Tight level control

Controller tuning is derived by first assuming that we apply a proportional-only controller. 

Let us assume that before the flow disturbance, the level is at steady state and at SP, i.e. i will be zero. Since the flow imbalance (f) will have existed for one controller scan interval (ts), the current error (in dimensionless form) is given by 

In order to bring the level back to steady state we need to restore the flow balance and so the controller must change the manipulated flow by the flow disturbance (f). In dimensionless form this means 

The tightest possible control would be to take this corrective action in the shortest possible time, i.e. the scan interval (ts). By combining Equations (4.9) to (4.11) we can derive the largest possible controller gain (Aimax)- 

Care should again be taken with the choice of engineering units. Controller scan interval ts) in most DCS is measured in seconds. So, if the flow range (F) is measured in m /hr, the result of this calculation should be multiplied by 3600 to ensure is dimensionless. If the 

---

Sump level. Usually, the bottom level manipulates the bottom flow or boilup. If the bottom sump is unbaffled, it is commonly used as a liquid source for both the reboiler and bottom product. In this case, for certain reboiler types (e.g., thermosiphon), the level must be controlled within a narrow range in order to supply a constant head to the reboiler. Level fluctuations may lead to unsteady reboiler action and column upsets (Sec. 15.3). If a baffle is installed, tight level control is normally not required, particularly if bottom product goes to storage. 

This last loop is the unusual element in the control structure. The tuning of most conventional level loops is simple because we use a proportional controller with a gain of 2. The level loop in the proposed stmcture is not conventional. In this application, we want fairly tight level control because the distillate/temperature loop depends on the vapor/level loop. In addition, the dynamics of the vapor/level loop contain some dynamic lags because the pressure loop is involved, that is, increasing reboiler heat input increases pressure, and the pressure controller increases condenser heat removal, which affects reflux-drum level. 

We may wish to apply very different tuning criteria. It may be more important to minimise disturbances to the manipulated flow than it is to maintain the level close to SP. This type of controller performance is known as averaging rather than tight level control. Averaging control can dramatically reduce the impact that flow disturbances have on a process. 

Its most obvious application is to feed surge drums. These are included in the process design specifically to reduce the effect of upstream flow disturbances on the downstream process. Installing tight level control in this situation makes the drum ineffective. 

The method used to tune the controller is very similar to that applied to tight level control. We start as before with a proportional-only controller. However, rather than eliminate the flow imbalance as quickly as possible we do so as slowly as possible. In this case the controller will take considerably more than one scan to make the correction, i.e. 

This, however, is just a first step in the controller design. Unlike tight level control we cannot retain such a proportional only controller. As we can see in Figure 4.8 the level, as designed, remains at the alarm limit set at 90 %. 

Three-element level control is most commonly applied to the control of water level in steam drums on boilers. However it is applicable to many other situations where tight level control is required and is made difficult by unusual dynamics. 

For some apphcations, tight level control is desirable. For example, a constant liquid level is desirable for some chemical reactors or bioreactors in order to keep the residence time constant. In these situations, the level controller settings can be specified using standard tuning methods. If level control also involves heat transfer, such as for a vaporizer or an evaporator, the controller design becomes much more comphcated. In such situations special control methods can be advantageous (Shinskey, 1994). 

Do the following, assuming that the flow transmitters and the control valve have negligible dynamics. Also assume that the objective is tight level control. 

The block diagram for the selector control loop used in the slurry example is shown in Fig. 16.16. The selector compares signals P and P2, both of which have the same units (e.g., mA or %). There are two parallel feedback loops. Note that Gy is the transfer function for the final control element, the variable-speed drive pump. A stability analysis of Fig. 16.16 would be rather complicated because the high selector introduces a nonlinear element into the control system. Typically, the second loop (pump flow) will be faster than the first loop (level) and uses PI control (although reset windup protection will be required). Proportional control could be employed in the slower loop (liquid level) because tight level control is not required. 

Note that averaging control can be used for Hp and Hp where tight level control is not required to smooth out the effect of disturbances, but not for Hp. 

One exception to the mle of averaging level control for surge drums is in the case of distillation column hold-ups. Averaging level control should not be used to control the reflux drum level or the reboiler sump level. Tight level control is required for these vessels to maintain the integrity of the column material balance so that changes in the reflux rate and reboiler duty will have the desired effect on product compositions and yields without introducing additional lag to the system. 

---