Variants classify process instances according to their structure. The relevant structure is the sequence of functions in a process instance.
Functions are identified by the value of the AT_OBJNAME attribute. Basically all other information is ignored from the EPC representation of the process instance, and only named functions and their antecedence relation are retained. This abstract view of an EPC is called a "summary".
There are two kinds of summary:
In addition, both kinds of summary contain information about which nodes occur as start and end nodes in an EPC. A start node is a node without predecessors; an end node is a node without successors. Among nodes forming a cycle, none is either a start or an end node. In a combined summary, a node is classified as a start respectively end node when it occurs at least once as a start or end node in the EPC.
Precise and combined variants
A precise variant is the set of process instances sharing the same precise summary.
A combined variant is the set of process instances sharing the same combined summary. It is more a coarse-grained than a precise variant. When two process instances belong to the same precise variant, they necessarily also belong to the same combined variant, but not vice versa.
Variant names
The name of a variant is referred to as a dimension ID, because it will be used as such in an EPC to record to which variant the process instance belongs. Or in other words, variants are represented as dimension values on process instances. The variant dimension has two levels, where the first (coarser) level contains the ID of the combined variant, and the second (finer) level contains the precise variant. The variant dimension can be queried from the PPM GUI similar to other dimensions.
Variants calculation
Variants are calculated during PPM-Import, after the attributes of an EPC have been calculated, but only if the variant dimension is registered for the process type of the EPC. They are not calculated for aggregated EPC.
The calculation results in a pair of numbers. The first of these numbers uniquely identifies the combined variant. It is prefixed with "v" and written into the attribute for the first level of the variant dimension. The second of these numbers is a serial number for the subordinate precise variant. It is not unique on its own and written into the attribute value of the second dimension level.
Existing values of attributes used within the variant dimension will be overridden by variant calculation.
Process instances for which the variant is unknown (for example, because the variant has not yet been calculated), have the precise value v0-0 (and thus a combined value v0).
More information about the usage of variant attributes during import and attribute calculation can be found in the documentation PPM Customizing.