We took one month of a UK pallet network's own TMS data and re-planned it under the operator's real constraints. Every consignment still delivered, in 41.7% fewer vehicles.
One month · all sites6,117 trucks dispatched
3,565 the freight needed2,552 moving air · −41.72%
A national pallet network, planned the way most networks are planned.
A multi-site UK palletised freight operator running a mixed fleet of 12- and 24-pallet vehicles: thousands of trips a month, scheduled site by site, day by day, under time pressure.
The work described here is a Flow Dynamics diagnostic: a fixed-scope engagement that takes a slice of the operator's own historical data and answers one commercially significant question with evidence rather than opinion.
The operator is anonymised. Every figure on this page comes directly from the diagnostic notebook run against their transport management system. Nothing has been adjusted for presentation.
02 · The question
"Are we running more trucks than the freight requires?"
One month of live operations was pulled from the TMS: 7,121 consignments totalling 66,772 pallets, delivered across 6,117 truck trips. Measured against the capacity of the vehicles actually dispatched, 26.5% of deck space ran empty: 38,890 pallet positions paid for and not used, in a single month.
A number like that invites an obvious objection: of course trucks run part-empty; deadlines force it. So before optimising anything, we tested whether the deadlines were really as tight as the planning assumed.
As operated
26.5% of deck unused
Modelled
9.6% of deck unused
Average deck utilisation across the network, drawn to scale on a 24-pallet trailer. Dashed cells are paid-for space moving air.
03 · The finding that made it possible
The slack the optimisation needed already existed.
We compared every consignment's due date against the date its trip actually started. If the network were genuinely deadline-bound, the two would sit tightly together. They didn't.
6.4 days
Mean gap between due date and actual dispatch. Measured, not assumed
FIG. 1 · Variance between due date and trip start, all sites, one month. Mean 6.39 days, σ 6.91. A conservative 6-day consolidation window was adopted for all modelling.
The headline saving below comes from an algorithm, but the permission for it came from the data. Most networks carry slack like this without knowing, because nobody has measured the gap between what the plan assumes and what operations actually do.
04 · The method
Five steps. Their data, their constraints, no leaps of faith.
1
Extract the baseline
One month of consignment, trip and vehicle data pulled directly from the operator's TMS: due dates, pallet counts, vehicle classes, trip assignments, delivery coordinates. No new systems, no integrations. The data already existed.
2
Clean and benchmark
Invalid rows and out-of-range loads removed; the as-operated month reconstructed exactly as it ran, so the comparison is against reality rather than a convenient straw man.
3
Measure the real tolerance
Due-date-to-dispatch variance quantified across every consignment. The observed mean of 6.4 days was rounded down to a 6-day window, so the model is only allowed flexibility the operation already exercised.
4
Re-plan under constraints
Consignments regrouped using an open-source constraint solver (Google OR‑Tools), respecting vehicle capacity (12- and 24-pallet classes), geographic proximity of delivery points, and the 6-day window. Every consignment must be delivered; nothing is dropped or deferred beyond its measured tolerance.
5
Compare like for like
The modelled month is reconciled against the baseline on consignment count and total pallets, both identical to two decimal places, before any fleet comparison is made.
05 · The result
The empty peak collapses into the full one.
The clearest way to see it is the shape of the network's loading, before and after. The baseline has two peaks: full trucks and nearly empty ones. The modelled plan absorbs the empty peak almost entirely.
FIG. 2 · Pallets per truck, as operated (left) vs modelled (right). Same consignments, same tonnage, same delivery obligations.
Metric
As operated
Modelled
Change
Consignments delivered
7,121
7,121
0.00%
Total pallets moved
66,771.7
66,771.7
0.00%
Truck trips
6,117
3,565
−41.72%
24-pallet trips
2,654
2,686
+1.21%
12-pallet trips
3,436
879
−74.42%
Wasted deck space (pallets)
38,890
8,240
−78.81%
The model barely touches the 24-pallet fleet. Three out of four 12-pallet trips were carrying freight that consolidation could put on trucks already going.
06 · Read it straight
What this does and doesn't claim.
This is a model run on historical data, and we'd rather you read it that way than as a promise.
What it shows
The freight, as it actually occurred, fit in 41.7% fewer trucks under the operator's own measured tolerances.
The flexibility used by the model was observed in live operations, not assumed.
The scale of the opportunity is large enough to survive heavy discounting by real-world constraints.
What it doesn't
It does not account for driver shifts, loading windows or customer-specific commitments. These will erode part of the modelled saving.
It is not an implemented result. Quantifying what survives contact with operations is precisely the job of the next stage.
If your TMS holds twelve months of history, it already holds the answer.
A Flow Dynamics diagnostic runs this analysis on your data, on your infrastructure. Nothing leaves your estate. Fixed fee, fixed scope, invoiced on completion. You end up with evidence, not a sales deck.
