Recycle snowball effects

Luyben (1993a) provided valuable insights into the characteristics of recycle systems and their design, control, and economics, and illustrated the challenges caused by the feedback interactions in such systems, within a multi-loop linear control framework. Also, in the context of steady-state operation, it was shown (Luyben 1994) that the steady-state recycle flow rate is very sensitive to disturbances in feed flow rate and feed composition and that, when certain control configurations are used, the recycle flow rate increases considerably facing feed flow rate disturbances. This behavior was termed the snowball effect. ... 

Luyben, W. L. (1994). Snowball effects in reactor/separator processes with recycle. Ind. Eng. Chem. Res., 33, 299-305. 

The important advantage of this strategy is that the reactor behaves as decoupled from the rest of the plant The production is manipulated indirectly, by changing the recycle flows, which could be seen as a disadvantage. However, it handles nonlinear phenomena better, such as for example the snowball effect or state multiplicity. Additionally, this strategy guarantees the stability of the whole recycle system if the individual units are stable or stabilized by local control. 

The second issue regards the optimal plantwide material balance. It is clear that the raw materials must be fed only in amounts required by the target production and selectivity. The control structure of fresh feeds should allow flexibility, within predefined limits, both in production rate and selectivity, but avoiding large variation of recycles that might upset some units (snowball effect). 

In CS3 (Figure 5.25), the hydrogen fresh feed is increased by about 9% from 348.5 to 380, while the recycle flow of phenol remains fixed to 220kmol/h. This control structure works well. Both the production of cyclohexanone and cyclohexa-nol is increased by about 4%, while phenol makeup increases with 8%. The purity of both products remains above 98%. A somewhat shorter transition time is obtained. The fact that hydrogen pushes the plant better than phenol is quite surprising, but it can be explained by the fact that there is no snowball effect on the gas-recycle side. 

In this section we explore two basic effects of recycle (1) Recycle has an impact on the dynamics of the process. Th e overall time constant can be much different than the sum of the time constants of the individual units. (2) Recycle leads to the snowball effect. This has two manifestations. one steady state and one dynamic. A small change in throughput or feed composition can lead to a large change in steady-state recycle stream flowrates. These disturbances can lead to even larger dynamic changes in flows, w hich propagate around the recycle loop. Both effects have implications for the inventory control of components. 

We call this high sensitivity of the recycle flowrates to small disturbances the snowball effect. We illustrate its occurrence in the simple example below, It is important to note that this is not a dynamic effect it is a steady-state phenomenon. But it does have dynamic implications for disturbance propagation and for inventory control. It has nothing to do with closed-loop stability. However, this does not imply that it is independent of the plant s control structure. On the contrary, the extent of the snowball effect is very strongly dependent upon the control structure used. 

However, we see in this strategy that there is no flow controller anywhere in the recycle loop. The flows around the loop are set based upon level control in the reactor and reflux drum. Given what we said above, we expect to find that this control structure exhibits the snowball effect. By writing the various overall steady-state mass and component balances around the whole process and around the reactor and column. wre can calculate the flow of the recycle stream, at steady state, for any given fresh reactant feed flow and composition. The parameter values used in this specific numerical case are in Table 2.1. 

Notice that the total rate of recycle plus fresh feed of A is flow-controlled. There is a flow controller in the recycle loop, which prevents the snowball effect. Sometimes the fresh feed of A is added directly into the reflux drum, making the effect of its flow on reflux drum level more obvious. The piping system where it is not added directly to the drum still gives an immediate effect of makeup flow on drum level because the flowrate of the total stream (recycle plus fresh feed) is held constant. If the fresh feed flow increases, flow from the drum decreases, and this immediately begins to raise the drum level. 

A stream somewhere in all recycle loops should be flow controlled. This is to prevent the snowball effect and was discussed in Chap. 2. 

Increasing the reactor temperature setpoint increases the production rate of vinyl acetate, so there must ultimately be net increases in all three fresh reactant feed streams. Oxygen and ethylene flows respond fairly quickly within about 20 minutes. However, the acetic acid feed actually decreases for the first 60 minutes in response to an increase in column base level. These results demonstrate the slow dynamics of the liquid recycle loop and illustrate the need for controlling the total acetic acid flow to the reactor so that the separation section does not see these large swings in load ( snowball effect"). The variability is absorbed by the fresh feed makeup stream. 

The handling of impurities in a VCM plant has been investigated by Dimian et al. (2001) by means of computer simulation. It has been demonstrated that selective chemical conversion of intermediate impurities can be used to prevent their accumulation and the occurrence of snowball effects in the separation units. The separation of impurities can be properly handled by exploiting the interaction effects through recycles. More details are given in the Chapter 17 (Case Study 3). 

Snowball effect designates the situation in which small variations of a flowsheet variable, generally a flow rate at the inlet or outlet of a process with recycles, generates large variations of streams around some units inside the process (Fig. 13.7). It is worthy to note that snowball is essentially a steady state effect, and not a dynamic one, because at finite disturbance the amplification remains finite. However, some units cannot tolerate large fluctuations in flows, particularly the distillation columns. Therefore, the designer should avoid snowball effects already at the conceptual stage. 

Figure 13.7 Increase in the recycle flow rate due to a snowball effect...

Chemical conversion is an effective way to counteract the accumulation of impurities due to positive feedback. Also, changing the connectivity of units may be used to modify the effect of interactions, for example by preventing an excessive increase in recycles due to snowball effects. Effective plantwide control structures may imply controlled and manipulated variables belonging to different but dynamically neighbouring units. The methodology to evaluate the dynamic inventory of impurities consists of a combination of steady state and dynamic flowsheeting with controllability analysis. This is used to assess the best flowsheet alternative and propose subsequent design modifications of units. Case Study 3 in Chapter 17 will present this problem in more detail. 

Snowball effects designate high sensitivity of recycle flows to small changes in input/output flows or in some parameters of units. The snowball is essentially a steady state effect, but the large magnitude of variations and the duration of transients can affect seriously the operation and control, leading eventually to plant upset. Therefore, the snowball effects must be avoided by design. 

An important phenomenon has been observed in the operation of many chemical plants with recycle streams. The same phenomenon has been observed and quantified in numerical simulation studies of industrial processes with recycles. A small change in a load variable causes a very large change in the flow rates around the recycle loop. We call this the snowball effect. 

Since Luyben identified the snowball effect (Luyben, 1994), the sensitivity of reactor-separator-recycle processes to external disturbances has been the subject of several studies (e.g., Wu and Yn, 1996 Skogestad, 2002). Recent work by Bildea and co-workers (Bildea et al., 2000 and Kiss et aL, 2002) has shown that a critical reaction rate can be defined for each reactor-separator-recycle process using the Damkohler number. Da (dimensionless rate of reaction, proportional to the reaction rate constant and the reactor hold-up). When the Damkohler number is below a critical value, Bildea et al. show that the conventional unit-by-unit approach in Figure 20.15 leads to the loss of control. Furthermore, they show that controllability problems associated with exothermic CSTRs and PFRs are resolved often by controlling the total flow rate of the reactor feed stream. 

Equation (20.19) shows that for values of Da much larger than unity, no snowball effect is expected. The snowball effect occurs as Da approaches a critical value of 1, but is eliminated by controlling the recycle flow rate, as shown next. 

Be aware of the snowball effect in a reactor-separator-recycle network and the importance of designing an adequate control system, which is presented in Sections 20J (Example 20.11) and 21.5 (Case Study 21.3). 

SNOWBALL EFFECTS IN THE CONTROL OF PROCESSES INVOLVING RECYCLE... 

---