Fleet Optimization & Routing Software for Plant Logistics
Fleet optimisation that keeps re-planning through the day: routes re-optimised against a frozen prefix of committed stops, driver hours, windows, cold chain and hazard built into every route, and cost per stop reported by route, driver and customer.
Schematic
Route build under constraints
How stops are sequenced against time windows and vehicle capacity.
Fleet optimisation for a day that keeps changing
MileSoft Fleet Optimization plans delivery routes for plant and distribution fleets and keeps re-planning them as the day moves. Orders mature after dispatch, windows shift and vehicles run late, so the optimiser runs repeatedly on a rolling horizon: stops already executed or communicated to a customer stay fixed, and only the free tail of each route is re-sequenced. Updated sequences reach drivers through a driver application with navigation, and dispatchers see the same map and can approve or override.
Vehicle capacity, driver hours, customer time windows, cold chain and hazardous goods are constraints inside the route model rather than checks after release, so the system never publishes a route a driver cannot legally or safely run. Multi-depot operation assigns each order to the depot whose vehicle can actually serve it best, which is not always the nearest one.
The cost model attributes fuel, distance, time and vehicle utilisation by route, driver and customer, and reports cost per stop as the primary figure — because a fleet can cut its mileage while its margin erodes, and cost per stop is what shows which customers, areas and windows are worth serving.
Fleet optimisation
Why a route plan made at 6 a.m. is wrong by 9, and why re-running it is not enough
Information arrives late. The freedom to act on it disappears early.
At six the planner faces an empty board: every stop is a free decision. By nine, three stops per vehicle have been served, two more have been promised to customers inside a stated window, and each driver has been shown a sequence. Served stops cannot be revisited; promised stops can move only by breaking a commitment someone relies on. What remains is a tail — and it shrinks through the day, so the capacity to absorb a new order falls exactly as new orders arrive.
How much this matters is a property of the business, not the software. A wholesale operation with every order in by 5 a.m. genuinely can plan once; an e-commerce operation with a noon same-day cut-off cannot. Plot your order confirmation times against dispatch time: the share that matures afterwards is the ceiling on what continuous re-planning can recover — measurable from your own timestamps in an afternoon, before talking to any vendor.
Commitment horizon
The frozen prefix: how far ahead a route should be committed
Freeze too little and drivers get churn they cannot follow. Freeze too much and new orders are stranded.
At each re-plan a vehicle's stops split into executed, committed and free. Only the free tail is re-optimised; committed stops are those whose planned arrival falls inside the commitment horizon. Updated sequences reach drivers through the driver app, and a driver never sees a stop they have passed reappear.
optimal horizon ≈ √( churn cost ÷ (late orders per hour × value of absorbing one) )
Stranding cost rises with the horizon, because more late orders miss the plan. Churn cost falls with it, because fewer communicated commitments are broken. The forms are stylised; the shape — a trade, not a setting — is what matters.
Worked example: a horizon from one day of order timestamps
Dispatch at 06:00, deliveries close at 18:00. 340 orders are confirmed before dispatch and 96 mature through the day — about 8 an hour — each worth a mean consolidation saving of 14. Churn is estimated at 450. Optimal horizon: √(450 ÷ 112) = 2.0 hours. The one-shot alternative is a 12-hour horizon: all 96 late orders stranded, costing 1,344 in lost consolidation, against a churn saving of 37.5.
The horizon belongs to the operator, not the optimiser. Stranding is a cost software can compute; churn is borne by drivers and customers, and no optimiser can see the point at which drivers stop following sequences. A driver quietly re-sequencing by their own judgement invalidates every later re-plan — which is why plan-against-actual comparison is part of the system.
Savings-aware waiting
Waiting is worth exactly the consolidation it might capture
So a remote route should be committed early and a dense one held open.
saving from joining stops i and j = cost(depot→i) + cost(depot→j) − cost(i→j)
The Clarke–Wright savings measure (1964). It is large when two stops are close to each other and far from the depot, and near zero otherwise.
Worked example: two late orders, both 40 km out
At 0.90 per km, stop P sits 3 km from an existing stop on a dense route: saving = 36 + 36 − 2.7 = 69.3. Stop R sits 55 km from the nearest stop on a rural route: saving = 36 + 36 − 49.5 = 22.5. The dense-cluster order is worth three times as much to absorb.
Holding both routes open for the same two hours pays the same churn on each and buys three times the value on one. A per-route horizon set from expected savings keeps the dense route open longer and commits the rural route at dispatch — capturing more of the valuable consolidations while disturbing fewer drivers, using geometry the planner already has.
Route constraints
Constraints that are not preferences: driver hours, cold chain, hazard
A route that breaks one of them is not a worse route. It is one that cannot be run.
Five constraints sit inside the route model rather than being checked after the plan is released: vehicle capacity, driver hours, customer time windows, cold chain and hazardous goods. A check applied after publication is a check somebody has to remember.
Driver hours
Regulation (EC) No 561/2006 caps daily driving at nine hours, extendable to ten twice a week, with a 45-minute break after 4½ hours. A route beyond that is one the driver may not legally complete.
Cold chain
A temperature excursion does not degrade a chilled consignment, it destroys it — so a sequence leaving it on board past tolerance is infeasible, not expensive.
Hazardous goods
Restricts which consignments may share a vehicle; it removes options rather than penalising them.
Multi-depot operation widens the same decision. An order is served from whichever depot has a vehicle with the capacity, constraint fit and room in its remaining tail — not necessarily the nearest depot.
Cost per stop
Optimise distance, but manage the fleet on cost per stop
The objective an optimiser minimises is not the metric the business should manage.
route cost = fixed vehicle cost + distance rate × km + hourly rate × (driving time + service time) · cost per stop = route cost ÷ stops
Worked example: two routes, one vehicle, opposite conclusions
A vehicle costs 120 a day fixed, 0.90 per km and 38 per loaded hour. Route A: 9 rural stops, 420 km, 6.5 h driving, 12 minutes per stop — 813 in total, 90.3 per stop. Route B: 34 urban stops, 140 km, 3.2 h driving, 14 minutes per stop — 668 in total, 19.6 per stop.
Route B travels a third of the distance and costs 82% as much: a mileage report flatters it enormously and a total-cost report only modestly. Cost per stop says what matters — those nine rural customers cost four and a half times as much to serve, which is a pricing and territory question, not a routing one. And Route B's largest cost is 7.9 hours of service time at 34 doors, which no distance optimisation touches.
Distance is the right optimiser objective: well-behaved, and the stops are largely given on the day. Cost per stop — reported by route, driver and customer — is the right business metric, because it decides which areas, windows and minimum order values are worth accepting. The failure it prevents is a fleet reporting falling mileage every year while its margin erodes.
Evaluating fleet optimisation software
Eight questions to put to any fleet routing software
Each is answerable by demonstration — including by us.
| # | Dimension | Ask the vendor to demonstrate |
|---|---|---|
| D1 | Hard constraints | Build a route exceeding a driver's legal hours. Is it produced, warned, or refused? |
| D2 | Explicit horizon | What is the commitment horizon, and can an operator change it? |
| D3 | Prefix respected | Re-plan mid-day. Are executed and communicated stops treated as fixed? |
| D4 | Churn reported | How many communicated stops did yesterday's re-plans move? |
| D5 | Savings-aware waiting | Is a remote route held open as long as a dense one, or committed earlier? |
| D6 | Depot choice | Is an order assigned to the nearest depot, or the one that can actually serve it best? |
| D7 | Cost per stop | Report cost per stop by customer, not just by route. Can it? |
| D8 | Plan against actual | Does the system compare the executed sequence with the planned one, and flag divergence? |
D8 catches the failure nobody sees. A driver re-sequencing by their own judgement invalidates every subsequent re-plan, and a system that never compares plan with actual will not notice.
On evidence: no published outcome is attributed to this module. It appears in a multi-client 3PL deployment whose stated last-mile problem was routes planned once at 6 a.m. and fuel cost per kilogram rising for two quarters — a description of the problem, not a measured improvement. MileSoft's return model assumes a 12% utilisation gain on a 20-vehicle fleet; that figure is only reachable where a meaningful share of orders matures after the first plan.
Standards referenced: Regulation (EC) No 561/2006, Incoterms 2020, ISO 28000, ISO/IEC 19987 and 19988, GS1 SSCC, ISO 22400-2, EN 16798-3 and cold-chain practice. For fleet optimisation explained, see fleet optimization; derivations are in the fleet optimisation method paper.
Features
Key Features
Built-in capabilities that set our software apart
Continuous re-planning
routes re-optimised on a rolling horizon as orders mature, windows shift and vehicles run late
Explicit commitment horizon
executed and communicated stops stay fixed; only the free tail is re-sequenced
Constraints in the model
vehicle capacity, driver hours, time windows, cold chain and hazardous goods
Multi-depot assignment
by vehicle availability, load fit and remaining route, not proximity alone
Driver application
re-sequenced stops pushed with navigation; dispatchers approve or override on the same map
Cost per stop
fuel, distance, time and utilisation attributed by route, driver and customer
Plan against actual
the executed sequence compared with the planned one
Benefits
Key Benefits
What your team gains from adopting our platform
Testimonials
What Our Clients Say
Trusted by industry leaders across manufacturing, HVAC, and logistics.
"The software features, such as real-time monitoring and alerts, have proved invaluable in optimizing machine operations. MileSoft team's professionalism, technical expertise, and dedication to meeting our requirements have been impressive."
FAQ
Fleet Optimization & Routing Software — FAQs
Answers to the most common questions about this product. Need more? Book a demo or reach out via Contact.
Why is a 6 a.m. route plan no longer enough?
What is a commitment horizon?
How do route changes reach drivers?
Can it handle multi-depot operations?
Which cost metric should we track?
What does fleet optimization software actually change?
Explore More
Related Products
Complementary tools teams often pair with this solution
Related Reading
Go deeper
Explore the long-form story behind this product.
The Frozen Prefix: Commitment Horizons, Order Maturation, and Why Cost per Stop Is the Only Last-Mile Metric
Architecture, Computational Methods, and a Modelled Scenario for MileSoft Fleet Optimization
Read the paperRead the full Fleet Optimization & Routing Software for Plant Logistics deep-dive
Read articleSee how MileSoft delivers in Warehouse
Read case studySee your savings in under a minute
Estimate the time and cost this tool could give back — free, instant, no sign-up.
Size the opportunity from your own order timestamps
Share a week of order confirmation times and routes. We will show how many orders matured after dispatch, the commitment horizon that fits, and cost per stop by customer.