Intelligent optimization of clamping design of PEM fuel cell stack for high consistency and uniformity of contact pressure

ABSTRACT

In the proton exchange membrane fuel cell (PEMFC), the uneven distribution of contact pressure on the gas diffusion layer (GDL) will lead to the increase in local contact voltage, thus increasing the ohmic loss of the stack, and will affect the porosity to change the reaction gas concentration. Therefore, it is essential to optimize the consistency and uniformity of contact pressure. In this paper, the bolt-encapsulated different scale stack models are established. The influence of the cell number on the distribution characteristics of the contact pressure on the outermost and innermost GDLs in different stacks is discussed. According to the GDL contact pressure distribution characteristics in the 6-stack, it is found that it can be used as the data extraction model for optimizing the clamping condition of the 80-stack. Subsequently, the optimal Latin hypercube design is used to randomly generate 280 data sets with different clamping conditions, and a data-driven surrogate model based on the Support Vector Regression is used to predict the contact pressure distribution characteristics of GDLs in the 6-stack according to changes in clamping condition. Then, the surrogate model is taken as the fitness evaluation function in Gray Wolf Optimizer, and the clamping condition of the 80-stack is optimized, which improves the uniformity and consistency of the contact pressure distribution on the GDL by 66.67%. This paper proposes a strategy to optimize the clamping condition of large-scale stacks based on the calculation and prediction of the contact pressure distribution on the GDLs in small-scale stacks.

KEYWORDS:

• PEMFC stack
• GDL
• contact pressure
• clamping condition
• data-driven optimization
• surrogate model

Nomenclature

aparameter in GWO

penalty factor in SVR

dwidth of the gasket (mm)

gkernel function parameter gamma

lthickness of the gasket (mm)

nsafety factor

r₁random vectors of [0,1] in GWO

r₂random vectors of [0,1] in GWO

${\hat{y}}_i$the predicted value of the response

$y_i$the true value of the response

$\bar{y}$average of the true value of the response

Fbolt forces/load (kN)

Dendplate thickness (m)

Xset of design variables

P_gasgas working pressure (MPa)

Greek Letters

μfriction coefficient

σgasket contact pressure (MPa)

σ_minlower limit of the gasket contact pressure (MPa)

ωvalue of the weight in GWO

Abbreviations

BPPBipolar Plate

CVCoefficient of Variation

FEAFinite Element Analysis

GDLGas Diffusion Layer

GWOGray Wolf Optimizer

MSEMean Square Error

PEMFCProton Exchange Membrane Fuel Cell

PSOParticle Swarm Optimization

RMSRoot-Mean-Square

SVRSupport Vector Regression

Acknowledgments

This work was supported by the National Key R&D Program of China (Fuel Cell Engine Integration and Control for Heavy Duty Trucks-Vehicle Integration of Fuel Cell Heavy Duty Trucks).

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Additional information

Funding

This work was supported by the National Key R&D Program of China (Fuel Cell Engine Integration and Control for Heavy Duty Trucks—Vehicle Integration of Fuel Cell Heavy Duty Trucks)..