Bismarck Louw developed a hybrid model for the Cummins QSB6.7 turbocharged diesel engine used in load-haul-dump vehicles. His central question was practical: how can a mine estimate useful engine health and performance indicators when only a limited set of ECU measurements is readily available?
A KPI-focused modelling strategy
The model combines physics-based equations with data-driven mappings. Engine speed, load and fuel consumption form the accessible input set. Data-driven components map those measurements to the internal quantities needed by calibrated physical sub-models, which then estimate thermal, pressure, power and efficiency indicators.
This structure preserves engineering meaning while avoiding the sensor burden and proprietary detail required by a fully parameterised engine model. It also keeps the output focused on quantities that can support condition monitoring and maintenance decisions.
Experimental validation
Bismarck designed and carried out controlled dynamometer experiments on a QSB6.7 engine. The test programme supplied data for calibration and independent evaluation across operating conditions. On the reported test set, the combined mean absolute percentage error was 6.21% for thermal measurements, 5.08% for pressure measurements and 4.12% for power-related metrics, including gross indicated power, brake power and brake-specific fuel consumption.
Contribution and continuation
The work provides a reusable route from sparse operational data to a broader, interpretable set of engine KPIs. Those estimates can support performance monitoring, predictive maintenance and later vehicle-level digital twins.
Bismarck completed his MEng with distinction in 2025. The dissertation was developed into an accepted journal article, and the experimental data and modelling foundation continue in I’yaaz Kala’s Simscape engine research and the wider RAMMS mobile-equipment digital-twin portfolio.