Volume 14, Issue 4 -

Application of data-driven soft sensors in manufacturing fields, for instance, chemical, pharmaceutical, and bioprocess have rapidly grown. The issue of missing measurements is common in chemical processing industries that involve data-driven soft sensors. Locally weighted Kernel partial least squares (LW-KPLS) algorithm has recently been proposed to develop adaptive soft sensors for nonlinear processes. This algorithm generally works well for complete datasets; however, it is unable to cope well with any datasets comprising missing measurements. Despite the above issue, limited studies can be found in assessing the effects of incomplete data and their treatment method on the predictive performances of LW-KPLS. To address these research gaps, therefore, a trimmed scores regression (TSR) based missing data imputation method was integrated to LW-KPLS to formulate trimmed scores regression assisted locally weighted Kernel partial least squares (TSR-LW-KPLS) model. In this study, this proposed TSR-LW-KPLS was employed to deal with missing measurements in nonlinear chemical process data. The performances of TSR-LW-KPLS were evaluated using three case studies having different percentages of missing measurements varying from 5 % to 40 %. The obtained results were then compared to the results from singular value decomposition assisted locally weighted Kernel partial least squares (SVD-LW-KPLS) model. SVD-LW-KPLS was also proposed by incorporating a singular value decomposition (SVD) based missing data treatment method into LW-KPLS. From the comparative studies, it is evident that the predictive accuracies of TSR-LW-KPLS are superior compared to the ones from SVD-LW-KPLS.

This work aimed to evaluate the interaction of human IgG in non-conventional adsorbents based on chitosan and alginate in the absence and presence of Reactive Green, Reactive Blue and Cibacron Blue immobilized as ligands. The adsorption was evaluated at 277, 288, 298 and 310 K using sodium phosphate buffer, pH 7.6, at 25 mmol L −1 . The highest adsorption capacity was observed in the experiments performed with no immobilized dye, although all showed adsorption capacity higher than 120 mg g −1 . Data modeling was done using Langmuir, Langmuir-Freundlich and Temkin classical nonlinear models, and artificial neural networks (ANN) for comparison. According to the parameters obtained, a possible adsorption in multilayers was observed due to protein-adsorbent and protein-protein interactions, concluding that IgG adsorption process is favorable and spontaneous. Using an ANN structure with 3 hidden neurons (single hidden layer), the MSE (RMSE) for training, test and validation were 13.698 (3.701), 11.206 (3.347) and 7.632 (2.763), respectively, achieving correlation coefficients of 0.999 in all steps. ANN modeling proved to be effective in predicting the adsorption isotherms in addition to overcoming the difficulties caused by experimental errors and/or arising from adsorption phenomenology.

Fuel cells are finding wide range of applications from small capacity cells used for portable electronic devices to large capacity stacks for automobile applications. Among the different types of fuel cells, passive Direct methanol fuel cell (DMFC) has many advantages because of its simplicity. This paper presents the effect of Anode Diffusion Layer (ADL) thickness and its porosity on the performance of a passive DMFC. A one-dimensional, non-isothermal model is developed in MATLAB environment for exploring the complex physicochemical phenomena taking place inside the passive DMFC, by considering both heat and mass transfer effects. Modeling studies have been carried out by varying the ADL thickness from 0.1 mm to 0.6 mm, and the ADL porosity from 0.3 to 0.8. The concentration distribution of methanol, water and oxygen inside the cell have been predicted and, the crossover of methanol and water across the membrane have also been estimated. It is observed that increase in thickness of the ADL decreases the methanol corss over. Further, the effect of ADL thickness and porosity on the anodic overpotential and cathodic overpotential have been estimated. It was observed that increase in ADL thickness as well as its porosity increase the overpotentials.

This paper presents a chance-constrained multiobjective optimization framework for the optimal design of gas transmission lines (GTL) in which the total annual cost (TAC) and operability index as the two conflicting objectives are minimized and maximized, respectively. The delivery flow rates are assumed to be random Gaussian variables since the customer demand can be uncertain in the design phase. Accordingly, a robustness measure is introduced in order to quantify the risk of the final design against delivery uncertainties. The proposed model is capable of determining the optimum design variables including the pipeline diameter and thickness along with the location of the compressor stations and their capacities. The effects of design parameters including pipe thickness, pipe diameter and customer demand uncertainty level on the design of a GTL are investigated from economic as well as the robustness point of view. It is shown that the design with the highest nominal pipe size (NPS) and least thickness possible is the optimal design when TAC is considered as the objective function while the design with both the highest NPS and thickness possible shows the highest robustness and flexibility against delivery demand fluctuations. The final decision about the optimal NPS and thickness should be made on the basis of the expected uncertainty of the customer future demand. The results also can quantitatively suggest the required over-design factor that must be considered in the design phase of the GTL for various compressor stations.

Steam reforming of biomass bio-oil is a technique of producing bio-hydrogen which is an important biofuel. Acetic acid is a major constituent of biomass bio-oil especially its aqueous phase. In this study, the thermodynamic analysis of the steam reforming of acetic acid was considered in conjunction with the utilising of a novel statistical approach. Response surface methodology was used to elucidate possible interactions of the process factors and be used to develop regression models for the prediction of percentage molar yield of each species given a known set of inputs. The correlations were validated for the prediction of % molar composition of the product chemical species in the product stream. These correlations are of great relevance as it affords quick predictions given a known set of factors.

Computer Aided Design of Chemical Process is now a well established eld in the design of new process as well as in the optimization, revamp and retrot of existing ones. The use of powerful process simulators available today allows the process engineer to simulate even an entire process, but the majority of process simulators have only classical unit operations. So, if the process has a non-classical unit operation it needs to be simulated using a suitable computer language and further to be linked to the simulator. In this paper we addressed the problem of including a new unit operation in a process simulator and how to use the virtual plant to optimize and to evaluate the environmental impact of a chemical process. We used the free chemical process simulator COCO to simulate two styrene process production plant. The firrst one uses as a reactor a conventional PFR that is available in the simulator. The second plant uses a membrane reactor that was simulated using the software Scilab that was embedded in COCO simulator by using the CAPE-OPEN protocol. Then, we used both virtual plants to develop meta-models of the processes by using experimental design and surface responses. These empirical models were used after to optimize the plants and the results shown that it is possible to increase the styrene productivity up to 27.32 kmol/h using a PFR reactor and up to 30.56 kmol/h using a membrane reactor. Finally, we calculated the Potential of Environmental Impact (PEI) for each process using the WAR algorithm and we shown that both processes have PEI very similar. Therefore, the route that uses membrane reactor has an advantage over the route that uses PFR reactor since it allows to obtain higher styrene productivities.

The Linear Kalman Filter (LKF) and Extended Kalman Filter (EKF) as correction-prediction observers are two common model-based filters, which are widely used to estimate the unmeasurable states and parameters in the chemical processes. The main disadvantages of LKF and EKF are low accuracy and reliability of estimation and high computational time, respectively. The main object of this work is the modification of conventional Kalman filter to cover the disadvantages of LKF and EKF. The proposed method is planned based on the systematic updating the Jacobian matrix of estimator applied on the nonlinear systems. In this regard, a linearity index is defined based on the characteristics of outputs to determine switching time between LKF and EKF to avoid unnecessary updating the Jacobian matrix. The performance of the proposed method is compared with the LKF and EKF considering four chemical benchmarks as the nonlinear state space models. Altought EKF and proposed filter present similar precision to estimate unmeasurable states, the proposed method has a lower computational time.

A novel three-dimensional computational fluid dynamics mass transfer (CMT) model in Eulerian–Eulerian frame work is deploys for investigating the concentration profiles, and trays efficiencies in multicomponent distillation columns. The proposed model is based on Maxwell Stefan equations, and CFD was employed as a powerful tool to model the hydrodynamics and mass transfer. The two phases are modelled as two interpenetrating phases with interphase momentum, heat and mass transfer. The Closure model is developed for mass interphase transfer rate in ternary mixtures. The predictability of the mass transfer behaviours of multicomponent can result in a more efficient and predictable design of distillation trays. Two non-ideal ternary mixtures were studied. The tray geometry and operating conditions are based on the experimental works of Kalbassi and the composition profiles, tray efficiencies, and point efficiencies of mixtures were presented. The obtained results were confirmed by the experimental data. The results indicate that the values of individual component tray efficiencies and point efficiencies for these multicomponent systems were considerably different which confirm the interactive nature of the mass transfer in multicomponent mixtures. These mixtures also illustrated different point efficiencies across the tray because of the composition dependency of these mixtures. The average relative error for the prediction of efficiencies is about 8 %, which indicates the accuracy of the model.