Assessment of solvent recovery efficiency in pilot-scale pharmaceutical manufacturing units

Thesis Abstract

Solvent use in pharmaceutical manufacturing incurs significant environmental and cost burdens, with inefficiencies in recovery processes contributing to higher emissions, solvent losses, and elevated production costs. This study addresses the persistent gap between nominal recovery targets and actual performance in pilot-scale pharmaceutical units, where variability in solvent streams, process integration, and equipment condition influence recovery efficiency. The aim is to quantify recovery performance, identify primary loss pathways, and propose empirically grounded improvements to enhance material efficiency, lower costs, and reduce environmental impact. Specific objectives are (1) to quantify solvent recovery rates for commonly used solvents (acetone, methanol, and dichloromethane) across three pilot-scale single-stage and two-stage recovery configurations; (2) to identify and rank drivers of recovery inefficiency, including process parameters, equipment fouling, and operational deviations; (3) to develop a predictive model linking process variables to recovery efficiency; (4) to evaluate the economic and environmental implications of proposed optimization strategies; and (5) to formulate actionable guidelines for scale-up to full production. A mixed-methods approach integrates quantitative process data with qualitative insights from operator interviews. The population comprises pilot-scale pharmaceutical manufacturing units located in three multi-product facilities. A purposive sample of 12 recovery trains across these facilities is selected, with 6 units employing single-stage distillation and 6 employing two-stage solvent recovery configurations. Data collection combines (i) process data logs collected over 24 weeks, including feed solvent composition, temperature, pressure, residence times, and condenser efficiencies, (ii) analytical measurements of recovered solvent by GC-FID and GC-MS to determine purity and recovery yield, and (iii) structured interviews with 18 process engineers/operators to capture tacit knowledge on operational deviations and maintenance practices. Instruments are validated through calibration standards, inter-laboratory comparisons, and pilot-test runs. Analytical methods include descriptive statistics to profile baseline recovery performance, ANOVA to compare recovery efficiency across configurations and facilities, multivariate regression to identify significant predictors of recovery yield, and a hierarchical linear model to account for nested facility effects. Process-level mass balance checks ensure data integrity, while sensitivity analyses test robustness of the predictive model under alternative scenarios. A cost-benefit analysis evaluates the economic implications by calculating savings from solvent recovery improvements using net present value (NPV) and return on investment (ROI) over a 5-year horizon. Environmental performance is assessed via solvent emission reductions and life-cycle impact indicators. Expected findings include statistically significant differences in recovery efficiency between single-stage and two-stage configurations, with two-stage systems achieving up to 15–22% higher overall recovery yields under typical feed compositions. Key drivers of inefficiency are anticipated to be condenser fouling, suboptimal reflux ratios, and inconsistencies in feed solvent ratios due to upstream process variability. The predictive model is expected to explain a substantial portion of variance (R-squared > 0.70) in recovery yield when incorporating solvent characteristics, feed flow rate, and equipment condition indicators. The study also anticipates a favorable economic signal, with estimated payback periods within 2–4 years for recommended optimizations, driven by solvent savings and reduced waste disposal costs, and a measurable reduction in volatile organic compound (VOC) emissions. The contribution to knowledge lies in providing a rigorous empirical assessment of solvent recovery performance at the pilot scale, quantifying inefficient pathways, and delivering a validated predictive framework to inform design and operation of recovery systems in pharmaceutical pilot plants. The study advances understanding of how process design choices and operational practices interact to determine recovery outcomes and offers evidence-based guidelines for improving solvent efficiency during scale-up. The main conclusion is that optimization of recovery configuration, systematic maintenance to control condenser performance, and active feed quality management can substantially elevate recovery efficiency, reduce costs, and lower environmental impact. Recommendations include adopting two-stage recovery with calibrated reflux control, implementing a preventive maintenance program focused on condenser cleanliness, and integrating the predictive model into plant digital twins to guide real-time operational decisions.