DATADVANCE, an independent developer of design space exploration and predictive modeling software, constructed an approximation model for Mitsubishi Motors to predict internal combustion pressure model parameters.
The modern automotive market requires manufacturers to balance between increasing customer demands, strict environmental regulations. To meets the regulations, lightweight and good efficiency vehicle are required. But sometimes these negatively affect NVH comfortability. For these interwoven attributes, using the numerical simulation, system modeling, and design space exploration become the best option automotive companies can find an optimal solution to meet all these requirements.
Calculating engine exciting force requires knowing internal combustion pressure at arbitrary operation regimes. And combustion pressure is changed repeatedly every development phase with calibration of control design value. But it needs to build a complex and large-scale engine model. To describe the combustion process, in this case, Mitsubishi Motors used very simple physical model and approximation model with well-known Wiebe function that depends on five internal parameters and polytropic index.
The simple engine model is also affected by several operational parameters, so parameters of Wiebe function model are required to adopt such variations. Approximation model can find out Wiebe function parameters at the arbitrary condition and consequently, it gives an accurate combustion pressure calculation in a simple model.
DATADVANCE’s flagship software product pSeven was used as a powerful and helpful tool to predict combustion model parameters. It allowed creating a single workflow that uses the available small number of experimental data as input, run combustion simulation in Simulink and implement approximation techniques to results.
As a first step, optimization of residuals was used to fit the parameters of a combustion model to existing experimental combustion pressure vs. crank angle curves. As the next step, an approximation model was created to predict these parameters at arbitrary regimes using known values, like intake and exhaust manifold pressure, fuel consumption, injection timing, O2 concentration, brake mean effective pressure and RPM.
The created approximation model will be further used in 1D engine simulations via export through m-file or FMI (Functional Mock-up Interface), a tool-independent standard to support both model exchange and co-simulation of dynamic models, in almost any system modeling software. Such approximation models are also very fast in terms of computation, so implementing them in the product development process leads to significant time savings.