How much mileage is saved by relocation of warehouses, or by a merger of transport operations? The live demo below - that is loaded if you scroll further - shows how Fleet Mileage Calculator helps to answer these questions.
It simulates a full year of transports, creating vehicle routes, per warehouse, per day.
Acts as a tactical analysis tool, not an operational transport planning tool.
To import data, click button 'Import | Export' above each input table, paste your data (tab or semicolon separated) in textarea, click button again.
You have not logged in. Changes to warehouse demo data are ignored. Create an account to be able to log in and run with your own customers data. You can adjust other input tables and parameters, and run the software, to check out how it works.
Warehouses
Import | Export textarea
WarehouseID is a free format alphanumeric field. Latitude and longitude are the result of geocoding your warehouse address(es).
You can change data directly in the input tables (except for table Warehouses if not logged in). Navigate with tab, shif+tab, arrow up, arrow down.
Option 1: Shipments per year (non-preferred option, option 2 is preferred)
Import | Export textarea
CustomerID is a free format alphanumeric field. You can define demand and shipment size both in volume (for example, pallets) and in weight (for example, kilograms), as sometimes volume, sometimes weight may be the limiting factor of what fits into a truck (truck capacity is specified for volume and for weight - under Parameters). Express demand and shipment size in the same unit of measurement as you express truck capacity, and make sure that shipment size does not exceed truck capacity. After entry, press below "Convert" button.
Yearly input
versus
Daily output
Shipments
Demand volume
Demand weight
Option 2: Shipments per day (preferred)
If you know what shipments were delivered on what day (1,2,3,...), then use this option, as it will bring more realistic outcomes.
Import | Export textarea
Kilometers
Miles
Truck capacity (volume)
Truck capacity (weight)
Working days per year
i
Distribution hours per workday(s)
i
Unloading minutes per delivery
Average speed (km/hour)
Average speed (miles/hour)
Distance circuity factor
i
Truck fill rate
i
From day
To day
i
Map will display routes of all warehouses on this last day.
CVRP runs
i
Progress
Current CVRP
Data entry errors
High level approximation
Approximation based on avg warehouse-customer distances, avg inter drop distances derived from auto-measuring delivery area sizes, shipment amounts and sizes, and transport parameter settings. Mileage is determined per warehouse per day, then accumulated.
Capacitated Vehicle Routing Problem solutions - Summary
Outcomes of vehicle routes constructed.
Capacitated Vehicle Routing Problem solutions - Route details
Fleet Mileage Calculator constructs vehicle routes, per day, per warehouse, by solving a Capacitated Vehicle Routing Problem (CVRP).
Each CVRP is solved using the well-known Clarke and Wright's method to construct an initial solution
According to Girard (2005) and based on the 14 original benchmark instances of Christofides, Mingozzi and Toth (1979), the average deviation of the solutions obtained by Clarke and Wright’s method (with respect to the best published solutions) is 7.68%. (Source: scientific article "A Simple and Efficient Perturbation Heuristic to solve the Vehicle Routing Problem", Sylvain Girard, Jacques Renaud,and Fayez F. Boctor, August 2005, Working Paper DT-2005-JR-1.)
Clarke and Wright's method has been improved by Stelling Consulting by:
Applying a new savings-pairs sorting criterium, that often brings better results.
Applying limited randomization in the selection process of the next savings-pair to be evaluated/used in route-merging. This makes each inital solution different, sometimes worse but also sometimes better than standard Clarke and Wright.
Each initial solution is improved by:
Optimizing stop sequence of each route = solving a travelling salesman problem, done by an integrated TSP solver.
Exchanging stops between routes.
Randomness makes each run outcome different. The more runs, the higher the chance of finding a better solution. Average deviation from optimal outcome is estimated at approx. 0-2%.
