Interview at Real-Time 2010 in Paris, France with Nicolas Orand from LMS Simulation.

So-called 1D multi-domain plant models have gained in popularity throughout the automotive industry, for pre-design and design purposes.  While their primary use is intended for designing sub-systems and validating full system performance, their role in the controls design process has become evident.  Sub-system integration to validate full system behavior requires the integration of controls; this integration is done at different steps of the design process, from MiL (Model-in-the-Loop) or SiL (Software-in-the-Loop) to HiL (Hardware-in-the-Loop).  The interest in this integration of plant model with controls model or code goes beyond the validation of the system’s behavior. Continuity in the design cycle and the accuracy of the models add to final product quality.  The re-use of the plant model for design through the integration and validation process will ensure the required accuracy, as well as the accessibility of the variables needed for controls. Overall, the industry, which is in need of more controls, has accepted that bridging the controls design process with the system design process allows for better understanding of the system. As a result, necessary design modifications are shifted to earlier in the design process reducing reliance on late fixes, often done through lengthy calibration and dedicated testing.

The deployment of such a process requires knowledge to be developed and methodologies to be implemented within various groups of an organization.  Amongst existing 1D system simulation tools that are dedicated to the simulation of physical behavior of various natures, there is no parity in capabilities for supporting such a methodology. As a result, many aspects must be considered.  These include (i) the ability to represent physical components at various levels of complexity, and therefore the capability of running real-time simulation while maintaining an acceptable degree of accuracy, (ii) the ability to support model reduction with the same accuracy constraint, (iii) the ability to link physical model to controls, defining IOs, and executing the model on a real-time target.  The industry often faces interoperability issues with different tools used to represent the many sub-systems that compose the full system. This aspect, as well as the issue of communalization of tools, will not be discussed in this presentation.

Of the three considerations discussed above, model reduction is of particular interest since it not only involves tool capacity, but also requires some knowledge of dynamic systems.  The presentation will introduce the above mentioned concepts as well as a proposed methodology for reducing plant model complexity while maintaining the required level of accuracy.  Several examples will be used to visually illustrate the techniques used, with a final example on a conventional vehicle including the model of an internal combustion engine that uses high frequency combustion law, an automatic transmission, a 2D longitudinal vehicle body, a driver and a mission profile linked to the engine and the transmission controllers that originate from a separate tool for controls design.  Finally, the methodology for exporting the plant model to the real-time target will be explained.

Author(s): Nicolas Orand,
Narrator: Nicolas Orand,

Download Video : LMS-interview-RT10-110910.mp4

MPEG4 Mobile Feed :

Related Products