In the framework of the transformations introduced by the Industry 4.0 concept, and the consequent digitization of companies, the use of real-time (RT) data and models for decision-making is becoming extremely relevant. This has led to the development and increased use of the so-called digital twins (DT) that integrate data, models, simulations and tools for decision making. They use an updated virtual representation of the processes in RT to help operating their associated systems. In the context of the process industry, the main difficulties that presents its implementation are related to:
a) The lack of standardization of its architecture and functionalities.
b) The generation of data-based models compatible with the physical behaviour of the process and the updating of models in RT according to the changing process conditions.
c) The presence of uncertainties and structural model error.
d) The large dimensionality that appears in problems of simulation, estimation and optimization, which makes difficult obtaining solutions in RT.
This project tries to contribute to reduce these difficulties and to facilitate the development of Digital Twins by developing research on:
1. Concept, architecture, functionalities and operations of DT in the process industry.
2. Methods for data based model development that takes into account physics principles making them able for extrapolation outside the region of the experimental data.
3. Methods for distributed simulation, that allows integration of large-scale models in RT.
4. Development of distributed optimization methods and its application to particular problems of:
a. Parameters and states estimation and dynamic data reconciliation, that will facilitate the update of models in RT.
b. Modifier adaptation and stochastic optimization, that will facilitate the use of RT optimization in decision making with structural process-model mismatch and uncertain environments.
In addition, the project will also contribute to the solution of three relevant industrial case studies using the methods above mentioned: a set of sterilizers of tuna cans (oriented to modelling), the operation of a set of wind farms (oriented to optimal operation) and the manufacturing of fiberwood panels (oriented to the operation of DT in a control room).
The case study of food sterilization will apply and test different tools and methodologies developed in the project to improve the efficiency of the process and the quality of the product. This is supported by ASMsoft S.L., a SMEs that specializes in systems integration and MES solutions.
The case of power and energy dispatch for wind farms will apply and test different tools and methodologies developed in the project to improve the operation in RT of hundreds of wind farms and it is supported by GTD Science, Infrastructures & Robotics S.L.U.
The case study of the fiberwood panels considers the implementation of a DT in the control room of the factory to help improving the operation of a large press and it is supported by Sonae-Arauco, one of the most important companies of the sector.
The partners of this coordinated project have been working in modelling, process control and optimization for many years with industrial partners. They present this project with the will of continuing previous research and contribute to the development and extension of concepts and methods oriented to the solution of the implementation of DT in the process industry.
This project is supported by MCIN/AEI /10.13039/501100011033 / FEDER, UE. Overall financial support is 428340€. This is distributed between each sub-project: 194931€ for PID2021-123654OB-C31 and 113982€ for PID2021-123654OB-C32 and 119427€ for PID2021-123654OB-C33.