I=
Pakistan Research Repository Home
 

Title of Thesis

Recent Trends In Time Series Modeling And Prediction Of Wind Data Statistical And Fuzzy Reasoning

Author(s)

Yasmin Zahra Jafri

Institute/University/Department Details
Department Of Statistics / University Of Balochistan, Quetta
Session
2008
Subject
Statistics
Number of Pages
237
Keywords (Extracted from title, table of contents and abstract of thesis)
Numerical, Experimental, Series, Effectiveness, Sherwood, Prediction, Reynolds, Wind, Dimensional, Statistical, Reasoning, Qualitatively, Fuzzy

Abstract
Spiral wound module is the most common type of membrane configuration used for water purification applications due to high packing density, cost-effectiveness and its availability from several manufacturers. A significant problem in spiral wound membrane is the occurrence of concentration polarization and fouling which reduces the flux through the membrane, increases the passage of solutes and increases module pressure drop. Various methods are employed to prevent concentration polarization and fouling and to improve the membrane performance. The methods include modification of leaf geometry and control of operating conditions such as feed composition, inlet pressure and permeate recovery. A common technique to enhance the back-mixing of the accumulated particles is the use of net-type spacers between two adjacent membrane layers. These spacers not only define the channel height and provide passage for the feed flow but also promote mixing by continuously disrupting the concentration polarization layer. This in-turn increases the product water quality and quantity and the membrane life. On the other hand, the spacers increase pressure drop and create dead zones where the fouling phenomenon can increase.
A number of experimental and numerical studies appeared in literature in recent years due to the importance of feed spacer in the spiral wound element. An in-depth understanding, of the flow and concentration patterns in membrane feed channels, however, was not developed experimentally due to difficulties associated in applying flow visualization and measurement techniques. The detailed and systematic CFD investigations were limited to two-dimensional studies and the effect of different parameters of spacer geometry on flow behavior and spatial distribution of shear stress and mass transfer at membrane surfaces was not examined.
In this thesis three-dimensional modeling is performed using commercial CFD package FLUENT. The work carried out in this thesis is a thorough and systematic analysis of the effect of feed spacer geometry which include parameters such as filament spacing, thickness and flow attack angle on the membrane process. The fluid dynamics modeling in these channels reveal complex flow behavior and shows considerable effect of spacer geometry on the shear stress distribution. The values of critical Reynolds number at which flow leaves the laminar regime and becomes transient are also indicated. These values are found to vary between 75–300 depending on the spacer geometry. Another important contribution is the identification of appropriate turbulence models for simulating flows in spacer-filled channels. It is also shown that the improper
selection of turbulence models can affect the results both qualitatively and quantitatively. An over-prediction of upto 70 % in pressure drop values may result with some models when compared to the direct numerical simulation (DNS) results. Furthermore, the flow situations in which the two-dimensional approximation is valid for these narrow channels have also been indicated. In mass transfer modeling the concentration profiles are predicted and the spacer performances are determined in terms of mass transfer coefficient / pressure drop ratios. The spacers which involve zigzag flow have been found to be suitable for spiral wound modules since they result in higher values of spacer performance with minimal stagnation regions. The transient simulations show the movement of vortices and variation of shear stress and mass transfer coefficient with time. Results related to the entrance region in membrane feed channels are included as well.
The effect of feed spacer is also investigated experimentally. The pressure drop, product flow rate and concentration are measured for spacers of different heights, spacings and flow attack angles. The effect of operating conditions like feed pressure and total dissolved solids (TDS) on product flow rate and salt rejection is studied. The effect on product flow is more significant as it reduces approximately 30 % when feed TDS is doubled while the salt rejection only decreases 3 %. The CFD results are compared with experiments and a reasonable agreement is noticed between the two results. The comparison of shear stress pattern (from CFD) with the fouling pattern (from experiments) indicates that high shear stress region coincides with the low fouling area and vice versa. The effect of spacer on membrane fouling is determined by operating the experimental unit for longer periods. The measurement of permeate flux and salt passage showed declination of 45 % and 20 % respectively with the passage of time. The correlations are also developed for friction factor, Sherwood number and product flux to predict fouling rate. The uncertainty analyses for measurements shows that mass transfer coefficient, pressure drop and salt rejection are sufficiently accurate to compare different spacer geometries and to validate the numerical results.

Download Full Thesis
2,760 KB
S. No. Chapter Title of the Chapters Page Size (KB)
1 0 CONTENTS

 

vii
95 KB
2

1

CRITICAL REVIEWS

1.1 Stochastic Time Series Modeling, Simulation & Prediction

1.2 Regression Analysis Time Series Modeling & Simulation
1.3 Chaotic Time Series without Rule Based Fuzzy logic(FL),Mackey Glass Simulation with FL and Prediction
1.4 Rule Based Fuzzy Logic Time Series Prediction,Modeling and Simulation
1.5 Artificial Neural Network Time Series (ANNTS) Modeling, Simulation & Prediction
1.6 Thesis Plan

1
206 KB
3 2 THEORIES ON TIME SERIES MODELING, SIMULATION & PREDICTION

2.1 Autoregressive Moving Average (ARMA) Models
2.2 Fuzzy Time series
2.3 Chaotic Time Series With out Rule Based Fuzzy Logic-and Mackey Glass Simulation
2.4 Rule Based Fuzzy Logic (FL) for Singleton & non- Singleton Type -1 FLS using Back Propagation (BP) Design
2.5 Neural Network Forecasting

22
238 KB
4 3 RESULTS AND DISCUSSIONS

3.1 ARMA, MTM, Non-seasonal ARIMA Models on Hourly Averaged Wind Speed (HAWS) Data
3.2 PRATS, PR, MLR, MLRATS on Climate Data(Humidity, Rainfall Maximum & Minimum Temperatures
3.3 Mackey Glass Simulation on HAWS & Prediction of Wind data
3.4 Singleton & non-Singleton Type-1 Back Propagation(BP) Designed Fuzzy (FL), Simulation & Prediction of HAWS
3.5 Feed Forward Back Propagation Neural Network (FFBPNN) Prediction of HAWS

39
204 KB
5 4 CONCLUSIONS & SUGGESTIONS ON FUTURE WORK

4.1 ARMA, Non-Seasonal and Seasonal ARIMA Models
4.2 Regression Analysis Time Series (RATS)
4.3 Fuzzy Logic (FL) without Rule and Mackey Glass Simulation
4.4 Singleton & non- Singleton type-1 FL with BP Design
4.5 FFBPNN

56
141 KB
6 5 REFERENCES AND APPENDICES

62


249 KB