Backmixing residence time distribution and micro-mixing

There is another practical method for estimating conversions in reactors with residence time distribution, for perfect micro-mixing, that is also applicable to other reaction orders. To this end the reactor is simulated by a model that consists of a cascade on N perfectly mixed equal reactors (section 3.3.3). The RTD-function of the cascade with total residence time x can be calculated  

16) is plotted in figure 7.4 for different vdues of N, It can be used for modelling a real reactor with a non-ideal residence time distribution. The RTD-funcdon can be found from tracer-response measurements. By fitting eq. (7.16) with the experimental RTD-curve the fictituous number N can be determined. Tliis number merely is a measure of the residence time distribution of the real reactor. Then the chemical conversion is calculated for the cascade with N reactors, on the basis of reaction kinetics, using eq. (3.50), for which the conversion in each successive reactor is calculated with eq. (3.49), or one of the previous equations. 

Other examples of black box models are presented in section 7,2,32, The influence of backmixing in tubular reactors is discussed in section 7.2.2. 

A first order chemical reaction is carried out in a homogeneous solution, in a reactor with a diameter of 3 m the liquid height is 7 m. The mean residence time of the liquid in the reactor is 10 minutes. The reactor is equipped with a stirrer, consisting of two inclined blade impellers mounted on one shaft. The 

What is a likely explanation for these findings And how can this be checked Probably the t l narrow reactor is not perfectly macro-mixed. It may behave as a cascade of N reactors in series. By more intensive mixing N will go down, and the conversion also. We know that for = 1, the conversion is 0.9. With eq. (3.51) we find that in the first situation, the residence time distribution in the big reactor corresponds to N 2. With the higher stirrer speed, N drops evidently to a value = 1.5. This seems plausible. It can be checked with RTD-measurements, and fitting the results with eq. (7.16). 

---