A computational test problem for the MAEROS aerosol model is used to illustrate the application of uncertainty/sensitivity analysis techniques based on Latin hypercube sampling and regression analysis to aggregation problems. The test problem involves a five-component aerosol in the containment of a pressurized water reactor. The following topics are investigated:

  1. Cray 1-S CPU time requirements to implement and solve the system of differential equations on which MAEROS is based
  2. effects on computational time and representational accuracy due to the use of different overall section boundaries and numbers of sections and components
  3. behavior of the aerosol and the variables that influence this behavior.
The analysis provides information in each of the indicated areas in a reasonably straightforward and efficient manner. The same analysis approach could be productively employed in similar investigations for other models. Furthermore, due to the extensive use of MAEROS and other aerosol models in the analysis of reactor accidents, the results obtained in the analysis are also of considerable interest.