Advanced analytical models for the mechanical design of filament wound composite pressure vessels for hydrogen storage
Topic(s) : Special Sessions
Co-authors :
Marie HONDEKYN (BELGIUM), Nazim ALI (BELGIUM), Wim VAN PAEPEGEM (BELGIUM)Abstract :
In the global evolution towards net zero emissions by 2050, hydrogen fuel cells are emerging as a compelling solution to electrify heavy-duty and long-haul vehicles. Hydrogen proves appealing for these applications thanks to its high gravimetric energy density, resulting in excellent power and capacity features. However, hydrogen possesses a notably lower volumetric energy density compared to most other fuels, necessitating storage at extremely high nominal working pressures of 70 MPa. Storage is typically done in type IV composite pressure vessels, in which a polymer liner ensures H2 impermeability and a composite overwrap bears most of the load. The overwrap is produced by filament winding, and consists of both hoop and helical layers to guarantee a good performance in both the circumferential and axial direction of the tank.
Two important aspects in the design of a pressure vessel are the reduction of carbon fibre, which is the largest cost-driver, and safe storage. While detailed numerical simulations and experimental testing are useful tools to study the tank under different failure modes, they are less suitable for parametric studies or optimization problems during the design step. With this purpose in mind, an advanced analytical tool was developed in the present work to make quick predictions of the mechanical performance of pressure vessels under both thermal and mechanical loading. The input of the analytical tool consists of the tank geometry, the composite lay-up, the material properties, and the loading conditions. In less than 1 second, the layer stresses, strains and displacements are produced as an output.
The analytical tool combines a 3D elasticity solution in cylindrical and spherical coordinates to reproduce the mechanical response of the multilayer composite cylinder and dome, respectively (Figure 1). Axial force and torque continuity are imposed at the dome-cylinder transition, as well as continuity of the radial displacement and rotation of the normal, which causes a local bending moment and shear force at the junction. The resulting closed-form equations were implemented in Python and the results were validated with Abaqus. Figure 2 displays the stresses in the fibre-direction (S11) through the thickness of a realistic composite overwrap subjected to a pressure of 10 MPa. A good correlation between Python and Abaqus can be observed for the S11 stresses, both through the thickness and for the variations towards the junction. The 3D elasticity approach generates also accurate results for S22, S33 and S12 stresses. Furthermore, the analytical tool was optimized with failure criteria to predict the burst pressure, and with rules of mixtures to study the influence of fibre volume variations through the thickness of the laminate, which are observed in real tanks due to compaction during filament winding. Lastly, the analytical tool will be further enhanced to include also fibre angle variations and different dome shapes.
Two important aspects in the design of a pressure vessel are the reduction of carbon fibre, which is the largest cost-driver, and safe storage. While detailed numerical simulations and experimental testing are useful tools to study the tank under different failure modes, they are less suitable for parametric studies or optimization problems during the design step. With this purpose in mind, an advanced analytical tool was developed in the present work to make quick predictions of the mechanical performance of pressure vessels under both thermal and mechanical loading. The input of the analytical tool consists of the tank geometry, the composite lay-up, the material properties, and the loading conditions. In less than 1 second, the layer stresses, strains and displacements are produced as an output.
The analytical tool combines a 3D elasticity solution in cylindrical and spherical coordinates to reproduce the mechanical response of the multilayer composite cylinder and dome, respectively (Figure 1). Axial force and torque continuity are imposed at the dome-cylinder transition, as well as continuity of the radial displacement and rotation of the normal, which causes a local bending moment and shear force at the junction. The resulting closed-form equations were implemented in Python and the results were validated with Abaqus. Figure 2 displays the stresses in the fibre-direction (S11) through the thickness of a realistic composite overwrap subjected to a pressure of 10 MPa. A good correlation between Python and Abaqus can be observed for the S11 stresses, both through the thickness and for the variations towards the junction. The 3D elasticity approach generates also accurate results for S22, S33 and S12 stresses. Furthermore, the analytical tool was optimized with failure criteria to predict the burst pressure, and with rules of mixtures to study the influence of fibre volume variations through the thickness of the laminate, which are observed in real tanks due to compaction during filament winding. Lastly, the analytical tool will be further enhanced to include also fibre angle variations and different dome shapes.