected to conduit 261 upstream of humidifier 270, through hydrogen bypass conduit 274 connected to valve 272, and through a three-way bypass valve 276 connected to conduit 261 downstream of humidifier 270. Flow through bypass conduit 274 is controlled by hydrogen pressure regulator 278. Alternatively, and for certain types of humidifiers, the humidifier may be bypassed by reducing or stopping the transfer of water to a reactant stream passing through the humidifier.
 Water in the hydrogen exhaust stream exiting stack 210 is accumulated in a knock drum 222, which can be drained by opening valve 223. Unreacted hydrogen is recirculated to stack 210 by a pump 224 in recirculation loop 225. The relative humidity of the hydrogen exhaust stream is measurable by relative humidity sensor 280.
 During normal operation, air (oxidant) is humidified in humidifier 270 then delivered to stack 210 via oxidant humidification conduit 262. Conduit 262 has an inlet end connectable to a compressor 230 and an outlet end connected to fuel cell stack 210. Flow through humidification conduit 262 is controlled by oxidant pressure regulator 231. Oxidant flow rate is measured by mass flow sensor 282 and oxidant pressure is measured by pressure sensor 284. If humidification of the oxidant stream is not desired, oxidant flow may be bypassed around humidifier 270 through a three-way valve 288 connected to conduit 262 upstream of humidifier 270, through oxidant bypass conduit 286 connected to valve 288, and through a three-way bypass valve 266 connected to conduit 286 downstream of humidifier 270. Flow through bypass conduit 286 is controlled by oxidant pressure regulator 290.
 Water in the oxidant exhaust stream exiting stack 210 is accumulated in reservoir 232, which can be drained by opening valve 233, and the air stream is vented from the system via valve 234. The relative humidity of the air exhaust stream is measured by relative humidity sensor 291.
 In coolant water loop 240, water is pumped from reservoir 232 and circulated through stack 210 by pump 241. The temperature of the water is adjusted in a heat exchanger 242. The coolant inlet and outlet temperatures are measured by temperature sensors 292, 294.
 The cold start capability and freeze tolerance of the system 200 can be improved by reducing the amount of water remaining within the flow channels 14 a and 16 a, and in the electrodes of the MEA of each fuel cell in the stack 210 upon cessation of operation and reduction of the stack core temperature to near or below the freezing temperature of water. As used herein, “freeze tolerance” refers to the ability of a fuel cell or fuel cell stack to maintain substantially the same performance after one or more freeze/thaw/cold start cycles, where the stack after being shut off is exposed to sub 0° C. temperatures for an extended period of time then is cold started below 0° C. or is thawed above 0° C. then started.
 On shutdown, the operating parameters of fuel cell stack 210 are selected so that stack 210 operates under a drying condition until the voltage drops below (or resistance increases above) a threshold level. A number of different operating parameters may be adjusted to change the operation of stack 210 from a wetting condition to a drying condition, such as air or fuel stoichiometries, temperatures, pressures, compositions, and relative humidities. A suggested shutdown sequence comprising a drying operation is as follows:
 (a) receive shutdown instructions;
 (b) turn off coolant pump 241 so that coolant flow is stopped (increases stack operating temperature);
 (c) actuate bypass valves 272, 276, 288 and 266 so that reactant supply to stack bypasses humidifier 270 (reduces the reactant inlet relative humidities);
 (d) adjust compressor operation to decrease oxidant supply pressure;
 (e) adjust fuel pressure regulator 278 to decrease fuel inlet pressure;
 (f) once stack resistance has exceeded (or the voltage has decreased below) a predetermined threshold value, shut off compressor 230 and close valves 221, 231, 278 and 290 (shuts off the fuel and oxidant supplies to the stack);
 (g) shut off hydrogen recirculation pump 224; and
 (h) open switch 218 (disconnects the stack from the external circuit).
 Steps (b) to (h) should be completed before the stack 210 overheats. Empirical testing can be performed to determine the maximum period of time for performing these steps before overheating occurs. Alternatively or in addition, the stack temperature can be monitored during the shut down operation; if the stack gets too hot, the coolant pump 241 can be reactivated.
 System 200 illustrated in FIG. 6 has a number of sensors to monitor various operating parameters during stack operation, including relative humidity sensors 280 and 291 located in the exhaust conduits downstream of the stack, reactant supply pressure sensors 271, 284, and inlet and outlet coolant temperature sensors 292 and 294. While these sensors are sufficient to carry out the drying operation (b) to (h) described above, additional sensors (not shown) are required if data for all the variables specified in the water flux equations (1) to (8) are desired, for example, mass flow sensors for the oxidant and fuel supplies, fuel and oxidant relative humidity sensors upstream of stack 100, fuel and oxidant pressure sensors downstream of the stack, stack current sensor, and oxygen and hydrogen concentration sensors. These sensors may be useful to precisely monitor the water flux in and out of the cell, so that a water management program can be carried out during stack operation to prevent excess water from building up in the stack. By carrying out such a water management program, the amount of water remaining at shutdown can be reduced, thereby reducing the need for drying operation. Such a water management program may also improve system performance and efficiency during operation, as the mass transport limitations associated with excess water accumulated in the fuel cell will be reduced.
 System 200 as illustrated in FIG. 6 is supplied with air from the compressor 230 and with pure hydrogen from a pressurized hydrogen tank 220. For greater output voltages, it is advantageous to supply fuel cells with more concentrated reactant streams and preferably with pure reactant streams (for example, pure hydrogen and oxygen reactants). This is an advantage because the presence of relatively large amounts of non-reactive components in the reactant streams can significantly increase kinetic and mass transport losses in the fuel cells. However, in certain applications it may be impractical to store and provide the desired reactants in pure form. In this connection, hydrogen may be supplied to system 200 by reforming a supply of methanol, natural gas, or the like on site or on board (not shown).
 The reformed hydrogen stream tends to contain some carbon dioxide generated as a result of the reforming operation. Air typically has a oxygen concentration of about 21%; the major component in the dilute oxidant air stream is nitrogen. Known approaches may be implemented in system 200 to increase the concentration of the reactant in the reformed fuel and/or air streams, that is, enrichment, to improve the performance of system 200. Such known approaches typically involve separating out a component from the reactant stream, including cryogenic, membrane, and pressure swing adsorption methods. In a cryogenic method, component separation is achieved by preferentially condensing a component out of a gaseous stream. In a membrane method, component separation is achieved by passing the stream over the surface of a membrane that is selectively permeable to a component in the stream. In a pressure swing adsorption (PSA) method, a gas component is separated from a gas stream by preferential adsorption onto a suitable adsorbent under pressure. A PSA apparatus (not shown) may be installed on the fuel supply conduit 261 between the fuel supply 220 and the stack 210 to provide an enriched fuel stream to stack 210. The PSA apparatus may also be installed on the oxidant supply conduit 262 between the air compressor 230 and the stack 210 to provide an enriched oxidant stream to stack 210. By controlling the degree of enrichment provided by the PSA apparatus, the fuel and oxidant concentrations can be controlled (φf, φo) to encourage the stack to operate under a drying condition during shut down.
 System 200 shown in FIG. 6 may be configured so that the oxidant and fuel stream pass through stack 210 in a concurrent flow arrangement. According to another embodiment of the invention, one of the fuel and oxidant streams may be reversed so that the oxidant and fuel streams pass through the stack 210 in a counter-flow arrangement (not shown). Such counter-flow arrangement will affect the temperature gradients in the stack 210. Temperature sensors (not shown) can be installed in the oxidant and fuel passages to measure the inlet and outlet oxidant and fuel stream temperatures, so that the effect of the temperatures on the water balance formulas can be determined. It may be desirable in embodiments of the present method, to intermittently reverse the reactant flow directions during operation or shut down. An example of apparatus and methods for reversing the relative flow directions of oxidant and fuel through a fuel cell stack is described in U.S. Pat. No. 5,935,726.
 It should be noted that the stack can be disconnected from the external circuit prior to starting a drying operation (for example, between step (a) and (b). In such case, an auxiliary power source such as a battery (not shown) is provided to power the various components in the system 200 (for example, air compressor, pumps, actuators, sensors). After the external circuit has been disconnected, Nw,p becomes 0 in equation (7) as the electrochemical reaction for all intents stops and no water is produced. Substituting dry oxidant and fuel flow rates No,in, Nf,in for oxidant and fuel stoichiometries, and equations (2) to (6) in equation (1), the following water flux equation is derived (wetting condition or water balance):
 Note that the primary difference between equations (7) and (9) is that the net water influx is reduced by elimination of the water production term Nw,p and that the reactant flow rates cannot be defined in terms of stoichiometries, since current is 0. Using equation (9), operating parameters can be determined that will cause the stack to operate under a drying condition; equation (9) can be verified by empirical testing using the same test methods that were applied to test equation (7).
 In a shutdown procedure where the external circuit is disconnected before a drying operation is performed, the voltage measured will be the open circuit voltage (or open circuit resistance if resistance is measured). Empirical testing can be performed to determine at what voltage drop (or resistance increase) should the drying operation be stopped.
 While particular elements, embodiments and applications of the present invention have been shown and described, it will be understood, of course, that the invention is not limited thereto since modifications may be made by those skilled in the art without departing from the scope of the present disclosure, particularly in light of the foregoing teachings.
FIG. 1 is an exploded side view of a typical solid polymer electrochemical fuel cell with a membrane electrode assembly interposed between two fluid flow field plates.
FIG. 2 is a perspective cut-away view of an electrochemical fuel cell stack.
FIG. 3 is a dimensionless representation of the net water flux in a fuel cell obtained under different operating conditions.
FIG. 4 is a graph showing the change in cell voltage and resistance over time for a fuel cell operated under different conditions.
FIG. 5 is a graph showing the change of resistance over time in a fuel cell stack operated under a drying condition.
FIG. 6 is a schematic diagram of a fuel cell electric power generation system incorporating a humidifier bypass purge system, actuators, and sensors that can cooperate to perform a controlled fuel cell drying operation at shutdown.