Highest Levels - With state-of-the-art automation procedures, it is possible to reach significantly higher levels of efficiency in chemical production plants. These objectives can be achieved within a short time and at moderate expense, usually with an excellent cost-benefit. The field of application is wide. In the control area, there are many benefits to using a model-based concept over PID control (proportional-integral-derivative-controller). In the management area, PIMS (plant information management systems) are gaining importance.

Model-Based Concepts vs. PID

The "normal" functions of an automation system are feedback and feed-forward control. This includes interlocks as well as sequential functions. Safety functions, alarms, process states and their visualization and means to act on the process are also a part of the basic function set of a DCS (distributed control system). Being able to identify improvement potential on this level requires sound knowledge of production processes. The team has to be able to choose the right tools, find the optimum in the relevant production process, and judge the robustness of the installed applications. Moreover, it is absolutely essential that people from the production environment be involved in basically every step of the automation process. If the project team fails to do so, the whole project is doomed to fail. This may sound trivial, but it really is not. Model-based approaches have been on the scene for 15 years now, but PID controllers have been there for more than 50 years.

There are two objectives to aim for. Process oscillation should be reduced significantly. Process transitions, that is, going from operating point A to B, should be accelerated without violating any given process restriction. Using a PID is like driving a car with an opaque windshield. The driver may use the side window and try to keep the car in the lane by maintaining a constant distance to the guard rails or some other road boundary. In this case, the driver is aware that he/she is turning only if this distance changes. This is definitely a complicated approach to driving a car. By contrast, using a model-based control concept is like seeing through a freshly wiped windshield.

The driver sees what is in front of him/her and he/she can take action early on. This is surely a better way to perform this given task.

Model-Based Control

The example of temperature control in a fermentation process shows the benefits of model-based approaches to the chemical industry. Periodically, water has to be injected into different fermentation tanks. For the water to remain aseptic, it must be heated to about 90°C. Since this temperature would kill the biomass in the fermentation reactors, the liquid has to be cooled to 30°C before it is injected. The faster this lower set-point is reached, the sooner water can be injected. This, of course, saves batch time.

After the model-based control is implemented, the lower set-point is reached at least five minutes earlier. Since there was a total of 60 injections per batch, the total of batch-time reduction amounted to at least 300 minutes. Needless to say, the economies gained from this automation project were exceptionally high.

Knowing Everything There Is To Know

With the help of PIMS, added value can be achieved on the MES (manufacturing execution system) level. In general, this means that process-related data and information are used to find optimization potential. This refers to International Standard (ISA 95) "puzzle parts" such as performance analysis, process management, and date acquisition only. It does not include workflow optimization. PIMS enable engineers to find a correlation or hidden information, which serves as the point of departure for optimization efforts. Beneficial categories are finding and understanding process contexts as well as identifying root causes and improvement potential. Two case studies show what this means.

The first example refers to a separation column in a plant in Asia. At times the temperature in this column would begin to oscillate. Despite intensive research, the staff could not find the cause. That was where a team from Europe came in - without actually having to travel to Asia. Using computer networks, the team members had access to all process data. They took all process data points around the column - about 35 points over a period of several days. Then they fed the information into a statistical program to find out which process points dynamically correlate with the examined temperature.

After a few minutes, they had first results. There was only one significant relationship: the level in the feed tank. But since the feed flow was constant, it became clear that the temperature problem was caused somewhere upstream. After a few hours of additional statistical evaluations, the cause was found three or four process steps upstream - where no one had ever looked. Afterwards, it was easy to solve the problem. The temperature in the separation column has been even ever since.