Overcoming process deadtime with a Smith Predictor

A controller equipped with an accurate process model can ignore deadtime. Deadtime generally occurs when material is transported from the actuator site to the sensor measurement location. Until the material reaches the sensor, the sensor cannot measure any changes effected by the actuator.


In this simplified rolling mill, the flattened material takes D = S/V seconds to move from the reducing rollers to the optical thickness gauge. Courtesy: Control EngineeringFor the purposes of feedback control, deadtime is the delay between the application of a control effort and its first effect on the process variable. During that interval, the process does not respond to the controller's activity at all, and any attempt to manipulate the process variable before the deadtime has elapsed inevitably fails.

Deadtime generally occurs when material is transported from the site of the actuator to another location where the sensor takes its reading. Not until the material has reached the sensor can any changes effected by the actuator be detected.

Consider, for example, the rolling mill shown in the "Simplified Deadtime Example" which produces a continuous sheet of some material at a rate of V (inches per second). A feedback controller uses a piston to modify the gap between a pair of reducing rollers that squeeze the material into the desired thickness. The deadtime in this process is caused by the separation S between the rollers and the thickness gage.

The controller in this example can compare the current thickness of the sheet (the process variable, PV) with the desired thickness (the setpoint, SP) and generate an output (CO), but it must wait at least D = S/V seconds for the thickness to change. If it expects a result any sooner, it will determine that its last control effort had no effect and will continue to apply ever-larger corrections to the rollers until the sensor begins to see the thickness changing in the desired direction. By that time, however, it will be too late. The controller will have already overcompensated for the original thickness error, perhaps to the point of causing an even larger error in the opposite direction.

How badly the controller overcompensates depends on how aggressively it is tuned and on the difference between the actual and the assumed deadtime. That is, if the controller assumes that the deadtime is much shorter than is actually the case, it will spend a much longer time increasing its output before successfully effecting a change in the process variable. If the controller is tuned to be particularly aggressive, the rate at which it increases its output during that interval will be especially high, and the resulting overcompensation will be particularly severe.

Detuning the controller

The preferred method for curing a deadtime problem is to physically modify the process to reduce deadtime. In the rolling mill example, this could be accomplished by moving the thickness gage closer to the rollers or by running the sheet at a higher velocity.

But if deadtime cannot be cured by relocating the sensor or speeding up the process, its symptoms can still be addressed by modifying the control algorithm. The simplest method is to de-tune the controller to slow its response rate. A de-tuned controller will not have time to overcompensate unless deadtime is particularly long.

The integrator in a proportional-integral-derivative (PID) controller is particularly sensitive to deadtime. By design, its function is to continue ramping up the controller's output so long as there is an error between the setpoint and the process variable. In the presence of deadtime, the integrator works overtime. Ziegler and Nichols determined that the best way to de-tune a PID controller to handle a deadtime of D seconds is to reduce the integral tuning constant by a factor of D2. They also found that the proportional tuning constant should be reduced by a factor of D. The derivative term is unaffected by deadtime since it only comes in to play after the process variable has begun to move.

The Smith Predictor uses a model of the process to compute what the process variable would look like without disturbances so it can estimate the disturbances and predict what the process variable would look like without deadtime. Courtesy: Control Enginee

De-tuning can restore stability to a control loop that suffers from chronic overcompensation, but it would not even be necessary if the controller could first be made aware of the deadtime, and then endowed with the patience to wait it out. That is essentially what happens in the famous Smith Predictor control strategy proposed by Otto Smith in 1957.

Removing deadtime from the loop

Smith's strategy is shown in the "Smith Predictor" block diagram. It consists of an ordinary feedback loop plus an inner loop that introduces two extra terms directly into the feedback path. The first term is an estimate of what the process variable would look like in the absence of any disturbances. It is generated by running the controller output through a process model that intentionally ignores the effects of disturbances. If the model is otherwise accurate in representing the behavior of the process, its output will be a disturbance-free version of the actual process variable.

The mathematical model used to generate the disturbance-free process variable consists of two elements hooked up in series. The first element represents all of the process behavior not attributable to deadtime. The second element represents nothing but the deadtime. The deadtime-free element is generally implemented as an ordinary differential or difference equation that includes estimates of all the process gains and time constants. The second element of the model is simply a time delay. The signal that goes in to it comes out delayed, but otherwise unchanged.

The second term that Smith's strategy introduces into the feedback path is an estimate of what the process variable would look like in the absence of both disturbances and deadtime. It is generated by running the controller output through the first element of the process model (the gains and time constants), but not through the time delay element. It thus predicts what the disturbance-free process variable will eventually look like once the deadtime has elapsed, hence the expression Smith Predictor.

Subtracting the disturbance-free process variable from the actual process variable yields an estimate of the disturbances. By adding this difference to the predicted process variable, Smith created a feedback variable that includes the disturbances, but not the deadtime.

So what?

The purpose of all these mathematical manipulations is best illustrated by the "Smith Predictor Rearranged" block diagram. It shows the Smith Predictor with the same blocks arranged to yield the same mathematical results, only computed in a different order. This arrangement makes it easier to see that the Smith Predictor effectively estimates the process variable (including both disturbances and deadtime) by adding the estimated disturbances back into the disturbance-free process variable. The result is a feedback control system with the deadtime outside of the loop.

A Smith Predictor effectively removes the deadtime from the loop and allows the controller to manipulate the predicted process variable as if there were no deadtime at all. Courtesy: Control Engineering

The Smith Predictor works to control the modified feedback variable (the predicted process variable with disturbances included) rather than the actual process variable. If it is successful in doing so, and if the process model does indeed match the process, then the controller will simultaneously drive the actual process variable towards the setpoint, whether the setpoint changes or a load disturbs the process. The deadtime becomes irrelevant.

Unfortunately, in the real world, those are big ifs. It is certainly easier for the controller to meet its objectives without having to deal with the deadtime, but it is not always a simple matter to generate the process models required to make this strategy work. Even the slightest mismatch between the process and the model can cause the controller to generate an output that successfully manipulates the modified feedback variable, but drives off the actual process variable into oblivion. There have been several fixes proposed to improve on the basic Smith Predictor, but deadtime remains a particularly difficult control problem.

Vance VanDoren, PhD, PE, is a contributing content specialist for Control Engineering. 

Key concepts:

  • Deadtime in a process can cause a controller to overreact to a disturbance or setpoint change.
  • If it isn't practical to wait for a time interval to pass, there are techniques designed to make changes more quickly.


Read more on control strategy and loop tuning below.

Ahmed , OH, Egypt, 02/18/15 10:47 PM:

it is deep knowledge of controller and meaning of deedtime.
LARRY , WA, United States, 02/19/15 09:53 PM:

Great job pointing out the potential benefits without minimizing the potential hazards. The application example was a good one, and has a hidden hazard to make matters worse: the material balance for a rolling process means that repositioning the drive roller changes the travel rate at the output side, so that the "time delay" is not exactly constant. The classic "Smith Predictor filter," as implemented in the 1950s, used analog non-minimum phase filters, which often were subject to sensitivity and stability limitations. We have options today to use the same architecture model, but replacing the Smith Predictor filter part of the process model with digital delay lines and digital filters. However, appropriate tuning of these can still be tricky. Why is there never a free lunch?
WILLIAM , CA, United States, 04/07/15 12:28 PM:

It seems one can also view the Smith Predictor as "feed-forward control", where the controller output is simply another measured "input disturbance" which the feed-forward equation responds to in an attempt to keep the thickness (controlled variable) from changing. If the predictor's "plant model" is perfect, it seems it would simply counter the control action until one dead-time has passed.
With today's digital controllers, it seems a simpler approach is that after each control change, wait one dead-time before acting on the sensor signal (i.e. common sense). This might be realized by simply setting the PLC cycle time (or task time) to match the dead-time.
Tom , Non-US/Not Applicable, Australia, 04/07/15 09:06 PM:

Also refer to F.G.Shinskey :
PID - deadtime control of
distributed processes.
Glenn , Canada, 04/11/15 12:52 PM:

A good article and timely for me as I just designed and commissioned 3 MSP (Modified Smith Predictor) controllers last week. MSP's have been common in the paper industry since the 1980's. An MSP predicts the effect of the process time constant in addition to the dead time. The final controller is simply gain X predicted error divided by process gain. If significant model mismatch is possible, I often add compensation for uncertainty. These controllers are very simple to design and implement and work well with both improved stability and speed.
The Engineers' Choice Awards highlight some of the best new control, instrumentation and automation products as chosen by...
The System Integrator Giants program lists the top 100 system integrators among companies listed in CFE Media's Global System Integrator Database.
Each year, a panel of Control Engineering and Plant Engineering editors and industry expert judges select the System Integrator of the Year Award winners in three categories.
This eGuide illustrates solutions, applications and benefits of machine vision systems.
Learn how to increase device reliability in harsh environments and decrease unplanned system downtime.
This eGuide contains a series of articles and videos that considers theoretical and practical; immediate needs and a look into the future.
Controller programming; Safety networks; Enclosure design; Power quality; Safety integrity levels; Increasing process efficiency
Additive manufacturing benefits; HMI and sensor tips; System integrator advice; Innovations from the industry
Robotic safety, collaboration, standards; DCS migration tips; IT/OT convergence; 2017 Control Engineering Salary and Career Survey
Featured articles highlight technologies that enable the Industrial Internet of Things, IIoT-related products and strategies to get data more easily to the user.
This article collection contains several articles on how automation and controls are helping human-machine interface (HMI) hardware and software advance.
This digital report will explore several aspects of how IIoT will transform manufacturing in the coming years.

Find and connect with the most suitable service provider for your unique application. Start searching the Global System Integrator Database Now!

Infrastructure for natural gas expansion; Artificial lift methods; Disruptive technology and fugitive gas emissions
Mobility as the means to offshore innovation; Preventing another Deepwater Horizon; ROVs as subsea robots; SCADA and the radio spectrum
Future of oil and gas projects; Reservoir models; The importance of SCADA to oil and gas
Automation Engineer; Wood Group
System Integrator; Cross Integrated Systems Group
Jose S. Vasquez, Jr.
Fire & Life Safety Engineer; Technip USA Inc.
This course focuses on climate analysis, appropriateness of cooling system selection, and combining cooling systems.
This course will help identify and reveal electrical hazards and identify the solutions to implementing and maintaining a safe work environment.
This course explains how maintaining power and communication systems through emergency power-generation systems is critical.
click me