Design, implementation and evaluation of a continuous flow bio-based solvent recovery system for petrochemical refineries

Thesis Abstract

Global petrochemical industries face escalating solvent losses, rising purification costs, and cumulative environmental burdens associated with conventional solvent recovery processes. In particular, legacy batch-wise solvent recovery units exhibit limited separation efficiency for azeotropic mixtures, variable feed compositions, and mismatch with fluctuating refinery throughput, leading to suboptimal solvent reuse, increased waste streams, and higher lifecycle emissions. This study aims to design, implement, and evaluate a continuous flow bio-based solvent recovery system integrated into a petrochemical refinery, enabling higher recovery yields, reduced energy consumption, and enhanced process sustainability. The objectives are to (i) develop a continuous flow solvent recovery configuration leveraging bio-based solvents with tailored polarity and low toxicity, (ii) model and optimize mass and energy balances using process simulation and experimental validation, (iii) implement a pilot-scale module within an operational refinery line to assess operability under real-world feed variability, and (iv) evaluate environmental and economic performance using cradle-to-gate lifecycle assessment and technoeconomic analysis. Theoretical grounding draws on the material balances and separation theory of multicomponent distillation with entrainers, supported by process intensification concepts and the Green Solvent Framework. The study tests hypotheses that (a) continuous flow integration with inline ultrafiltration and membrane-assisted pervaporation improves overall solvent recovery by at least 15% relative to incumbent units, (b) the use of bio-based solvents reduces total process energy intensity by 8–12% and lowers volatile organic compound emissions, and (c) a robust control strategy maintains product purity within ±2 wt% across feed variability ranges. Methodologically, the research follows a mixed-methods design combining experimental, computational, and field-based evaluation. The population comprises refinery solvent streams and recovery units at a mid-to-large-scale petrochemical complex. A pilotsection consisting of a continuous feed purifier, membrane-assisted separation module, and a bio-based solvent recovery column was constructed and tested with 20 representative feeds over 6 months, yielding 1200 experimental runs. Data collection employed inline spectroscopic process analytics (FTIR, NIR) for real-time composition, gas chromatography–mass spectrometry (GC-MS) for impurity profiling, and calorimetric measurements for energy consumption. Instrument validation included calibration against standard reference materials, with repeatability checks across three independent operators. For statistical analysis, regression models quantified relationships between feed composition, residence time, and solvent recovery yield; ANOVA assessed performance differences across solvent systems; and Monte Carlo simulations evaluated sensitivity to feed fluctuations. Process modeling utilized Aspen Plus for material and energy balances, incorporating a dynamic model of the heat integration network and a model predictive control (MPC) scheme to maintain target purity. Life cycle assessment (LCA) followed ISO 14040/44, comparing the baseline and the new system across cradle-to-gate boundaries, while the technoeconomic analysis (TEA) considered capital expenditure, operating costs, and payback period under three market scenarios. Key expected findings include demonstrable improvements in solvent recovery efficiency and purity, quantified energy savings, and reduced environmental impact. Specifically, anticipated outcomes are recovery yields enhancement by 18–22% relative to the existing unit, energy intensity reductions of 10–14%, and a corresponding decline in process emissions by 20–30% per tonne of solvent recycled. The study also anticipates that the MPC-driven control strategy will sustain stable product quality despite feed variability, with average process downtime below 1% of annual runtime. The anticipated contribution to knowledge spans (i) a validated design framework for continuous flow bio-based solvent recovery in refineries, (ii) empirical data on the performance and operability of membrane-assisted separations integrated with pervaporation in live refinery conditions, and (iii) a comparative LCA and TEA demonstrating the environmental and economic viability of bio-based solvent systems in petrochemical contexts. Overall, the research is expected to provide a scalable blueprint for retrofitting existing solvent recovery units with continuous flow bio-based solvents, delivering measurable improvements in sustainability, cost competitiveness, and regulatory compliance. Recommendations include standardizing feed characterization protocols, expanding pilot testing to diverse refinery configurations, and pursuing collaborative efforts with solvent producers to optimize bio-based solvent formulations for wide industrial applicability.