Statistics - Normalizing datasets to identify outliers

I am working with another freelancer developing a database of monitoring well sample results. I have approximately 50 locations (wells) and we routine sample each well for approximately 50 different parameters. This means I have approximately 2500 datasets. Each dataset currently consists of about 40 samples but each continues to grow every time new samples are collected, analyzed and added to the database. I am looking for someone who can help me develop programmable statistical testing such that we can normalize each set and be able to assess if any new sample is 'statistically different' from the set. I have created histograms for each dataset and most do not appear to be normally distributed. The entire process needs to be somewhat automated so I am envisioning some type of 'and..if' scenarios where we test to see if the dataset is normal, if not test for chi squared, if not test for log normal etc (Please understand my statistical skills are week so the examples listed may not be appropriate at all and are intended only to explain how I think we might proceed. I expect the successful freelancer will work with the freelancer I have already hired to develop these statistical tests in VBA programming