With an active blocking back calculation less traffic arrives than without blocking back calculation.
The reason lies within the somewhat contradictory combination of a static assignment and the dynamic procedure blocking back (Dynamic assignments do not take blocking back into account, DUE uses an own procedure). The traffic load on links with a capacity bottleneck and ensuing links is smaller, because only the traffic that got through these links during the blocking back 1st phase is counted. Capacity and a static assignment should share the same time scope, typically 'a day', or 'an hour'. Therefore traffic not able to get along capacity reduced links will reach its destination, but beyond this time scope.
The links before the capacity bottleneck get a mixed traffic volume, counting the traffic that got through + the traffic stocked on these links. The stocking capacity is defined in a link attribute (Edit link -> Congestion -> Stocking Capacity). Traffic exceeding this stocking capacity is blocked back onto the preceding link or links.
The PrT paths list shows, that all traffic was loaded onto the paths, thus left the origin but not necessarily reached its destination: Lists -> Paths -> PrT Paths