The challenges of ever increasing combustion engine complexity coupled with the introduction of new and ever more stringent emissions regulations place a unique strain on the time available during the base engine hardware development and calibration phase of the product development cycle. Considering state of the art gasoline engine architecture (dual variable valve timing, direct injection with turbocharger) it is common to have at least 12 degrees of freedom as system inputs. The understanding of interactions and inter-dependencies of these inputs is therefore key in optimising the performance of the engine.
MAHLE Powertrain has developed a process using a global Design of Experiment (DoE) technique based on Gaussian processes that can be used to accurately model and optimise many aspects of an engine’s performance. The output of this optimisation is then coupled with the proposed calibration architecture, which enables key calibration models to be completed early in the product development program.
In this paper careful application of this methodology has been applied to a high specific output, 3 cylinder gasoline internal combustion engine which has allowed for the engine development test program to be condensed through not only reducing the number of test points but also increasing the overlap between base engine development and calibration activities. Furthermore, when applied to specific performance attributes the process has highlighted optima that would otherwise have been potentially overlooked as part of a conventional test program.