In brief
This is an example of the second step of working with us: an analysis on historical data before any trials. A simulation on a wholesale network’s real data from one month shows the potential to cut the fleet by a third for the same work. We do not name the client’s industry, and we have scaled all absolute figures to a fleet of 100 vehicles so that the client cannot be identified. The percentages are as in the analysis.
Input data
- Dates and addresses of all deliveries over a full month, scaled to about 71,000 deliveries to about 4,800 customers.
- Regular and occasional customers.
- The fleet’s actual work over the same period: 100 vehicles after scaling.
Simulation assumptions
The simulation takes into account:
- travel speeds on the road network measured from GPS data,
- time spent at each point, including loading and unloading,
- vehicle capacity and volume limits,
- availability of vehicles and drivers,
- the time windows within which deliveries to customers must fit.
Results
Scaled to a fleet of 100 vehicles:
| Measure | Today | With automated dispatch | Change |
|---|---|---|---|
| Vehicles | 100 | 67 | −33% |
| Working hours per month | about 25,600 | about 17,400 | −32% |
| Kilometres per month | about 419,000 | about 381,500 | −9% |
| Deliveries per available working hour | 2.77 | 4.09 | +48% |
Cutting the fleet by a third proved possible in each of the five weeks of the month.
Limitations
The simulation compares planned routes with the fleet’s actual work. It does not include breakdowns or other disruptions during the day, which is why the next step is trials at selected locations.
What next
We are currently working with the client on the next steps, which are intended to lead to a rollout.