High-level control in cement production

Attaining ‘Industry 4.0’ has been an essential task of the industry for years. Recently, terms such as ‘big data’ and ‘AI’ (Artificial Intelligence) have been used heavily in many fields. AI combined with big data is hoped to provide solutions to long-standing problems, not only in automation. It might therefore be surprising to learn that fully autonomous mill operation (including the use of AI) has been taking place since 2009. This article will briefly summarise how the methods of High-Level Control (HLC) have been used in the cement industry in the early 2000s and how they manage to control ever more complex closed-loop-controlled processes where standard controllers fail.

To accelerate the integration of advanced technologies in the cement industry, some business consultants proposed to ‘copy-paste’ Industry 4.0 solutions from chemical plants/refineries and apply them in cement plants. A recent example is a report regarding the first successful conversion of regular plant control to AI control which qualified it as a breakthrough. Caution is needed here – the capabilities of AI are still limited as its history shows.

AI is a very broad term and it is difficult to find a definition of the concept on which everyone agrees. In the wider sense, it can be defined as a branch of computer science dealing with the simulation of intelligent behaviour in computers, i.e. the capability of a machine to imitate intelligent (human) behaviour.

Technically speaking, the majority of the AI-systems which are used in industry today are data-driven algorithms. The basic principle of these algorithms is relatively simple, but they gain their capabilities from huge amounts of data, high repetition rates of calculations and multiple interconnections.

The usage of AI for many tasks is not a new idea. The development of faster computers with the possibility to store and process huge amounts of information (Big Data) makes the use of AI possible and reasonable. Deep Learning, which is itself a part of Machine Learning, makes use of multi-layered artificial neural network (ANN) to learn from Big Data and search for patterns that could be used for problem solving after some training of the ANN.

Knowledge based automation including fuzzy logic and analytical methods such as, for example, Model Predictive Control (MPC) are also part of AI (Figure 1). Depending on the task, different AI methods are more applicable than others. Nowadays it is clear that there is no one-size-fits-all AI module for the cement production process like in refineries. Fuzzy logic is applicable for closed-loop control of technical processes with a moderate number of variables and data for which a control strategy can be expressed. It is a good choice for processes, where safe operation in critical situations is mandatory. Neural networks are used for recognition of hidden patterns for processes for which a control strategy cannot be expressed and that have a high numbers of input variables. MPC is a good choice for well-understood processes for which a mathematical model is available. Optimisation is possible if the model can be calculated faster than real time. On-line adaptivity is seriously not practicable in cement production.

Figure 1. High Level Control relies on AI.

How AI started in cement production

A short discussion with AI-solution providers made it clear that this ‘revolution’ in cement plant control systems is simply operating with MPC and soft sensors. Powitec Intelligent Technologies (Germany) used self-adapting MPC and machine learning in 2001. It was the first comprehensive black-box controller that operated a rotary kiln fully autonomously for more than 24 hours without manual interaction. The core of the system was an image-processing camera analysing the main burner flame and an online prediction of free lime. Various AI components were used as far back as 2002. Using adaptive MPC, the fluctuating energy input to the kiln and calciner was adjusted automatically. Pioneers in the use of AI during these days were the cement producers LEUBE (Austria) and Maerker Zement (Germany). Soon after, companies such as ABB, FLSmidth, Pavillion, KIMA Echtzeitsysteme and Rockwell entered the market with similar model-based controllers. Today, many companies have returned to the more robust fuzzy logic control. Predictive models are used for soft sensors. The reason for this is the serious difference in the cement manufacturing process compared to other common production processes.

Figure 2. Clinker bed in a pet-coke and coal-fired rotary kiln – a unique, complex process with numerous varying conditions.

Special process conditions

Clinker production is a complex operation. The so-called ‘multi-dimensional non-linear process model’ of a kiln or a mill has failed to model the real systems adequately. The real kiln or mill unit is subject to wear and other natural variations, which models so far have failed to represent. Roughly speaking, the behaviour of a kiln or mill tomorrow will be different from its behaviour today. There were attempts at implementing software features, such as self-adaption and self-learning, but a host of changes have to be considered: liners, balls, chutes, feeders, valves, refractory, fuels, and the raw materials. And then there are even further changes to consider: the quarry components and additives, the fuel calorific value, water and ash changes, the change of coal and pet coke particle size distribution from their mills and the related change of combustion (ignition point, burn out, same shape, etc.). All these changes can affect the quality of clinker and cement – a significant challenge for a controller. If a multi-dimensional process model is fed with signals that have a drift, or are not stable, it fails. It is possible to teach these models with ANN in order to catch the drift and to adjust the model accordingly. The question is how fast the process changes. It can happen that an ANN based model has to be re-trained regularly which is impractical for many manufacturing processes. Consequently, this means that for ‘drifting’ systems MPC will not be the best solution. In such cases, a rule-based system is needed to control the process when MPC or/and ANN results are not realistic.

Assisted by modern modules of AI, these systems reach a new level in automation. This article will detail some plants which operate their mills fully autonomously over several days. The ‘auto-pilot’ is not limited to smooth operation conditions only. KIMA’s MILLMASTER allows for fully automated start and stop of the mill, automatic recovery after emergencies and switching between cement types. The following case studies will briefly discuss some plants that increased their performance using KIMA’s SMARTCONTROL. The software-platform was supplied globally in nearly 200 rule-based fuzzy-expert systems known as MILLMASTER, which use all AI modules that are shown in Figure 1.

High Level Control as a combination of different AI modules

In 2008, KIMA Echtzeitsysteme (the previous name of KIMA Process Control) published an article about a project to supply 30 SMARTCONTROL packages for ball mills (including the SMARTFILL fill-level measurement system) to a selection of Holcim group plants in Eastern and Central Europe. After the commissioning, and later during steady operation of these plants, the development of MILLMASTER continued separately in the Holcim group as well as in KIMA Process Control. New designs of the human machine interface, programming logic, and new software modules were developed to follow new trends in automation. The aforementioned case studies show the results and report the experience of the current users of this product.

KIMA’s SMARTFILL is a precise fill-level measurement system for ball mills. It measures structure borne sound loss-free directly on the mill shell and converts this signal into fill-level information (see Figure 9). SMARTFILL can not only be integrated with KIMA’s MILLMASTER but also with the control software from LafargeHolcim, HeidelbergCement, and BuzziUnicem. SMARTFILL has made KIMA the market leader in this instrumentation field, as it provides robust and drift-free process signals, which can be used for automated operation with MILLMASTER in combination with other process variables. Sensors like the V-SENS (vibration sensors) and the newly developed T-SENSOR (a contactless torque sensor) build upon the success of SMARTFILL and make HLC also available for VRMs and combined grinding circuits with roller presses as elaborated in the next section.

High Level Control of ‘difficult systems’

Figure 3. A sketch of a typical grinding circuit, consisting of a roller press and a ball mill.

Figure 3 shows a combined grinding circuit, consisting of a roller press and a ball mill. Such a grinding circuit for cement is quite widespread because of its advantageous specific electrical energy demand and the product quality. From the viewpoint of control theory however, this a difficult system, which consists of three subsystems (roller press, ball mill chamber 1, ball mill chamber 2), each of which is critical, as they are part of a respective feedback loop (via the separators) with individually different delay times (or time constant). The feedback loop enables each of these subsystems to oscillate on their respective resonance frequency, which is determined by the aforementioned time constant. Even worse, these time constants are nonlinear functions of clinker quality (grindability), which is never guaranteed to be constant. Conventional PID-controllers are not able to handle such systems that are prone to oscillations.

To better understand the complexity of the combined grinding circuit, such a system can be compared to three pendulums, which are coupled by springs with different stiffness as shown in Figure 4. The time constant of each pendulum is determined by its mass and the length, and the coupling is determined by the stiffness of the connecting springs. In the ideal case, this system is excited (shifted) by a constant raw feed, and each pendulum moves to a new equilibrium state and remains there.

Figure 4. Mass-spring model of the combined grinding circle with a roller press and a two-chamber ball mill.

In practice, however, the excitation by raw feed is not constant, the length of each ‘pendulum’ changes in time and the stiffness of the springs change in time also. The result is a system, which is oscillating permanently at varying frequencies and amplitudes (see Figure 5).

Figure 5. Excited spring-mass system.

The task of a closed loop control system is now, to adjust the excitation (i.e. raw feed), the resonance frequencies (length of pendulum, i.e. transport speed of bucket elevator, conveyors belts and air slides) and the stiffness of the coupling springs (i.e. mass flow from roller press into chamber 1 and from chamber 1 into chamber 2). Controlling such systems is a difficult task and cannot be performed by using a single PID-Controller. According to KIMA’s long lasting experience, such a system can be controlled successfully with MILLMASTER.

Case study 1: Fully autonomous mill operation at EQIOM Ciments

Autonomous mill operation using modules of AI is no longer science fiction. In 2009 MILLMASTER was implemented in a VRM automation project in Northern France. At its plant in Dunkerque, EQIOM Ciment’s Loesche type LM 46.2+2 S VRM mainly produces slag cement, and the mill is controlled by a MILLMASTER system. Every weekend (Friday afternoon to Monday morning) the plant is operated completely unmanned and MILLMASTER runs the mill fully autonomously.

Figure 6. Vertical Roller Mills usually have also a big potential of automation and optimisation.

“The MILLMASTER system is used daily and gives us the opportunity to concentrate on performance optimisation while the mill is running. It is also faster than an operator when it comes to protecting the equipment in case of important changes in mill behaviour. It would be hard to run without this expert system for a long period.” Pierre Vonstein, Operations Manager for North and Normandy Grinding Stations, EQIOM.

Figure 7. MILLMASTER Cement type configuration screen for CEM II/B-M (V-L) 42.5 N.

Case study 2: Increased production at Fabrika Cementa Lukavac

A key advantage of the MILLMASTER system is that it can be configured in such a way that the operators do not see much from the system. Just the ‘on/off’ switch allows them to start or stop the ‘auto-pilot’. Switching it on is only for one purpose: increased production. A representative example is Fabrika Cementa Lukavac D.D. in Bosnia and Herzegovina. In 2018, the system was installed on a 65 tph design ball mill, which usually reached a base line of 67 tph as Emir Cilimkovic (Process Manager) reported. Here are the ‘before-and-after-results’ for the two cement types produced in the plant:

Figure 8. View on the Lukavac plant, Bosnia and Herzegovina.

Confirmed production increase

• >10% production increase (see Table 1).
• Additional results: Reduced fluctuation in quality compared to the baseline situation.
• Less wear of grinding media (compared with operation without MILMASTER) and mill liners because of higher material level in the mill.
• The optimal filling level was confirmed by means of the crash stop method.

Figure 9. The SMARTFILL device on the mill with chamber 1 and chamber 2 sound sensors.

Case Study 3: Increased production at Opterra Karsdorf, Germany

After the merger of Lafarge and Holcim, plenty of plants were told to switch off their former expert systems LUCIE (Lafarge) and MILLMASTER (Holcim), as they left the group. Both expert systems required a certain level of support by technology centres, and experts visited on a regular basis to secure successful day-to-day operation. Following this, a "couple of well-known suppliers were asked to equip these plants with an alternative software that offers the same performance as the previous systems, but — if possible — with easier handling from the plant site and without the necessity for regular maintenance from external resources. In 2015, KIMA Echtzeitsysteme was awarded a contract to equip all ball mills at the plants that had been acquired by CRH in Germany. While there were a few concerns about the small — albeit well established — supplier, KIMA, it was also known that this company had previously equipped some 30 Holcim plants in Eastern Europe with SMARTCONTROL systems, such as MILLMASTER and KILNMASTER. The former Lafarge plant in the small eastern German town of Karsdorf (some 50 km away from Leipzig) was also equipped with the MILLMASTER expert system. In total six cement mills, four of them centre discharge mills, received an individual software package and parametrisation to reach the challenging optimisation guarantees of between 5% and 8% of production increase or, alternatively, 4% to 6% of specific energy demand reduction. It is also worth mentioning that the on-site commissioning of a MILLMASTER for a ball mill usually takes no longer than five days. The rest of the work is executed remotely via a VPN connection.

Table 1.

To conclude, all performance goals of the project at Karsdorf were reached. And, last but not least, the expert systems found total acceptance by the operators, as Tim Fröhlich – the performance engineer in the plant – reported.

Conclusion

Most people believe that AI consists only of self-learning ANN that feed on Big Data to automatically control complex processes. This is not accurate. As already mentioned, different AI methods should be chosen depending on the task. It seems that the complexity of the clinker burning process makes it a bad candidate for pure ANN control, not least because of the wear of the involved equipment which changes the system that the neural network was initially trained for. The significant components of AI being used today are similar to those used 10 or 20 years back. But today, the computers have become much faster and have access to large amounts of historic data.

Read the article online at: https://www.worldcement.com/special-reports/30102020/high-level-control-in-cement-production/