Search

In a supply chain system, it is very important to forecast the changes in the market in order to maintain an inventory level that is just enough to satisfy customer demand. However, demand forecasting is responsible for the so-called “bullwhip effect”, the exaggeration of demand fluctuation toward upstream nodes. In this study, we present a minimum variance control (MVC) approach to solve this problem. First, customer demand trends are described by a general ARIMA model. The customer demand forecast is used to determine two inventory targets:  inventory at hand and inventory on the road. The changes in inventory level are predicted using a model of material balance and information flow. The order policy is obtained by minimizing the errors between predicted inventory levels and set points and using a function that penalizes large changes in orders. Two controller parameters, the penalty cost factor and the relative weight between the changes in inventory and absolute inventory, are used to optimize the excess inventory and backorder subject to the constraint of no “bullwhip”. Simulation results show that this approach can track customer demand and maintain a proper inventory level without causing a bullwhip effect, whether the customer demand trend is stationary or not. Furthermore, the performance of the MVC is found to be superior to those of other approaches such as order-up-to-level, proportional and integral (PI) control, and smoothing order policy found in the literature.

In a reactive distillation column in which the reaction is kinetically controlled, the steady-state temperature and composition profiles will change whenever there is a change in the production rate or the catalyst deactivates. Because of changes in the residence times on the catalytic stages, conversion changes. This poses a problem for controller design because the controller set points must be changed when disturbances in the feed flow rate and catalyst activity occur. For a column used for the synthesis of n-butyl acetate, we were able to locate points in the temperature and composition profiles that are relatively insensitive to throughput rate changes and catalyst deactivation. Reboiler duty is used as the manipulated variable in the temperature control loop. The feed rates of n-butanol and acetic acid are controlled, with the set point of the acetic acid flow loop ratioed to that of the n-butanol flow loop. The set point of the ratioed control is used as the manipulated variable in an inner cascade loop to control the acetic acid composition inside the column. This inner composition cascade loop guards against feed flow sensor bias in the feed-ratio control. The value of this composition set point is manipulated by an outer cascade loop that controls the composition of the bottom product. This outer cascade loop accounts for changes in reaction residence time relative to reaction time due to production rate changes or catalyst deactivation. The proposed scheme was able to reject various disturbances and maintain the desired high-purity and high-conversion operation.

Recent advances in reactive distillation have led to renewed interest in the design/redesign of many esterification processes. For the esterification of amyl acetate, design alternatives exist that offer a gradual transition from the conventional recycle structure to reactive distillation. The phase behavior of the quaternary mixture immediately excludes the possibility of the conventional recycle structure, and in this work, we explore the designs of two different configurations, a coupled reactor/column and reactive distillation column , for the amyl acetate process. The total annual cost (TAC) is used to evaluate the economic advantages of different designs. Systematic design procedures were devised for the coupled reactor/column system, as well as for the reactive distillation process. The results clearly indicate the economic advantage of reactive distillation for the amyl acetate process. Unique characteristics of amyl acetate process include a significant two-liquid zone and high-purity water from the aqueous phase of a decanter. This leads us to a single-end-composition control problem. However, limit cycles can occur if the net reaction rate and production rate are not properly balanced or if the feeds are not stoichiometrically balanced. Therefore, a second composition loop is used to infer possible imbalances. Dynamic simulations indicate that reasonable control can be achieved for very complex reactive distillation dynamics.

Existing forms of adaptive resonance theory, e.g., ART2 and Fuzzy ART, employ similarity-based vigilance measures and contrast enhancement that is analog in nature. They use the “fast” or “fast-commit-slow-recode” learning rules, which do not guarantee convergence of clustering results. Therefore, they are not suitable for process sensor pattern monitoring which required geometrically based classifications. A modified version of the adaptive resonance theory, DART, was developed. DART uses a distance-based vigilance measure, a contrast enhancement procedure that is around the center of the prototype instead of around the null input, and the Kohonen learning rule to ensure convergence when accepting inputs that are highly correlated dynamically. The necessities of such modifications were demonstrated using a simple mathematical example:  the Leonard−Kramer problem. The ability of DART to isolate different faults from operation history and to monitor operation in an adaptive manner for a complex plant is demonstrated using the Tennessee−Eastman problem. Although the process exhibits highly nonlinear dynamic behavior, DART is able to obtain classifications that are geometrically based in the sensor pattern space and are closely associated with various fault origins. Given such classifications, the nonlinear nature of the movements of this complex process in the sensor pattern space can be easily visualized. Therefore, dynamic operation can be closely monitored, and prewarning for imminent shutdown can also be provided.

In this work, bifurcation analysis and dynamic simulation were used to investigate the optimum conventional control strategy of an isopropyl alcohol (IPA), cyclohexane (CyH), and water (H2O) heterogeneous azeotropic column. Steady-state process analysis shows that the optimal operation point should be located at a critical reflux, a transition point at which the distillation path switches from a route that passes through the IPA + H2O azeotrope to one that passes through the IPA + CyH azeotrope. A good control strategy must be able to maintain a steady column temperature profile that shows a plateau near 70 °C to ensure passage around the IPA + CyH azeotrope. An inverse double-loop control strategy is proposed based on principal component analysis. This scheme is capable of maintaining the desired column temperature profile given all kinds of feed disturbances, thus keeping the product IPA purity at the desired level.

The adsorption isotherms and the effective diffusivities of benzene in activated carbon pellets in the presence of subcritical and supercritical carbon dioxide were measured from the sorption curves in a spinning basket autoclave at temperatures 308, 318, and 328 K and pressures up to 163 bar. The adsorption isotherm data showed that no condensation of benzene in activated carbon occurred at the given temperatures and pressures. The effective diffusivities measured were observed to be concentration-dependent and to decrease with increasing pressure. At high-pressure operations, the effective diffusivities were found to be invariant with temperature.

Adsorption of carbon dioxide near its critical point on DeGussa IV activated carbon is investigated in this study. A volumetric method was used to measure the adsorption/desorption isotherms at 284, 300, 305, 310, and 314 K over a large pressure range. At subcritical temperatures, adsorption isotherms display a discontinuity at the vapor pressure of carbon dioxide, and desorption hysteresis is observed. However, there is no desorption hysteresis if adsorption is terminated before vapor−liquid transition occurs. At supercritical temperatures, adsorption isotherms display a plateau, and the excess decreases beyond the critical pressure. No hysteresis occurs during the desorption process. The adsorption isotherms can be represented very well by the simplified local density model.