Search Results

Predictive physical modeling is an established method used in the development process for automotive components and systems. While accurate predictions can be issued after tuning model parameters, long computation times are expected depending on the complexity of the model. As requirements for components and systems continuously increase, new optimization approaches are constantly being applied to solve multidimensional objectives and resulting conflicts optimally. Some of those approaches are deemed not feasible, as the computational times for required single predictions using conventional simulation models are too high. To address this issue it is proposed to use data-driven model such as neural networks. Previous efforts have failed due to sparse data sets and resulting poor predictive ability. This paper introduces an AI Toolchain used for data-driven modeling of combustion engine components. Two methods for generating scalable and fully variable datasets will be shown.

A new phenomenological CI combustion model was developed. Within this model the given injection rate may contain an arbitrary number of injections during one cycle. Another target was a short computation time of one second per cycle on average. The new approach should also have the ability to simulate a wide engine spectrum from passenger-car engines through to marine engines. The ignition delay is calculated separately for each single injection. In this way the model depicts the influence of pilot injections on the ignition delay of proximate injections. Each pilot injection is modeled as a single air-fuel mixture cloud with air entrainment. The burn rate of the pilot injection is modeled as a function of flame propagation and of the current local excess air ratio. If the local excess air ratio becomes too lean the pilot combustion stops or does not start at all. Main and post-injections are calculated by means of a slice approach.

Two combustion models are presented: A quasi-dimensional approach, based on the injection shape and an empirical model. Both models have computation times of less than one second per cycle. The quasi-dimensional approach for CI combustion discretizes the injection jet in slices. Pilot-injections are modeled as separate zones. The forecast capability and the limitations of the model are discussed on the basis of measurements. Mentioned above the base of the quasi-dimensional model is the injection rate. Often it is difficult to obtain these data. There is therefore another empirical approach for combustion, which does not need the injection rate as input. Both models have to be calibrated. This can be done by an automatic calibration tool on the basis of the advanced Powell method. The differences and advantages compared with other optimization methods are shown. Emission-simulation models are highly important in simulating CI engines.