Neglect Level Control at Your Peril

This first article in a four-part series examines reset mode tuning issues.

By Cecil L. Smith, Cecil L. Smith, Inc.

1 of 3 < 1 | 2 | 3 View on one page
This article first appeared in our sister publication Chemical Processing.

The level in a process vessel often isn't viewed as a critical variable to control. Especially in surge vessels, controller performance typically is considered satisfactory if no high- or low-level trips occur. Vessel level's effect on the corporate bottom line usually is nil. So, it's easy to see why vessel level controls receive little attention.

However, neglecting level control carries definite risks. Indeed, loss of level control has contributed to three major industrial accidents (see: "Don't Underestimate Overfilling's Risks").

Processes consist of a collection of unit operations. The various flow streams that interconnect these unit operations propagate variance from one to the next. This variance ideally should diminish as it spreads but the nature of some unit operations actually amplifies it.

Control loops also can amplify variance, either through improper operating objectives or poor tuning. The flow between some unit operations depends upon a level controller's output to a final control element on a flow stream. Two factors make level loops a potential source of variance:

  1. 1. Variance in any stream in or out of a unit operation leads to variance in vessel level, to which the level controller responds by changing the flow of either an inlet or a discharge stream.
  2. 2. The inherent nature of level processes complicates controller tuning, especially for integral or reset mode. Excessive reset action (i.e., reset time too short or reset rate too fast) results in a cycle in the flow. In the conservatively tuned level loops applied to surge vessels, the slowly responding controller produces a cycle with a period of hours. This is where level differs from other loops — reducing the controller gain or sensitivity doesn't eliminate the cycle but merely increases its period.

Figure 1 shows a vessel with four feed streams and one product stream. The level controller influences just the discharge stream. The only difference between controlling using a feed stream instead of a discharge stream is directionality (increase-increase versus increase-decrease). In Figure 1 the final control element is a valve but a pump with a variable frequency drive often is a viable, and possibly preferable, alternative.

The controller in Figure 1 translates variations in vessel level to variations in discharge valve opening and, hence, discharge flow. Maintaining level within a given proximity to its set point requires certain changes in discharge flow. Generally, the tighter the level is controlled, the larger the necessary variations in discharge flow.

Now, let's consider some characteristics of a feedback controller:

  • It decreases the variance in the control error by increasing the variance in the controller output, which translates to higher variance in the flow through the final control element.
  • It shifts variance from one variable to another; it doesn't reduce total variance.
  • An improperly tuned feedback controller can significantly raise total variance.

Although instrument technicians often are responsible for tuning level controllers in a plant, propagation of variance is a process issue that's most appropriately addressed by process engineers.

Most level processes are integrating/ramp/non-self-regulated, the primary exception being gravity flow applications. When the level controller is on manual, with an integrating process:

  • Vessel level changes at a rate proportional to the imbalance in the material balance (total flow in minus total flow out).
  • Changes in level (and, thus, head) don't affect the discharge flow and, consequently, the imbalance in the material balance.
  • The rate of change in level remains the same as the level increases or decreases.

When the level doesn't directly impact any flow in or out, the dynamic characteristics of the process act as an integrator. The integrator in the reset mode of a controller coupled with an integrator in the process can have adverse consequences.

Starting from an equilibrium state (total flow in equals total flow out), any upset results in a ramp change in level, hence the term "ramp process." If the upset conditions persist, the ramp continues until the level reaches a limiting condition, usually in the form of a high- or low-level process trip. When no control actions are taken, such processes don't seek an equilibrium, hence the term "non-self-regulated process."

1 of 3 < 1 | 2 | 3 View on one page
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments