Sources of variation plot with batch conditions (Q802)

The information in this article applies to: 
SIMCA 13
Symptoms: 

For a model combining batch level data created from the batch evolution model and qualitative batch condition data, the Sources of Variation plot does not display the batch condition data but is confused by them and potentially crashes when modifying the plot using Properties.

Workaround: 

Generally speaking it is a disadvantage to model two unequally large blocks as X. To avoid this in the described case, create a hierarchical top model based on Batch conditions and compressed batch evolution data instead.

Suggested work process: 
  1. Create a hierarchical batch model for each set of batch evolution data in your project (On the Batch tab, click Create Hierarchical Batch Models). One set could be process data, another In Process Control (IPC) data. A model could also be built for Batch conditions if these are numerous and correlated.  
  2. Create a hierarchical top model at batch level based on score values from the multivariate model(s) created in step 1 and the Batch conditions in original form or as hierarchical score variables. 

The model type in step 1 can be PCA, PLS or OPLS. If OPLS is applied there will only be one score vector (t1) to use in the hierarchical top level model. 

 
Status: 

Fixed in SIMCA 14.

Date modified: Jan 29, 2015
Article type: 
Problem