Interactive Visual Analysis of Multiple Simulation Runs Using the Simulation Model View: Understanding and Tuning of an Electronic Unit Injector

Abstract

Multiple simulation runs using the same simulation model with different values of control parameters generate a large data set that captures the behavior of the modeled phenomenon. However, there is a conceptual and visual gap between the simulation model behavior and the data set that makes data analysis more difficult. We propose a simulation model view that helps to bridge that gap by visually combining the simulation model description and the generated data. The simulation model view provides a visual outline of the simulation process and the corresponding simulation model. The view is integrated in a Coordinated Multiple Views ;(CMV) system. As the simulation model view provides a limited display space, we use three levels of details. We explored the use of the simulation model view, in close collaboration with a domain expert, to understand and tune an electronic unit injector (EUI). We also developed analysis procedures based on the view. The EUI is mostly used in heavy duty Diesel engines. We were mainly interested in understanding the model and how to tune it for three different operation modes: low emission, low consumption, and high power. Very positive feedback from the domain expert shows that the use of the simulation model view and the corresponding ;analysis procedures within a CMV system represents an effective technique for interactive visual analysis of multiple simulation runs. We also developed new analysis procedures based on these results.

A visual analytics framework for spatio-temporal analysis and modelling

“…Thus, Demšar et al (2008) employ coordinated linked views and clustering for exploration of a geographically weighted regression model of a spatio-temporal phenomenon. Matković et al (2010) support users in exploring multiple runs of a simulation model. Visualisation and interaction can reduce the overall number of simulation runs by allowing the user to focus on interesting cases (Matković et al 2011).…”

Interactive visual analysis of families of curves using data aggregation and derivation

“…One option is to use complex interaction methods and visualizations, and retain data in its original form. Com- plex interactions include, for instance, the angular brush in parallel coordinates [9] and the angular line brush in the curve view [16]. Conversely, new data can be synthesized during the analysis.…”

“…There is one injector/pump unit per cylinder in the cylinder head assembly. We analyzed [16] the Delphi E3 Diesel Electronic Unit Injector (EUI) [7], an advanced unit injector with two independent, fast-response precision actuators that can change the injection pressure level and adjust the fuel delivery timing and duration, allowing a very flexible choice of fuel injection characteristics. 2880 different simulation runs were computed and analyzed in the case study [16].…”

“…The angular line brush [16] is proposed as a method for the intuitive brushing of curves based on the angle. As the curve view already has its own type of brush, the line brush, which is proven and well accepted, it seemed as a best solution to improve it to be able to brush curves with a specific slope interval.…”

“…Our work is motivated by case studies we have done with domain experts from different fields including engineering [13,14,16] and medicine [15]. We discovered that many details of the data sets from those very different problem domains can be represented as families of curves [14].…”