Advanced Simulation Techniques for Efficient Valve Design Optimisation

Traditional finite element analysis (FEA) remains a cornerstone of valve engineering design, but it is inherently limited. By focusing on discrete load cases, it defines behaviour at the edges of a design rather than across it. For valve systems subject to complex, multi-variable loading and multiple failure modes, this leaves critical gaps in understanding, particularly when designs are pushed beyond their original intent.

By investing in hardware, software and personnel, PDL continue to build on its extensive knowledge and experience in enabling rapid simulation-driven design optimisation. This raises some important questions for engineering teams:

·       Do you fully understand which requirements drive your design decision making?

·       Do you understand the capabilities of your current designs beyond their original intended use?

·       Do you wish to optimise your new designs to reduce cost, complexity and timescales?

Traditional finite element analysis (FEA) methods consider individual design cases. Whilst this produces bounding cases useful for understanding behaviour at the edges of the design space, and establishes safe limits, it leaves the majority of the design space unexplored. Hence, it doesn’t give any insight into the performance away from the discrete points sampled, and cannot be used to predict behaviour when the design requirements are pushed beyond their original intent.

For components which experience complex loading from multiple sources, or have multiple failure modes, this behaviour can be highly non-linear. A high number of simulations is required, adding significant time and cost to your engineering workflow. PDL use advanced analysis methodologies to fully characterise the system performance, understand the input factors that drive critical outputs, and provide insight into how best to optimise design whilst maintaining project timescales.

This starts with Parameter Sensitivity, exploring which design variables are critical to the required outputs. This can be applied to variable loading to fully understand existing products, variable geometry parameters to optimise new designs, or a combination of both in parallel. Once the parameter sensitivity is understood, the PDL team use a statistical Design of Experiments (DoE) approach to optimally sample the design space between specified minimum and maximum values of the design parameters. This reduces simulation time by reducing the number of design points required to create accurate response surfaces.

PDL use Ansys optiSLang for process integration and design optimisation which automates the required analyses. The advantage of optiSLang over other traditional response surface generators (such as DesignXplorer) is that it allows simulations to be ran in parallel resulting in an up to 8x speed increase over other similar analysis tools. Combined with our dedicated high-performance solving capacity, this means that full system characterisation can be completed in a similar timeframe to that which would previously be required for traditional analysis.

From the model outputs, response surfaces are fitted to each output parameter, showing the behaviour over the full the design space in a single image. In parallel, reduced order models are built from the simulation data, replacing computationally expensive FEA models with lightweight, accurate surrogates. The response surfaces and reduced order models allow for robust design analysis of interim points in near real time with no further simulation work required, allowing the design space to be fully understood and future requirements freely investigated.

PDL look forward to continuing to share developments in advanced simulation with BVAA members. For further information on how these methods can be applied to your designs, contact the team: fea@pdl-group.com

Search related valve / actuator articles:   PDL Solutions (Europe) LtdIssue 104CFD & FEA