- Posted by Okechukwu Anosike on July 20, 2009
- [Permalink]
Staying abreast of best practices isn’t enough. You also need to know how to implement the fundamentals.
The classical approach to optimization is to consider one variable at a time, but most of the time this might give misleading results because it takes no account of the way the variable interacts with each other. When this is considered on a scale up of a laboratory experiment one has to note that if scaling up is considered in terms of surface area, density, viscosity, surface tension scale up are normally governed by parameters and variables increase by a squared function and when considering scaling up by volume, parameters and variables normally increase by a cube function.
There are different methods of process optimization but I will be talking about the use of statistical methods for achieving process optimization. This method of optimization falls into two categories namely the empirical search method such as simplex and the experimental design method commonly known as design of experiment, DOE.
Simplex method and DOE
In the simplex method, one might want to investigate the combined affect of X different factors, an initial set of X+1 experiments are run in which the conditions span an appropriate of X factor. The comparison of the results then allows a new, potentially more interesting, set of conditions to be computed- by rejecting the worst point from the set and reflecting it through the centriod of the remaining points. By repeating this process, the reaction conditions are inexorably shifted onto a more optimal region. Using this method or the modified simplex / super modified simplex method, this can be a very powerful tool for rapid location of optimal conditions.
Experimental designs (DOE) this method is useful for locating optimal conditions. This process normally provides valuable information about the effects of each variable on the process.
Tags: plants, scale up, optimization
Category: Chemical Engineering, Plant Scale Up's, Process Engineering
Actions:
E-mail | Comments (16)
Share With: Twitter | FaceBook