Calculate Dead-Time By Visually Inspecting Your Step Test Data
Adults can learn a thing or two from children. That’s especially true when it comes to matters of time. Whereas adults view it as fashionable to show up to a social event 15 or more minutes late, children can’t wait to go and be a part of the action. When at a theme park adults typically their time strolling from place to place while children know that each minute of delay can cost hours longer in line. Simply put: When it comes to time every second matters. That viewpoint seems highly relevant to process control, control loop tuning, and calculating Dead-Time. Dead-Time is generally the model parameter that’s easiest to calculate when tuning a PID controller. In most cases Dead-Time can be estimated through simple visual inspection of the associated step test data. It represents the amount of delay between the change in Controller Output (CO) and the initial response of the Process Variable (PV). Easy as it may be to compute, Dead-Time presents significant challenges in terms of achieving optimized control loop performance. For that reason it is often referred to as the killer of control. Several details about Dead-Time and its calculation that are worth noting include:
- Law or Relativity
Always consider Dead-Time relative to the Process Time Constant. As noted in an earlier post the Process Time Constant serves as the “clock of the process” and it describes the speed with which the measured PV responds to changes in CO. As the Dead-Time value grows relative to the Process Time Constant it becomes increasingly difficult to achieve tight control.
- This and That
Dead-Time is a combination of several things. Transportation lag – the time it requires for material to travel from one location to another – plays a part. So too does instrument lag – sometimes called sample lag – which is the time needed to collect and process a measured PV sample. Higher order processes can naturally appear to respond sluggishly and they also factor into the Dead-Time value.
- Read the Signs
As its name implies Dead-Time is a time-based model parameter just like the Process Time Constant. Proper calculation of Dead-Time can only result in a positive value that’s characterized in units of time – whether milliseconds, hours, or something in between. A value that is either negative or lacking appropriate units of time should be the first sign that something is amiss.
- Much Too Much
If the Dead-Time value is greater than that of the Process Time Constant, then the controller will seem impatient and react in advance of the PV response. The controller’s tendency to respond prematurely can induce an unstable, oscillatory condition. The Smith Predictor is a multivariable control strategy for addressing such situations involving excessive Dead-Time.
Dead-Time can inhibit the effective and efficient control of production processes and taking appropriate steps to account for Dead-Time is essential. For starters, the location of process instrumentation should be considered thoughtfully as it limits the negative effects of Dead-Time. Next, when analyzing step test data and modeling the process dynamics it’s key to calculate the Dead-Time value accurately. Doing so enables the PID controller to perform its job. Although tuning PID control loops can be challenging using generally available software tools, at least one has proven to consistently handle processes with long Dead-Time values.