"TUNE_OL.EIK" and "TUNE_CL.EIK" PID Loop Tuning Aides for PID Loops in Automated Logic Eikon FBs.	

V1.0     26-OCT-1995

Eikon Files developed by Clay Loving, West Coast District Sales Manager, ALC.
Documentation reviewed and edited by Mike Sorrells, Training Manager, ALC

These files are intended to provide assistance in setting up the initial values for PID Loops within ALC Eikon Graphic FBs.  Presently, the preliminary step-by-step instructions from below are also located at the bottom of the Eikon files for viewing while in Eikon.

The 2 Eikon files are for Open Loop Tuning and Closed Loop Tuning, respectively.

Common to all PID auto tuning methods are the rules of parameter estimation as developed by Ziegler-Nichols some forty years ago. The rules themselves were developed by empirical simulation of different systems. Rules exist for closed loop stochastic evaluation as well as open loop. Both require a transfer function of the given system. As previously mentioned, HVAC systems are generally first order. But, in the case of hot wax, room control, and other systems with damping, the system is represented as a second order transfer function.

Open loop tuning proves to be a very good pre-tuner procedure. Open loop tuning is based on the PID regulator being halted, a step change in the output (actuator), and then a statistical analysis of the process response. The notable variables are the delay time before the process responds, the maximum rate of change once the process is in transition, the magnitude change of the process, and lastly the time constant for the process. Although these variables can be determined without the use of a transfer function for the process, this function is required for the closed loop tuning.

Closed loop tuning can be regarded as a fine tuning of the process parameters. In closed loop tuning, the integral and derivative contribution are switched out. The transfer function then provides a gain schedule which will yield a closed loop oscillation of the process. The amplitude and frequency of the oscillation observed are then utilized to further refine the previous PID regulator parameters. The gain schedule is ideally a relay function, but the operator can negate this feature. Although the gain schedule function reveals more specific information concerning the true static gain of the system, the interpretation requires greater skill.



Open Loop Tuning Procedures

Using the trending capability of SVW, the empirical graphic representation of the system response will provide the necessary information for pre tuning the control regulator.

1. The first step is to set the trending to the smallest increment possible that will still provide a full step response of the system on a single trend screen. Typically, a 1 second trend increment will be appropriate. The points to be trended are the controlled variable, E.g., duct static pressure, and the control device, E.g., vane actuator.
 2. Identify the control device safe limits of operation. For example: a 0-10 vdc actuator may not begin to stroke until 2 vdc, and at 9 vdc may deliver too great a static pressure to the ductwork. Therefore, a range of 2 - 8 vdc would be a safe range of operation for this tuning activity.
 3. Lock the control device to a value which will both present a load to the system and also represent one end of the stroke, E.g., 3 vdc on the vane actuator. 
 4. After the controlled variable has stabilized, lock the control device to the high limit of safe operation.
 5. Selecting the controlled variable and the control device for trending, wait for the system to again reach a stable level. By adjusting the display range of the controlled variable, and if necessary repeating the above steps with a different trend interval, the open loop step response should fill the trend screen. Obtain a print out of the graph.
 6. Using the Eikon PID tuning program in simulation mode, utilize the graph to establish the necessary parameters.
- Process Noise: During periods where the control device was locked and the system was stable, what was the maximum deviation of the controlled variable, i.e., fluctuation in variable when the control device is not changing?
- Process Variable Response (k): Enter the total magnitude change of the controlled variable during the step change. 
- Maximum Rate of Response (slope): Calculate the maximum slope of the response (rise/run). Note that this is a very important factor and care must be taken to be accurate.
- Time Constant of Response (L+T): Measure the total elapsed time from the graph from the point at which the step response initiated to the point in time where the maximum slope line intercepts with the horizontal line drawn through the final stabilized system response.
- Process Control Setpoint: Enter the expected controlling setpoint of the system.
 7. Once the parameters are entered in the simulation mode, the controller settings are calculated. Note that the far right hand column provides a recommendation for controller type, i.e., Proportional, Proportional + Integral, or Proportional + Integral + Derivative. If none of the settings are recommended, then refer to Non PID Solutions.
 8. After entering the PID parameters from the Eikon Simulation into the gfb, monitor the control loop stability. Utilizing both the trend and the status page PID information, determine if further tuning is required. For closed loop tuning, refer to that section. For further incremental tuning, complete the next step.
 9. The status page representation of the PID controller provides much of the information needed to further tune the controller. For evaluation of the performance of the control, the trend graph will identify cycling of the controlled variable and any associated cycling of the control device. 
- Proportional Only Control: Enter a Bias (0-100%) that is equivalent to the low end of the control device spring range, E.g., 20% for the above example (2 vdc of a 0-10 vdc range = 20%). Incrementally increase P gain 25% to reduce error between setpoint and controlled variable. Likewise, decrease by same amount if cycling persists.
-  Proportional + Integral Control: The ideal status page indicators are a very small influence of the proportional gain in comparison to the influence of the integral gain. Enter a Bias as with Proportional Control only, above. Turn ON Hold I Error unless the controlled variable setpoint changes from day to day. For further tuning of the parameters, systematically reduce the P gain by 25% until its' contribution to the output is very small in relation to the contribution of the I gain. If cycling around setpoint persists, reduce the I gain by 25% until cycling is negligible.
- Proportional + Integral + Derivative: Use the same strategy as with PI plus the notion that the derivative influence ought not be greater than the proportional component. 

Non PID Solutions

The PID regulator is not ideal for all applications. The Eikon PID tuner is likewise not appropriate for tuning all PID applications. Therefore, an understanding of when not to use them is needed.

The PID regulator assumes several characteristics about the controlled system which may not be true. First, that the controlled system is a first or second order system, i.e., the system responds with near proportional characteristics after any damping. Second, that the characteristics of the system do not change over time, E.g., how fast the control device reacts or, the maximum rate of change of the controlled variable. As well, the Eikon tuner has an assumption that the ratio L/T is between 0.1 an 1. When less than 0.1, the control would favor a higher order control method. When greater than 1, the control needs to compensate for dead time. Although a PID regulator can still be used (PID for  less than 0.1 and P only for greater than 1), cycling of the controlled variable may not be avoidable.

Common alternatives to PID regulators include on/off control, incremental control, and gain scheduling. On/off control, which is analogous to ALC's If Color micro block, is appropriate when controlling a gas furnace or single stage dx. These systems are themselves not a modulating control device and as such do not lend themselves to a modulating controller.

Incremental control is ideal in cases of pulse width modulation (PWM) where an incremental change in magnitude and or polarity causes the control device to move in a given direction and speed. Because there is no feedback from the control device, incremental control can only be used where the control device can tolerate being signaled open or close when it is already fully stroked. Because deadband is  required to prevent undue cycling, incremental control like on/off control cannot be used in situations where a precise setpoint must be achieved. Incremental control can easily be achieved with Eikon through the use of a Ramp Up/Down micro block.

Gain Scheduling is an approach which addresses both systems that have long apparent delays as well as higher order and high gain characteristics. Using multiple PID regulators where one may have only that of a P gain versus the other having P, I, & D gains may be beneficial when the response of the control device behaves differently when first approaching setpoint then when controlling at setpoint. Using a single PID regulator and changing the setpoint as a function of the change in controlled variable has a similar net effect.



Closed Loop Tuning Procedures

Using the Trending capability of SVW, the empirically graphic response of the system will provide the necessary information for tuning the control regulator.


 1. The first step is to set the trending to the smallest increment possible that will still provide a full step response of the system on a single trend screen. Typically, a 1 second trend increment will be appropriate. The points to be trended are the controlled variable, E.g., duct static pressure, and the control device, E.g., vane actuator.
 2. Identify the control device safe limits of operation. For example: a 0-10 vdc actuator may not begin to stroke until 2 vdc, and at 9 vdc may deliver too great a static pressure to the ductwork. Therefore, a range of 2 - 8 vdc would be a safe range of operation for this tuning activity. Since closed loop tuning does by nature not require direct involvement by the operator, monitoring of the cycling to identify whether or not the safe limits are being exceeded is required.
 3. Disable the I and D gains by setting them to zero. 
 4. Introduce a step change to the loop setpoint to bring it to a value somewhere near the mid-point of the controlled variable's normal operating range. 
 5. Adjust the proportional gain so that steady oscillations about the setpoint occur. The oscillations are referred to as ultimate oscillations since any increase in the proportional gain would lead to instability. Also not the period between successive oscillations. The values obtained are referred to as the ultimate gain and ultimate period respectively. By adjusting the display range of the controlled variable,  the closed loop response should fill the trend screen. Print out a copy of the graph.
 6. Using the Eikon PID tuning program in simulation mode, utilize the graph to establish the necessary parameters.
- Ultimate Gain (A): Enter the total magnitude change of the controlled variable during the oscillation. Note that for true cycling, the controlled variable will not have plateaus.
- Signal Amplitude (D): This represents the control devices' cycling amplitude. Note that for steady oscillations the amplitude and frequency will be consistent.
- Ultimate Period (U): Measure the total elapsed time from the graph for the period of the cycling controlled variable.
- Process Setpoint Change: If a relay setpoint change method was used, enter the amplitude of change.
- Control Signal Range: This is the spring range of the control device.
7. Once the parameters are entered in the simulation mode, the controller settings are calculated. Note that the far right hand column provides a recommendation for controller type, i.e., Proportional, Proportional + Integral, or Proportional + Integral + Derivative. If none of the settings are recommended, then refer to Non PID Solutions.
8. After entering the PID parameters from the Eikon Simulation into the gfb, monitor the control loop stability. Utilizing both the trend and the status page PID information, determine if further tuning is required. For further incremental tuning, complete next step.
9. The status page representation of the PID controller provides much of the information needed to further tune of the controller. For evaluation of the performance of the control, the trend graph will identify cycling of the controlled variable and any associated cycling of the control device. 
- Proportional Only Control: Enter a Bias (0-100%) that is equivalent to the low end of the control device spring range, E.g., 20% for the above example (2 vdc of a 0-10 vdc range = 20%). Incrementally increase P gain 25% to reduce error between setpoint and controlled variable. Likewise, decrease by same amount if cycling persists.
-  Proportional + Integral Control: The ideal status page indicators are a very small influence of the proportional gain in comparison to the influence of the integral gain. Enter a Bias as with Proportional Control only, above. Turn ON Hold I Error unless the controlled variable setpoint changes from day to day. For further tuning of the parameters, systematically reduce the P gain by 25% until its' contribution to the output is very small in relation to the contribution of the I gain. If cycling around setpoint persists, reduce the I gain by 25% until cycling is negligible.
- Proportional + Integral + Derivative: Use the same strategy as with PI plus the notion that the derivative influence ought not be greater than the proportional component. 


For more information at this time, please contact someone within your company who attended the 1995 Technical Update Seminar or contact Mike Sorrells at (770) 429-3000.