INES – Interface between Experiments and Simulation
The goal of INES (Interface between experiments and simulation) is the development of software tools to facilitate the data-supported modeling and simulation of chemical plants. The aim is to obtain optimally parameterized models. Therefore, reliable data should be selected from an existing database in order to realize meaningful model adjustments.
This adjusted model then is available for optimizing the process design:
- No outliers
- Stationary intervals
- Mass balances satisfied
Removal of outliers is done interactively by defining a window around the median of a data series proportional to the median absolute deviation. This allows the user to define an outlier-detection according to the working context (accuracy of the instrumentation, known errors). Stationary intervals are found by a heuristic segmentation of the data series. In this approach, break points are set such the averages between neighbouring time-intervals differ significantly. For these intervals, a statistical analysis for stationarity can be realized.
Mass balances are checked by an interactive configuration of a data reconciliation. For control volumes defined by the user, mass balances are calculated if data redundancy is given; furthermore, a componentwise data reconciliation is done. In the case of non-redundant data, the software makes suggestions which quantities have to be measured in order to achieve redundancy. The results of the data reconciliation are used to detect possible gross errors in the data.