Testing of system models coupled to autonomous and human operators is something we do a lot of at Claytex.
The system models need to be both representative of the characteristics and behaviour of the real world components, in addition to being fast and robust enough to simulate with minimal overruns.
Claytex have been working in this field for many years and have acquired a significant amount of experience in this area. During the development of all of our Modelica libraries, we always keep in mind the real time capabilities of the models. This improves their robustness, so that no significant intervention is required by the customer or ourselves to facilitate these features.
On the cooling side, since 2007 we have built up a library of thermo-fluids components within the Claytex library, which was later based on the Modelica.Fluid library. This ultimately gives access to enhanced and bespoke components within Claytex.Fluid that can also be complemented with Modelica.Fluid components. Similarly we have based the Claytex.Media library on the Modelica.Media library base fluid definitions but developed real time capabilities on top of them.
To help with the analysis of the models we can enable the animation properties of the fluids circuits:
Video: Example of optional visuals for the real time cooling system models
Fixed Time Step Size
Of course you can run these models with variable time step solvers. However, when running on Hardware-in-the-Loop and Driver-in-the-Loop environments, it is usually required for these models to run with fixed step solvers. When considering the fixed simulation step size required for cooling models, there are several factors to consider, in particular whether the model will be required to run alongside other components in the system. For example a vehicle model with multi-body suspension and powertrain, as well as what typical transients will be experienced. Splitting or co-simulation of physical systems that in the real world are fully integrated with one another is not something we recommend for the applications we get involved with. The advantages of running the co-simulation models at different integration step sizes is usually outweighed by the communication related efforts. We have also developed and come across lapsim models that require a single systems model for the calculations, strengthening the single system model requirement.
Most of the vehicle models we run on the simulators run at 1ms time steps, and so we will normally also make sure the fluids models can run at such time steps.
Application and Performance
We recently concluded a project for a cooling system for a customer vehicle where the cooling of all main powertrain components was required, with multiple heat exchangers and fluid circuits. Replaying vehicle performance race track lap test data into the cooling system test bench running at 1ms time step yielded the turnaround times shown below. This leaves more than enough room for adding a multi-body suspension vehicle model and powertrain to still be able to turnaround under 1ms time steps.
Figure 3: Example of turnaround time for a hybrid vehicle cooling system running at 1ms time step with Euler mixed inline integration.
It is worth noting that the simulation performance of these models is not limited to real time. They can simulate much faster than real time particularly when using variable step solvers which also facilitates optimisation applications.
If you would like to know more about the library and how you can access it, please get in touch.
Written by: Alessandro Picarelli – Engineering Director
Please get in touch if you have any questions or have got a topic in mind that you would like us to write about. You can submit your questions / topics via: Tech Blog Questions / Topic Suggestion