All exceptions have one common property: the function in the 1/r(U) - plot shows either an infinite value at zero conversion or at least an increase with lower conversion. This behaviour causes that the area under the function gets higher than the area of an equivalent rightangle, as it is valid for the CSTR. That means that the space time gets here longer in not back-mixed reactor types. The reaction type seems to get extremely slow, when there is nearly no conversion (at the beginning of the reaction, regarded as a batch). And that is the reason for this 'special type of reactions':
The common behaviour of these reactions (first 3 items!) is that they are either not running, "when there is not something existing that will be produced in the reaction course", or that something is 'growing in a sort of induction period'. And under these circumstances it is clear that a total initial backmixing brings 'back' a certain level of this 'initiating' compound. Therefore it is better to take a CSTR for the 'starting phase' of such a reaction. But it is important to realize that there is an optimum space time at an optimum conversion and all chosen higher conversion levels produce increasingly worse space times. Therefore it is necessary to take a serial reactor combination of e.g. a CSTR and a TFR. Another possibility is the choice of a loop reactor, where the amount of back-mixing can be adjusted with the recirculation factor.
take your browser back for previous text or: