Doctor of Philosophy (PhD)
New Page 1
A general dynamic model for continuous EPDM polymerization in which crosslinking and gel formation are attributable to reactions between pendant double bonds has been developed. A pseudo-kinetic rate constant method is introduced to construct a moment model for a pseudo-homopolymer that approximates the behavior of the actual terpolymer under long chain and quasi-steady state assumptions. The pseudo-homopolymer model is then used as the basis for application of the numerical fractionation method. The proposed dynamic model is capable of predicting polydispersities and molecular weight distributions near the gel point with as few as eleven generations, and in the post-gel region with as few as five. The overall molecular weight distribution (MWD) of the sol was constructed by assuming a Schulz two parameter distribution for each generation.
A parameter selection procedure is proposed to determine the kinetic parameters that can be estimated from the limited plant data. The procedure is based on the steady-state parameter output sensitivity matrix. The overall effect of each parameter on the measured outputs is determined using Principal Component Analysis (PCA). The angles between the sensitivity vectors are used as a measure of collinearity between parameters. A simple algorithm which provides a tradeoff between overall parameter effect on key outputs and collinearity yields a ranking of parameters by ease of estimation, independent of the available data. Its nonlinear and dynamic extensions are also developed and tested to address the nonlinearity and dynamics of the parameters' effects on the outputs. The key kinetic parameters determined by the parameter selection procedure were estimated from steady-state data extracted from dynamic plant data, using a newly developed steady state detection tool.
A hierarchical extended Kalman filter (EKF) design is proposed to estimate unmeasured state variables and key kinetic parameters of the EPDM kinetic model. The estimator design is based on decomposing the dynamic model into two subsystems, by exploiting the triangular model structure and the different sampling frequencies of the on-line and laboratory measurements directly related to the state variables of
each subsystem. Simulation tests show that the hierarchical EKF generates satisfactory predictions even in the presence of measurement noise and plant/model mismatch.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Li, Rujun, "Dynamic modeling and parameter estimation for an ethlyene-propylene-diene polymerization process" (2003). LSU Doctoral Dissertations. 2655.
Kerry M. Dooley