of a particular function type are of the same variable type. This

eliminates the need for more than one representation of a function type

in a computer program. It also avoids having to determine during

solution which output variable type would be required for a particular

set of function index values.

 This restriction is a reasonable one to make, and actually is

widely used in accepted solution procedures for typical modeling

problems in chemical engineering. One shortcoming might appear to be

the inability to assign different outputs to the equations describing

different sections of a process, such as a distillation column. The

rectifying and stripping sections are often treated separately and this

restriction would seem to prevent that. This is not the case, however,

since it is possible to describe the rectifying and stripping sections

as separate problems with the appropriate connection equations. This

has the added advantage that different solution procedures can be

derived for the two sections.

 It should be noted that each output variable in the example

problem had the same number of indices as the function for which it was

the output. Clearly, z could not have been chosen as the output for

either f or g since there are six equations of each type, but only five

variables of type z. If the number of indices on a function type

exceeds the number of indices on a variable type which occurs in that

function, the total number of functions can be made to exceed the total

number of variables by choosing appropriate index limits. For this

reason, a variable type cannot be assigned as the output for a function

type which has fewer indices than the function type.