Developing powerful, environmentally-friendly diesel engines that can reduce emissions and fuel consumption requires increasingly complex logic functionality and ever-more complicated control devices in the engine control module. On top of this, engine prototypes have to go through more and more rounds of testing, raising costs and the time invested. A major problem for engineers is gaining access to testing facilities. They are often fully booked – a curse when trying to shorten development cycles. To develop control concepts and calibrate control parameters in parallel with engine development, experts at the Esslingen Systems Technology/Automotive Steinbeis Transfer Center looked into ways to simulate and integrate function and control developments.
In a joint project carried out with Glasgow’s Caledonian University, Steinbeis specialists designed a “virtual test stand” as a development environment for putting diesel engine control unit software through its paces. The aim of the cooperative development was to create a diesel model that would be more accurate and versatile than phenomenological models (0D), yet still faster in reacting flexibly to new projects. The solution should also reduce simulation time and improve userfriendliness by linking with established development tools.
To save development time and reduce future development risks, the research team tapped into commercially available simulation products. The environment developed was based on BOOST, an engine cycle simulation code made by AVL List which is used for computational fluid dynamics (CFD) in diesel engines. For the control function and prototype language, the team used a package called MATLAB/SIMULINK made by The MathWorks.
In a series of experiments on the virtual test stand, the team simulated the function of a 3 litre V6 cylinder diesel engine featuring a common rail direct fuel injection system with two turbochargers. The turbochargers worked in parallel, maximizing performance and dynamic properties. A known drawback with this setup is that the characteristic of the less effective turbocharger reduces towards the limitations of the pump, similar to the effect of ageing. To get round this, a variety of control concepts were investigated for compensating the amount of intake air in both turbochargers, using tools such as industrial and adaptive PID controls, FUZZY control systems and Smith predictors with artificial neural network models (ANN) as black box observers.
There were many physical similarities between the parallel simulations on the virtual test stand and real engines. This was especially the case in the pulsating air system, triggered in the crankshaft angle, typical with the latest technology and this compares well with results from phenomenological models. Compared with a real engine and a real car, rapid prototype development of air controllers confirms that, even if there are no prototypes available for testing, the parallel function development of electronically controlled diesel engines can facilitate improved engine performance, reduce development time and minimise development costs as originally hoped.