(189a) Statistically Driven Axial Activity Variations in Graded Catalyst Beds: Loading From Pre-Measured Single Tube Aliquots

Statistically Driven Axial Activity Variations in Graded Catalyst Beds: Loading From Pre-Measured Single Tube Aliquots

E.M. Calverley, P.M Witt, J.D. Sweeney, Core R&D, The Dow Chemical Company, Midland MI

Graded activity profiles in multi-tubular packed bed reactors are often created by diluting active catalyst pellets with inert pellets of similar size, shape and density.  This scheme allows careful tuning of the axial activity profile to balance the loss of driving force as reactants are consumed.  Loading mixtures of inert and active particles into a reactor is a statistical process which leads to unavoidable axial variability in catalytic activity at the scale of individual pellets.  The fact that this produces tube to tube variability in commercial reactors is well recognized in the industry, but this phenomenon has not been well explored in the literature.

In earlier work, we showed that the effects of this statistical variability are significant for commercially relevant tube to particle diameter ratios and extents of dilution, when catalyst mixtures are loaded from large reservoirs and the loading statistics are described by the binomial distribution.  In the present work, we extend the analysis by modeling the performance of a reactor loading scheme in which pre-measured aliquots of catalyst – inert mixture are prepared, such that each zone in each tube has precisely the desired number of active pills.  The statistical variability in the axial activity profile for this scheme is described by the hypergeometric probability distribution which reflects the outcomes of Bernoulli trials from a finite population.  Simulations of reactor tubes with activity profiles generated by random sampling from the hypergeometric probability distribution were performed for a range of tube to particle diameters and catalyst dilution schemes to assess the magnitude of tube to tube variability assignable to statistical variability. 

Loading from pre-measured aliquots produces axial variability sufficient to produce substantial tube to tube differences in performance, though tubes loaded in this way perform significantly better than tubes loaded from a large reservoir such as a drum of catalyst.  Smaller catalyst particles (larger tube to particle diameter ratio) and smaller extents of catalyst dilution reduce the axial activity variability, and we show how build a map of inherently stable combinations of pellet size and catalyst dilution.