Optimization of Multibed Pressure Swing Adsorption Processes Dragan Nikolic ´, † Eustathios S. Kikkinides, † and Michael C. Georgiadis* ,‡ Department of Mechanical Engineering, UniVersity of Western Macedonia, SialVera & Bakola Str., 50100 Kozani, Greece, and Department of Engineering Informatics and Telecommunications, UniVersity of Western Macedonia, Karamanli & Lygeris Str., 50100 Kozani, Greece This work presents an optimization framework for complex pressure swing adsorption (PSA) processes including multibed configurations and multilayered adsorbents. The number of beds, PSA cycle configuration, and various operating and design parameters have been systematically optimized using recent advances on process optimization. The Unibed principle has been adopted relying on the simulation over times of only one bed while storage buffers have been used to model bed interactions. A novel state transition network (STN) representation is employed for the efficient simulation and optimization of the processes. Two large- scale multicomponent separation processes have been used to illustrate the applicability and potential of the proposed approach in terms of improvement of product purity and recovery. Results indicate that significant improvements can be achieved over base case designs. 1. Introduction Separation of gas mixtures by pressure or vacuum swing adsorption (PSA, VSA) has become a common industrial practice in the area of small- to medium-scale air separation, small- to large-scale gas drying, small- to large-scale hydrogen recovery from different petrochemical processes, and trace impurity removal from contaminated gases. The main reasons for an increased interest in such processes are the lower energy requirements and capital investment costs compared to the traditional separation processes. Hence, the development of optimization strategies for the design and operation of simple and complex PSA processes are of paramount importance in improving process performance. Smith and Westerberg 1 developed an approach to determine the optimal schedule of a PSA process. The optimization problem was formulated as a mixed-integer nonlinear program- ming problem which calculates the optimal schedule based on a given set of operating steps and constraints (e.g., steps that require bed interconnections, continuous operating of the compressor or continuous production). The basic model has been extended to support more flexible schedules, that is to determine the best subset and sequence of steps on the basis of a given set of operating steps. The main assumption of this work is that the sequence of operating steps was predetermined and the objective was mainly concerned with the derivation of an optimal schedule. In a subsequent publication, 2 the same authors employed an optimization approach on the simple integral model of a PSA process and applied it on the process of hydrogen purification from a hydrogen/methane waste stream. Nilchan and Pantelides 3 introduced a formal optimization framework of PSA processes. They introduced a novel math- ematical programming approach to the optimization of general periodic adsorption process which comprises the rigorous mathematical model of an adsorption bed, periodic boundary conditions, cycle timing constraints, and bed interaction con- straints. Two different techniques for determining the cyclic steady state (CSS) have been applied: dynamic simulation (DS) and complete discretization (CD) (both in space and time). Jiang et al. 4 proposed an algorithm to accelerate the CSS convergence by using a Newtonian-based method with accurate sensitivity calculation to achieve fast and robust convergence. They applied the method to two different single-bed oxygen VSA cycles and chose to maximize oxygen recovery using the tank pressure, gas valve constant (from the bed to the tank), and adsorption and desorption times as optimization decision variables. The same authors developed a robust simulation and optimization framework for multibed PSA processes 5 and investigated the effect of different operating parameters on process performance using a five-bed eleven-step PSA config- uration for the separation of hydrogen from a H 2 /CH 4 /N 2 /CO/ CO 2 mixture. To simulate the behavior of a five-bed PSA configuration, two different approaches were employed: Unibed and Multibed. The Unibed approach assumes that all beds undergo identical steps so only one bed is needed to simulate the multibed cycle. Information about the effluent streams is stored in data buffers, and linear interpolation is used to obtain information between two time points. The Multibed approach considers a multibed process as a sequence of repetitive stages within the cycle. A black-box and a simultaneous tailored framework to solve the optimization problem was presented. In the black-box approach, sensitivities are calculated at CSS by using the perturbation method, while in the simultaneous framework either by perturbation or direct sensitivity approach by the DASPK package. A trade-off between H 2 purity and recovery was observed. Mendes et al. 6 developed an efficient and robust cyclic adsorption processes simulator and proposed a systematic theoretical optimization method to design and optimize small- and large-scale PSA units. The system of partial differential equations was solved by an adaptive multiresolution approach, ensuring great stability and accuracy of the simulated solution. The proposed optimization procedure was applied in a case study involving oxygen production from air. This work was then extended 7 to optimize more complex cycles, analyze different types of adsorbents and different operating conditions. A performance analysis is presented for two different types of adsorbents using the Skarstrom cycle with one equalization step (PSA and VSA). * To whom correspondence should be addressed. Tel: +30 24610- 56523. Fax: +30 24610-56501. E-mail: mgeorg@otenet.gr and mgeorg@ uowm.gr. † Department of Mechanical Engineering. ‡ Department of Engineering Informatics and Telecommunications. Ind. Eng. Chem. Res. 2009, 48, 5388–5398 5388 10.1021/ie801357a CCC: $40.75  2009 American Chemical Society Published on Web 04/23/2009