I guess an approximate Integral function may be used to replace the actual Integral:
1. choose a large-enough Upper Limit of your Integral function: it is usually that the approximate integral value is close to the accurate value, when its Integral Upper Limit of is large enough;
2. choose a resonable Discretisation methods supported by gPROMS, like non‐uniform-interval Discretisation methods ( the code is like TRANSFORM( LOG, 4.0) ] ), which may greatly decrease computation load;
3. check the accuracy of the approximate Integral function;
4. if the accuracy of the approximate Integral is good enough, then it could be used in your model