The following tip is from the ISA book by Greg McMillan and Hunter Vegas titled 101 Tips for a Successful Automation Career, inspired by the ISA Mentor Program. This is Tip #93
If you look at a process flow diagram, you quickly realize that the flows are designed to move up and down in unison. The ratios of the flows are maintained. If you double production rate, you can double the process flows as a first approximation. The utility flows for heating and boilup will probably double as well. For cooling, the flows will need to increase but the ratio may change because the heat transfer coefficient changes with cooling flow, and cooling water supply temperature may change with cooling water return temperature. As a result, you can readily recognize that the most common feedforward signal is flow and that flow feedforward control is really flow ratio control.
Controllers have feedforward multipliers and feedforward summers built-in to the PID algorithm. There is a bumpless transfer when feedforward action is turned on or off. In the literature, a traditional analysis shows that the intercept of a plot of manipulated flow versus feedforward flow is zero. The conclusion is that a feedforward multiplier should be used to correct for slope. In practice, my associate Terry Tolliver, world’s best distillation control expert, and I found that a feedforward summer was a much better practical solution and that the built-in feedforward did not provide the necessary visibility and flexibility.
First of all, there are scaling problems when you use a multiplier. Second, by doing this, you introduce a nonlinear gain in the output that is proportional to flow. A cursory analysis might indicate that this is beneficial, because the process gain is inversely proportional to flow for temperature and composition control. However, for vessels and columns, the process time constant is also inversely proportional to flow. Because the controller gain is proportional to the ratio of time constant to process gain, the effect of flow on the controller gain via the process is cancelled. For plug flow volumes that are deadtime dominant, such as static mixers, the time constant is negligible and there is a slight benefit from a feedforward multiplier. The benefit is marginal because the controllers on these volumes use more integral action than gain action due to deadtime dominance (the reset time factor decreased from 4 toward 0.5; Tip #92).
Last, most of the error in a flow measurement is an offset that is best corrected by a feedforward summer and the slope (ratio) is best corrected by Operations. The summer output like the model predictive controller bias can be historized and analyzed for improvements. The desired ratio must be accessible and the current ratio must be visible to Operations. For the start-up of unit operations, particularly for reactions and separations, flows are normally put on straight ratio control. The temperature or composition controllers are not put in automatic until the process equipment has reached normal operating conditions.
Concept: A ratio station is needed to provide operations with the interface needed to start up and take control of equipment for abnormal operation. The interface and the historization of the actual ratio and desired ratio enable analysis and continuous improvement. A feedback correction from a process controller should provide a positive and negative bias to the ratio station output. When the bias is zero, the feedforward signal is perfect. Feedback control must take over if the feedforward signal is not usable because of flowmeter problems or a flow rate beyond the rangeability of the flowmeter.
Details: Filter the feedforward flow measurement so that noise does not cause the control valve to dither; rapidly cycle. Use a ratio station for the operator to set the desired ratio and see the actual ratio. Prevent zero or negative flows from causing bizarre ratios. Bias the ratio station output by a process feedback controller to correct for feedforward errors. Scale the feedforward bias to provide a correction that covers the desirable operating range. For example, if the operating point is 50 percent, a -50 percent to +50 percent bias should be provided by the feedback controller to adjust the manipulated flow from 0 to 100 percent.
Use cascade control so the effect of valve nonlinearity is removed by a secondary flow loop for the manipulated flow. Set up the ratio and bias station so the feedback controller can be put in manual to stop feedback correction but not stop feedforward action. Allow the loop to run with only feedback action and no feedforward. If the feedforward flow is beyond the flowmeter’s rangeability, go on total feedback control. If the manipulated flow is beyond the flowmeter’s rangeability, go off of cascade control and onto direct throttling of the control valve. When not in cascade control, schedule the process controller gain and the ratio to account for the nonlinearity introduced by the control valve installed characteristic. Consider whether a smart feedforward (Tip #92) is needed.
Watch-out: Orifice flow measurements get noisy below 25% of the maximum flow. Vortex meter signals may drop out below 10 percent of the maximum flow. In the start-up of boilers, the drum level controller may have to directly manipulate the feed water valve until the feed water flow is above the flowmeter’s low rangeability limit. Turning on three element level control, that is, feed water to steam flow ratio control, may need to wait till the steam flow is above the flowmeter’s low rangeability limit. Make sure a flow ratio is not used as a process variable for a controller. A flow ratio is too nonlinear for PID control.
Exceptions: Poor flow measurement selection, installation, or maintenance can cause flowmeters to be too erratic or unreliable for flow ratio control.
Insight: Temperature and composition loops should be able to switch as necessary between feedforward only, feedforward plus feedback, feedback only, cascade, and single loop control.
Rule of Thumb: Set up a ratio station, feedforward summer, and cascade control system for maximum flexibility and visibility to address start-up and abnormal conditions and measurement limitations.
Hunter Vegas, P.E., holds a B.S.E.E. degree from Tulane University and an M.B.A. from Wake Forest University. His job titles have included instrument engineer, production engineer, instrumentation group leader, principal automation engineer, and unit production manager. In 2001, he joined Avid Solutions, Inc., as an engineering manager and lead project engineer, where he works today. Hunter has executed nearly 2,000 instrumentation and control projects over his career, with budgets ranging from a few thousand to millions of dollars. He is proficient in field instrumentation sizing and selection, safety interlock design, electrical design, advanced control strategy, and numerous control system hardware and software platforms.