Batch level dataset and model (BLM) are wrong in some cases (Q903)
If you have opened this knowledge base article because of a warning message in SIMCA there is a possibility that you have a problem with your batch level models. We recommend you follow the workaround described below to be on the safe side.
The problem occurs when you have two or more datasets and the variables in them are to some extent the same and you generate a batch level dataset with raw data or raw data statistics. There is then a risk that some variables in the batch level dataset contain misaligned data. There are several ways to end up with datasets that fulfil the described criteria, e.g. importing phases separately and only a few variables are common between the different phases or generating variables in only one of the datasets before creating the batch level dataset. You will also get the same type of misalignment in the batch level dataset if your datasets have the same variables but in different order and you re-arrange the dataset order in the workset dialog before generating you batch evolution models (BEM).
The values for complete variables in the batch level are wrong. The displayed values come from another variable, e.g. all the temperature variable values come from the pressure sensor. Verify that the first value(s) in each variable in the batch level dataset are showing the right sensor values.
Fixed in SIMCA 15.0.2, released June 2018. Delete all batch level datasets and generate them again in SIMCA 15.0.2 or later.