Advanced Process Control for Cement Production

Related Vendors

TURCK GmbH

Italvacuum, Srl

ProMinent GmbH

Siemens AG

Functional Principles of the Mill Control Systems

The new mill control systems works in three steps. First, the neural soft sensor records a total of nine process input variables and uses these to predict the cement fineness. These input variables are:

• separator speed;

• quantity of fresh material;

• swirl flap position;

• four variables representing the fill level of the mill: circulating elevator flow, mill drive performance, acoustic sensor chamber 1, acoustic sensor chamber 2

• mill temperature; and

• recipe type.

The fineness prediction from the neural soft sensor is compared with the values measured by the lab. The compared value for “fineness” (quality) and the measured fraction of “rejects” (oversize material) form the control parameters for the MPC. The values are optimized by changing the feed quantity and the separator speed. This is done via predictive calculation of the relevant control procedures based on a complete process model to bring the control parameters as near as possible to the desired setpoints. External setpoint settings are thus calculated for the lower-level individual controllers to optimize the future behavior of the plant over a specific time period.

Conclusion: Payback in Three Months

By determining the setpoints of the individual controllers more accurately than before, the MCO system ensures optimum throughput, energy consumption, and product quality, and maintains stable production conditions that can be adapted to the current mill situation at any time. A further advantage is that the time lag between a change to the system inputs and the corresponding response at the system output is significantly reduced. The result is a more uniform grinding process which optimizes the throughput of the mill while retaining product quality. It also makes the plant operator’s work easier. Energy consumption per tonne of cement has fallen, and the service life of the plant’s mechanical components has increased. Compared to operation without an expert system, the use of Sicement IT MCO achieved a performance boost of 5–8 percent, and the investment paid off within just three months.

In contrast to the old control system that was based solely on the Blaine value (a measure of the degree of fineness) as the input variable, Sicement IT MCO takes its inputs from parameters that are set much further upstream. As well as avoiding time lags, this also makes it much easier to collect and compare input data, and so paves the way for further optimization in the future based on an improved understanding of the grinding process. In contrast to other new systems, Sicement IT MCO can also be fully integrated into the plant and the existing PCS 7 control system, so there are no additional servicing and maintenance costs for the plant operator.

Following the success of the project to optimize control of the first ball mill, Siemens has now started work on the second mill, and the two other mills will follow in due course.

* T. Kopetzky is Project Manager, R. Wieser is Marketing Manager, and T. Walther is Director Competence Center Cement, all with Siemens AG Industry Sector. G. Hefter is Production Manager, Südbayerisches Portland-Zementwerk Gebr. Wiesböck & Co. GmbH, Rohrdorf/Germany.

(ID:27777410)