Scilab and time-to-market reduction
By Jean-Pierre Bovée, Sanofi
While the academics world has long welcomed the virtues of processes physical modeling versus the fine control of production and wider knowledge of the determinants of performance and quality, reality is far from the predictions in the field.
Typically the first step is modeling, conducted using collected data and transferred to a simulation environment. Even if this allows offline evaluation of the model, it is an online evaluation that is intended. Beyond this stage of model validation, if one approaches the control of the process, which may involve complex models, the available tools to implement the corresponding algorithms are automatons or driving systems which were not designed for these purposes. At this particular stage one most often finds technological barriers that slow down the test of the concept. This transposition of strategies developed in the simulation environment is difficult, costly and sometimes impossible.
Hence the idea of a direct coupling between the industrial environment and the simulation environment: data arrives directly to the simulation environment and the latter returns the control commands via the developed algorithms.
The coupling of an environment that can test the validity of models and their control by relatively complex algorithms was conducted between Scilab and industrial automatons to overcome the steps of translation / implementation in the automatons controllers. This opens the way for an accelerated strategies test based on process modeling, one of the major pathways of quality and costs control during this period of industrial history.
Sanofi_JPBovee_ScilabTEC2012.pdf 1.81 MB