Modeling, cross-validation, and optimization of a shipboard integrated energy system cooling network

This work presents the modeling, cross-validation, and thermodynamic optimization of a shipboard integrated energy system (IES) cooling network to assess (1) the practicality of a dimensionless graph-based model for scalability; (2) the reliability of different statistical metrics in cross-validation; and (3) the effects of design and operational parameters on the system thermal performance. The IES comprised two 2.5 MW induction motors representing a ship propulsion system which operated at various torques over time resulting in different heat generation profiles. A dimensionless first-law graph-based model was formulated with the lumped capacitance approach, and the unknown physical parameters were estimated using the ant colony optimizer with three different calibration metrics. The validated model was used to distribute the total heat exchanger inventory N optimally across the IES by minimizing the peak system temperature. Subsequently, the study demonstrated the robustness of the optimal heat exchanger area allocation against N and the 20 x reduction in N with a tenfold increase in the motor cooling power.