Search

Heterotrophic cultivation can be a preferred option for commercial microalgal biorefinery owing to its significantly higher productivity and lipid content leading to high value products. However, l...

In this study, the effects of ship motion on the CO2 absorption performance of the offshore columns (i.e., the separation columns installed on offshore floating units) are investigated. Specifically, a pilot scale column with 0.4 m diameter and 4 m packed height is used, and NaOH/air system is implemented to measure the CO2 absorption performance of the column under both vertical and offshore conditions. The experimental results obtained from the vertical conditions are used to validate the mathematical model which can compute the effective surface area of the separation columns. Then, such model is used to produce the reference values for the experimental results obtained from the various offshore conditions, where tilt angle, roll motion conditions, liquid load, gas factor, liquid surface tension and liquid viscosity are used as the design variables. The reduction factor is proposed, which can describe the change in the absorption performance of the offshore columns with respect to the important factors. From our study, the tilt angle was shown to be the most important factor, and the mass transfer efficiency was reduced by 8.4% in average (compared to the value obtained under the vertical condition) when the tilt angle was 6°. It was also shown that the roll motion can enhance the absorption performance of offshore columns under certain operating conditions, and the maximum improvement (6.7%) in the mass transfer efficiency was achieved when the liquid load, roll motion amplitude and period were 14 m3/m2 hr, 8° and 45 s, respectively.

In this paper, a parallel neural network architecture is proposed to improve the performance of neural-network-based nonlinear principal component analysis. There exist two typical approaches for such analysis: simultaneous extraction of principal components using a single autoassociative neural network (also known as autoencoder), and sequential extraction using multiple neural networks in series. The proposed architecture can be obtained by systematically pruning the network connections of a fully connected autoassociative neural network, resulting in decoupled neural networks. As a result, more independent (i.e., less correlated) principal components can be obtained than the simultaneous extraction approach. The proposed architecture can be also viewed as a rearrangement of multiple neural networks for the sequential extraction in a parallel setting, and thus, the network training becomes more efficient. Simulation case studies are performed to illustrate the advantages of the proposed architecture, and it was shown that it is particularly beneficial for deep neural networks.

Microalgal biomass is considered to be a promising renewable energy source to replace fossil fuel given its simple growth mechanism and high productivity per unit area. Nevertheless, microalgal bio...

In this article, the statistical process monitoring problem of the Tennessee Eastman process is considered using deep learning techniques. This work is motivated by three limitations of the existing works for such problem. First, although deep learning has been used for process monitoring extensively, in the majority of the existing works, the neural networks were trained in a supervised manner assuming that the normal/fault labels were available. However, this is not always the case in real applications. Thus, in this work, autoassociative neural networks are used, which are trained in an unsupervised fashion. Another limitation is that the typical dataset used for the monitoring of the Tennessee Eastman process is comprised of just a small number of data samples, which can be highly limiting for deep learning. The dataset used in this work is 500-times larger than the typically-used dataset and is large enough for deep learning. Lastly, an alternative neural network architecture, which is called parallel autoassociative neural networks, is proposed to decouple the training of different principal components. The proposed architecture is expected to address the co-adaptation issue of the fully-connected autoassociative neural networks. An extensive case study is designed and performed to evaluate the effects of the following neural network settings: neural network size, type of regularization, training objective function, and training epoch. The results are compared with those obtained using linear principal component analysis, and the advantages and limitations of the parallel autoassociative neural networks are illustrated.

The formulation of a kinetic model for a complex reaction network typically yields reaction rates which vary over orders of magnitude. This results in time scale separation that makes the model inherently stiff. In this work, a graph-theoretic framework is developed for time scale decomposition of complex reaction networks to separate the slow and fast time scales, and to identify pseudo-species that evolve only in the slow time scale. The reaction network is represented using a directed bi-partite graph and cycles that correspond to closed walks are used to identify interactions between species participating in fast/equilibrated reactions. Subsequently, an algorithm which connects the cycles to form the pseudo-species is utilized to eliminate the fast rate terms. These pseudo-species are used to formulate reduced, non-stiff kinetic models of the reaction system. Two reaction systems are considered to show the efficacy of this framework in the context of thermochemical and biochemical processing.

A key feature of amine-based post-combustion CO2 capture process is a wide operating range induced by periodic load changes in power plants, which necessitates flexible operation. One possible approach to enhance the operational flexibility is to design a reliable controller that can effectively regulate the process over the operating range. To this end, in this study, a robust model predictive controller is designed by analyzing the dynamic characteristics of a post-combustion CO2 capture process. Specifically, gap metric analysis is performed to analyze the sensitivity of the process. From this analysis, optimal operating conditions are identified by evaluating similarity among the dynamics around different operating conditions. Then, a single linear model predictive controller is designed on the basis of the linear approximation of the original nonlinear model at the chosen conditions. Finally, the effectiveness of the controller is illustrated through a case study on an example CO2 capture process.

This research suggests a three-stage model framework for the design of a microalgae-based biofuel supply chain to meet the goal of economic commercialization. First, the design stage decides the spatial layouts and dimensions of each scale of biorefineries and economic analyses are done to estimate the capital costs and operating costs for different design options. Using the spatial dimensions determined in the first stage, the second stage selects the candidate locations for the biorefineries using an geographic information system (GIS) based site evaluation. This stage screens the available land area for the biorefineries, and also reduces the computational burden of the latter stage. In the mathematical optimization stage, a mixed-integer linear programming optimization model is formulated to make multi-period strategic and tactical decisions of the supply chain under the total cost minimization objective. Since the formulated problem is computationally intensive, a two-stage decomposition solution strategy is proposed to solve the problem in a reasonable time. The model framework is demonstrated through a case study cast in Texas, U.S., with a time horizon of ten years, using the suggested decomposition method. As a result, the minimum fuel selling price (MFSP) of microalgae-based biodiesel is calculated as $10.92/galbiodiesel, which is about three times higher than the current biodiesel price $3.51/galbiodiesel. To reduce the cost, the strategies of biomass storage and maximum delivery are investigated to deal with the productivity fluctuation. In the scenario analysis, the underutilization of production capacities is alleviated resulting in reduced MFSP of $7.90/galbiodiesel, $7.92/galbiodiesel, and $7.43/galbiodiesel respectively. Clearly, for economic feasibility, the production cost should be further reduced by developing more cost-efficient technologies and integrating high-value coproducts into the biorefinery portfolio.

In this study, nonlinear control of high duty counter-current heat exchanger is considered. The dynamics of this system can be captured by first order hyperbolic partial differential equations (PDEs), which are stiff due to the high rate of heat transfer. The potential of multi-time scale dynamics is discussed, and model reduction is performed using singular perturbations to obtain non-stiff PDE models which are valid in each time scale. Three controllers are designed and compared in the simulation case study: input/output linearizing controllers on the basis of original model and reduced order model, and a proportional-integral controller. The nonlinear controller designed using the reduced order model showed a superior performance compared to the others, especially in the presence of large modeling errors and disturbances.