Floating-pressure control is achieved by using a valve-position controller to drive the control valve to an economically desirable position such as 90-percent open, by manipulating the set point of the pressure controller. This forms a cascade system with the VPC as the primary controller, as it has a fixed set point. However, pressure is the more important variable, in that it must be controlled, whereas valve position is simply optimized.
To minimize upsets to the pressure loopespecially where there are multiple valves in the system, from which one is selected for controlthe VPC should not have proportional action. Integral-only control is recommended, with the integral time set relatively long, compared to the integral time of the pressure controller. Additionally, the VPC does not require access by the operator for startup or emergency operationin fact, it can be a blind controller. The operator can override its action at any time by placing the pressure controller in local set or in manual.
To prevent windup of the VPC when its loop is opened, external-reset feedback is recommended. Positional PI control algorithms calculate output from deviation multiplied by proportional gain, plus a reset-feedback term, which is normally the controller output lagged by the integral time constant. When the control loop is open, this positive feedback causes integral windup. Primary-controller windup can be prevented in a cascade system by taking its reset-feedback from the secondary controlled variable rather than from the primary controller output (which is the secondary set point). In this way, opening of the pressure-control loop also opens the reset-feedback loop, thereby preventing primary windup. It also delays primary integration by the response of the secondary loop, but this is favorable, and even adapts primary integration for any variation in secondary-loop dynamics. Recognize that the secondary controller must have the integral mode, because any secondary-loop offset will cause primary-loop offset.
Some controllers use a back-calculation of the integral-feedback term to prevent windup when the secondary loop is open. While this does prevent windup, it does not provide the smooth transition of the integral time constant, and can result in erratic behavior and even offset under noisy conditions.
F. Greg Shinskey, Process Control Consultant, North Sandwich, New Hampshire.
Measurement Problems on a High Level
"We have a level measurement problem with a differential pressure (DP) level transmitter that doesn't seem to track level properly. The vessel is a glass lined reactor, with an internal agitator and baffles. The only other available opening is a 1-in. port on the top of the vessel. The vessel is 60-in. in diameter and 12-ft. high. There are cooling coils filled with Dowtherm, and a glass-wool-and-aluminum-lag insulating jacket that is about 6 in. thick. We are having real problems on high level, and the process upsets if the vessel is too full.
You have not described the type of differential pressure level transmitter used, its elevation, the level range which should be detected, the filled or dry nature of the reference leg, or the locations of the d/p pressure taps on the reactor; therefore, one can only speculate:
The top connection of the d/p transmitter must be above the maximum level ever expected to occur in the reactor. To be on the safe side, it is best to connect it to the vent line on top of the reactor.
If you have a dry leg application, the problem can be with the range elevation adjustment. In case of a wet leg application, the source of the problem can be the range depression setting.
If it is possible that the reactor contents can boil, the use of a DP transmitter is the wrong level detector selection. This is because it detects the weight and not the height of the liquid column. The two will coincide only if the liquid density is constant. Therefore, it is possible that, during a boiling episode, the bubbles will cause the liquid density to drop and the DP cell will indicate a low level, when in fact the level is high and the liquid is boiling over.
If agitation causes the d/p measurement to cycle, this pressure cycling needs to be averaged. This is likely to occur at all levels, not only at high ones.
If the DP measurement is OK at all levels except high ones, and all you need is an alarm or interlock actuation when the level reaches some maximum, than you can use the 1 in. connection on the top of the reactor to insert a high level probe into the reactor. If the reactor contents are conductive, you can use a conductivity probe. If the contents are nonconductive, a capacitance probe can be used.
When the reading is incorrect at high levels, the level transmitter span or Upper Range Value is likely in error. The transmitter span may be calibrated 0-100-in. H2O. 100-in. is the URV. A DP transmitter used as a level transmitter needs the span or URV calculated. The URV is the height of the actual level measurement multiplied by the liquid specific gravity (SG). If the desired measurement is 100-in. high and the SG was 0.8, then the URV would be 80-in. H2O. If your SG used in the current calculation is 0.9 and the actual SG is 0.8, then a URV of 90-in. H2O is actually 80-in. H2O. This error would cause the level indication to read lower than the true level. At full level, the DP would actually be 80-in. H2O, but because the transmitter is calibrated for 90-in. H2O, the transmitter would send a signal of 89% and might cause an overflow. Best course? Measure the liquid's SG and recalculate transmitter calibration. If the transmitter is not located at the same height as the bottom tap, then the Lower Range Value will not be 0-in. H2O and will also need recalculation. If the SG varies, and you want to always read a level that is higher than the actual level, then use the lowest SG in your calculations. If the SG varies a lot, try using the more accurate level reading from a good radar level transmitter which measures the real height of the liquid and is not affected by changing SG.
Brian Smith, NOVA Chemicals (Canada) Ltd.
Need an Expert Opinion?
Send your knotty control and automation questions to Bela Liptak and he will either answer the questions himself or bring in specialists from his cadre of Instrument Engineers Handbook, Fourth Edition co-authors. Questions and answers will be published online periodically on ControlGlobal.com as they are received. Each month, BÃ©la will pick the best question, and publish it, along with the answer in CONTROL. If your question is chosen, a prize (of the editorial staffs choosing) will be awarded to the submitting winner.
Instrument Engineer's Handbook, 3-Volume Set
Instrument Engineer's Handbook, Volume I (Measurement and Analysis)
Instrument Engineer's Handbook, Volume II (Control and Optimization)
Instrument Engineer's Handbook, Volume III (Software and Networks)