A Rolling Horizon MILP Framework for Strategic Well Drilling Planning in Semi-Arid Regions: Integration of Spatial Interference Constraints and Resource Allocation

Abstract Groundwater mobilization through well drilling is essential for agricultural sustainability in semiarid regions facing increasing water scarcity. This study addresses the sequencing of drilling and commissioning operations for irrigation wells under regulatory spatial interference constraints—specifically, the prohibition of simultaneous operations within 200-m exclusion zones mandated by national water codes. We formulate a mixed integer linear programming (MILP) model integrating the complete well lifecycle (drilling, activation, operation) with capacity limitations and operational precedence. Given computational intractability for realistic instances (up to 1.8 million binary variables), we propose a hybrid rolling horizon method combining relax-and-fix temporal decomposition with sliding local reoptimization and lightweight metaheuristics. Computational experiments on instances calibrated with Tunisian operational data (12 programs, 487 wells, 2015–2023) demonstrate (1) near-optimal performance with median optimality loss below 1% against proven optimal solutions on benchmark instances; (2) 4.5%–6.5% improvement in discounted water coverage over sequential heuristics reflecting current institutional practices; and (3) scalability to 110 wells over 180 periods while maintaining average MILP gaps below 3%. Sensitivity analysis identifies robust parameter configurations ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper D equals 15" display="inline" overflow="scroll"> <mml:mi>D</mml:mi> <mml:mo>=</mml:mo> <mml:mn>15</mml:mn> </mml:math> days, <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="gamma equals 0.4" display="inline" overflow="scroll"> <mml:mi>γ</mml:mi> <mml:mo>=</mml:mo> <mml:mn>0.4</mml:mn> </mml:math> ) balancing solution quality and computational tractability. The contribution lies not in algorithmic novelty—the techniques employed are established—but in formulating a previously unaddressed problem class, adapting solution methods through problem-specific calibration, and providing comprehensive empirical validation. The framework offers actionable decision-support for public water authorities, with demonstrated gains translating to approximately <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="140,000 normal m cubed" display="inline" overflow="scroll"> <mml:mn>140,000</mml:mn> <mml:mtext> </mml:mtext> <mml:msup> <mml:mrow> <mml:mi mathvariant="normal">m</mml:mi> </mml:mrow> <mml:mrow> <mml:mn>3</mml:mn> </mml:mrow> </mml:msup> </mml:math> additional water supply per 80-well program. Results support evidence-based planning aligned with SDG 6 (clean water) and SDG 2 (zero hunger) in semiarid regions worldwide.