Summary : this story shows the ModeliScale consortium project: motivation, challenges, outcomes and perspectives.

The challenges

Partners especially worked these 3 objectives:

• To properly process multi-mode models (electric switches, thermodynamic phases, valves & pumps start, doors in building etc…) with varying structure & dimension, handling of mode changes
• To properly initialize models to specify the consistent initialization of large Modelica models
   and for related tools to correctly initialize them
• To analyze, compile and run very large models (compilation + execution of large scale Modelica models (> 10E6 equations), large state vectors, multiplicity of modes (> 10E115)

State of the art

The physics is well described with Modelica behavioral equations, but no Modelica tool is currently able to analyze, compile, initialize and execute correctly such above- described models. Additionally, the supporting tools provide insufficient diagnostic support.

More precisely and related to these high technical needs, we have to deal with  :

• multi-physics DAE (Differential Algebraic Equation) systems,
• exponential curve of combined states & modes combination,
• complexity of multi-physics modelling of energy systems,
• interoperability of these systems when aggregated into a larger view.

In order to progress, there is a need for new concepts from research.

Scientific work

 Partners worked on a large set of topics to address these challenges.

Hereafter the main ones :

• Structural Analysis of DAE models : original approach to handle multi mode models; resulting in a Inria prototype coupled to Dymola, and 1st scale of models analyzed
• Consistent Initialization of multimode systems : to use the structural analysis for a correct initialization; resulting in a Inria experimental prototype
• Better support of non-smooth systems : to support stiffness and fast changes in models (examples : electric components, contacts, valves, use of controllers); resulting in a complementarity systems method from Inria
• Proper Initialization of systems: to correctly initialize models (initial parameters and states) ; resulting in a Data assimilation approach prototyped by 3DS and tested by EDF
• Large-Scale systems : to support of larger systems thousands of components, >10e6 equations with multiple modes; resulting in several enhancements in Dymola (compiler, large DAEs with nonlinear algebraic loops, sparse systems, parallel execution, Jacobian computation, diagnosis support) and simplified fast library prototype for electric network load flow calculation
• Co-simulation of systems : to support large and/or distributed models execution; resulting in FMI standard enchancements, with support by Dymola for FMU generation and EDF/DACCOSIM for FMI based simulation

Insight on results

Outcomes of the scientific work have been prototyped by Inria, Dassault Systemes in CATIA Dymola and EDF in DACCOSIM NG. The consortium partners both made use of these and provided additional results from their own.

Numerous and promising outcomes with already valuable results:

• Many outcomes related to CATIA Dymola technology:
  • Coupling with Multi modes analysis prototype is tested on small models, with promising results
  • New Initialization method prototyped and tested
  • Scalability improved,  with objective still to reach the next order of magnitude at city level
• Some of these features are already included in the latest 2021 release
• Contribution to FMI standard with features to better support co simulation
• Several of topics are shared in scientific papers

Software results will be more detailed in the next episode.

Stories SystemsEngineering Modelica FMI DYMOLA CATIA  EnergyGrids ​​​​​​​

ModeliScale 3DEXPERIENCE ​​​​​​​

Next episode 

https://r1132100503382-eu1-3dswym.3dexperience.3ds.com/#community:6/post:4DBligMiT8iaxl2iyd04ow

Previous episode

https://r1132100503382-eu1-3dswym.3dexperience.3ds.com/#community:6/post:75smyMLPSV6kFWZYbv-bgg