A project focused on creating a digital model describing the electrical, mechanical, thermal and chemical behaviour of a power transformer.
As much as possible, the algorithms in this model are based in international standards and guidelines from IEEE, IEC, and CIGRE.
This digital twin is fully open source distributed under the MIT license and implemented using the e-lang engineering programming language.
The best way to visualise, interact and experiment with this digital twin is using a code editor, we suggest using VSCode and installing the e-lang extension.
A digital twin like this has many use cases in asset managment, for example:
- failure mode identification
- condition assessment
- root cause analysis
- operational decision-making
All these files are located in the src folder of this repository.
At a high level, PTDT is implemented in four main layers that allows it to interact with the external world. Both ingesting data and producing results. These layers are:
- Ingestion
- 6A1C Models
- Algorithms
- Results
The following diagram illustrates these how these four layers operate as a pipeline of data flow.
A brief summary of the responsibilities of each module or layer is described below.
The ingestion layer heps us acquire raw data, clean it up, transform it as necessary and validate it ready to be assigned to our models.
The models layer is incharge of specifying the data structures for all the compoments of the twin as well as more abstract (non-physical) concepts. Instances of these models will contain the data representing the current state of the power transformer.
The algorithma layer contains all the calculations, formulas, functions, rules and procedures that described the behaviour of a particular transformer.
The results layer prepares outputs that can be consumed by external systems, whether it be a report, a visualisation or another software system. This would be the point of integration of this digital twin to the rest of a digital asset management ecosystem.