In recent years, Integral Powertrain has been developing methodologies using Design of Experiments techniques to improve calibration robustness while significantly reducing the required testing resource. These processes are relevant to a number of applications such as engine hardware characterisation, catalyst light-off development and, most significantly, engine dynamometer mapping.

Design of Experiments (DoE) is a method used to determine the response of a system to various combinations of inputs. Using statistical analysis techniques a test matrix can be defined to fully characterise the system response while minimising the amount of data collection required. Careful selection and processing of the data points also allows interactions between various inputs to be characterised.

Many modern engine management systems contain physical models of the engine operating envelope. This makes full characterisation of the system essential, in order to produce robust calibrations for all operating conditions. Traditional engine mapping processes require data to be collected for all combinations of inputs, known as the ‘Full-factorial’ approach. This leads to costly and time consuming testing exercises for modern engines with high feature content, such as; twin variable valve timing, direct injection and turbocharging. The resulting large dataset can be also be difficult to manipulate using traditional data analysis techniques.

The methods developed by IP, employ the Mathworks Model Based Calibration (MBC) and Calibration Generation (CAGE) toolboxes, for which Integral Powertrain was one of the earliest adopters. In the example of engine dynamometer mapping, a small amount of screening data is collected to define the engine boundary conditions. A test matrix is then created using DoE methods. A single dataset is collected on the engine dynamometer from which multiple models can be generated for, amongst others, engine torque, spark efficiency, exhaust back pressure, over-fuelling etc. The EMS strategy is replicated in CAGE and the MBC models used to fill the appropriate calibration tables directly. This calibration can then be easily validated on the engine resulting in significant savings in testing time. The models can also be easily augmented and reprocessed if hardware changes are subsequently made, thus negating the need for a complete new testing exercise.

Integral Powertrain, a Mathworks Connections partner, has used these techniques to produce series production calibrations for a number of leading global vehicle manufacturers.