Walmsley, Timothy GordonWalmsley, MichaelUdugama, Isuru A.Hall, Keegan2026-01-132026-01-132025https://hdl.handle.net/10289/17868Industrial energy use is one of the most significant contributors to global greenhouse gas emissions. One of the most effective strategies for reducing heat demand is optimising the design of heat exchanger networks (HENs). While automated synthesis methods like mathematical programming have long promised optimal designs, their industrial adoption remains limited. Key barriers include expensive software licenses, complexity of models and a disconnect between proposed thermal savings and real-world cost savings, especially for non-continuous processes. This thesis addresses these challenges by developing two complementary open-source tools that bridge the gap between advanced process integration methods and practical industrial implementation. The novel contributions in this thesis are delivered across two primary streams: (1) the development of OpenHENS, an entirely open-source tool for synthesising cost-optimal HENs using a robust multi-stage solution strategy, and (2) a machine learning-based surrogate modelling approach for predicting utility system performance under new heat recovery or plant configurations. Together these contributions allow engineers to investigate multiple heat recovery options and accurately evaluate the operational cost savings. OpenHENS combines a novel three step strategy to systematically reduce model complexity and generate a broad set of structurally diverse near-optimal heat HENs. Engineers then apply their judgement in selecting a design that aligns with real-world constraints such as spatial layout, capital budgets, and controllability, factors often too complex to model directly. OpenHENS is publicly available as a Python-based open-source tool and is designed to be accessible to engineers without prior experience in mathematical programming or coding. When validated on thirteen common benchmark problems, OpenHENS consistently returned solutions within 8% of the lowest known total annualised costs reported in the literature using commercial optimisation software. To support credible evaluation of energy efficiency projects, this thesis also develops a surrogate modelling framework tailored for large, non-continuous industrial sites where utility system behaviour is influenced by variable production, operator decisions, and equipment constraints. Trained on high-resolution plant data, the model captures non-linear, time-dependent system behaviour and is used to predict fuel consumption, cogeneration, boiler steam generation, and equipment-level steam demand across full-year operational periods. The approach is demonstrated through the evaluation of a hot water network retrofit at a pulp and paper mill, where standard costing methods were found to overestimate annual savings by NZD $9.8 million. The surrogate model also enables fair cost allocation prior to the design phases by quantifying the marginal steam cost (MSC) at each plant. Results showed that the hot water network retrofit had a MSC of -$10 per tonne, whereas the pulp dryer and paper machine had the highest MSCs around $20 per tonne. Beyond HEN evaluation, the surrogate model supports strategic scenario analysis for energy decision-making. The model was used to assess the implications of shutting down paper production on steam balances, asset utilisation, and site-wide fuel costs. The results revealed unintended consequences, including increased natural gas usage and instability due to low boiler turndown. Because the model reflects actual operational patterns embedded in historical data, these insights underscore the need for improved utility system operation. Additional ‘what-if’ scenarios demonstrated the potential for significant cost reductions by integrating new power generation technologies that recover surplus steam from biomass combustion.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.Thesis