Pulp and paper 101: the basics of process control.

Process control comprises the regulation of a series of steps or operations toward a desired result or product. Generally, major economic forces are involved, with a desire to produce on-specification product at the lowest possible cost. Hence, process control is a key contributor to good economics.

The most common "workhorse" regulator is the PID controller. The PID setpoint represents where we want the process variable to be. The error is defined as the difference between the setpoint and the process variable: setpoint--process variable = error.

* The P-element provides Proportional error correction. Here, the controller output is proportional to the error or change in the process variable.

* The I-element provides Integral error correction. Here, the controller output is proportional to both the amount and duration of the error signal.

* The D-element provides Derivative error correction. Here, the controller output is proportional to the rate of change of the error signal.

Correct loop tuning plays an important part in commissioning a stable PID controller with close to zero error at all times. There is no room in this column to discuss all the available techniques for loop tuning. A web search on the subject of "Loop Tuning" yielded 404,000 hits, which is probably more information than anyone can process!

Even with well-tuned controllers, why do some mill managers state that their process runs extremely well, but they are not convinced of gaining any benefits from process control? Figure 1 illustrates the problem. The jagged line between points 1 and 2 represents process performance with no control.

With good control (between points 2 and 3), we are able to reduce the variability significantly, but we are also running at the same target (e.g. chemical and energy usage).

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

In most process areas, operators would be responsible for moving to a more economic target as indicated at point 3. Unfortunately, the operator may be quite reluctant to move targets closer to operating limits. The solution is to have an "economic optimizer" that automatically moves the target closer to operating limits (constraints) and generates sizeable savings in chemical or energy usage.

Figure 2 shows a realistic automatic target move on a lime kiln resulting in annual energy savings of approximately US$ 250,000. The lime kiln depicted is controlled using a multi-variable model predictive controller with a linear programming optimizer. Multi-variable means manipulating (in this case) three variables and controlling four variables within limits. "Model predictive" means that the controller predicts its future manipulations to ensure that process variables remain within pre-defined constraints.

The model predictive controller learns what the process model is during a sequence of small up/down pulses called pseudo random binary sequences (PRBS). During these PRBS tests, the process maintains the same average and does not cause any long-term bumps (which operators do not appreciate.)

Multi-variable model predictive control (MVMPC) is a relatively new technology that is demonstrating outstanding results. Some MPC packages have the ability to adapt a new process model in the background while using the original model for control. In the event that process changes have been made over time resulting in a deviation between process predictions (where the process should be according to the model), a decision can be made to switch to the adapted model.

MVMPC has been applied to virtually all pulp and paper processes and is now becoming available embedded in distributed control systems (DCS). The ability to have the process model developed by an objective analysis tool is a godsend for engineers. Prior control technology required that an engineer had to derive first principal control models (e.g. heat transfer equations, etc.)

MVMPC may appear to be a far cry from the PID controller--but is it really? What can be more basic than for mill engineers to have access to a tool whereby they can make a large impact on the bottom line for their mill by improving quality and reducing costs?

About the Author: Torsten Wesslen is pulp and paper application consultant for Invensys, Duluth, Georgia, USA. Contact him by email at: [e-mail].