Improved quadratic cuts for convex mixed-integer nonlinear programs

Lijie Su, Lixin Tang, David E. Bernal, Ignacio E. Grossmann

Abstract: Highlights•Scaled quadratic cuts proposed for Outer Approximation and Partial Surrogate for convex MINLP.•Scaled quadratic cut proved to be tighter underestimate than tangent cut for convex functions.•Scaled quadratic cuts integrated with hybrid cuts, multi-generation cuts in OA and PSC methods.•Six solution methods, OA-QCUT, OA-MQCUT, OA-HCUT, OA-MHCUT, PSC-QCUT and PSC-MQCUT, are developed.•Numerical results demonstrate the effectiveness of solution methods with scaled quadratic cuts.