Ocean thermal energy conversion (OTEC) systems utilize from the difference between the temperatures of surface and deep water and drive a thermodynamic Rankine cycle for electric power generation. The generated power depend on the temperature of the surface water as warm source and due to the variation in the temperature of surface, the output power of the OTEC system frequently changes. This uncertainty nature results in the variation in the generated power and so, integration of large-scale OTEC generation units to the power system is a challenging problem and new techniques must be developed for studying the effects of resources on the power system. Therefore, the balance between generation and consumption is important and from reliability point of view, spinning reserve must be scheduled to prevent load curtailment in the events such as forced outages of generation units, transmission lines and so on. In a power system containing large-scale OTEC power plants, the uncertainty nature of these plants must be considered in the reserve scheduling and for this purpose, a reliability model considering both failure of composed components and variation in the output power, is developed and for determining a suitable multi-state model, fuzzy c-means clustering technique and XB index is utilized. Then, the proposed multi-state model is used for spinning reserve determination of a power system containing OTEC plants using of the modified PJM method. Numerical results associated to RBTS and IEEE-RTS present the effectiveness of the proposed technique for operation studies of power systems containing OTEC systems.