This is an interesting question, but I think can easily get a little too philosophical as to what 'modeled directly' means or what a software was 'designed to model' that the code writers were probably deliberately trying to avoid.
I read the key sentences in the sections in 90.1 where 'Exceptional Calculation' is mentioned as (in 90.1-2013):
11.4.5: "Where no simulation program is available that adequately models a design, material, or device...", and
G2.5: "When the simulation program does not model a design a design, material, or device..."
(emphasis mine)
Both Chapter 11 and Appendix G give detailed descriptions of what an accepted 'Simulation Program' are and the features they shall have. DOE-2 is explicitly called out in both places, and EnergyPlus explicitly in Appendix G. Therefore in my opinion, the use of EnergyManagmentSystems or use of curves or dummy zones would not necessarily fall under the rubric of Exceptional Calculation Method, since they are done within the capabilities of the simulation program, and are "adequate" (at least in the sense that, if the Rating Authority finds something fishy in the results they can request more information on the inputs, justification, etc.).
So I think the ultimate issue here is: both ECB and PRM provide a framework for demonstrating building performance, but leave the option open to the Rating Authority to define its own specific standards in both approving the simulation program and the calculation methods. If the RA doesn't think the way you're using EMS (or the use of EMS in general) is an acceptable capability of the simulation program EnergyPlus, that's fine, but then you can follow the Exceptional Calculation Methodology to make your case for its use (which they can then accept or reject).
Edit: in summary: ¯\_(ツ)_/¯
Are you asking the definition in 90.1 or in LEED?