Edited by Emily Guenther, associate content manager, Control Engineering, concepts Jim Ford, PhD, PE, is a senior consultant at Maverick Technologies.
Control is better with closed-loop, and when the time between changes in the process, variable and automated corrective action is kept to a minimum. That’s the key difference between open- and closed-loop control. Otherwise, it’s open-loop controls, may not be performing optimally. The key point is that the logic solver, be it a human being, a proportional-integral-derivative (PID) controller, or a refinery optimizer, must be making decisions in real-time to provide the best control performance. Hopefully by Tuesday, the planning department will catch up. Prices may have changed significantly since Friday, so once implemented, refinery operation may be far from where it should be for Monday’s market conditions. The optimizer solution that was implemented Monday is probably using older feed and product prices from Friday because no one was onsite since Friday to manually change it. Since setting optimizers in refineries and chemical plants still uses open-loop control, the external variables may change, but the process will continue running where it was set.
The control room operator enters the key operating setpoints to "optimize" the unit. Resulting key operating targets typically are communicated to the control room via an email or spreadsheet-but not closed loop-and only after someone in the planning department has reviewed the optimizer solution to verify its reasonableness and feasibility. How do such external inputs get into the optimizer, and how do the optimizer-generated targets get loaded into the control system?Įxternal inputs usually are entered manually by designated personnel in the company’s planning department. Optimizers require external inputs, such as feed and product prices, supply-and-demand figures, fuel costs, etc., to calculate optimum operating targets. Many refineries and chemical plants use optimizers to determine targets for key operating variables, such as charge rate, product specifications, reactor severity, etc. If the operator at the Indonesian tank farm isn’t constantly watching the tank level and operating the valve, how closely will the process variable track its target? It will degrade in proportion to the amount of dead time. One approach examines the standard deviation of the controlled variable over a long period.
The quality of control of a loop with extensive dead time or lag can be measured. The man was in charge of monitoring the site glass and moving the valve manually as necessary. While the differences between the two options sounds clear, sometimes it can be complicated.įor example, a chemical/process engineer was walking at a refinery’s tank farm in Indonesia, which was separated physically from the refinery, He came upon an odd-looking tank equipped with only a crude site glass level indicator, and a man sitting on a stool next to a hand-valve on a line exiting the tank. This could be a situation where an automatic loop is placed in manual. An operator sets a valve or other actuator at a given point and it stays there until it is changed manually again. On the other hand, open-loop control is done manually. There is a tendency to think more about closed-loop applications where the manipulated variable is adjusted automatically to maintain a setpoint.
In some applications, a process designer will have to decide if a given function should be run with open- or closed-loop control.