Supplier Diversification Under Geopolitical Uncertainty: A Decision Model for Resilient Oil Procurement

Geopolitical shocks in 2025 showed how quickly a low-cost crude source can become unavailable or expensive to deliver. The Israel and Iran confrontation threatened traffic through the Strait of Hormuz and pushed freight and war-risk premiums sharply higher, renewed Houthi attacks in the Red Sea revived Cape of Good Hope diversions, and naval drone strikes interrupted loadings at the CPC terminal near Novorossiysk. These events were route-specific and partly correlated, and they arrived in a year of soft benchmark prices, when procurement decisions guided by delivered price alone appeared safe. This paper develops a decision model that helps refiners and national oil companies decide not only which suppliers to buy from but how much to take from each, comparing alternatives on four criteria: purchase cost, transport risk, lead time and supply security. The model has three linked stages. The Analytic Hierarchy Process (AHP) converts the priorities of a procurement committee into consistent criteria weights; TOPSIS ranks candidate suppliers and assigns them to strategic, flexible and opportunistic tiers that guide contract type; and a scenario-based mean and Conditional Value-at-Risk (CVaR) linear programme allocates monthly volume subject to supplier caps, chokepoint exposure limits, a lead-time ceiling and a minimum portfolio supply-security score. Lead time is priced through in-transit carrying cost and late-arrival penalties, and disruptions are generated with a Gaussian copula so that Middle East and Black Sea outages can occur together and coincide with spikes in replacement prices. The model is applied to a stylised western Mediterranean refinery buying 6.0 million barrels per month from eight origins and is stress tested under five regimes calibrated to 2025 events. Norway, the United States and Brazil rank highest, yet the optimal portfolio (United States 30%, Kazakhstan 20%, Brazil 20%, Norway 15% and Libya 15%) also takes capped volumes of discounted, lower-ranked crude, showing that rankings and allocations answer different questions. The recommended portfolio costs only USD 0.07 per barrel more in expectation than a pure cost minimiser, removes Hormuz exposure and raises the portfolio security score to 7.50. Under compound stress it lowers the 95% CVaR of delivered cost by USD 4.47 per barrel, about USD 26.8 million in a severe month, while single sourcing reaches a tail cost of USD 89.46 per barrel. Re-optimising once a regime shift is recognised reduces compound-stress CVaR further, from USD 77.13 to USD 75.89 per barrel, and sensitivity analysis shows that most of the protection comes from the security floor and chokepoint caps rather than from the risk-aversion parameter. The results indicate that resilience is inexpensive when it is bought through explicit, auditable constraints, that supplier rankings are better used to set contract tiers than volumes, and that pre-agreed rebalancing triggers offer practical protection for procurement desks. The framework is transparent and solvable with standard tools, although its inputs are illustrative and should be replaced with firm-specific data before use.