We present a new mass-transfer model for simulating industrial nylon-6 polymerization trains.
In this model, both diffusion and boiling (bubble nucleation) contribute to mass transfer. With
this model, we are able to simulate widely differing production technologies using identical mass-transfer parameters, along with identical models for fundamentals such as phase equilibrium,
physical properties, and polymerization kinetics. To illustrate, we simulate the direct-melt process
and the bubble-gas kettle process. The direct-melt process builds up the polymer molecular weight
and removes nearly all residual caprolactam monomer by employing, under vacuum, a wiped-wall evaporator and a rotating-disk finisher. The bubble-gas kettle process, on the other hand,
injects inert gas bubbles through the melt at nearly atmospheric pressure to build up the polymer
molecular weight but does not significantly reduce the caprolactam level because the diffusion
coefficient is so low. We validate our process models using commercial train performance data
at different production rates, including the first known validation of a dynamic rate-change
simulation for industrial polycondensation trains. Model predictions quantitatively agree with
product quality data such as formic acid viscosity (FAV), polymer end-group concentration, and
water extractables. The prediction errors for the direct-melt process are 2.81%, −3.13%, and
−3.06% for FAV, water extractables, and amine end groups, respectively. For the bubble-gas
kettle process, the prediction errors are −17.2%, −17.0%, and −7.49% for extrusion FAV, washed-and-dried FAV, and water extractables, respectively. These errors are much lower than the ca.
−50% errors obtained by existing advanced models for devolatilization.