Development and Validation of an Automated Numerical Flow Bench Workflow for Rapid Prototyping of Next Generation Combustion Systems

The development of next-generation combustion systems for marine applications, driven by the urgent need for decarbonization and enhanced efficiency, is often constrained by time-intensive design-validation cycles. While Computational Fluid Dynamics (CFD) is a powerful tool for ana lysing in-cylinder flow, its manual setup and long runtimes create a significant bottleneck for rapid prototyping. This thesis addresses this challenge by developing and validating a robust, automated workflow for numerical flow bench simulations, specifically applied to the gas exchange system of a Wärt silä 31 engine conducted through a flow bench test rig. The core of this work is the PyFlowBench package, a Python toolkit designed to automate the entire CFD pipeline, from parametric mesh generation and case setup using experimental data to simulation execution and post-processing. An extensive series of Reynolds-Averaged Navier-Stokes (RANS) simulations were performed us ing this workflow and rigorously benchmarked against a comprehensive experimental dataset. The results demonstrate two primary achievements. First, the automated workflow successfully reduced simulation turnaround time, enabling consistent and scalable analysis. Second, the val idated CFD model showed strong agreement with experimental data, particularly in high-flow regimes where measurement uncertainty was minimal, with mass flow rate predictions gener ally falling under a ±5% deviation. Crucially, the simulations also served a powerful diagnostic role, identifying and quantifying significant experimental uncertainties stemming from sensor placement and instrumentation limitations. The study also identified the need for further un certainty quantification of experimental measurements and numerical modelling. Further, it was pinpointed that the adiabatic wall assumption was a key source of temperature prediction er rors. Ultimately, this work delivers a validated and automated platform that significantly accelerates the design-analysis cycle. It lays the essential groundwork for future advanced studies, including Large Eddy Simulations (LES), Conjugate Heat Transfer (CHT), and the development of Machine Learning surrogate models, paving the way for the rapid optimisation of more efficient and cleaner combustion systems.