"We look at the entire value chain"
The digitalisation of the steel industry presents major challenges for plant engineering. think.steel spoke with Katja Windt, Member of the Managing Board (CDO) at SMS group, about the benefits of digitalisation for steel producers, predictive maintenance and sustainability. Digitalisation also brings to light aspects that have not yet been given much consideration – for example, safety. You can read why this is the case in the following interview.
think.steel: What is meant by the term “services”?
Katja Windt: As SMS group, we engage with our customers by establishing long-term partnerships. Our aim is to maintain and enhance the performance of our customers’ plants throughout their entire life cycle. Together with our customers, we develop integrated solutions that are specifically tailored to their individual requirements. Following the modular principle, we combine products and services from the three areas of
electrics & automation, digitalisation and technical service. We focus on key performance indicators such as plant availability, product quality, productivity and delivery reliability, as well as on sustainability and occupational safety. By combining our integrated services with performance-based business models such as Equipment-as-a-Service or similar, we become a long-term partner for our customers and help them to focus on their core competencies.
How do customers engage you?
In greenfield projects, we are an integral part of the plant through the ‘digital-ready’ principle. Everything from a single source. Our digital tools cover the entire value chain, utilise data seamlessly retrieved from the automation systems and provide direct input for the overarching digitalisation of the relevant service. In brownfield projects, we develop projects in collaboration with the customer: the aim is to identify specific KPIs and problem areas, develop solutions and then resolve them in partnership with the customer. Particularly in the brownfield approach, we work on a use-case basis and build bespoke solutions that address specific pain points and KPIs.
An integrated mill consists of six or more plant segments or units, such as the liquid phase, casting and rolling. Do you consider them separately or as a whole?
We manufacture products that span the entire value chain. The steel industry always faces a major challenge with order sequencing, i.e. the sequence of orders for the rolling process. There are strict and complex rules, which make planning and control difficult. For the customer, for example, it is important to capitalise on the potential in intermediate stock levels – where buffering is used to adjust the sequences. There is considerable potential not only within the individual units of the plant, but also between them – for example, between the hot rolling mill and the cold rolling mill. At each of these units, the staff are well versed in how to combine their batches. Quite often, the sequencing of orders – for example, at the hot rolling mill and subsequently at the cold rolling mill – is not optimised in terms of cost and lead time. This means that, at the cold rolling mill, the staff there have to start the sequencing process all over again. In a ‘golden batch’ analysis, we analysed the energy flow across the entire plant process, rather than per unit. – For example, which production orders should be combined into a single batch to ensure they are produced as efficiently as possible in terms of energy? It is often the case that each plant unit optimises only for itself, resulting in bottlenecks between them. Coils or slabs end up in storage – which is costly.
What digital potentials are we talking about in concrete terms?
It is always of high importance for the customer to reduce downtime of the equipment. To ensure that, we first analyse the customer’s problems, e.g. at the electric arc furnace to figure out the reasons why there are downtimes occuring. Essential is to identify with the domain experts (e.g. metallurgists, electrics&automation experts) the relevant parametres. With new AI approaches such as pattern recognition we identify further relations of combinations of parameters which can be a cause for the downtime. Next step is to derive as a counter measure parametre adaptations and to develop digital applications to predict such situations early enough for counter acting prior further downtime occurs. By bundling hardware, automation and software, and services into one solution, our customers can fully benefit from reliable data and insights which are converted into actionable items and direct operator guidance. The procedure is strongly supported by our SMS Data Factory which structures data from the automation level for further digital analysis.
There are customers who are already highly digitised. How do you proceed here?
No matter what level of digitalisation our customers have already reached, we provide further development potential to them – be it product quality improvement, asset health solutions or predictive production planning. We are always one step ahead and providing latest technologies. This is the for example the holistic planning approach. Here we provide across plants optimised sequencing of orders in steel plants for reduced delivery times and higher yield as well as lower work in progress (WIP) levels. Furthermore, the following areas are of importance: Autonomous production and the development of real-time security applications through industrial 5G. Autonomous production means that the control and automation processes can be carried out mainly without workers intervention. Steel mills are dangerous places for the health of people, which means that in principle employees should not enter unsafe areas. For example, via a message on the helmet or goggles, we can inform the worker when he or she enters an unsafe area and identify this unsafe area with digital applications in advance..
How can predictive maintenance be classified within digitalisation?
Predictive maintenance is based on 4.0 logic. The predecessor, industry 3.0, still implies, that one analysis why errors have occurred and try to avoid it next time. In 4.0 logic, we try to predict the error with pattern recognition to prevent it before it occurs. Pattern recognition means that one always evaluate different parameters in combination with each other. They then form high dimensional correlations, i.e. under which circumstances the error occurred. The next time a similar constellation is identifiable, the parameters are adjusted prior the failure takes place. This way, for predictive maintenance we can better foresee whether a part will fail e.g. in two or in three weeks. On the one hand, the goal is to prevent failures that occur ad hoc. On the other hand, the goal is to predict the lifetime of a part more accurately – hopefully then the analysis will show that enhancement of the lifetime of a component in planning is possible – which will result in cost reduction. The better the learning cycle of such AI algorithms is, the more accurate one can predict and the better one can prevent failures and with that to increase the overall performance in service delivery and yield at the customer plant.
How far can working with the data go?
During plant operation, we ensure a stable production process, maintain and increase plant performance, and help our customers to make data-based decisions. Therefore, based on profound data analysis including AI algorithms like pattern recognition, we turn data into value. With our knowhow, we can correctly interpret the data obtained to derive forecasts and, above all, give recommendations for action. Such “actionable items” can either be fed directly back into the automation system in a closed control, or they are passed on to the operator as a recommendation for action. Figuring out root causes, SMS group has advantages compared to pure software companies because we already know the processes of how to produce steel in detail. We have metallurgists and other experts from E&A, mechanics and digital in house. Based on our longtime experience as a company of 150 years age, we have already modelled steel production with sophisticated models based on math and physics. Added to this, is the experience at various customer’s sites. Overall, we know the main influencing parameters and how to interpret them even in high-dimensional parameter space with the help of pattern recognition and transdisciplinary experts. This means, that if we have the data available, we know which parametre to take into account and which not, how to analyse data and to derive countermeasures to provide the best solution to our customers.
How can the digital twin be addressed in concrete terms?
For different purpose different types of digital twins exist. For example for validating our new automation software we perform so called integration tests. Besides testing, our automation allows our customers to get acquainted and trained with the new system before it went towards real use. We design digital twins for all of our plants. Furthermore, this virtual plant is connected to the hardware elements of the corresponding control panels of the equipment which ensures that our customers can train their employees in both ways: virtually and by using the real control panels.
What is particularly important for you to transport on the subject of “services”?
It is particularly important for us that we as SMS see ourselves as partners of our customers and try to offer solutions across the entire plant from mechanics to electrics & automation and digitalisation, which enable the customer to control processes optimised in terms of product quality, costs, particularly energy costs and asset health as well as due date reliability.