TIP 1101-02

ISSUED – 1997
© 1997 TAPPI

The information and data contained in this document were prepared

by a technical committee of the Association. The committee and the
Association assume no liability or responsibility in connection with the
use of such information or data, including but not limited to any
liability or responsibility under patent, copyright, or trade secret laws.
The user is responsible for determining that this document is the most
recent edition published.

Tuning guide for basic control loops

Scope

The overwhelming majority of controllers used in the pulp and paper industry are simple PID controllers. For
most control applications, a well designed and properly tuned PID control loop is all that is needed to achieve the control
objectives. The majority of the PID control loops are tuned by a “trial and error” procedure. Some people actually
develop a “knack” for tuning using this procedure. Unfortunately, most people do not. Until recently, if they used a
tuning procedure at all, it was likely to be a quarter-amplitude-damping method such as the Ziegler-Nichols Method (1).
However, many technicians engineers refrain from using this method because of the oscillations (cycling) that occur and
the poor stability margin.
This Technical Information Paper (TIP) describes a more applicable tuning method called “Lambda Tuning”
which has been adopted s the preferred controller tuning method in the pulp and paper industry (2). It is based on simple
process models that can be obtained by a sequence of simple “bump” tests while the loops are in manual mode.
The step by step tuning procedures covered in this TIP were developed by the author as part of committee
assignment for the System Design Subcommittee of the Process Control Committee.

Safety precautions

There are no safety precautions.

Content

Discussion

There are some very good reasons for using the Lambda tuning method. Among these are:

1. It results in smooth control without any cycling.

2. It allows the tuner to choose the close-looped time constant called Lambda (
) which specifies the closed-loop
performance of the loop.
3. It allows multiple loops (such as flows) to be coordinated by selecting the same
for each loop. This is important
in many blending systems.
4. It allows two interactive loops to be significantly decoupled by choosing the
of one loop (the less important one)
to be significantly larger than the
for the second loop.
5. The maximum non-resonant bandwidth for a loop can be selected for disturbance rejection tuning.
6. It facilitates the tuning of cascaded loops.
The Lambda tuning method is based on simple process models that can be obtained from a sequence of simple
“bump” tests with the loop in “manual” mode. The response of the process variable is recorded. The process gain (Kp),

Approved by the Process Control Committee of the Process Control, Electrical & Information Division
TAPPI
TIP 1101-02 Tuning guide for basic control loops / 2

time constant () and deadtime (D) are estimated from this response. The process can usually be divided into two
classes: self-regulating processes (flows, temperatures, consistencies, most pressures, etc.) And integrating processes
(tank levels, steam pressures, boiler drum levels, etc.).

Procedure for tuning loops (non-integrating process)

STEP 1: Place the loop in manual. Make a small change in the controller output (say 5%). Discuss the size of
the move you want to make with the operator. After the process variable has reached a new steady state
value, move the controller output in the opposite direction by the same amount (say 5%).
STEP 2: From the trends obtained in STEP 1, estimate the process gain, the deadtime and the process time
constant.
STEP 3: Select the simple process model from Table 1 that best fits the response obtained in STEP 1.
STEP 4: Select the desired closed-loop time constant (
) you would like. It should be larger than the sum of the
deadtime plus the time constant as a general rule.
STEP 5: Calculate the controller gains based on the model selected in STEP 3. Enter these numbers into the
controller. Place the loop back on automatic. Make changes to the setpoint or load and observe the
response. Adjust the proportional gain to modify the speed of response of the loop if needed.

Example 1: Tuning flow control loops

A first order plus deadtime transfer function adequately represents the dynamics of most flow processes for tuning
purposes. A dynamic simulation package (VISSIM) was used to simulate the process and the controller.
The flow loop was placed in manual mode and a +5% step change was made to the controller output ( CO) at
t = 10 seconds and a -5% step change was made at t = 60 seconds. The flow response is shown in Figure 1.
The process gain can be calculated:

The process time constant can be estimate:

 = 10 seconds (approximately)

There is no apparent deadtime.

D = 0 Seconds

The flow transmitter is ranged 0 - 2000 GPM so the range (R) is 2000. The normalizing coefficient (K) can be
calculated:

Selecting the first order plus deadtime process model from Table 1, Kc, and KI can be calculated using the tuning
equations:
3 / Tuning guide for basic control loops TIP 1101-02

From the “bump” test response the process parameters are estimated:

Kp = 30, K = 0.5,  = 10, and D = 0

First
was selected to be 10 seconds, then 20 seconds.

Kc KI

10 .666 6.0

20 .333 6.0

The loop was placed in auto and the setpoint was stepped from 600 GPM to 1200 GPM. The closed-loop flow
responses for
= 20 seconds are shown in Figure 2. Notice only Kc had to be changed as
was varied.

Example 2: Tuning consistency control loops

A first order plus deadtime (FOPDT) model adequately represents the dynamics of most consistency loops. Note
there are tow FOPDT models in Table 1. One is for a deadtime dominant process ( << D) and the other is for a process
time constant dominant process (D << ). In this example the deadtime is dominant. The tuning equations for this
model are:

In this example, the open-loop response is omitted for brevity. The process parameters for this example are:

D = 10 Seconds,  = 3 Seconds

Kp = .03, K = 25

The closed-loop time constant (
) was selected as:

= 10 Seconds

The proportional gain can be calculated:

The integral gain can be calculated:

TIP 1101-02 Tuning guide for basic control loops / 4

The controller is tuned using these values and then placed in automatic. The consistency setpoint is stepped from
3.0 to 3.4 at t = 0. The response is shown in Figure 3.

Procedure for tuning level control loops

There are dozens of tank level control loops in every pulp and paper mill. The process model is simply an
integrator plus process gain. In most level control loops the deadtime is not significant. The tuning procedure is:

STEP 1: Place loop in manual mode (process should be in steady state-level constant)
STEP 2: Make a step change in the controller output (CO) and record the level response.
STEP 3: Calculate the process gain from the level response:

STEP 4: Select the allowable level variation (ALV). The ALV is the maximum variation (%) from setpoint that
could be tolerated under regulatory operation. Use the largest ALV possible so that the tank can absorb
flow disturbances.
STEP 5: Calculate Kc and KI using the tuning equations from Table 1.
STEP 6: Insert tuning values and return loop to automatic operation. Make step change to load and observe
response.

Example 3: Tuning tank level control loops

An integrator plus a process gain is a good model for most tank levels. The level control loop was switched to
manual mode. At t = 120 seconds a +10% step change was made to the controller output (CO). At t = 240 seconds
a -10% step change was made. Another -10% step change was made at t = 360 seconds and a +10% step change was
made at t = 480 seconds. The controller output history and the level response are shown in Figure 4. The process gain
is calculated as:

Using the tuning equations shown in Table 1, the tuning gains can be calculated.

Let ALV = 33.3%

5 / Tuning guide for basic control loops TIP 1101-02

Note that
was calculated in this example. Selection of ALV fixes
. It represents the recovery time after a step
change in load has occurred. Calculation of integral gain:

The tuning parameters are entered into the controller and it is returned to the automatic mode. At t = 0 the load
is increased by +20%. The level control response is shown in Figure 5. Note that the recovery time (
) is 80 seconds
and level falls approximately 5% before recovering. An ALV of 33.3% means that a load change of 100% would cause
a maximum level variation of 33.3%. This 20% load change would result in a maximum level variation of 6.7%. Since
there is some integral action the variation is actually 5%.

Example 4: Tuning a chip bin level control log (integrator plus deadtime)

There are a number of processes that can be adequately represented by integrator plus deadtime model (drum
level, chip bin level, etc.). The tuning procedure for these loops is the same as for a tank level control loop except for
the selection of tuning equations from Table 1.
The open loop response to estimate the model parameters is shown in Figure 6. The model parameters are
estimated:

Process Gain: Kp = 0.002

Deadtime: D = 120 seconds
Range: 0 - 100% (K = 1)

Using the appropriate tuning equations shown in Table 1, the tuning parameters can be calculated:

These tuning values are entered into the controller and it is returned to the automatic mode. At t = 100 seconds
a -20% load change (decrease in chip meter speed) is made. The Chip Bin Level response, Chip Meter RPM (%Range)
and Chip Bin Feedrate (%) are all shown in Figure 7. Note that the recovery time () is about 360 seconds (6 minutes)
and the level variation is 10%.

Control valve nonlinearities

A significant number of control valves have excessive stiction, backlash or hysteresis causing the control loop to
cycle. In some mills, as many as 20% of all control valves have been found to have poor dynamic characteristics which
result in poor control performance.
TIP 1101-02 Tuning guide for basic control loops / 6

If you have followed the tuning procedure for flow loops as outlined in the paper and the flow cycles when the
control is placed in the “auto” mode, suspect stiction. Place the control loop back in “manual” mode and make a series
of small moves (about 0.25% each) while recording the flow. A series of steps increasing the valve position, followed
by a series of steps decreasing the valve position, should be performed. This will allow you to determine the stiction and
backlash. If the backlash is greater than 1.0%, then you have found the cause of the cycling. The problem can be caused
by one or more of the following: packing, improper actuator size, linkage problems in the positioner, incorrect
installation, or the valve was improperly designed.
Reducing the integral gain (KI) will often allow the valve to be used until it can be repaired or replaced. However,
it will have a greatly reduced response speed.

Keywords

Consistency, Control, Stock chests, Control systems, Environments

Additional Information

Effective date of issue: March 11, 1997

Task Group Chairman: Fred Thomasson.

References

1. Ziegler, J. G. and Nichols, N. B., “Optimum Settings for Automatic Controllers,” Trans. ASME, Nov., 1942, pp
759-768.
2. Sell, N., editor, Bialkowski, W. L., Thomasson, F. Y., contributors, Process Control Fundamentals for the Pulp
and Paper Industry, TAPPI Textbook, to be published by TAPPI PRESS, 1995 (TAPPI).
7 / Tuning guide for basic control loops TIP 1101-02
TIP 1101-02 Tuning guide for basic control loops / 8
9 / Tuning guide for basic control loops TIP 1101-02
TIP 1101-02 Tuning guide for basic control loops / 10
11 / Tuning guide for basic control loops TIP 1101-02
TIP 1101-02 Tuning guide for basic control loops / 12