This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

During a product's regular Engineering and Current Product Support (CPS) Lifecycle, situations appear where quick and reliable decisions need to be made. Simulation can be a helpful tool to support the decision-making process and provide important information that can influence decision quality. In many engineering environments, manual processes must be executed and accompany the evaluation of numerical results. Usually, a work request system exists that includes creating process input and output documents, reviews of tasks and results, and a queue for the analysis work to be executed. The execution of such a process is usually time-stretching to deliver results and influences decision-making regarding urgent tasks that usually need a decision within a few hours or days. Highly standardised simulation work can hence be automated and sequentially executed, even if different simulation disciplines or tools are involved, using suitable Multi-Disciplinary Optimization (MDO) and Simulation Process and Data Management (SPDM) Software Tools. A workflow prepared in this manner can be executed quickly even by a non-expert user and is only limited by the simulation runtimes. It allows for delivering results quickly and increases the chances of influencing the decision-making process in urgent engineering tasks. An example of such a process is presented, focusing on the analytical standard work carried out on the membrane of a Diesel Exhaust Fluid (DEF) pump supplying flow in a Selective Catalytic Reduction (SCR) dosing system. Here, the membrane performance is evaluated by analysing specific properties, including the stress condition, strain hot spots, and the final pump flow performance. A FEA simulation model is utilised to analyse the structural behaviour. Another FEA simulation is set up to investigate the pump flow performance, which obtains the volume due to membrane displacement as a function of piston stroke and fluid pressure. This data is then plugged into a 1D system simulation of the pump, driven with varying rotational speeds, to obtain the final flow condition. The first step towards successfully implementing the concept was parameterising the CAD model of the membrane geometry. Afterwards, the FEA simulation models were parameterised based on the most relevant inputs and outputs and automated with Python scripts for different operations (e.g., result calculation, data transfer, and generating and exporting required contour plots and animations). The next step was integrating the simulation, pre-processing, and post-processing tools in a single environment using an MDO tool. Three workflows were created to achieve this: two investigation-specific (structural behaviour and flow condition) back-end workflows and a front-end workflow nesting the back-end workflows. The users interact with the front-end workflow while the back-end workflows conduct all the necessary operations, ensuring simplicity. A SPDM tool deployed on Cummin's intranet was then used to democratise the simulation workflows along with their dependencies (e.g., scripts and simulation files). Since multiple users were planned to execute the workflows, their versioning and traceability could also be maintained in the SPDM tool. It also enabled the sharing of computational resources by allowing users to run the process on different workstations and the in-house HPC. Since the workflows were implemented as a team project, anyone in the team can now run the workflows. This user interface is accessible via a web browser on the intranet, and anyone with the granted access can run the front-end workflow. The post-processed results are also made available to users on their web browsers to enable them to conduct their necessary investigations.