Experiment 9: Optimization of the energy consumption and operation of a coil coating industrial line (COATWIN)

Motivation of the Experiment

Considering the context of the fourth industrial era, Industry 4.0, REPLASA has started its transformation towards a smart factory, testing new technologies like Artificial Intelligence and Cloud Computing. Among the different strategies designed to be implemented in the forthcoming years, an intelligent energy management system for the production lines was defined as one of the main lines of work, due to the high cost that is associated to the energy consumption (that can be as high as 5% of the total product cost). Reducing energy consumption in the production line could have a high impact in the cost of the manufactured product and thus in its competitively in the market. For this reason, achieving a significant reduction of the energy consumption has been established as a priority for the company.
The energy consumption of the line is associated to electricity (mechanical and electronic parts of the line), natural gas (heating and cooling elements of the line), and compressed air (pneumatic parts of the line). The management of the parameters (such as speed of the line, temperature of the oven, intensity of the cooling system) of these elements is complex and are different depending on the final product. Currently, REPLASA does not have any tool (example: a simulation model) allowing to change the production parameters and assess the impact of those changes in the energy consumption. In general, the optimization of the coil coating line is a real challenge for the company and there is not a simulation or model of it. Within this context, in its Digital Transformation plan REPLASA has identified the potential of Digital Twins for the optimization of production lines. For this particular need, a Digital Twin with Artificial Intelligence capabilities would allow REPLASA to create a twin of the coil coating line and, when properly trained, it has the model energy consumption, make simulations and lead to decentralized decisions.

Purpose of the Experiment

The plan of TEKNOPAR is to use the Cloud and HPC resources from the DIGITBrain Platform, and later on deploy the services (data, model and algorithms) to be developed within the proposed project on to the Cloud of the DIGITbrain platform or on premise server of REPLASA. It will be necessary to train the algorithms using the HPC resources and utilize MarketPlace of the Digital Agora to publish the results achieved.
The IIoT platform will be verified and validated for a new set of assets (sensors, actuators, PLCs, data sources, etc.) and will further be tested for scalability, security and flexibility in a different industrial setting. TEKNOPAR’s machine learning library will include new algorithms for predictive maintenance used to estimate RUL of the machinery of line, TEKNOPAR’s synthetic data generator (DATA-GEN) will be used in a new setting as an authoring tool with Docker virtualization means. A completely new cognitive digital twin for the production line will be generated, additional features in visualisation are expected to be implemented.

Coil coating industrial line

Technical Impact

REPLASA does not have any tool allowing to change the production parameters and assess the impact of those changes in the energy consumption so the development of an intelligent system, based on AI techniques and machine/deep learning will allow REPLASA to control the consumption in real time while establishing the best production parameters to optimize the consumptions (gas, electricity and compressed air). Moreover, it will allow to make simulations and predictions of such consumptions, allowing decision making.

Currently there are no sensors installed on the coating line (will be installed during the experiment). The information REPLASA can get from the PLC comes from the data of the frequency convertors of the motors, but no data collection or storage occurs. Therefore, the design and implementation of a digital twin will mean a step forward in the company's digitization process. Starting from scratch, the possibilities for future growth are high, also giving rise to potential collaborations with TEKNOPAR both in the short and medium term. By this, we can say that for TEKNOPAR, the main impact includes a software product innovation opportunity: the cognitive digital twin developed for the coating line will be a new product/service developed by TEKNOPAR, which will enhance the existing software elements (the required sensors/data sources will be new additions to the existing IoT platform). The new algorithms and models may be added to existing machine learning predictive maintenance and energy optimization library of TEKNOPAR. In addition, all the components in the project will run on the Cloud or on premise server, and the Quality attributes (performance, flexibility, portability, scalability) of the existing blocks are aimed to be improved in the proposed project.


Expected Economic Impact

The experiment accelerates energy efficiency transformation in the existing manufacturing line by enabling the right information and right technology to be available at the right time and in the right form, empowering smart energy efficient decision-making and consequently resulting in a significant reduction of energy along the production process with substantial cost savings derived from optimized operation.
In terms of the evidence and answers to the driving question, these will be provided by the energy consumption of the line. Currently, the consumption is about 1000Mwh/year in electricity and 600- 650tn of natural gas per year. Thanks to the experiment, it is expected to reduce energy consumption in about 10-15%, thus helping to significantly reduce the carbon footprint and lower the cost of energy. Therefore, a decrease in energy consumption means producing at cheaper costs, and therefore increasing the competitiveness of the company against the competitors. Therefore, we are making it possible for the product to revalue in the markets where REPLASA is already present, and even offering the possibility of opening to others in which today it is not feasible to compete in price, thus increasing company sales. In addition to that, for TEKNOPAR since a completely new cognitive digital twin for the production line will be generated, additional features in visualization are expected to be implemented. To scale up after the experiment, TEKNOPAR will include manufacturers in the same sector of and utilizing similar technologies like REPLASA in TEKNOPAR’s target customers. Both local, and European markets will be targeted. Also, the experiment has also business development benefits for both entities, as it will allow them to collaborate for the first time and explore other potential collaborations.

Project Partners



is the end-user
in this


serves as the data model
& algorithms


is the Digital Innovation Hub with the role of experiment supervisor.