Advanced Process Control for Cement Production

Related Vendors

TURCK GmbH

Italvacuum, Srl

ProMinent GmbH

Siemens AG

Requirements For Optimized Control

To optimize the energy consumption of the mill, Siemens has developed a new mill control system. The first pilot project involved one of the four ball mills at the Rohrdorf plant. Dating from 1965, this mill consists of two chambers: the first chamber is filled with large steel balls, and the second with smaller balls. The ball mill has a length of 11.6 m and a diameter of 3.8 m, and it has a rotational speed of 15.4 rpm. The drive power is 2,400 kW, and the grinding capacity is 60 t/h.

One of the most important quality-relevant parameters in cement grinding is the fineness of the product. Samples are taken every hour, analyzed in the lab, and the resulting particle size analysis used to correct the process. Acoustic sensors, meanwhile, record the filling level of the mill.

The idea behind the new control system was that it should record and adjust the quality parameters automatically. A further requirement was to replace the existing optimization system and to integrate the new system into the existing control system without any significant overhead. This demanded software that uses a knowledge-based approach and current plant data — an expert system — to predict the plant’s quality parameters and control them automatically, thus relieving the plant operator of these tasks.

Another aim was to achieve maximum throughput with the desired level of grinding fineness.

Architecture and Implementation

The result is a system known as Sicement IT MCO (Mill Control Optimization) from Siemens, based on components from the advanced process control (APC) library of the Simatic PCS 7 control system. Using APC, even complex relationships between process parameters and plant variables can be described mathematically and used to control the plant automatically and flexibly.

In this case, a neural “soft sensor” records the process input variables and predicts the fineness of the cement leaving the ball mill. To reduce process deviations and to stabilize the grinding process, a model-based predictive controller (MPC) is used; this contains a complete model of the process dynamics with all interconnections. Together, the neural soft sensor and the MPC system provide accurate and responsive control of the complex grinding process.

The MCO system was implemented in three phases. First, the plant was analyzed to identify the existing input and output variables and examine the quality of the existing control solutions. In the second phase, plant tests were carried out to determine the step responses and collect production data. Once the data had been analyzed, the models for the neural soft sensor and the MPC system were developed. The third phase included the software engineering needed to integrate Sicement IT MCO into the existing Siemens PCS 7 control system.

The Sicement IT MCO system was commissioned without shutting down the ball mill. Initially the new controller ran in the background, shadowing the existing system. Once the new system had shown that it could handle all the required functions, the old system was switched off. The system was then refined to produce optimum results for each of the eight recipe types defined in the specification.

(ID:27777410)