Time (hours)

Figure 8.12 CS3 feed rate disturbances with and without QR/F ratio.

from about 6.6 mol% to less than 3 mol%. This is still very large compared to the specification of 0.5 mol%, but keep in mind that the disturbance is very large (a 50% step increase in feed flowrate), which is probably much larger than that to which an industrial column would typically be subjected.

Feed Composition Changes Perhaps more important are the performances of the three alternative control structures to disturbances in feed compositions, since feedforward is seldom an option to handle these disturbances. Figures 8.11b and 8.11d compare the three alternative control structures for two step changes in feed composition. At time equal to 0.2 h, the feed composition is changed from 40 to 45 mol% propane and from 30 to 25 mol% isobutane. Then at time equal to 3 h, the feed composition is changed from 45 to 36 mol% propane and from 25 to 34 mol% isobutane.

The performances of structures CS1 and CS2 are quite similar because in both the reflux flowrate is essentially constant since only feed composition is changing. There is some steady-state deviation from the desired product purities. Increasing propane content in the feed with a fixed stage 8 temperature produces (somewhat counterintui-tively) a higher level of isobutane impurity in the distillate and a lower level of propane impurity in the bottoms.

The response of structure CS3 shows some important results, both steady-state and dynamic. From the analysis discussed earlier and presented in Figure 8.6, we would expect the constant reflux ratio strategy used in CS3 would not handle feed composition changes as well as the constant reflux flowrate strategies of the other two structures, at least from a steady-state perspective. Figure 8.6 shows that a flowrate slightly higher than design reflux (4% higher for the high-distillate case and 10% for the low-distillate case) would give on-specification product purities over the range of feed composition. To achieve the same product purities with a constant reflux ratio strategy would require operating with 35% and 24% higher reflux ratios than the values at the design feed composition for the two cases. Thus it is no surprise that CS3, with its reflux ratio fixed at the design value, does not do as good a job in maintaining product purities for feed composition disturbances.

The results in Figure 8.11 show that an increase in propane concentration in the feed produces only a small steady-state shift in distillate purity when CS3 is used, which is less than that produced by the other two control structures. However, the change in the bottoms purity is larger than that produced by the other structures. The same occurs for a decrease in feed propane composition.

The results shown in Figure 8.6 are for the case when both distillate and bottoms purities are maintained. The results in Figure 8.11 reveal an asymmetric behavior in which distillate purity changes little but bottoms purity changes drastically. One fundamental reason for this is our selection of a rectifying section tray to control. Had we selected a tray in the stripping section, bottoms purity would be better maintained at the expense of larger changes in distillate purity. These results suggest that the method for selecting the control tray temperature depends on which product is more important.

Of course, this problem of steady-state shift in product purity for feed composition changes could be solved by using a cascade composition/temperature control (CC/TC) structure. Keep in mind, however, that the reflux ratio would have to be fixed at the highest value needed to handle the range of feed compositions.

The results shown above are for the high-distillate case. Similar results were obtained for the small-distillate case. However, the dynamics for the feed composition disturbances are much slower because the changes made in the propane concentration of the feed are much smaller (changed from 4 to 5 mol% C3).

8.2.5 Dual-Composition Control

The control structures studied up to this point all use single-end inferential control. A single tray temperature is controlled with the objective of maintaining a temperature profile in the column that we hope will hold product purities close to their specifications. This goal was achieved with varying degrees of success, depending on the control structure used and the disturbance. If tight product composition control is required, a dualcomposition control structure can be used. However, this would require two online composition analyzers, which are expensive and require high maintenance.

To illustrate the improvement in control that is achieved by using dual-composition control, the CS3 control structure is augmented by two composition controllers, one controlling propane impurity in the bottoms (CCxB) and the second controlling isobutane impurity in the distillate (CCxD). Figure 8.13a shows the Aspen Dynamics flowsheet. Three-minute deadtimes are used in the composition loops.

The controller output signal from the CCxD controller changes the ratio of the distillate to reflux. It is tuned using the relay-feedback test and Tyreus-Luyben tuning (KC = 1.8 and tj = 54 min with a composition transmitter span of 5 mol% isobutane). The controller output signal from the CCxB controller changes the setpoint of the stage 8 temperature controller, which in turn changes the ratio of the reboiler heat input to feed flowrate. It is tuned using the relay-feedback test and Tyreus-Luyben tuning (KC = 1.0 and tj = 37 min with a composition transmitter span of 2 mol% propane).

Setting up the multiplier for the reboiler heat input to feed ratio is a little tricky. The multiplier uses metric units: flows in kmol/h and heat in GJ/h. So the steady-state value of the second input to the multiplier, which is the temperature controller output signal, is calculated:

QR(GJ/h) (1.038 x 106 Btu/h)(1.055 GJ/106 Btu/h) _ 0 0241

Figure 8.13b shows the controller faceplates. Note that the flow controller on the vapor distillate (FCD) has a remote setpoint (on "cascade") coming from a multiplier ("ratio") whose two inputs are the reflux flowrate and the CCxD controller output signal. The temperature controller is also on "cascade" with its setpoint coming from the CCxB controller.

Figure 8.14 demonstrates the effectiveness of this dual-composition control structure. Figure 8.14a shows how product purities vary in the face of the same scenario of feed flowrate disturbances and feed composition disturbances used previously. A comparison of these results with the CS3 results given in Figure 8.11 reveals a very significant reduction in product quality variability, both dynamically and at steady state. Both products are returned to the specifications, even for feed composition disturbances.

Figure 8.14b shows the changes in other key variables. The solid lines represent feed flowrate disturbances; the dashed lines, feed composition disturbances. Note that the CCxB composition controller changes the setpoint temperature for the feed composition

Cascade Cascade

Figure 8.13 (a) Dual-composition control structure; (b) dual-composition faceplates.

Cascade Cascade

Figure 8.13 (a) Dual-composition control structure; (b) dual-composition faceplates.

disturbances, shifting it lower than the design 128°F for higher propane compositions in the feed and higher for the lower propane compositions in the feed. Likewise the reflux ratio is adjusted by the CCxD controller from its design value of 2.63 so that the distillate purity is maintained for feed composition disturbances.

The variability in the vapor distillate flowrate is still much less than with the other control structures, so even with dual-composition control the downstream unit is not subjected to large and rapid disturbances.

Feed rate changes

Feed comp changes

Feed rate changes

Was this article helpful?

0 0

Post a comment