CFD based evaluation of in-­cylinder mixture quality in a hydrogen internal combustion engine

Abstract

Internal combustion engines are by far the most common power source used nowadays: from automobiles and vehicles to electrical power generation, etc. Fossil fuels (e.g., diesel and gasoline) are generally used, which raises serious environmental concerns: damaging the environment, harming the health of humans and other living beings. The exhaust gases generated by the combustion of fossil fuels represent a real and significant risk. Alternatives are being studied nowadays, such as power cells and electric engines, but also natural gas or hydrogen based internal combustion engines.

This work is part of a series of tools that GDtech is implementing using OpenFOAM to be applied in the development of new internal combustion engines using either hydrogen or natural gas as fuels. The first tool developed is a re-meshing procedure to simulate a complete engine cycle. While this work provides the tools to evaluate the quality of the mixture within the cylinder.

The aim is to provide quantitative criteria to analyse the mixture formation, both spatially and in time, as well as the capacity to decide the optimal situation for combustion. Furthermore, the criteria can be used as comparison tools to benchmark injection nozzles or strategies with respect to the mixture formation. Additionally, the criteria were found to be better to evaluate the dependence between the mesh resolution and the flow results than traditional cylinder’s mean variables. The criteria are general and can be applied for any (gas) mixture

The tools are developed in Python using the built-in capabilities of ParaView and are used here to post-process cold RANS simulations and one detached eddy simulation obtained with OpenFoam. The tools are applied to 5 different injection nozzle geometries that introduce hydrogen inside a constant volume cylinder without considering the piston’s dynamic. The same cylinder’s volume is used for all the nozzles.

The difficulty to validate the results is the principal limitation. Flow visualisation remain of utmost importance. But new technologies related to image processing through artificial intelligence may present an opportunity to reconstruct experimental results and post-process them in a similar way and with the criteria defined in this work.