echnology has recently made a giant leap forward. Our phones are now smarter than ever, our computers seemingly anticipate our every need, and our home devices can regulate room temperature, turn on the lights, set daily schedules, plan our commute, conjure up virtual-reality environments, and help improve our health. Autonomous cars are on the verge of large-scale commercial deployment, and self-driving trucks and other forms of transport will follow them shortly. Our lives are changing fast, and one senses that the acceleration curve of innovation has never been steeper.
The ubiquitous cloud
Powered by large server arrays that we call the “cloud,” many of today’s consumer technologies exhibit a level of intelligence that is both remarkable and overwhelming. We may take it for granted that the contacts on our mobile phones are updated automatically and that we can dictate search queries instead of typing them in letter by letter. Yet the underlying complexity of these tasks is something we usually only have a vague idea of. Exactly how cutting-edge electronic products build up their intelligence is hard to grasp—even if it appears that their continual connection to the cloud must have made a vast contribution to the recent increase in performance. This is why it can be fascinating but also somewhat intimidating to use cutting-edge smart devices.
Smart devices are defined by three characteristics. First, they have a means of connecting to a network or to other smart devices using data-transfer technologies and protocols such as mobile broadband, Wi-Fi, and Bluetooth. Secondly, they are often small in size but carry sufficient processing power to enable complex operations as well as interaction with one another and the Internet. Thirdly, they are capable of collecting input from the human end user in a variety of forms (text-based input, speech, pattern recognition, etc.) and provide information and entertainment in a multitude of forms. As for the form factor of today’s smart devices, they range from smallsized smartphones to larger hand-held tablets to surface computers, TV sets, and household appliances. They are also an integral part of the smart steel plant of the future, for instance in the form of smart sensors.
It is not only consumer technology that has made giant advancements. The same is true for steel-production solutions. They too have become “smarter,” and they too have reached a level of sophistication that is both awe-inspiring and occasionally intimidating—even to seasoned industry professionals who have continuously worked hard to stay up-to-date with innovations made in the steel industry. Terms such as “Industry 4.0,” “digitalization,” and the “Industrial Internet of Things” have tried to capture some of the pivotal aspects of the recent technological progress. But so far, these keywords have fallen short of clarifying the ultimate goal of the initiatives associated with them.
The future is clear
In this issue of Metals Magazine, it is our goal to thoroughly demystify these terms. Rather than talk of an abstract “Industry 4.0,” we at Primetals Technologies would like to discuss the path we see leading into the digitalized future of steel production. It is clear to us that this path encompasses a multi-dimensional, comprehensive digital integration of all production equipment found in a typical steel plant. What is crucial for meeting this target is not only interconnected components but also innovative software that makes the complexity arising within a production facility more manageable.
As in other industries, the accumulation of a plant’s production data by use of smart sensors is only the first step. It is a prerequisite for the creation of a “digital twin” of the respective facility, which then takes transparency in steel production to the next level. In the end, the target is to enter a world in which steel is made with fewer resources on all fronts—from energy requirements to raw-material expenditures to labor costs.
Steel producers will also be more flexible. No longer will the paradigm of “producing more for growth’s sake” be the reasoning of choice. Instead, producers will be able to re-adjust their production setup with such ease that very small lot sizes can be achieved and customization can reach new levels of depth. Seamless documentation of the manufacturing process will ensure that all end products not only meet their specifications but that their full production history is known in detail, which is a decisive factor for complete and lasting customer satisfaction.
The term “digital twin” refers to a detailed digital representation of physical assets including the processes that these assets allow to be executed. Importantly, a digital twin has to dynamically reflect any changes made to its actual, physical counterpart. This is done by bringing together a semi-modular base structure, advanced process models, machine learning, software analytics, and artificial intelligence with clearly defined targets. The strength of a digital twin is that it continuously learns and updates itself ideally in real time, utilizing the combined information from a variety of sources. Besides sensor data, a capable digital twin relies on expert knowledge gathered from skilled human operators, on the technical expertise of specialized engineers, on findings stemming from similar digital twins, and on the larger environment that the digital twin may be a part of. Of course, historical data also plays a big role in advancing a digital twin’s depth. When applied to a steel-production facility, the main benefit of introducing a digital twin is that any envisaged changes to the physical production chain can be tested and evaluated beforehand in the digital domain at zero risk.
Where are we now?
So if this is where we are going, then where are we now? Let us make a brief assessment. When it comes to process automation, the steel industry has been ahead of other sectors for decades. While most steel producers have employed and refined “Level 1” and “Level 2” automation for years now, many companies belonging to the discreet-manufacturing sector are lagging behind in terms of automation. And that is really just one example of an industry that has some catching up to do.
Automation in steel production is special in that it is not only based on process models but also on abstractions of what happens inside the liquid steel itself—particularly at the stage of solidification. Thermodynamics play a huge role in this context, and the related mathematical models are indeed as complex as one would expect them to be. The next step of course is to use three separate but linked types of models simultaneously: a model of the steel as it is being transformed, a model of the entire steel-production process from the ironmaking stage to the final treatment, and a model of the complete plant, as built, including “invisible aspects” related to equipment that is not a direct part of the production chain.
The smart plant
With these three classes of models in place, the groundwork has been laid for a steel plant to become an “intelligent” plant. What we mean by this term is the following: A smart plant “knows” the state it is in, thanks to condition monitoring, and can make adjustments based on its findings with elaborate automatic functions. Artificial intelligence will increasingly find room to grow in this area, despite being in its infancy today. The plant of the future also facilitates “smart work,” which involves all staff members being given the information they need to execute the tasks at hand. Predictive maintenance ensures that production stops are proactively avoided—and that they can be scheduled far in advance, if necessary.
The smart plant we envision relies on three dimensions of integration. The first dimension is that of horizontal transparency and interconnectedness. This means that, for instance, the rolling section of a steel-production plant “knows” precisely what was done at the earlier casting stage, and can therefore react and adjust its parameters accordingly. Depending on the raw materials used, the entire plant has to be run in a specific manner. The smart plant can accommodate a variety of different input-material choices. In other words, it allows producers to benefit from greater “input-material flexibility.”
A unique feature of the advanced automation systems developed for the steel industry is that three classes of models can be used simultaneously and in conjunction with one another. This is an area in which Primetals Technologies has done significant pioneering work. Among the first class of models are those that capture the overall setup of a metals-production plant. A model belonging to this group encompasses the core plant layout including the production equipment of all process stages. The second class relates to the processes taking place inside the plant. These types of models are detailed abstractions of what is done by any given piece of equipment over the entire length of the production chain, and at any given time. The third class is arguably the most complex of the three. Its models are representations of what goes on inside the product itself as it is being created and formed. They capture the changes that occur in the liquid steel when it is being cast and when it solidifies. Models of the third class extend to the rolling, further processing, and coiling of the steel, thereby making tremendous complexity much more manageable. It is the parallel existence of these three classes of models that make
the steel industry stand out.
The second dimension is about time. Every piece of equipment in a steel plant has a particular story, which usually begins with the basic engineering at the planning stage of the respective unit. Once the new facility is built, or, in the case of a revamp, once the new components are added, a period of maintenance-supported operation follows. The smart plant ensures that the overall lifetime of all equipment is maximized, and that any service work is scheduled for such times when the impact on productivity is lowest.
The Digital Unity
A combination of three software layers represents the third dimension of integration in a smart plant. At Primetals Technologies, we have chosen to partner with German-based PSI to provide our customers with a state-of-the-art production-management system (PMS). This software solution dynamically plans and tracks all plant activity at every step of the production chain. Customer orders are transformed into instruction sets that are then executed in a highly optimized way.
The PMS is accompanied by Primetals Technologies’ own Through-Process Optimization (TPO), which scans the entire production process for non-conformities and identifies areas of improvement. When used in a smart plant, TPO can dramatically accelerate a producer’s time to market with cutting-edge, high-value steel grades.
Finally, the Maintenance and Asset Technology (MAT) of Primetals Technologies assists and simplifies the maintenance efforts in a smart plant. This system features comprehensive, built-in application knowledge that generates actionable items for the maintenance staff to carry out. All suggestions are devised in accordance with a customized maintenance strategy for optimum efficiency. Advanced analytics give further insight into the inner workings of the plant and into maintenance activities in particular.
It is no secret that artificial intelligence (AI) is still at the beginning of its development. However, the impact it could have on our society once it reaches a more advanced stage is already a much-discussed subject. The core idea behind the concept of AI is that machines are brought to a stage where they can perceive their environment and take action to maximize the probability of a certain outcome. In consumer electronics, AI can be seen at work in speech recognition, the anticipation of Internet search queries, the automatic display of context-related information, and many other applications. When it comes to steel production, the engineers of Primetals Technologies have already started to implement AI technology in a selection of established data-evaluation processes. Eventually, AI is expected to greatly facilitate the analysis of all production steps within a steel plant. For this to become reality, (smart) sensors must continually collect large amounts of data at all stages of the manufacturing route and for all of the involved equipment. While predictive maintenance has already made great advancements in steel production, AI will push the boundaries of what is possible and significantly increase overall efficiency as well as the reliability of the production process.
Arguably one of the most impressive characteristics of a smart plant is the possibility to run “virtual startups” within a dedicated software environment called “Maintenance and Simulation System” (MSS, see pg. 60). The MSS lets engineers test new equipment configurations so that only the most efficient setups will be implemented in the real plant. Highly detailed abstractions of some of the steel plant’s physical components serve as the basis for the MSS. These abstractions are sometimes introduced into “cyber-physical systems.” These systems are bridges between the physical world and the digital domain, and represent aspects from both areas. They are a reflection of the immense complexity at work in a steel plant, and point the way to the future. Besides startup simulations, the MSS can greatly ease the development of new production strategies, for instance, when a change in the core product mix is desired.
The smart plant is a major step forward for metals production. Among its main benefits are the possibility to tailor end products to even more specific customer requests. Smaller lot sizes can be achieved without a significant loss in productivity. Everything that is produced at the plant is constantly monitored, tracked, and evaluated, resulting in seamless production documentation for every single order. The smart plant frees up producers to obtain input material from a multitude of sources, as the manufacturing chain can be more easily adjusted to accommodate changes in raw-material quality. Efficiency is improved on many levels. The associated optimization ranges from a reduction of energy consumption to a more effective distribution of human labor. Importantly, the smart plant introduces significant advancements in terms of lowering the carbon footprint and increasing environmental friendliness.
Two of the main characteristics of a smart plant are that the facility can determine its own state, and that it can learn from its own assessments and analyses over time. Both aspects can only be achieved with a holistic approach toward the steel-production process. It is precisely the perfect alignment of all components at work in the facility—by means of their interconnectedness—that can make the smart plant a reality.
Our Metals Orchestra
At Primetals Technologies, we have devised the concept of the “Metals Orchestra” to illustrate both the finely tuned interaction of all production equipment and its automated operation by the introduction of one unifying metaphor. Just as an orchestra consists of many proficient players, a smart plant comprises numerous different components that must all act in harmony to deliver a standout performance. While it is essential that all players can reach similarly high levels of virtuosity, they also need to be optimally directed to effectively complement each other. This is where, in an orchestra, the conductor comes in. The same is true for a smart steel plant—but here, there are no fewer than three conductors doing their job in perfect unity: the PMS, TPO, and MAT, the software and know-how systems we mentioned earlier.
More needs to be said about these conductors, and the same is true for the many individual players of the “orchestra of steel.” Over the course of the next 64 pages of this issue of Metals Magazine, we are therefore presenting a selection of those technologies that ideally prepare a steel plant of today for the world of tomorrow. These components and software solutions will take a steel plant to the next level—and make it one of the first smart, self-learning plants of the future.
- Steel producers can obtain raw materials from different sources and easily adapt to variations in quality
- Higher degree of customization of orders
- Smaller lot sizes without a significant loss in productivity
- Seamless documentation of production history for every single end product
- Higher degree of plant optimization leading to a more efficient operation
- Fully transparent processes and workflows
- Improved environmental friendliness
- Lower energy consumption