This is not possible to do this directly for 4 or more components which require the number of trays in the column to be specified and then its performance calculated.
As with multicomponent shortcut distillation, we specify the recoveries of a light and heavy key component. However, it is also necessary to specify the recovery of the non-key component.
If we are separating three components, A, B and C in decreasing order of volatility, there are two possibilities.
Here A is the light key, B the heavy key and C the nonkey.
Here B is the light key, C the heavy key and A the nonkey.
The value of nonkey recovery is significant and may need to be experimented with, as discussed below.
The program carries out two stage-to-stage calculations, one from the top and the other from the bottom of the column. The equilibrium stage compositions are displayed on the triangular diagram. The feed stage lies where the top and bottom composition profiles intersect.
In the McCabe-Theile binary design the optimum feedstage position is located from the top and bottom operating lines and the q-line. There is no corresponding criterion for the ternary calculation. The feed stage position and thus total number of stages depend on the nonkey recovery required.
It is not immediately obvious what the nonkey recovery should be. Almost by definition it is something that the designer is not usually interested in. However, the behaviour of the column can be quite sensitive to its choice. It should obviously be larger tha that of the key component which leaves the column with it, how much so will depend on how close boiling the two species are.
Too large a specified nonkey recovery will result in a column with more stages than necessary to give the specified light and heavy key recoveries. However too small a specification can also result in too many stages, and may actually cause a pinch occurs which prevents one of the profiles from ever reaching the feed point. It is therefore usually necessary to try different nonkey recoveries and note their effect.