Of all of Germany’s economic sectors, the manufacturing industry has by far the greatest energy consumption. Some 39 percent of the primary energy used in the German economy flows into the production of goods, followed by energy supply (28 percent) and transport (12 percent).
Germany’s manufacturing sector alone has an annual consumption of around 4000 petajoules – more than most countries consume in total! By way of comparison: the whole of Argentina including all its economic sectors and private households consumes just under 3600 petajoules, the Netherlands around 3500. If shown in terms of joules, each of these figures would be followed by no less than 18 zeros. Such orders of magnitude quickly make it obvious, that even the smallest savings in the production sector can have a truly enormous impact.
Despite all the uncertainty about how best to protect the climate, one thing is quite certain – if Germany wants to achieve its climate goals, manufacturing companies will have to drastically reduce their energy consumption. With its Climate Protection Programme 2030, the German government is giving them incentives to do so. However, the acquisition of new efficient machinery is a long-term investment, and hardly a single company is in a position to keep its technical equipment continually up-to-date with the latest energy-saving standards. One ray of hope: Perhaps they will not need to, thanks to intelligent control technology. To allow existing plants and equipment to work automatically in an energy-saving mode, Fraunhofer IPK in cooperation with ÖKOTEC Energiemanagement GmbH has developed the technology solution EnEffReg.
Some of Germany's biggest manufacturing companies partnered up with the EnEffReg research project to bring new energy-saving technology into circulation: Bayer in Berlin, thyssenkrupp Steel Europe in Duisburg Hamborn, and Daimler at its works in Berlin Marienfelde. In close cooperation with these leading companies, research partners ÖKOTEC and Fraunhofer IPK have developed a method that calculates ideal set points from energy measurement data and automatically transmits them to the relevant machines. Not only does this identify the most efficient operating mode, but also directly programs the machine accordingly. The EnEffReg technology was tested on technical supply systems that have a particularly high energy consumption.
Control of the systems, on the one hand, is based on the energy efficiency software EnEffCo®, also developed in collaborative research between Fraunhofer IPK and ÖKOTEC, and on the other on a key performance indicator methodology commissioned by the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety. In operation, EnEffReg extracts an extensive range of measurement data from a system in order to learn how the system behaves in terms of efficiency. It can reset networking subsystems and learn to autonomously take changes into account. As Prof. Dr. Jörg Krüger, head of the Automation Technology division at Fraunhofer IPK explains, »In the EnEffReg project, we are specifically focused on automatically boosting the energy efficiency. To this end, we do not just need to be able to measure, but also to control the machines within closed loops. We are going a step further than previous to reach a much higher level of performance.«
Availability and safety are the highest priorities in the automated regulation of systems. Only when the stable operation of a system is ensured can the software set it according to energy efficiency criteria.
Fraunhofer IPK scientists have developed a Three Stage Inspection Procedure that includes reviewing each individual component, then the connections between components, and finally the respective situation-dependent operational requirements. The proposed control can be implemented only after these three review stages showed that it will not endanger the safe operation of the system.
This data-driven approach puts the automated regulation of the system on a safe footing. However, for industrial users it is important to have a good idea of how the learning system operates. They are calling for models with a maximum level of transparency. In order to meet this requirement, the EnEffReg team used a method recently developed by American scientists: Sparse Identification of Nonlinear Dynamics. This method predicts the dynamic behavior of a system by respectively assorting a transparent combination of weighted mathematical functions from libraries. Hence, users can easily track any decision made by the artificial intelligence behind EnEffReg.
Novel methods for visualizing the energy efficiency factors of machines that were developed in the course of the project also play a vital role in ensuring transparency. As Knut Grabowski, head of the research project at ÖKOTEC, remarks: »We now have a totally new solution for scientists’ and engineers’ dream of a graphic representation of multidimensional interdependencies – and one, moreover, that can also be used for a great range of application problems independently of EnEffReg.«
The project brought plenty of new insights and even first positive results for Bayer, thyssenkrupp and Daimler. Dr. Tilman Dombrowski from Bayer comments:
»The rapid changes made in the operations of our supply technology enabled us to immediately spot a technical defect in the installed sensor system. Without monitoring energy efficiency, we would have noticed this at a much later point in time.« At a Daimler plant, the measurement values revealed that the position of an installed sensor had been documented incorrectly.
In the recooling plant of a steel mill, thyssenkrupp energy manager Hans-Peter Domels achieved up to 15 percent energy savings with the help of the developed software. »We tried out new modes of operation which make better use of the cool temperatures of the cooling towers and thus take the strain off the cooling units,« he says. »This effect would not have become so apparent, if we had not been able to bring all the measurement data together in one single point and visualize it.«
Carsten Klemm, energy manager at the Mercedes-Benz Berlin Plant, states, »We have now proven the general applicability of optimization in actual operations. But we have still got a lot of work to do before we can extend it to other systems for higher energy savings.«
As always when it comes to automation, this project too had to ask itself just how far industry is prepared to go with fully automated solutions. Do companies prefer systems that merely make recommendations for optimization instead of direct intervention in the working of machines?
Daimler’s Peter Voß does not think so: »I would see a recommendation-based assistance system more as a interim solution. Operating personnel need clear heads to monitor the mode of operation and address any problems should they occur.« He would certainly be positive about any further automation of energy optimization.