RWTH Aachen University uses filament winding alone for pressure vessel dome reinforcement

Researchers Marcus Welsh (left) and Jannes Duhr (right) holding a demonstrator manufactured in the tubular composites laboratory at ITA. Source (All Images) | ITA of RWTH Aachen University

Composite pressure vessels (CPVs) for hydrogen storage are critical but costly components in fuel cell systems, with the composite overwrap accounting for roughly 57–67% of total vessel cost depending on operating pressure. Reducing that material burden — by providing targeted reinforcement in the dome region — has become a central focus of CPV development. ongoing research at the Institut für Textiltechnik (ITA) of RWTH Aachen University (Aachen, Germany) is investigating whether filament winding alone can deliver targeted local dome reinforcement, potentially eliminating the extra manufacturing steps that have so far complicated dome-specific reinforcement strategies.

Figure 1. Winding types used to manufacture filament wound pressure vessels.

The design challenge stems from the load distribution inherent to cylindrical pressure vessels (Fig. 1): hoop (circumferential) stresses are twice the magnitude of axial stresses, and the dome regions require specific reinforcement beyond what standard polar or helical winding provides. A 2013 U.S. Department of Energy (DOE) design report had identified “doilies” — discrete carbon fiber strips placed in the dome prior to winding — as one approach to providing that local reinforcement, but the 2015 follow-up report dropped this innovation from the baseline design due to process complexity concerns and their impact on high-volume manufacturability.

Since then, two notable industry approaches have emerged to revisit the problem. Cevotec GmbH (Unterhaching, Germany) has a fiber patch placement (FPP) system — the Samba Pro PV-1 — that places thin, oriented fiber patches on the liner dome before winding. Separately, Taniq B.V. (Rotterdam, The Netherlands) integrates automated fiber placement, filament winding, and automated rubber winding within a single robotic cell, allowing local dome reinforcement to be applied at any point in the laminate layup sequence. Both approaches, however, introduce either part-transfer operations between machines, additional tool changes or dedicated fiber cutting and placement steps.

Figure 2. Manufacturing demonstration of a filament wound dome reinforcement layer  

Figure 3. Local dome reinforcement wound on both domes with a transition layer connecting both layers.

ITA’s research takes a different approach: making use of the inherent stickiness of towpregs during the filament winding process to wind what is termed a “hybrid layer.” Unlike conventional polar, helical, or hoop layers — each of which has both endpoints in either the dome or cylinder region — the hybrid layer has one endpoint in the dome and one in the cylinder. Because towpregs combine fiber and partially cured resin into a single intermediate product, the tacky resin provides friction during winding, enabling non-geodesic trajectories that would not be stable using the standard wet winding process. The result is a local reinforcement layer that can be integrated anywhere within a laminate layup without machine changes, tooling changes, or fiber cutting and re-feeding operations.

Feasibility has been demonstrated on a laboratory scale in ITA’s tubular composites facility, but engineering challenges remain. Chief among them is material buildup at the cylinder endpoint: When multiple hybrid layers are wound, staggering those endpoints will likely be necessary to avoid excessive local thickness accumulation and potential structural integrity concerns. The distance of the cylinder endpoint from the dome transition — a parameter sensitive to winding tension, tow bandwidth and vessel diameter — also requires optimization. Minimizing that distance saves material but increases the risk of fiber instability and slippage.

Next steps will center on finite element modeling using laminate geometry generated with the winding program from Rheinmetall Invent GmbH (called RHWind, Neckarsulm, Germany). In the publicly funded cooperative project H2Lorica with Rheinmetall, ITA expanded the software’s capabilities to manufacture hybrid layers and export manufacturable laminate geometries to finite element software for high-fidelity progressive damage simulations. The goal is to determine whether hybrid layers can provide sufficient local reinforcement to justify the replacement of select helical layers — and if so, to use those findings to guide further manufacturing trials that optimize the cylinder endpoint placement and processing window.