Generic FMU-Visualization

The “functional mock-up interface” offers the possibility to exchange simulation models across different simulation software. The “functional mock-up unit” is a compiled simulation model which can be accessed through standardized interfaces. Whatever simulation tool generated the model, the created FMU represents the dynamic behaviour. Thereby, the underlying model structure is not traceable and hence, any animation or visualization possibilities of the original model are lost. With the help of a visualization specification, a generic visualization scene based in the FMU, can be created. This kind of visualization specification can be generated by the “OpenModelica” compiler for any kind of Modelica model. The visualization software “OMVis” is able to create the visualization based on an FMU and the corresponding visualization specification.

workflow for the FMU based visualization and simulation wit OMVis

Performance Analysis

For an efficient simulation of complex models, a precise analysis of Modelica-Compiler and simulation runtime is essential. These examinations can reveal slow algorithms or inefficient memory management. The centre for Information Services and High Performance Computing (ZIH) is involved in the development of the analysis tools “Score-P” and “Vampir”. Hence, the possibility to create traces with the help of these tools was established.

 

An example is depicted in the following figure. This is a trace of a parallel simulation using 4 processor cores. Besides the evaluation of the runtime behaviour of each single thread and equation, the cache behaviour was investigated and displayed. Furthermore, it is possible to get performance data without the operation of these complex tools. Therefore,  precise measuring functionalities are implemented in the simulation runtime, which facilitate a comprehensive analysis.

 

The next figure depicts the various segments of the simulation runtime of a model computation which has been created with internal measuring functionalities. The pie chart enables a hierarchic analysis of each section. As can be seen, the actual calculation “runSimulation” consists basically of the calculations “evaluateODE” and “evaluateAll_wo_ODE”.

 

 

Figure 1. trace of a parallel simulation
Figure 2. performance analysis

Automated Parallelization

Modelica offers profound possibilities to describe physical systems. Be it multi body systems, hydraulic or electric circuits – Modelica enables us to describe various systems. It is the compiler’s task to transfer models into mathematical and calculable models, regardless of their physical context. This claim to universality is to be maintained in the context of parallelizations. The user should be able to model as is usual; the mathematical processing is up to the compiler.

 

Three-dimensional Simulation of Spiral Springs

Spiral springs are used in various technical systems and despite its low costs they have a big influence on the efficiency of the applied component. Actually, they are essential for the functionality of the system.

Therefore it’s necessary to predict the performance of the spring even in early stages of development. This is achieved through dynamic simulation.

Highly detailed models are relevant in order to attain conclusions about the behaviour of a spring. This high level of detail leads to enormous calculation times of various hours.

Within the HPCOM project, methods and tools are developed to reduce the simulation time applying high performance computers. This saving of time enables the design of an optimized spring for its specific use case.

 

 

Model of a Spiral Spring