Ammonia has received in recent years significant attention as potential carbon free fuel. However, its combustion properties limit its direct application for both providing heat and in power generation through gas turbines. Ammonia cracking is one potential solution to circumvent the problem by producing hydrogen. When using the ammonia in gas turbines, it is possible to heat integrate the endothermic decomposition reaction with the exhaust gas from the gas turbine. Thermodynamic and kinetic limitations have however a major impact on the achievable ammonia conversion. Based on the consideration of these limitations, this paper presents a detailed investigation of key design parameters affecting the overall process efficiency utilizing both an equilibrium reactor model and a reactor model based on detailed kinetics and heat transfer. Ammonia decomposition should occur at sufficiently high pressure to avoid a) the com-pression energy demand for achieving the pressure of the combustion ch... [more]

Anaerobic digestion (AD) across Europe and the UK faces increasing economic and operational pressure from volatile feedstock supply under climate extremes. Existing stochastic programming (SP) approaches for feedstock planning often rely on limited historical observations and/or simplify yield uncertainty in ways that miss the joint, non-linear response of crops to weather variability, thereby understating downside supply risk. We develop an integrated decision-support framework that links climate uncertainty to AD procurement planning by coupling mechanistic crop simulation, generative surrogate modelling, and stochastic optimization. First, APSIM is used offline to generate a mechanistic yield knowledge base across weather trajectories and discrete planting-density choices. Then, a conditional GAN (CGAN) is trained to produce non-parametric joint yield samples for multi crops conditioned on scenario features and management, enabling fast Monte Carlo evaluation. At last, these samples... [more]

Black-box optimization (BBO) deals with problems involving functions that are either unknown, imprecise, or costly to evaluate. Current BBO methods encounter multiple challenges, such as high computational demands from excessive function evaluations, difficulties in handling complex constraints, lack of theoretical convergence guarantees, and unstable performance due to significant variations in solution quality. This work presents a machine learning-powered feasible path (MLFP) framework for general BBO problems involving complex constraints. An adaptive sampling strategy is first proposed to explore optimal regions and pre-filter potentially infeasible points, thereby reducing the number of evaluations. Machine learning algorithms are utilized to build surrogates for black-box functions. The feasible path algorithm is integrated to accelerate theoretical convergence by updating only independent variables instead of all variables. Computational experiments demonstrate that MLFP can ra... [more]

Reliable multimodal monitoring in crystallization processes remains challenging due to heterogeneous PAT signal quality, sensor drift, asynchronous sampling and nonstationary noise. This work presents a machine-learning-assisted fusion framework that integrates multimodal PAT alignment, estimation and physics-guided regularisation to generate coherent concentration and particle-size trajectories. A mechanistically informed simulation platform is developed to produce synthetic Raman, FTIR, FBRM and image-based crystal size data with realistically simulated drift, heteroscedastic noise, dropouts and distortion patterns. Sensor reliability is inferred through a Random Forest model trained on variance-normalised discrepancies and quality metrics, which allows the dynamic adjustment of channel contributions. Across modalities, the Random Forest achieves MAE values of 0.03-0.20 for probability-type indicators and shows stable explanatory power for variance-inflation factors on particle-size... [more]

Environmental sustainability is increasingly recognized as a critical consideration in pharmaceutical development, yet it is rarely incorporated at the scale of molecular-level design. This study introduces a strategy to predict cradle-to-gate indicators that can be flexibly incorporated into multiple early-stage molecular prioritization scenarios. A dataset of 150 pharmaceutical-relevant molecules was compiled, with each molecule described by structural descriptors, thermophysical properties, and ReCiPe endpoint indicators representing human health, ecosystem quality, and resource scarcity. A dual-branch multi-task model combining graph-based and descriptor-based representations was trained to predict these three endpoint indicators. Model performance was evaluated through validation metrics, local sensitivity analysis, and SHAP-based interpretability. A case study with solubility-based feasibility constraints was then used to illustrate how different sustainability weighting schemes... [more]

Large-scale mixed-integer programs (MIPs) typically cannot be solved by standard solvers with reasonable computational cost. MIP-heuristics decompose large-scale monolithic mixed-integer programs into polylithic programs such that they can be solved with reasonable computational cost at the price of loosing their optimality certificate. The decomposition is steered by hyperparameters that impact the solution quality and the computational cost diametrically. The proper selection of the hyperparameter values is a black-box optimization problem which is mostly solved by grid search or random search. In previous publications the authors proposed a novel hyperparameter optimization method based on Bayesian optimization and studied a use case from the PSE domain. Computational studies showed that the BO-based algorithm is superior for objective functions with few optimal solutions.This contribution generalizes the description of the MIP-Heuristic Optimization Problem (MIP-HOP) and the comput... [more]

This work aims at verifying the CO2 absorption capture model using monoethanolamine (MEA) solvent developed by the U.S. DOE's Carbon Capture Simulation Initiative (CCSI) and performing uncertainty propagation of mass transfer, liquid hold-up and reaction kinetics properties in the complete model, which includes absorber and stripper columns. The verification of the Aspen Plus CCSI model, based on pilot plant data from the National Carbon Capture Center (NCCC) for a CO2 flue gas concentration between 7 and 11% (mol) allowed uncertainty quantification (UQ) analysis for four different selected operational points using Monte Carlo Simulation (MCS), where low liquid mass transfer parameters exhibited an impact on calculation convergence. Gaussian Processes (GP) surrogate model was implemented, followed by a sensitivity analysis in order to correlate the most sensitive parameters with studied outputs.

Biological systems are governed by coupled interactions between intracellular metabolism and bioreactor operation that span multiple time scales. Constraint-based metabolic models are widely used to describe intracellular metabolism, but repeatedly solving the optimization problem at each time step in dynamic models introduces numerical challenges related to infeasibility and computational efficiency. This work presents a multi-scale modeling framework that integrates genome-scale, constraint-based metabolic models with dynamic bioreactor simulations. Intracellular metabolism is described using positive flux variables in a parsimonious flux balance analysis, and the resulting embedded optimization problem is replaced by a neural network surrogate. The surrogate provides a smooth approximation of the embedded optimization mapping and eliminates repeated linear program solves during simulation. The approach is demonstrated for fed-batch fermentation of Escherichia coli, in which the surr... [more]

This work presents a data-driven surrogate modeling framework for predicting distillation behavior assuming an infinite number of stages and distillation limits informed by residue-curve topology and pinch-point feasibility analysis. The framework provides a direct mapping from feed composition and distillate-to-feed ratio (D/F) to distillate and bottom product compositions, making it suitable for flowsheet synthesis and optimization applications. The approach combines three components: a classifier that identifies feasible singular-point splits, a boundary regression model that predicts D/F limits separating pure- and mixed-product operating regimes, and a neural network that interpolates product compositions in the intermediate regime. The method is demonstrated for the ternary system ethanol, benzene, and water at 1 atm using data generated from rigorous vapor-liquid-liquid equilibrium analysis. Results show that the framework provides reliable predictions for pure splits while reta... [more]

Stochastic gene regulatory networks exhibit complex dynamics that require efficient methods for parameter inference and uncertainty quantification. In this work, we propose a surrogate modelling framework that combines a partial integro-differential equation (PIDE) formulation with polynomial chaos expansions (PCE) to efficiently approximate the stochastic dynamics of gene expression models under parametric uncertainty. The approach represents the time evolution of low-order statistical moments as polynomial functions of uncertain kinetic parameters, enabling fast evaluations and tractable inference. The method is demonstrated on a self-regulating gene network, achieving accurate parameter estimation and a reduction of approximately two orders of magnitude in computational cost compared to direct PIDE-based optimisation.

Industrial decarbonization has heightened interest in electrifying major chemical processes, but existing planning methods typically assume fixed electricity prices and overlook how industrial power use affects the grid. This work introduces a grid-aware optimization framework that captures two-way interactions between industrial electricity usage and the power flows within the grid. We use the DC Optimal Power Flow (DC-OPF) model to generate Locational Marginal Prices across refinery demand levels and embed a surrogate reflecting the relationship between the power demand and the prices into an operational optimization problem for a partially electrified refinery. The surrogate model is embedded within the optimization problem using disjunctive reformulations and off-the-shelf packages such as OMLT (Optimization and Machine Learning Toolkit). In a case study considering an oil refinery with installed electric boilers, electrolyzers, H2 storage, and post-combustion carbon capture infras... [more]

As industries embrace the digitalization of Industry 4.0, the abundance of process data creates new opportunities to optimize industrial control systems. Traditional Proportional-Integral-Derivative (PID) controllers often require manual tuning to address changing conditions. This paper introduces an automated, adaptive PID tuning method using historical data and machine learning for a continuously evolving, data-driven approach. The method centers on training a surrogate model using historical process data to replicate real system behavior under various conditions. This enables safe exploration of control strategies without disrupting live operations. An RL (Reinforcement Learning) agent interacts with the surrogate model to learn optimal control policies, dynamically responding to the plant's state, defined by variables like operational conditions and measured disturbances. The agent adjusts PID parameters in real-time, optimizing metrics such as stability, response time, and energy... [more]

As the demand for resources continues to grow, pipelines have become critical for transporting water, fossil fuels, and chemicals. Monitoring pipeline systems is essential, as leaks can lead to severe environmental damage and safety hazards. This study aims to develop a pipeline leakage detection system based on digital twin technology and Physics-Informed Neural Networks (PINNs). By embedding physical principles, such as the continuity and momentum equations derived from the Navier-Stokes equation, into the neural network's loss function, the model can predict pressure and flow dynamics with high accuracy while adhering to physical constraints. PINNs are particularly advantageous as they require minimal data, maintain physical consistency, and provide reliable interpretations, making them well-suited for addressing pipeline safety challenges. The model is designed to simulate fluid dynamics under normal operating conditions, with deviations in prediction errors signaling potential lea... [more]

Pressure-swing distillation (PSD) is a frequently applied method to separate pressure-sensitive azeotropic mixtures; however, its energy demand is very high. In continuous mode, PSD is performed in a system consisting of a high- and a low-pressure column. If the composition of the feed is between the azeotropic compositions at the two pressures, it can be introduced into any of the columns, leading to two possible column sequences. Depending on the feed composition, one of the sequences is optimal whether in terms of energy demand or total annual cost (TAC). In the present work, surrogate model-based optimization is applied to determine the optimal TAC value as a function of the feed composition between the azeotropic ones. As a first step, the column sequence with feeding into the high-pressure column is studied here. The mixture to be separated consists of water and ethylenediamine, which form a maximum-boiling azeotrope. The columns are modeled separately and a large number of simul... [more]

A Digital Twin (DT) is a purposeful digital representation of a physical entity that employs data, algorithms, and software to enhance operations, making it possible to e.g., forecast failures, or evaluate new designs through the simulation of real-world scenarios. DTs are enablers for real-time monitoring, simulation, and optimization. However, traditional simulation DTs often rely on complex, non-linear mechanistic models with high computational demands, complex structures, and a large number of specific parameters and thus pose quite a challenge to maintainability. Surrogate models, on the other hand, are simplified approximations of more complex, higher-order models. These approximations are typically built using data-driven approaches, such as Random Forest Regression, facilitating faster simulations, simpler adaptation, and quicker deployment. This study analyzes the complexity of mechanistic and surrogate modeling approaches in the context of DTs to aid model selection. A model... [more]

Hydrogen-powered heavy-duty trucks have a high potential to significantly reduce CO2 emissions in the transportation sector. Therefore, efficient hydrogen storage onboard vehicles is a key enabler for sustainable transportation, as achieving high storage densities and extended driving ranges is essential for the competitiveness of hydrogen-powered trucks. Cryo-compressed hydrogen (CcH2), stored at cryogenic temperatures and high pressures, emerges as a promising solution. This study presents a comprehensive dynamic thermodynamic model that is capable of simulating the tank system across all operating conditions and, therefore, enables thermodynamic analysis of drive cycles. The core of the model is a differential-algebraic equation system that describes the thermodynamic state of the hydrogen in the tank. Additionally, surrogate models based on artificial neural networks are applied to efficiently describe quasi-steady-state heat exchangers integrated into the tank system. Several use... [more]

The food-energy-water nexus (FEWN) has been receiving increasing interest in the open literature as a framework to address the widening gap between natural resource availability and demand, towards more sustainable and cost-competitive solutions. The FEWN aims at holistically integrating the three interconnected subsystems of food, energy and water, into a single representative network. However, such an integration poses formidable challenges due to the complexity and multi-scale nature of the three subsystems and their respective interconnections. Additionally, the significant input data uncertainty and variability, such as energy prices and demands, or the evaluation of emerging technologies, contribute to the system’s inherent complexity. In this work, we revisit the FEWN problem in an attempt to elucidate and address in a systematic way issues related to its multi-scale complexity, uncertainty and variability. In particular, we provide a classification of the sources of data and te... [more]

Post-combustion carbon capture technologies have the potential to contribute significantly to achieving the environmental goals of reducing CO2 emissions in the short term. However, these technologies are energy and cost-intensive, and the variability of flue gas represents important challenges. The optimal design and optimization of such systems are critical to reaching the net zero and net negative goals, in this context, the use of computer-aided process design can be very effective in overcoming these issues. In this study, we explore the implementation of carbon capture technologies within an industrial complex, by considering the pooling of CO2 streams. We present an optimization formulation to design carbon capture plants with the goal of enhancing efficiency and minimizing the capture costs. Capital and operating costs are represented via surrogate models (SMs) that are trained using rigorous process models in Aspen Plus, each data point is obtained by solving an optimization p... [more]

Machine learning presents opportunities to improve the scale-specific accuracy of mechanistic models in a data-driven manner. Here we demonstrate the use of a machine learning technique called Sparse Identification of Nonlinear Dynamics (SINDy) to improve a simple mechanistic model of algal growth. Time-series measurements of the microalga Chlorella Vulgaris were generated under controlled photobioreactor conditions at the University of Technology Sydney. A simple mechanistic growth model based on intensity of light and temperature was integrated over time and compared to the time-series data. While the mechanistic model broadly captured the overall growth trend, discrepancies remained between the model and data due to the model's simplicity and non-ideal behavior of real-world measurement. SINDy was applied to model the residual error by identifying an error derivative correction term. Addition of this SINDy-informed error dynamics term shows improvement to model accuracy while maint... [more]

Explicitly incorporating the effects of chemical phenomena such as chemical pretreatment and mineral scaling during the design of treatment systems is critical; however, the complexity of these phenomena and limitations on data have historically hindered the incorporation of detailed water chemistry into the modeling and optimization of water desalination systems. Thus, while qualitative assessments and experimental studies on chemical pretreatment and scaling are abundant in the literature, very little has been done to assess the technoeconomic implications of different chemical pretreatment alternatives within the context of end-to-end water treatment train optimization. In this work, we begin to address this challenge by exploring the impact of pH control during pretreatment on the cost and operation of a high-recovery desalination train. We compare three pH control methods used in water treatment (H2SO4, HCl, and CO2) and assess their impact on the operation of a desalination plant... [more]

Design problems for process systems engineering applications often require multi-scale modeling integrating detailed process models. Consequently, black-box optimization and surrogate modeling have continued to play a fundamental role in mission-critical design applications. Inherent in surrogate modeling applications, particularly those constrained by “expensive” function evaluations, are the questions of how to properly balance “exploration” and “exploitation” and how to do so while harnessing parallel computing in techniques. We devise and investigate a one-step look-ahead GRASP heuristic for balancing exploration and exploitation in a parallel environment. Computational results reveal that our approach can yield equivalent or superior surrogate quality with near linear scaling in the number of parallel samples.

Process simulation problems often involve systems of nonlinear and nonconvex equations and may run into convergence issues due to the existence of recycle loops within such models. To that end, surrogate models have gained significant attention as an alternative to high-fidelity models as they significantly reduce the computational burden. However, these models do not always provide a guarantee on the prediction accuracy over the domain of interest. To address this issue, we strike a balance between computational complexity by developing a data-driven branch and prune-based framework that progressively leads to a guaranteed solution to the original system of equations. Specifically, we utilize interval arithmetic techniques to exploit Hessian information about the model of interest. Along with checking whether a solution can exist in the domain under consideration, we generate error-bounded convex hull surrogates using the sampled data and Hessian information. When integrated in a bran... [more]

Regarding the difficulty of extracting fault information in the faulty status of UAV (unmanned aerial vehicle) engines and the high time cost and large data requirement of the existing deep learning fault diagnosis algorithms with many training parameters, in this paper, a small-sample transfer learning fault diagnosis algorithm is proposed. First, vibration signals under the engine fault status are converted into a two-dimensional time-frequency map by multiple simultaneous squeezing S-transform (MSSST), which reduces the randomness of manually extracted features. Second, to address the problems of slow network model training and large data sample requirement, a transfer diagnosis strategy using the fine-tuned time-frequency map samples as the pre-training model of the ResNet-18 convolutional neural network is proposed. In addition, in order to improve the training effect of the network model, an agent model is introduced to optimize the hyperparameter network autonomously. Finally, e... [more]

The process design of hot air reflow soldering is one of the key factors affecting the quality of PCBA (Printed Circuit Board Assembly) component products. In order to improve the product quality during the design process, this paper proposes a robust optimization-based finite element simulation analysis method including significant influencing factor screening, robustness evaluation, robust optimization, and reliability verification for the reflow soldering process. The simulation model of the reflow soldering process temperature field based on experiments is constructed and validated. Sensitivity analysis is used to select important influencing factors, such as the last five set temperature zones (T5 to T9) in the reflow oven and the thermal properties of materials such as PCBs (printed circuit boards), BGAs (ball grid arrays), and solder paste, as well as noise factors like the heating environment during the soldering process. Several surrogate models are used to construct the respo... [more]

Precise and efficient calculations are necessary to accurately assess the effects of thermal protection system (TPS) uncertainties on aerospacecrafts. This paper presents a probabilistic design methodology for TPSs based on physics-informed neural networks (PINNs) with parametric uncertainty. A typical thermal coating system is used to investigate the impact of uncertainty on the thermal properties of insulation materials and to evaluate the resulting temperature distribution. A sensitivity analysis is conducted to identify the influence of the parameters on the thermal response. The results show that PINNs can produce quick and accurate predictions of the temperature of insulation materials. The accuracy of the PINN model is comparable to that of a response surface surrogate model. Still, the computational time required by the PINN model is only a fraction of the latter. Considering both computational efficiency and accuracy, the PINN model can be used as a high-precision surrogate mo... [more]