Multiway Partial Least Square For Modeling And Monitoring Of Poly Methyl Methacrylate Reactor
Polymerization is the process in which monomer units are combined by chemical reaction to form long chains monomer called polymer. The output variables such as molecular weight and conversion are unavailable online thus they need to be measured through laboratory analysis and moreover only a limited...
Saved in:
Main Author: | |
---|---|
Format: | Thesis |
Language: | English |
Published: |
2014
|
Subjects: | |
Online Access: | http://eprints.usm.my/46260/1/Syarina%20Binti%20Osman24.pdf http://eprints.usm.my/46260/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Polymerization is the process in which monomer units are combined by chemical reaction to form long chains monomer called polymer. The output variables such as molecular weight and conversion are unavailable online thus they need to be measured through laboratory analysis and moreover only a limited sample are made during the process. The unavailability output variables leads to the development of inferential model which in this study used Multiway Partial least squares (MPLS). MPLS model is one of the famous Chemometric method which able to monitor and predict the process output properties. This inferential model prepared a place where output properties such as molecular weight can be predicted based on operating data such as reactor and jacket temperature and coolant flowrate. This inferential model requires a large amount of data which accumulated during previous processes. These data are considered as normal process which produced desired output properties. The present study focuses on the prediction the output properties of Methyl Methacrylate (MMA) polymerization such as monomer conversion, the weight average molecular weight (Mw) and the number average molecular weight (Mn). This is due to the nature of these properties since these output properties are not available online. This work helps to monitor the properties and it can be used to detect off specification product. The idea behind the additional model is that this model gives a prediction value at specific time during the process. |
---|