Like .dat format, relational data can be read without any option settings. Vensim determines the delimiter from the file type (tab or csv), but everything else is fixed.
Just like the export format, data is assumed to be laid out with one data point per row, using a column structure like:
<variable name> <subscript 1> <subscript N> <time> <value>
<variable[subscripts]> <time> <value>
Columns should be delimited by tabs or commas, matching the .tab or .csv file extension.
You can include a header row, but as long as it lacks numbers, it will be ignored.
If subscript dimensionality varies, the columns need not be aligned. The first text field Vensim encounters on a line will be read as the variable name. Any subsequent text is interpreted as a subscript. The first number encountered is time, and the second is the variable value at that time. Any line deviating from this will be ignored.