Variance minimization in resource leveling for self-financing project portfolios: a convex MIQCP approach

Mirkhorsandi Langaroudi, Seyed Mahdi; Khosravi, Hossein; Davoodi, Alireza; Movahedifar, Seyed Mojtaba

doi:10.22124/jmm.2026.31598.2845

Variance minimization in resource leveling for self-financing project portfolios: a convex MIQCP approach

Articles in Press

Document Type : Research Article

Authors

¹ Department of Civil Engineering, Ne.C., Islamic Azad University, Neyshabur, Iran.

² Department of Civil Engineering, Hakim Sabzevari University, Sabzevar, Iran

³ Department of Mathematics, Ne.C., Islamic Azad University, Neyshabur, Iran

⁴ Department of Civil Engineering, Ne.C., Islamic Azad University, Neyshabur, Iran

10.22124/jmm.2026.31598.2845

Abstract

This study addresses the critical trade-off between financial returns and operational stability
in capital-intensive project portfolios. We propose a novel convex Mixed-Integer Quadratically
Constrained Programming (MIQCP) framework that unifies Net Present Value (NPV) maximization, strict self-financing, and direct resource variance minimization. Unlike existing non-convex or
heuristic models, our approach endogenizes flexible phasing strategies and introduces a dual-buffer
mechanism to protect both liquidity and resource capacity. By exploiting the positive semi-definite
properties of the quadratic constraints, we ensure global optimality for portfolios with 50+ activities. Computational results reveal a significant ”constrainedness” effect, where tighter financial and
precedence constraints accelerate convergence by pruning the search tree. Findings demonstrate that a negligible NPV sacrifice (< 2%) yields disproportionate gains in resource stability (> 8%), providing a high-fidelity decision-support tool for managing internal capital markets under high volatility.

Keywords