P-only control

So what can we do in this case If column operation requires that we stick to this control structure, feedforward control will help to reduce the swings in distillate flowrate. However, better plantwide performance can be achieved if we can switch the control structure to one in which the vapor distillate is not used to control pressure. One possible alternative is shown in Fig. 6.27. Condenser cooling is used to control pressure, reflux flowrate controls reflux drum level (with P-only control), and the flowrate of the vapor distillate is ratioed to reflux flowrate. With this structure we allow the disturbances that the column energy... 

Initially use proportional-only controllers in all loops except flow7 controllers, where the normal tight tuning can be used K = 0.5 and T = 0.3 minutes). Set the gains in all level controllers (except reactors) equal to 2. Adjust the temperature, pressure, and composition controller gains by trial and error to see if you can line out the system with the proposed control structure. If P-only control cannot be made to work, PI will not w7ork either. When stable operation is achieved, add a little reset action to each PI controller (one at a time) to pull the process into the setpoint values. 

FIGURE 15.24 The effect of on the response of a P-only controller for a first-order process to a setpoint change. Note that is increased from 1 to 3. 

When choosing between P-only, PI, or PID controllers, one should consider the dynamics of the combined actnator/process/sensor system. For conventional control loops in the CPI, about 93% are PI controllers, 2% are P-only controllers, and 5% are PID controllers. The following guidelines can be nsed to choose the proper controller mode based on process dynamics and control objectives. 

P-only control is used for processes that are not sluggish and for which some degree of offset is acceptable. A sluggish process is characterized as a process that does not respond qnickly to changes in the manipulated variable (i.e., not a first-order-like response). Typical applications are level control and pressure control. Many control loops should use P-only controllers bnt instead use typical PI or PI with a relatively small amount of integral action, since most operators do not want offset from setpoint. 

The pressnre sensor is qnite fast, whereas the process (change in pressure for change in vent valve stem position) and the actnator are generally the slowest elements in the feedback system therefore, this is also a relatively fast-responding process. The P-only controller can be used if offset elimination is not important, and a PI controller can be used when offset elimination is important. 

Using setpoint changes, increase in small increments nntil the response meets the tnning criterion. (See Fignre 15.45, which is based on a 1/6 decay ratio.) For tnning a P-only controller, the tnning procedure is completed. 

P-only controllers with Kc = 2 are used for both reflux drum levels as suggested in Luyben and Kc = 10 is used for both base bottom levels for faster dynamics of the internal flow of the overall process and also for faster increase or decrease of entrainer makeup into the system. The PI settings of the top pressure control loops for both columns are set... 

Once Ku and are known, Ziegler and Nichols provide simple calculations for the derivation of tuning constants. These are, for a P only controller... 

Another published method frequently quoted is that by Cohen and Coon (Reference 6). It too uses the quarter decay criterion and is presented as sets of formulae based on process dynamics. For P only control... 

Figure 16.10 shows the closed-loop responses for the Smith predictor (0 = 2) and PI control (0 = 0). The controller settings are the same as those developed in Example 16.2 for 0 = 0. A comparison of Fig. 16.7 (dashed line) and Fig. 16.10 shows the improvement in performance that can be obtained with the Smith predictor. Note that the responses in Fig. 16.10 for 0 = 2 and 0 = 0 are identical, except for the initial time delay. This closed-loop time delay results from the numerator delay term in (16-22). The process model G (5-) is second-order and thus readily yields a stable closed-loop system for P-only control, but not necessarily for PI control (cf. Example 14.4). 

A proportional plus integral controller will give a response period that is longer than a P-only controller but much shorter than an 1-only controller. Typically, the response period of the process variable PV under PI control is approximately 50 per cent longer than for the P-only (1.5th, Figure 4.11). Since this response is much faster than I-only, and only somewhat longer than P-only control, the majority (>90 per cent) of controllers found in plants are PI controllers. The equation for a PI controller is... 

The PI controller gain has an effect not only on the error, but also on the integral action. When we compare the equation for a PI controller (Equation 4.14) with that for a P-only controller (Equation 4.11) we see that the bias term in the P-only controller has been replaced by the integral term in the PI controller. Thus, the bias term for PI... 

The Kc and 71 are used to adjust the PI controller gain to give the loop a desired response. Suppose 71 = oo, which would result in = 0, regardless of the value of Kc- In effect, the response would be that of a P-only controller with a period equal to and a sustained error. Although 71 = oo is not realizable, it can be set to a very large number in min/repeat to minimize the integral action. 

Although the response period of a loop under PI control is only 50 per cent longer than that for a loop under P-only control, this may in fact be far too long if in is as large as 3 or 4 h. In order to increase the speed of the response, it may be necessary to add an additional control mode. 

Figure 4.16 shows a comparison of the control-loop response to a load upset for both P-only and PD control. The response of the measurement PV under PD control is faster and results in a smaller offset than the loop under P-only control. This faster response is due to the addition of the derivative action. 

In a PI controller, in order to minimize the integral action, Tj was made a large number. This makes the integral gain approach zero, and the controller then behaves essentially like a P-only controller. However, in the PD controller, even by setting Ti to a very small value, there is still the possibility of a sizeable derivative contribution if there is a noisy input, i.e. if dCV/dr is large. 

The addition of the derivative mode in the PID controller provides a response similar to that of a P-only controller, hut without the offset because of the integral action. Therefore, a PID controller provides a tight dynamic response, but since it contains a derivative block it cannot be used in any processes in which noise is anticipated. 

Flow When dealing with a flow loop, P-only control can be used with a low controller gain. For accuracy, PI control is used with a low controller gain and high integral action. Derivative action is not used, because flow loops typically have very fast dynamics and flow measurement is inherently noisy. 

Level Levels represent material inventory that can be used as surge capacity to dampen disturbances. Hence, loosely tuned P-only control is sometimes used. However, most operators do not like offset, so PI level controllers are typically used. 

Now consider the case, shown in Figure 7.18, of a tank with a P-only controller and some valve hysteresis. All valves have some hysteresis, but excessive valve hysteresis typically occurs when the valve sticks as it tries to open and close. This can happen for a number of reasons, including overtightened packing, etc. 

Without integral action, P-only controllers typically never operate at the set point, so there is always an offset between the level and its set point. While this is actually what provides some of the benefits described above, many operators dislike seeing this sustained offset and resist its use. 

---