Emerson’s Mircea Lupu shared his thoughts on Ovation system algorithms for handling processes with long lag times—a control challenge that vexes many.
Mircea opened noting that control algorithms are at the heart of any control system to provide stable and accurate response and robustness against disturbances and process changes. Tuning these algorithms to achieve the desired performance can be laborious.
Proportional-Integral-Derivative (PID) control is the workhorse algorithm but has difficulty with processes with time lags. There are solutions to address these lags but setting and maintaining tuning can be tedious. From a loop stability standpoint, the integral gain in a PID loop works against the time delay.
The derivative gain in a PID helps if tuned in a proper range. Too little or too much causes oscillation in the loop. In summary, proportional gain should be decreased, integral reset time should be increased with time delay and derivative gain may increase with time delay, but too much may destabilize the loop.
A Smith-Predictor (PREDICTOR) algorithm uses an internal model of the process to remove time delay from the control loop. The control algorithm will act on the predicted process variable rather than the actual (delayed) process variable. If the prediction is good, larger proportional gains and smaller reset times can be used. This increases the stability and robustness and improves transient performance. The performance of the loop is only as good as the prediction.
The Ovation Advanced Process Control (APC) Toolkit is an extension of the Ovation system that allows advanced algorithms to be added to an advanced controller.
Predictive algorithms solve an optimization problem with constraints to compute the manipulated variable (MV). This algorithm is ideal for supervisory control or setpoint generators for fast (low-level) feedback loops. This control block does multivariable process control—multiple input and multiple output (MIMO) applications.
This predictive algorithm is based on a linear quadratic optimization function with optimization constraints.
Mircea also discussed control with a cascaded loop configuration. In this arrangement the primary (upstream) controller drives the setpoint of the secondary (downstream) controller. This type of control strategy is recommended for processes with slow dynamics in which a relatively fast process has to be manipulated to control the slow process. This isolates a slow control loop (outer loop) from non-linearities by using an inner (fast) loop.
The disadvantages of cascaded control loops are the added measurement (PV) for the inner loop and having to tune the additional control algorithm. When tuning these loops, the inner loop has to be faster than the outer loop at least by 3X. Start by tuning the inner loop first and then the outer loop.
Mircea summarized his presentation reiterating the control challenge of processes with long lag times, such as those found in water & wastewater treatment facilities. You must first understand the dynamics of the process to design a stable and robust control strategy. Alternatives to PID control include PREDICTOR algorithms and model predictive control. The Ovation DPATUNE tool helps for model estimation. Finally, proper tuning techniques are important, especially in cascaded control loops.