Data model incorporates CO₂ emissions, resource consumption and social factors into production planning
How sustainable is our industry really? Until now, when people talked about efficiency, they mainly focused on how well machines are utilized, how often they break down and how fast they produce. But what about the CO₂ emissions caused directly and indirectly? What about resource consumption or social responsibility?
A research team at TU Wien (Vienna, Austria), in cooperation with Fraunhofer Austria Research GmbH, has developed a new evaluation system that can be used to logically model production processes in order to measure and optimize sustainability. They have published two studies in IFAC-PapersOnLine.
An Overall Sustainable Equipment Effectiveness (OSEE) index is intended to help reconcile major social goals with everyday operational decisions in practice. This will also make it easier for industry to meet European sustainability requirements, which will play an increasingly important role in the future.
Sustainability as a corporate goal
"Efficiency plays a decisive role in all areas of production engineering," says Prof. Fazel Ansari, Head of the Production and Maintenance Management Research Division at TU Wien. "Of course, you want to use raw materials as sparingly as possible, consume as little energy as possible and have as few machine failures as possible.
"But until now, this has only been driven by the need for economic efficiency. However, ecological and social sustainability are independent goals in their own right; they are values that must be anchored in corporate objectives. We therefore wanted to develop a system that not only reflects economic aspects but also ecological and social aspects and makes them tangible, comprehensible and optimizable for corporate management."
Today, for example, it is common practice to define an "Overall Equipment Effectiveness Index" (OEE)—a measure of how effectively equipment is used in relation to its theoretical optimal utilization. Ansari and his team propose instead an OSEE index—for "Overall Sustainable Equipment Effectiveness."
This index incorporates parameters such as a machine's energy consumption, direct and indirect CO2 emissions, consumption of raw materials, lubricants or water, waste production and the service life of components. At the same time, social factors are also taken into account: What are the working conditions like during operation and maintenance? What are the ethical standards along the supply chain? How well does knowledge transfer work within the company? Does training provide sufficient opportunities to build up the necessary expertise?
"We have found that companies already have a great deal of data that can be used to answer such questions," says Fazel Ansari. "Unfortunately, it is often not used, or not used in the best way," says Theresa Madreiter, doctoral student at TU Wien and research associate at Fraunhofer Austria.
For example, a lot can be learned from sensor data from production facilities, operating data from central production control can be incorporated, as well as staff working hours and attendance data, knowledge documentation and training data, and even experience reports and feedback from maintenance personnel.
The factory in the computer
All this data is then used to recreate a multi-layered AI model of the production processes. "We create a network of data and experiential knowledge to show which work steps, machines and people are dependent on which other work steps, machines and people. This allows us to see how different activities influence each other, how a failure at one point affects other processes—and what impact a particular change has on the sustainability of the overall process," explains Ansari.
Although Fazel Ansari and his team have been working with artificial intelligence for years, the model was not simply implemented as a neural network, as is common with large language models. "It is very important to us that the results of the model are comprehensible and explainable step by step. That is why we are not using an LLM in this case, but a Bayesian network, which allows the significance of each individual measure to be logically understood."
Application to the metal manufacturing industry
The system not only provides a key figure for the sustainability of processes, it also allows concrete diagnostics of current processes and a well-founded forecast of how specific changes in production will affect the complex network of workflows. Fazel Ansari's team is currently working with companies in the metal manufacturing industry—an area where even small improvements can have a major impact on sustainability.
More information:
Theresa Madreiter et al, From OEE to OSEE: How to reinforce Production and Maintenance Management Indicator Systems for Sustainability?, IFAC-PapersOnLine (2024). DOI: 10.1016/j.ifacol.2024.08.121
Theresa Madreiter et al, Sustainable Maintenance: What are the key technology drivers for ensuring Positive Impacts of Manufacturing Industries?, IFAC-PapersOnLine (2024). DOI: 10.1016/j.ifacol.2024.09.232
Provided by Vienna University of Technology
