Process optimisation: never the same again

By applying recently developed AI and ML algorithms in process optimisation, production processes that in the past have not had adequate solutions can now be improved.

Siemens has chosen to develop a modular digital software platform with a large choice of applications for the cement industry. The platform is flexible and is continuously being expanded and updated based on the feedback and requirement requests received from customers. New ideas and developments can be implemented in an agile and efficient manner. The platform, called Sicement Operations, offers an extensive catalogue of applications. In essence, the system works like a smartphone. Once installed on the server, users can select which applications they want to install and use. New applications can be added and installed at any time, while existing subscriptions can be stopped or altered.

Sicement Operations is a cloud-based web platform built to host applications intended to improve the processes in cement plants. It uses the relevant data available in real-time from various sources such as condition monitoring systems, the process control system, and others. Zooming in on the cement production itself, a lot of process data is already available in the existing DCS and Process Historian systems.

It is therefore very important that the automation system has an architecture that allows immediately accessible real-time, deterministic, and transparent transmission of this data, even at very high communication rates. For this reason, upgrading to an industrial ethernet network, such as Profinet, is strongly recommended.

Since the platform is capable of accessing all relevant information related to the kiln, process optimisations, based on AI and ML, can be easily implemented.

AI for kiln prediction

In the first version of AI for the kiln, it was possible to predict the future behaviour of the kiln. The system graphically displayed the course of the kiln’s critical parameters over the next 15 and 30 minutes. Based on these predictions, the operator was able to adjust the correct settings to keep the kiln constantly operating within its optimal boundaries. The accuracy of the predicted values was very high from the start (for some even more than 85%), and the longer the system was in service, the more accurate the predictions became.

The system also systematically learned how the operator would set their new setpoints under various conditions and after a period of operation there was so much information in the database such that the system could not only predict what the future values would be, but it could also suggest the optimal values for the kiln parameters. It was then up to the operator to decide whether to implement these new values or not.

Each kiln has its own personal behaviour and only someone who has been working with the kiln day in and day out, for many years, is able to determine the optimal settings, often based on intuition. By using artificial intelligence algorithms, the system learns from these experienced operators ensuring their knowledge is not lost, allowing the younger generation to learn from their years of experience.

Enjoyed this article so far? To read the rest and start your free trial to World Cement, click here.

Read the article online at: https://www.worldcement.com/special-reports/22072022/process-optimisation-never-the-same-again/