The fact that your constraint is an equation will make the feasible region nonconvex. You can approximate it within a MILP, but to get the exact relationship would require a nonlinear solver.
------------------------------
Paul Rubin
Professor Emeritus
Michigan State University
------------------------------