Optimization of heat exchange networks via differentiated initial structure-guided evolution strategy for heat transfer units

Heat exchanger network （HEN） optimization represents a challenging mixed-integer nonlinear programming （MINLP） problem with a vast solution space, where the selection of initial points critically influences optimization performance and requires proper distribution across the search domain. This study developed a novel approach employing three complementary strategies for generating differentiated initial structures: a virtual node method utilizing stream temperature differences to guide optimal heat exchanger matching through virtual temperature settings, a flow-splitting technique that allocates stream nodes proportionally according to heat capacity flow rates, and a heat exchanger assignment approach that redistributes thermal loads within pre-connected structures. Together, these methods created individualized initial configurations while ensuring uniform distribution of heat exchange units throughout the solution space. Comprehensive testing across multiple case studies demonstrated the effectiveness of this approach, achieving optimized annual costs of 2 891 084.9 USD·y⁻¹, 1 412 538 USD·y⁻¹ and 1 940 386 USD·y⁻¹ while significantly enhancing both search space coverage and optimization efficiency compared to conventional methods. The results confirm the superior performance of this structured initialization strategy in addressing the complex challenges of HEN optimization.