Material resource planning is an integral part of supply chain management. The tasks in the supply chain need materials and resources to be executed, thus, allocating resources correctly is an important part of task scheduling. Specifically, construction tasks for subsea wells require the use of resources, such as rigs, and planning the schedule of these operations involves the sizing of various materials and services necessary for their execution. This study is motivated by real-life scheduling planning from a large Oil and Gas company that estimates the demand for materials and services stochastically due to the uncertainties associated with the tasks in their start dates and durations. The calculation of the demand is subject to the current schedule that the company has and a set of rules that indicate allocation conditions, logistics parameters, disembarking conditions, and dependencies to allocate the tools and services needed for each task and estimate their quantity and how many days they will be used. These sets of tools and rules can change depending on the user and their operation knowledge. Additionally, the company uses a large number of scenarios, which results in extremely high computational times and impacts operational decision-making. In this context, scenario reduction could assist the company in its decision-making process. The methodology proposed in this work evaluates and identifies representative scenarios of uncertainty in strategic planning schedules of offshore rigs in order to reduce the number of scenarios used in the calculation of the demand for tools and services. With the use of unsupervised techniques, such as k-means and hierarchical clustering, we identified a subset with the most representative scenarios for the scenario reduction. The Wasserstein Distance and graphical visualization were used to measure the representativeness of the selected scenarios and find the best subset. Moreover, the scenario reduction subset was also used to analyze the impact of the reduction in the demand calculation. The Agglomerative Clustering with Ward Linkage (hierarchical clustering) obtained the best clustering evaluation and representativeness metrics, resulting in a selected subset of 782 scenarios. To find a minimal representative set of scenarios, the best clustering method and the Wasserstein Distance were used, resulting in a number of 343 scenarios. This presents a reduction of 84 percent in the execution time of the demand calculation, with the highest error of 11 percent in the demand calculation.