Method for individualizing an operating strategy of a fuel cell stack, control device

Abstract

The invention relates to a method for individualizing an operating strategy of a fuel cell stack, comprising a stack of several technically similar fuel cells composed of several cell components and differing in detail due to manufacturing tolerances, based on which the operating behavior of the fuel cell stack is individualized. According to the invention, to account for the individuality of the operating behavior, a predictive model of the fuel cell stack is created, individually parameterized taking into account at least one individual characteristic of the fuel cell stack based on the different manufacturing tolerances and / or the stacking sequence, fed into a system simulation, and with the aid of this, an operating strategy for the fuel cell stack is specified. The invention further relates to a control device which is designed to implement the fuel cell stack-specific operating strategy.

Description

[0001] The present invention relates to a method for individualizing an operating strategy of a fuel cell stack having the features of the preamble of claim 1. Furthermore, the invention relates to a control device. State of the art

[0002] Fuel cells consist of various components. The core of a fuel cell is the membrane-electrode assembly, which typically consists of a membrane coated on both sides with a catalyst material. Gas diffusion layers can be applied to either side of this membrane, provided they are not part of the membrane-electrode assembly. These layers are enclosed by monopolar or bipolar plates, which contain flow distribution structures for the reacting fluids and a coolant. The individual components of a fuel cell are typically either simply stacked and pressed together or connected by lamination, welding, and / or sintering.

[0003] During the production of cell components, small deviations from the nominal dimensions occur due to process-related manufacturing tolerances, for example in bore diameters, coating thicknesses, and the like. These lead to pairs of technically identical cell components actually producing different results. Consequently, fuel cells of the same nominal design exhibit different operating behavior within a tolerance range.

[0004] To scale the generated electrical energy, several fuel cells are combined into a fuel cell stack. The individual fuel cells in a fuel cell stack typically do not experience the same operating conditions. For example, cells located at the edge of a fuel cell stack experience different thermal boundary conditions than those located in the center. For space-saving reasons, the supply and discharge lines for the reacting fluids are often installed on the same side. This results in increased pressure loss for the cells located on the opposite side. As a result, these cells are supplied with a lower stoichiometry and experience a lower coolant flow.In conjunction with the individuality of each fuel cell, fuel cell stacks exhibit individual operating behavior due to the manufacturing tolerances of the cell components that determine their operating behavior, as well as the individual stacking sequence within a tolerance range. This individual operating behavior includes differences in the optimal operating point with regard to efficiency, service life, and operational stability.

[0005] Individually varying operating behavior of fuel cell stacks results in a majority of fuel cell stacks operating under suboptimal conditions when applying a generic operating strategy designed for a fictitious medium-sized fuel cell stack. This results in lower efficiency, a shorter service life, and lower operational stability than would be technically feasible. The present invention therefore addresses the problem of individualizing the operating strategy of a fuel cell stack.

[0006] To achieve this object, the method having the features of claim 1 is proposed. Advantageous further developments of the invention can be found in the subclaims. Disclosure of the invention

[0007] A method is proposed for individualizing the operating strategy of a fuel cell stack, comprising stacking a plurality of technically similar fuel cells composed of multiple cell components and differing in detail due to manufacturing tolerances, which determine the individual operating behavior of the fuel cell stack. According to the invention, to account for the individuality of the operating behavior, a predictive model of the fuel cell stack is created, individually parameterized, taking into account at least one individual characteristic of the fuel cell stack based on the different manufacturing tolerances and / or the stacking sequence, and fed into a system simulation. With the aid of this model, an operating strategy for the fuel cell stack is specified.

[0008] The predictive model can be a model of a generic fuel cell stack or can already contain individual characteristics that characterize the individual fuel cell stack. However, the essential customization step occurs in the parameterization of the model. In this process, at least one constitutive equation and / or boundary condition is defined in the model based on, preferably, a comprehensive set of characteristics, but at least one individual characteristic, of the fuel cell stack. The term "individual characteristic" encompasses any variable that can vary during the production of technically similar fuel cell stacks, fuel cells, or cell components. Some examples of individual characteristics are listed below, although this list is not exhaustive: With regard to the catalyst-coated membrane, the term "individual characteristic" includes, for example, membrane thickness, sub-gasket thickness, electrode coating thickness, coating defects, and the like. With regard to the gas diffusion layers, the term "individual characteristic" sometimes includes stiffness, thickness, electrical conductivity, permeability, or gas diffusion coefficients. With regard to the bipolar plates, the term "individual characteristic" includes, for example, channel geometry, deviation from the nominal dimension, coating quality, wetting angle, and others. Furthermore, thicknesses, widths, and mechanical properties of seals and geometric deviations of ports are also considered.

[0009] With regard to the stacking and bracing process of the individual cells to form a fuel cell stack, the term individual characteristics includes variables such as geometric deviations, stack curvatures, bracing forces and others.

[0010] By parameterizing the predictive model to account for at least one individual characteristic and using it in a system simulation, the individual deviations of the operating behavior of the fuel cell stack under consideration from the nominal operating behavior are made accessible to computational analysis methods. This provides the advantage that an actual optimal operating condition can be used to derive an operating strategy, whereas otherwise only a presumed operating condition optimal for an average fuel cell stack would be available. The result is an operating strategy that is more appropriate for the individual fuel cell stack, thus operating the individual fuel cell stack at a more efficient operating point and thus more efficient use of resources.

[0011] In a further development of the invention, it is proposed that the at least one individual characteristic of the fuel cell stack be abstracted into at least one parameter and used to parameterize the predictive model. This preferred embodiment is advantageous because it allows an individual characteristic whose significance is primarily qualitative in nature, for example, coating quality, to be given a quantitative expression in the form of a parameter. This enables its implementation in the predictive model and thus an approximation of its predicted operating behavior to reality.

[0012] Furthermore, it is proposed that the operating strategy be optimized to maintain a fuel cell stack-specific operating window to maximize efficiency with a specified service life to be achieved. This preferred embodiment is advantageous because, for most applications, a specified service life cannot be undercut, and the concept of optimal operation usually includes the exploitation of the maximum possible efficiency. The operating window to be maintained for this purpose is taken from the parameterized predictive model for each fuel cell stack. Ensuring compliance during operation by appropriately anchoring it in the individual operating strategy is advantageous for most applications.

[0013] In a further development of the invention, it is proposed that the operating strategy be optimized to comply with operating limits in order to ensure operational stability. Operational stability may include the definition and maintenance of a minimum permissible stoichiometry and / or maximum permissible holding times if these are undershot, the definition and maintenance of a minimum inlet humidity of the supplied air as a function, for example, of the process temperature, and the like. Ensuring operational stability in this way is essential for most applications, given that fuel cell systems rarely operate exclusively at the design operating point. Partial load or overload operation are often included.An appropriate anchoring of operational stability in the form of operating limits within the operating strategy extends the advantages of individualization to these load ranges that do not correspond to the design point.

[0014] It is further proposed that the operating strategy be optimized to adhere to operating limits to ensure the desired service life. In most applications, adhering to a specified service life plays a crucial role. Load ranges that significantly shorten the service life should generally be avoided. Anchoring this principle within the operating strategy in the form of operating limits represents a sensible way to limit permissible operating ranges for most applications.

[0015] It is further proposed that at least one individual characteristic be recorded during the manufacturing process of the individual cell components and / or fuel cells. Preferably, a large majority of the individual characteristics are recorded in the proposed manner. This form of recording is advantageous because options for recording tolerance deviations at the manufacturing level are typically already implemented as quality assurance measures. This preferred embodiment thus draws on existing infrastructure, avoids the often laborious measurement of the necessary variables after the manufacturing process, and creates uniformity between data recorded for quality assurance and for individualizing the operating strategy.

[0016] Furthermore, it is proposed that at least one parameter of the predictive model be determined during and / or after a conditioning process of the manufactured fuel cell stack as part of test bench observations, whereby the at least one parameter is either obtained by recording the at least one individual characteristic as a parameter or is determined numerically by algorithm-supported approximation of the model behavior to the test bench observations. In general, not all parameters of the predictive model can be determined via individual characteristics measured during the manufacturing process. Some parameters, for example, those related to electrochemical kinetics, contact resistances between adjacent layers, or water transport properties of the membrane, cannot be measured at the component level and must be recorded by other means.This preferred embodiment accomplishes the detection based on test bench measurements during or after the conditioning phase of the fuel cell stack.

[0017] This is particularly advantageous because, for manufacturing reasons, test bench measurements are performed after the cells have been stacked into a fuel cell stack. During these tests, many parameters that were not yet known at the time of cell production and stacking are determined. Using these parameters to parameterize the predictive model minimizes the workload and the stresses on the fuel cell stack associated with some testing procedures.

[0018] Further outstanding model parameters can be determined using algorithmic optimization, for example, with the help of machine learning models such as nonlinear regression models, Gaussian processes, or neural networks. This is achieved by initializing the outstanding parameters using the aforementioned algorithmic optimization in such a way that the model approximates the real operating behavior of the fuel cell stack observed on the test bench.

[0019] This preferred embodiment is advantageous because by including behavior of the fuel cell system that can only be measured on the test bench, whether by quantitatively recording further individual characteristics or by qualitatively approximating the model to observations, a higher degree of individualization of the predictive model and thus also a higher degree of individualization of the resulting operating strategy is enabled.

[0020] Furthermore, a control device is proposed that is configured to implement the fuel cell stack's individual operating strategy. The operating strategy is typically implemented through automated fine-tuning of the operating parameters during ongoing operation according to the objectives or principles pursued by the operating strategy. Since a control device for adjusting operating parameters during ongoing operation is typically available, for example, in a so-called fuel cell control unit, the proposed control device can be implemented by expanding such an existing fuel cell control unit to implement the individual operating strategy.

[0021] A preferred embodiment of the invention is explained in more detail below with reference to a figure. The figure shows a schematic representation of a method according to the invention. Detailed description of the figure

[0022] The figure shows a schematic sequence of a preferred embodiment of a method according to the invention which serves to individualize an operating strategy of a fuel cell stack.

[0023] In a first process step S1, individual characteristics of the fuel cell stack under consideration are determined, which are due to manufacturing tolerances or other individual characteristics of fuel cell components.

[0024] This includes, for example, the catalyst-coated membrane, the gas diffusion layer, and the bipolar plates. The individual characteristics are recorded as parameters. This process step (S1) is performed during the manufacturing process of the individual components.

[0025] In a further process step S2, individual characteristics of the fuel cell stack are determined, which are attributable to the stacking and bracing process of the fuel cells. This includes, for example, fuel cell stack curvature and bracing forces. The individual characteristics are recorded as parameters. This process step S2 is carried out during the stacking and bracing process.

[0026] In a further process step S3, individual characteristics of the fuel cell stack are determined, which are most easily captured during test bench measurements. These individual characteristics are recorded in the form of parameters. This includes, for example, cell voltages, currents, temperatures, input conditions, contact resistances between the layers, water transport parameters of the membrane, and the like. The individual characteristics are recorded in the form of parameters. This process step S3 is carried out on a test bench during or after the conditioning process of the fuel cell stack.

[0027] Using the parameters obtained in process steps S1 to S3, a predictive fuel cell stack model is parameterized for each fuel cell stack individually in a further process step S4. Parameters not covered by previously obtained parameters are then numerically determined in a process step S5 using algorithmic optimization, for example, nonlinear regression, so that the model represents an operating behavior that approximates the observed operating behavior of the fuel cell stack on the test bench as closely as possible.

[0028] In a subsequent process step S6, the parameterized model is implemented in a system simulation, and a fuel cell-specific operating strategy is generated based on one or more operating objectives. For example, the operating strategy is based on an operating window individually preferred by the fuel cell stack to maximize efficiency with a specified service life to be achieved. The operating strategy derived from these operating objectives is made available via a control device during operation of the fuel cell stack.

Claims

[1] Method for individualising an operating strategy of a fuel cell stack, comprising a stack of several technically similar fuel cells composed of several cell components and differing in detail due to manufacturing tolerances, on the basis of which the operating behaviour of the fuel cell stack is individualised, characterized by that in order to take into account the individuality of the operating behavior, a predictive model of the fuel cell stack is created, individually parameterized taking into account at least one individual characteristic of the fuel cell stack based on the different manufacturing tolerances and / or the stacking sequence, fed into a system simulation and with the help of this, an operating strategy for the fuel cell stack is specified. [2] Method according to claim 1, characterized bythat at least one individual characteristic of the fuel cell stack is abstracted into at least one parameter and used to parameterize the predictive model. [3] Method according to claim 1 or 2, characterized by that the operating strategy is optimized to maintain a fuel cell stack-specific operating window to maximize efficiency for a given service life to be achieved. [4] Method according to one of the preceding claims, characterized by that the operating strategy is optimized to comply with operating limits in order to ensure operational stability. [5] Method according to one of the preceding claims, characterized by that the operating strategy is optimized to comply with operating limits in order to ensure the desired service life. [6] Method according to one of the preceding claims, characterized bythat at least one individual characteristic is recorded during the manufacturing process of the individual cell components and / or fuel cells. [7] Method according to one of the preceding claims, characterized by that at least one parameter of the predictive model is determined during and / or after a conditioning process of the manufactured fuel cell stack in the context of test bench observations, wherein the at least one parameter is obtained either by recording the at least one individual characteristic as a parameter, or is determined numerically by algorithm-supported approximation of the model behavior to the test bench observations. [8] Control device designed to implement the fuel cell stack-specific operating strategy.

Images