The solution
Tourism Holdings Limited ran about 4,000 recreational rental vehicles across 14 locations in Australia and New Zealand. Demand at one branch could exceed supply even while similar vehicles sat elsewhere.
A blunt fix would have been to reposition more vehicles by rule of thumb. CSIRO instead built VASS and D-VASS: one system for the static daily schedule and another for live booking decisions.
The model combined substitutions and relocations, minimizing their costs rather than handling them as exceptions. That mattered because substituting a vehicle type could please today's customer while damaging tomorrow's availability.
The result was a scheduling discipline embedded in THL operations: allocation became a network-flow decision across branches and vehicle classes, not a branch manager's scramble.
What it achieved
Treat substitutions and relocations as one schedule
Why it worked
It recognized that similar vehicles are partially interchangeable, not identical.
It priced relocation and substitution in one model instead of optimizing each branch locally.
The dynamic tool made the algorithm useful at the sales counter, not just overnight.
The system turned customer-facing availability into a central fleet-allocation problem.
What can be applied
In rental networks, capacity is not just count by branch. Similar assets can substitute, but only if software prices relocation, customer fit, and future bookings together.
Aftermath
The Interfaces paper says CSIRO supplied VASS and D-VASS and that THL embedded the static and dynamic scheduling systems into its operations.
FOLLOW THE EVIDENCE