Physical models in systems engineering
Model-Based Systems Engineering (MBSE) has many faces and goes far beyond SysML. In particular, SysML can only be recommended for physical models with reservations. The following is about the possible alternatives to SysML for modeling physical systems.
Sense and purpose of MBSE
With "classic" MBSE, e.g. with SysMLis primarily about mastering complexity. Classic MBSE creates the Traceability from requirement to system element, and Encapsulation. However, it is difficult to depict the behavior with sufficient fidelity. Sure, we can describe discrete states well with state charts or activity models. But it becomes more difficult with continuous processes. SysML provides Parametrics ready. There is also the SysML Extension for Physical Interaction and Signal Flow Simulation (SysPhS). You can do a lot with it.
These approaches should be seen as extensions to SysML rather than core features. For example, it is possible to use corresponding model elements export from Cameo to Simulink to simulate them there. However, the Simulink model is subordinate to the SysML model.
This approach is perfectly practicable if the physical aspects only occur selectively. However, if they are leading, then it can make sense to turn the tables.
Tools and notations for physical models
The alternative to SysML is the use of other notations, which are often specified by the tool. One of the better-known tools is Matlab/Simulink. Simulink provides elements of countless domain-specific modeling languages, from electronics, electrics, pneumatics and many more. This allows very precise models to be created quickly.
What is missing at first is the encapsulation and hierarchical structure of the model, as well as traceability. However, all of these can be supplemented with appropriate techniques. Furthermore, many SysML modeling tools have interfaces to Simulink.
The approach of Valispace. Here, the decomposition of the system is clearly in the foreground. In contrast to SysML, the parametric linking of the system elements is simple and designed for physical systems. This concept also results in traceability to the requirements, at least if these are quantifiable. For example, by adding up the masses of the individual components, it can be ensured that the specified maximum weight is not exceeded. Valispace comes from the aerospace industry and is based on the needs of this industry. However, this does not rule out its use in other domains.
Domain is in the foreground
It becomes even more specific, for example, with PathWave Design. This software focuses on systems consisting of verified hardware for high-frequency applications. Their freelance employee Don Dingee has wrote an interesting articlewho looks at MBSE from the perspective of a digital twin. In particular, he argues that generic MBSE approaches simply do not have enough precision to realize the added value of a digital twin.
And this brings us back to the initial assertion that classic MBSE is primarily about mastering complexity. PathWave, on the other hand, is accurate enough for practical simulations. This in turn increases confidence in the system to be developed at an early stage, as the following image shows:
Conclusion
MBSE can take many forms and there are many different notations. Physical models can be realized with SysML. However, it is worth understanding the goals of modeling in detail. If it is more about simulation and digital twin and less about mastering complexity, then there are probably more effective approaches.
