Cloud Model-Based Decision Making for Deep-Water Sand Control Method Selection

Selection of sand-control methods for deep-water reservoirs must reconcile quantitative measures—such as reservoir pressure, grain-size distribution, cost and payback period—with qualitative criteria like sand-control performance, durability, applicability, safety risk and environmental impact. To bridge the gap between these indicator types and their inherent uncertainties, this paper develops a multi-criteria decision-making method that integrates cloud-model theory with an improved Analytic Hierarchy Process and objective weighting. Expert judgments expressed through linguistic variables are quantified via the cloud model’s expectation, entropy and hyperentropy parameters, replacing the traditional 1–9 AHP scale to better capture fuzziness and randomness. Objective weights are derived from a deviation maximization method, after which a novel fusion algorithm combines subjective and objective weights into a unified composite. Both quantitative data and cloud-model outputs form a normalized decision matrix, which is then evaluated by a projection-based optimization to determine each scheme’s closeness to the ideal solution. Application to a case study on the KT gas field—an unconsolidated sandstone reservoir prone to sand production—demonstrates the method’s effectiveness. Five candidate methods (premium screen, expanding screen, open-hole gravel packing, high-speed water packing and frac-and-pack) were scored and ranked, with projection values ranging from 0.4874 to 0.6832. The frac-and-pack approach achieved the highest closeness and was selected as the optimal solution, in agreement with expert recommendations. This example confirms that managing qualitative uncertainty alongside quantitative data yields clear, reproducible rankings and practical decision support. The contributions of this work include a cloud-model extension of AHP for rigorous uncertainty treatment, a weight-fusion methodology harmonizing subjective and objective indicators, a cloud-generator procedure for quantifying qualitative inputs, and projection-based scheme selection—offering new tools for robust multi-attribute decision making in petroleum engineering.