Model concept, solver and details - WITNESS
Modelisation concept
WITNESS applies complex system of systems (1) simulation techniques to IAM modelling , applying a system engineering approach starting from the high level systems (economy, energy, resources, climate, population…) and their interactions, and applying fixed point algorithms to find equilibrium between systems (see here poster presented on that topic at IAMC'24)
From an implementation point of view, we will define namespaces within which sub-system models will interact and exchange. An ontology is also defined to formalize all modules and parameter's names.
In the main namespace, all interfaces exposed by each of the main systems are automatically coupled using the ontology semantic and multidisciplinary analysis and optimization (MDAO) computational methods (MDF per system in most cases).
Going down recursively refining each system is similar, you will define sub-namespaces for each sub-system you identify to further refine it if needed, and you will define your current systems interfaces to let them be automatically be coupled through the ontology in your current namespace.
This methodology allow for easy refinement of models as the models interconnexions are prescriptive (expose interfaces in specific namespaces) but not descriptive (connections handling is automated an holistic within namespace).
You can also decide to replace in some area the driver running multidisciplinary analysis (MDA), by a controller running multi-disciplinary optimization (MDO) of you want to orient the result vs an objective or want to respect specific constraints.
For example in the basic WITNESS framework, an MDO controller is optimizing ventilation of available investment from Macro-economy to various Energy production techniques in Energy system, in order to maximize energy production while minimizing emissions under resources and materials contraints (more details here).
Solver
Solving the system of systems in performed by the SoSTrades simulation platform.
The model themselves are written in Python language and wrapped in an SoSTrades discipline that integrate model code, documentation, interfaces...
Models not written in Python or requesting specific setup (libraries, licenses, computing infrastructure...) can be integrated through a REST API (or any available python mechanism).
Coupling of the models is performed using an extended / improved version of GEMSEO library, which is providing a state of the art set of applied mathematics standard solvers implementation - and to which SoSTrades adds namespaces, any data type support, advanced controllers, cloud scalable execution...
The platform facilitate usage of full adjoint based optimization by consistently managing gradients, that makes optimization cost almost independent of the number of parameters used. Those gradient also enable sensitivity analyses and end-state of transition path stability assessment.
The platform itself can be used through a REST API to setup / run / analyse a scenario, and generate a synthesis HTML widget that can be embedded in some external dashboard.
(1) Charaf Eddine Dridi, Belala Faiza. System of Systems Modelling: Recent work Review and a Path Forward. - ICAASE’20, Oct 2020, Constantine, DZ, France. ffhal-04227335f
Note: The documentation of WITNESS is 'in preparation' and is not yet 'published'!| Corresponding documentation | |
|---|---|
| Previous versions | |
| No previous version available | |
| Model information | |
| Model link | |
| Institution | Open-Source for Climate (OS-C), N/A, https://os-climate.org/transition-analysis/., Linux Foundation (LF), N/A, https://www.linuxfoundation.org/. |
| Solution concept | Systems dynamics based approach |
| Solution method | OptimizationSimulation-based optimization |
| Anticipation | |