Computers & Chemical Engineering最新文献

The wake effect reduces the total power production of wind farms. This paper presents a method for wind farm power optimization through wake effect reduction. The proposed method optimizes the yaw angle offsets and de-rating settings of all turbines to maximize total power generation. The optimization approach is gradient-based, with gradients at each iteration obtained through system identification using field test data, eliminating the need for physical models. In system identification, test signal design, model estimation and model validation problems are solved in a systematic manner; in the gradient-based optimization, in order to achieve fast convergence, methods for initial value and initial step-size determination, variable step-size iteration and iteration termination are developed. The method is verified using the FLORIS wind farm model developed by National Renewable Energy Laboratory (NREL), USA. The studied wind farm consists of 80 wind turbines configured similarly to the Horns Rev I offshore wind farm in Denmark. The result of the developed optimization method is highly consistent with those obtained using FLORIS's built-in optimization tool.

Carbon capture, utilization, and storage supply chains (CCUS) play a pivotal role in achieving sustainability targets but necessitate meticulous risk identification and mitigation measures. Traditional safety assessments often occur post-design, constraining proactive risk management efforts. Hence, there is a pressing need to optimize safety performance during the design stages. To address this challenge, a framework for evaluating and optimizing CCUS supply chain safety performance using inherent safety index system (ISI) is introduced. Recognizing the trade-offs between total cost, environmental impact reduction, and risk mitigation, our approach considers multi-objective optimization to concurrently address these sustainability objectives and generate a Pareto set of solutions. Utilizing the augmented $\epsilon$-constraint method, we applied this framework to optimize CCUS networks and develop sustainable designs across three key objectives. The method was applied to a CCUS system that includes various CO₂ utilization pathways to minimize the total annual cost, CO₂ emissions, and safety risks. The resulting Pareto surface illustrates unique network configurations, each representing a distinct trade-off scenario. Through a case study, we optimized a CCUS network to achieve economic, environmental, and safety objectives. The most economically viable design, with a total annual cost of $97 million and a 40 % net carbon reduction, prioritizes CO₂ utilization for value-added products, while limiting CO₂ sequestration. Conversely, safety-focused designs shift utilization towards safer routes, including CO₂ sequestration and algae production. The proposed framework offers a systematic approach to developing sustainable CCUS supply chain designs, balancing economic viability, environmental sustainability, and safety.

The feeding process of the ethylene cracking furnace necessitates the synchronized adjustment of multiple controlled factors. The process mainly relies on operators to do it manually, which is burdensome and may lead to significant variations in coil out temperature (COT) due to the differing expertise of operators. This paper proposes a method for learning the feeding strategy of the ethylene cracking furnace using offline reinforcement learning. The agent learns and optimizes the operating strategy directly from datasets, eliminating the need for sophisticated process simulator modeling. In addition, the advantage function is incorporated into the Twin Delayed Deep Deterministic Behavioral Cloning (TD3BC) algorithm, which enables the agent to acquire more effective operational experience. The proposed method is initially evaluated using benchmark datasets. Further, the proposed method is validated through comparative experiments on a feeding process validation model, demonstrating superior rewards and outperforming manual operating experience as well as other offline reinforcement learning methods.

The Simulated Moving Bed (SMB) is a continuous chromatographic separation process that operates on the principle of counter-current movement between the solid and liquid phases. Due to periodic switching of feed and product ports across numerous connected columns, adjusting SMB operating variables such as feed and product flow rates and switching time to achieve desired separations is challenging. While equilibrium theory can help narrow the search space, obtaining essential information such as accurate adsorption isotherms is crucial. This requirement, combined with often highly stringent production specifications, makes it challenging to identify even a feasible operating condition, let alone an optimal one. Trial-and-error-based approaches are often impractical as reaching cyclic steady state can be time-consuming, and any waste produced during this period can lead to significant economic losses. While rigorous dynamic models are available, they are computationally intensive and often do not accurately mirror actual process behavior. To address these challenges, the use of Bayesian Optimization (BO) is proposed to sequentially approach optimal SMB operation. Furthermore, it is suggested to employ the simpler True Moving Bed (TMB) model as a prior for the BO, which significantly accelerates convergence. This approach is demonstrated on an SMB process for cresol separation. Initially, the effectiveness of the BO using the TMB model is examined to gain insights into its behavior. Subsequently, we apply BO to the rigorous SMB model, informed by prior knowledge from the TMB model. Our results show that the developed BO framework rapidly converges to the optimal operating parameters that satisfy the purity constraints. We examine the efficiency improvements over various search algorithms and highlight the advantages of using the TMB model as a prior.

The motility-induced phase separation (MIPS) phenomenon in active matter has been of great interest for the past decade or so. A central conceptual puzzle is that this behavior, which is generally characterized as a nonequilibrium phenomenon, can yet be explained using simple equilibrium models of thermodynamics. Here, we address this problem using a new theory, statistical teleodynamics, which is a conceptual synthesis of game theory and statistical mechanics. In this framework, active agents compete in their pursuit of maximum effective utility, and this self-organizing dynamics results in an arbitrage equilibrium in which all agents have the same effective utility. We show that MIPS is an example of arbitrage equilibrium and that it is mathematically equivalent to other phase-separation phenomena in entirely different domains, such as sociology and economics. As examples, we present the behavior of Janus particles in a potential trap and the effect of chemotaxis on MIPS.

Hydrogen refueling stations (HRS) can cause a significant fraction of the hydrogen refueling cost. The main cost contributor is the currently used mechanical compressor. Electrochemical hydrogen compression (EHC) has recently been proposed as an alternative. However, its optimal integration in an HRS has yet to be investigated. In this study, we compare the performance of a gaseous HRS equipped with different compressors. First, we develop dynamic models of three process configurations, which differ in the compressor technology: mechanical vs. electrochemical vs. combined. Then, the design and operation of the compressors are optimized by solving multi-stage dynamic optimization problems. The optimization results show that the three configurations lead to comparable hydrogen dispensing costs, because the electrochemical configuration exhibits lower capital cost but higher energy demand and thus operating cost than the mechanical configuration. The combined configuration is a trade-off with intermediate capital and operating cost.

A particle size-based Smoluchowski coagulation and fragmentation equation was solved in the free and open source process modeling package CADET. The WFV and MCNP schemes were selected to discretize the internal particle size coordinate. Weights in these schemes were modified to preserve and conserve the zeroth and third moments for size-based equations. Modified propositions and proofs for the scheme are provided. Analytical Jacobians were derived and implemented to reduce the solver’s runtime. A two-dimensional Smoluchowski coagulation and fragmentation equation with axial position as external coordinate was formulated and discretized to support simulations of continuous particulate processes in dispersive plug flow reactors. Five 1D and four 2D test cases were used to validate the implementation and benchmark the solver’s performance. The runtime, L1 error norm, L1 error rate, particle size distribution moments up to sixth order and several scalar metrics were analyzed in detail.

This study introduces a machine learning framework tailored to large-scale industrial processes characterized by a plethora of numerical and categorical inputs. The framework aims to (i) discern critical parameters that influence the output and (ii) generate accurate out-of-sample qualitative and quantitative predictions of production outcomes. Specifically, we address the pivotal question of the significance of each input in shaping the process outcome, using an industrial Chemical Vapor Deposition (CVD) process as an example. The initial objective involves merging subject matter expertise and clustering techniques exclusively on the process output, here, coating thickness measurements at various positions in the reactor. This approach identifies groups of production runs that share similar qualitative characteristics, such as film mean thickness and standard deviation. In particular, the differences of the outcomes represented by the different clusters can be attributed to differences in specific inputs, indicating that these inputs are potentially critical to the production outcome. Shapley value analysis corroborates the formed hypotheses. Leveraging this insight, we subsequently implement supervised classification and regression methods using the identified critical process inputs. The proposed methodology proves to be valuable in scenarios with a multitude of inputs and insufficient data for the direct application of deep learning techniques, providing meaningful insights into the underlying processes.

Data-driven inverse optimization (IO) aims to estimate unknown parameters in an optimization model from observed decisions. The IO problem is commonly formulated as a large-scale bilevel program that is notoriously difficult to solve. We propose a derivative-free optimization approach based on Bayesian optimization, BO4IO, to solve general IO problems. The main advantages of BO4IO are two-fold: (i) it circumvents the need of complex reformulations or specialized algorithms and can hence enable computational tractability even when the underlying optimization problem is nonconvex or involves discrete variables, and (ii) it allows approximations of the profile likelihood, which provide uncertainty quantification on the IO parameter estimates. Our extensive computational results demonstrate the efficacy and robustness of BO4IO to estimate unknown parameters from small and noisy datasets. In addition, the proposed profile likelihood analysis effectively provides good approximations of the confidence intervals on the parameter estimates and assesses the identifiability of the unknown parameters.

The process systems domain is undergoing the fourth industrial revolution, which is helping industries digitize and optimize their production techniques. Concurrently, the field of data-based modeling has been expanding, leading to the proposal of many fault detection models. However, the rapid expansion has created gaps in the field. For instance, Smart Manufacturing (SM) methodologies have yet to be incorporated into undergraduate chemical engineering education. Additionally, only a few developed fault detection models have been deployed for real-time usage and practical applications. This study takes a crucial step toward bridging the two mentioned gaps by enabling undergraduate students to learn SM techniques and developing a safe and controlled academic environment for deploying fault detection models. The demonstration is implemented on a shell and tube heat exchanger, taught in a senior year laboratory course, using the Smart Manufacturing Innovation Platform (SMIP). The implementation provides an easily customizable pipeline for SM applications involving human-in-the-loop decision-making on a real-life hardware system. Actual data from heat exchanger equipment is used to train and compare the performances of several state-of-the-art fault detection models, including fully connected, convolutional, and recurrent neural networks. Current work also presents tutorials on deploying models for practical real-time applications using the SMIP. The overall architecture is a plug-and-play package that will motivate students to learn about SM and catalyze their interest in developing and deploying fault detection models using real-world data.