I had 2 columns in a table, let's say "One" and "Two", where One has integer values and Two string values. In my model I created a calculated column Three to concatenate these values. Because One and Two are actually redundant and never used on its own, I decided to delete those columns and save some space (the tables contained just over 2.2 million records).
So I started with updating my source view to concatenate the values in T-SQL, but when I refreshed the table in Power Pivot I got the error:
"Unable to convert a value to the data type requested in table x for column Three"
This error occured because I renamed the column in the view One and I already had that column present in my PP earlier (although I deleted it before opening the Table Properties). So my column One was previous INT, but would now be VARCHAR, however PP keeps this setting in mind although the column was deleted.
I found an answer on StackOverflow:
My solution was to recreate the column One in my PP and leave it empty (this would get the data type INT), and then add the new column One from the source in the Table Properties. This gives a warning that the column is already present in your workbook, but solves it by renaming the new column to "One 1" when refreshing the changes. After that you can delete the old column One and rename "One 1" to One and you're done!
So in short:
- Recreate the column (One) you deleted
- Then add the newly added column from your source with the same name
- PP renames the new column to "One 1"
- Delete the old column (One) again and rename "One 1" to One, or the desired name