I wanted to publicly acknowledge you for your help. I used an EXCEL add-in to model things.
The traffic I am modeling is split roughly 50-50 between voice and data. If I look at my current situation, suppose I have 200 Erlangs of traffic ging into a single trunk group engineered for 3% peak blocking (70% Retrial)… I wind up needing 214 circuits. If I split into two trunk groups, one for voice at 3% blocking, one for data at 10% blocking (70% Retrial on both), I wind up needing 112 circuits for voice and 104 circuits for data, or 216 circuits total. If I do what I thought to do originally, namely, route everything into anTrunk Group engineered for 10% blocking and then overflow only the voice calls to another trunk group so that the overall GOS for the vice calls was 3%, I wind up with 201 circuits and 4 circuits in the overflow group, for a total of 205. Finally, if I do what you suggested, route voice into a voice only group and then overfow it into a common group, I wind up with 68 circuits in the first group and 134 circuits in the second group, for a total of 202 circuits. Your suggestion is obvious to me now, since it has me using two more symmetrically-sized trunk groups… but I don’t think I would have thought of it on my own.