Abderrahim Abbas

Modeling an industrial reverse osmosis unit

A simplified model was used to predict the performance of hollow fiber RO membranes. The model is based on the solution-diffusion membrane mass transport model and takes into account the effect of the pressure drop in the fiber bore and in the fiber bundle, the variation of the bulk solution concentration on the shell side of the fiber bundle and the concentration polarization. The results of the model showed reasonable agreements with the experimental data for B9 membranes and the data from an industrial RO plant based on B10 membranes.

Use of fluid instabilities to enhance membrane performance: a review☆

Abstract This paper presents a review of some of the key methods of generating flow instabilities that have been implemented in membrane separation processes by various workers during the past few years (mainly the 1990s). The types and effects of incorporation these techniques are discussed based on the experimental and theoretical investigations in different applications, particularly, production of potable water from natural fresh water sources, water desalination, and purification of biological and organic solutions and suspensions. The main types of instabilities that are discussed are Dean vortices in helical hollow-fiber membranes, air sparging, novel backflushing techniques and cyclic feed operation. There is a substantial amount of data in the reviewed literature in support of improved performance and fouling prevention in the presence of flow instabilities.

Predicting the performance of RO membranes

Theoretical and experimental recoveries are compared for spiral-wound and hollow-fiber membranes. The effect of ignoring concentration polarization and pressure drops is studied. Ignoring concentration polarization and pressure drops results in significant overestimation of the recovery. With spiral-wound membranes pressure drops were found to be less significant. In addition, a solution method based on integration was used for hollow-fiber membranes. Each of the water and salt fluxes is defined as an implicit function of two dimensions, namely length and radius, and this function is integrated over the whole of the membrane. The integration method resulted in recovery predictions closest to the average of the experimental data. The agreement between predicted and experimental salt rejection was less pronounced, the theoretical values being higher than the experimental.

Performance decline in brackish water Film Tec spiral wound RO membranes

In this paper, long-term operational data from a medium scale RO (reverse osmosis) plant based on brackish water FilmTec spiral wound RO membranes are presented and analysed. It was found that over a period of 500 days of continuous operation, the water permeability coefficient for the overall plant decreased by about 25% and the salt rejection dropped by 1.9%. Simple formulae representing the long-term performance of the plant were developed. These expressions were found to fit the data with very good accuracy.

Long-term performance of an industrial water desalination plant

The long-term performance of a medium-scale industrial spiral wound reverse osmosis (RO) water desalination plant was studied. Operational data were analysed for a period of 1500 days as a basis for evaluating the performance variation with time. A theoretical model based on the solution-diffusion mass-transfer theory and concentration polarization was employed to extract the water and salt permeability coefficients. The results indicate that after over 4 years of continuous operation, the water permeability coefficient declined by about 39% and the average salt permeability coefficient increased by about 60%. Periodic membrane cleaning schemes were effective in partially restoring the water and salt permeabilities.

Permeate recycle to improve the performance of a spiral-wound RO plant

This paper presents the results obtained from a simulation study on the effect of permeate recycle on the performance of a spiral-wound RO plant. The model used was developed for an actual small scale plant consisting of three stages. The first stage consists of two parallel vessels whereas the second and third stages contain a single vessel each. Each vessel houses three 8-inch spiral-wound brackish water FilmTec membranes. Part of the product water is recycled and mixed with the fresh feed. The model incorporates the solution diffusion mass transport theory, concentration polarization, along with the characteristics of the spacer for the estimation of the mass transfer coefficient and pressure drop. The use of these concepts together with the recycling of a portion of the permeate resulted in a model composed of a highly coupled system of nonlinear equations. The results of the simulation showed that recycling part of the permeate and mixing it with the feed was beneficial in terms of reducing the concentration polarization and improving the quality of the final product. However, as expected, this resulted in reducing the production rate. By using a recycle ratio of 25%, the concentration of the product was reduced by 15% compared to the case of no recycle, at the expense of 22% reduction in the production rate.

Flux enhancement of RO desalination processes

Abstract The simple model proposed by Kennedy et al. was modified so as to improve its ability to predict the periodic performance operation of RO systems. The modified model was validated using data from the literature as well as data obtained from experiments in this work. The predicted permeate rates compared well with the experimental values for the considered forcing functions: sine and asymmetric square waves. Periodic and steady-state experimental runs were performed on a spiral-wound RO membrane system at three average pressures: 20, 25 and 30 bars. For the case of periodic operation, the operating pressure was varied according to an asymmetric square wave having a period of 5 min. The duration of each run was 30 min. The effect of the time split, γ, of the forcing wave on the permeation rate as well the salt rejection was investigated. The results showed that periodic operation leads to improvements in the permeation rate. For the tested pressures, the overall maximum water production rates occurred at a time split approximately equal to 0.65. It was also found that cyclic operation may lead to small reductions in the salt passage.

On the performance limitation of reverse osmosis water desalination systems

The performance of Reverse Osmosis (RO) water desalination processes was investigated using a simulated brackish water plant based on spiral-wound membranes. A semi-rigorous mathematical model was employed in the simulation to calculate the water and salt fluxes at any point along the filtration channel. The effects of two key parameters, namely the transmembrane pressure and membrane surface area, on the performance of the process were investigated. Some insights into the performance limitation of RO processes were obtained. The rapid increase in the osmotic pressure of the brine was found to be the main factor which limits the performance of the plant. The study has also revealed that any attempts to develop new membranes for brackish water desalination which withstand higher operating pressures than 4.5 MPa will not result in significant gains in the plant performance. For seawaters having high salinity, the development of membranes, which operate at a pressure of up to 10 MPa, will significantly improve the membranes productivity.

Dynamic matrix control (DMC) of rolling mills

A dynamic matrix control (DMC) algorithm is used for the control of the exit strip thickness of a simulated rolling mill. The DMC is tuned using the recently proposed method by Cooper and colleagues [17 18]. The performance of the DMC is analyzed and compared to those obtained from the classical PI controller and Smith predictor. The sensitivity of the DMC to changes in the plant parameters is also investigated. The simulation results show that the DMC offers a very good performance and that it is much more robust than the Smith predictor and not as robust as the conventional PI controller.

CONTROL OF ALUMINUM ROLLING MILLS

The aluminum rolling mills suffer from the presence of large time-delay-to-time-constant ratios. Conventional proportional–integral–derivative (PID) controllers and their variants do not provide adequate performance when used to control these types of processes. Advanced control strategies are needed. The Smith predictor (SP) is a relatively simple control scheme that is used for time delay compensation. In this paper, the SP has been used to control a rolling mill, and its performance has been compared with that obtained from a conventional proportional plus integral (PI) controller. In addition, the sensitivity of the SP to modeling errors and changing plant conditions was investigated. As expected, it was found that, for the perfect model case, the SP provides superior performance as compared to the classical PI. The superiority of the SP is maintained as long as the modeling errors are not too large. The PI algorithm is much more robust than the SP control scheme.