This document provides an overview of PID control basics and tuning. It begins with explaining why understanding PID is important and then covers common control techniques like manual, on/off, and closed loop control. It defines PID terms like proportional, integral, and derivative control and how they work together. The document discusses process dynamics around dead time and lag. It then provides guidance on manually tuning PID loops and using auto-tune functions. It concludes by listing some Yokogawa products that incorporate PID control capabilities.
In manual control, an operator monitors the difference between the process variable and the setpoint.
The operator then makes changes to the control output to reduce or eliminate the error.
ON/Off control is the simplest form of feedback control. In this type of control, the output is driven from fully closed to fully open depending on the relationship of the process variable to the setpoint and hysteresis.
Every one listening today has experience with the thermostat in their home which is a common example of simple on off control.
PID control is a form of closed loop control. In closed loop control, the process variable (PV) is measured and compared to the desired value called setpoint (SP). The controller changes it’s output, or manipulated variable (MV), until the measured variable equals the setpoint.
Because process dynamics vary greatly and the controllers are made to be universal, controllers muse be TUNED to match the process.
For example, let’s say we have a large tank of liquid for which we are controlling the temperature. When the output comes on, if there is no change in temperature for 10 minutes, then the dead time of the process is 10 minutes.
In a direct acting process, as the PV increase towards the SP, the output also increases. A common direct acting process is chiller being used for air conditioning.
In a reverse acting process, as the PV increases towards the SP, the output decreases. Flow control, furnaces, and pressure are all examples of direct acting processes.
The concept of PID control and the terms associated with it will be the same whether we are talking about simple single loop controllers, more advanced programmable controllers, PLC’s and distributed control systems.
Put simply:
The proportional band adjusts the output amplitude.
The integral reduces or eliminates any error between the setpoint and process variable.
And the Derivative monitors the rate of change of the error in an attempt to anticipate process upsets.
Proportional Band is expressed as a % of the full operating span of the controller and is centered around the SP assuming that the manual reset is set to 50%.
For example, an operating range of 0-1,000° with a P of 5%, would equal a P of 50° straddled 25° above and below the setpoint.
Let’s say that this is a reverse acting process like a furnace with a SP of 500. If the PV drops to 475° or below, the output will go to 100%. Similarly, if the PV rises to 525 or above, the output will go to 0%.
The output will stay at these extreme values until the PV re-enters the proportional band between 475 and 525. When the temperature is within the proportional band, the output is PORTIONAL to the error.
Using proportional only control, once the optimum proportional band is set, the process variable will be offset from the setpoint.
Manual reset can them be used to shift the control output to compensate for the offset.
Manual reset is available on all Yokogawa controllers, with a default to 50%.
Using manual reset doesn’t provide very stable control. As parameters in the environment, or process system changes through the day, the process variable can drift.
To prevent having to constantly change the setpoint, we need an automatic reset.
The smaller the integral number the more often the proportional action we be repeated.
If integral is too small, the process variable will oscillate through the set point and create erratic control action.
If the number is too large, the action will be sluggish and unable to compensate for process upsets.
The integral number should be approximately 5 times the dead/lag time of the process variable.
The integral term continues to increase or decrease the output until a zero error condition is obtained.
Sometimes referred to as reset windup, integral windup occurs when the integral action continues to add to the control output past the operational range of the valve, variable speed drive, heater, etc.
(READ LAST POINT ON SLIDE)
Derivative can simply be thought of as a prediction of the error in the future based on the time set.
Gradually reduce P from a larger value. When the PV begins to oscillate, stop tuning and increase P slightly.
Gradually reduce I from a larger value. When the PV begins to oscillate, as with P, stop tuning and increase I slightly.
Gradually increase D from a smaller value. When the PV begins to oscillate, stop tuning and lower the value slightly.
Temperature loops typically have considerable dead time.
Fast loops are defined as having little or no deadtime or lag. Most fast loops are configured as a PI controller., proportional and integral terms only. This is because the derivative term is an overshoot/undershoot suppression function and naturally contributes to instability in a fast control loop.
Sometimes Auto Tune can’t be used or shouldn’t be used. They include:
Fast responding processes such as flow rate and pressure control
A process that does not allow the output to be turned on and off even temporarily
Processes in which product quality can be adversely affected if the PV values fluctuate beyond their allowable range.
Temperature processes where there are adjacent temperature zones that affect each other, for example a plastic extruder.
Let’s give Auto Tune a try. Again, I am simulating a furnace application using a Type K thermocouple and 4-20mA control output to simulate a gas valve. The temperature range is set to 100-600°F.
One problem that you will encounter when tuning is SP overshoot.
Setpoint Ramp - Some controllers and control systems have a feature called setpoint ramping. This feature allows a change of setpoint to be made in either engineering units/min or hour.
Using this feature reduces the size of the error between the SP and the PV at each ramp thus requiring less output to be used. This in turn will help to prevent overshoot.
Fuzzy Logic monitors the deviation for evidence of the danger of overshoot.
Once this danger is sensed, the setpoint is temporarily changed to a somewhat lower value (sub-SP).
Once the danger of overshoot appears reduced, the function returns the effective SP
gradually to the true SP.
