Dynamic simulation of granular filtration
Filtration has been the research subject for many years, yet there is still a lack of established and dynamic applicable models which could describe the complex mechanisms governing the clogging of filter columns. Clogged bed equations developed by far were entirely empirical, and the clogged bed...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
2016
|
Subjects: | |
Online Access: | http://hdl.handle.net/10356/67623 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Summary: | Filtration has been the research subject for many years, yet there is still a lack of established and
dynamic applicable models which could describe the complex mechanisms governing the
clogging of filter columns. Clogged bed equations developed by far were entirely empirical, and
the clogged bed circumstance was not fully apprehended. Through literature studies, it was also
concurred that minimal studies has been done to evaluate the clogging of filter grains using
Computational Fluid Dynamic (CFD) modules. To bridge the research gap, a comprehensive
study of the clogging of filter column was carried out. This study serves to achieve 2 objectives:
i) to design a numerical model which could predict closely the change in specific deposit and
head loss incurred in a filter column throughout the filtration stage. The model could then be
employed as a tool to estimate the maximum run length of the filter column, and ii) to
demonstrate that CFD software can serve as an accelerated approach to comprehend the
clogging mechanism of a filter column. For the first objective, theoretical and empirical
equations were included in the development of a MATLAB model which allows the close
prediction of the filter performance parameters as filtration progresses. The predicted time to
limiting head obtained using the MATLAB model deviates slightly from the run length of a
pilot-scale filter by 0% to 3.2%. For the second objective, more than 50 domains representing
different clogging conditions were integrated into a CFD software. The simulated values for
head loss deviates slightly from theoretical equations by 0% to 10% and was found to reside
between the head loss values calculated using 2 empirical equations. In addition, methods to
optimize the quality of mesh elements in the CFD software are proposed. Further works on
enhancing the performance of the MATLAB model through the incorporation of an automatic
iterative process are suggested. The CFD model could also be improved by minimizing the gap
size near the domain walls. |
---|