Logistics Simulator

Simulate first,
then dispatch

Four simulations along the rolling operation. The initial values are examples: set the sliders to the scale of your fleet. The calculation paths sit beneath each simulation.

From analysis to a testable simulation

A freight carrier rarely loses its money in one blow. It loses it rolling: with every empty kilometre, every hour at the ramp, every driver who quits. Three stages turn the rolling loss into a calculation.

Stage 1
Analysis
Looks back and shows where kilometres and hours remain today: in empty runs, waiting times and departures.
Stage 2
Forecast
A model per lane, ramp and team states which week, which waiting time, which departure becomes likely, with a stated error instead of gut feeling.
Stage 3
Simulation
You change a control variable and see what the intervention would lead to, before it touches the fleet.
Control

The fourth stage happens inside the operation: no model dispatches a route, negotiates a time window or holds a staff conversation. After introduction of a measure, the simulated trajectory is compared with telematics, dispatch and personnel data. What the measurement refutes gets replaced.

What one percentage point of empty-run ratio costs

Every empty kilometre causes variable costs. At the initial values of this page, just two percentage points less already carry six figures. You set the achievable reduction yourself. Only the consequences are calculated.

Every slider is an assumption, not a fact. All initial values are fictional examples: no customer data, no industry figures, not the values of the sample carrier from the six modules.

Annual km of the fleet
Empty-run ratio today
Variable cost per empty km
Achievable reduction (percentage points)
Savings per year over the reduction (marker: set value)
Empty km today
per year
Empty-run costs today
per year
Kilometres saved
per year
Savings per year
at variable costs

Calculation: empty km = annual km × ratio. Empty-run costs = empty km × variable cost per km. Kilometres saved = annual km × reduction. Savings = kilometres saved × variable cost. Deliberately no separate diesel calculation, because the diesel sits inside the variable per-kilometre rate.

Control

The reduction does not arise in the chart but in dispatch: return loads, triangular runs, partner network. The model shows on which lanes empty kilometres arise systematically; which load a truck takes is decided by dispatch.

The hour at the ramp

The simplest calculation on this page: every result can be estimated in your head.

Arrivals with waiting time per year
Avoidable dwell time per arrival
Dwell cost rate
Realisable share
Dwell costs today versus remaining
Avoidable hours
per year
Cost of these hours
per year
Realised hours
per year
Savings per year

Calculation: avoidable hours = arrivals × minutes / 60. Costs = hours × dwell cost rate. Savings = costs × realisable share.

Control

The model forecasts the waiting time per ramp and time window; the conversation with the customer about better time windows is led by dispatch, not by the software.

What a smaller forecast error is worth

Misplanning costs as a distribution instead of a single number: 500 simulated years per scenario, once with the current forecast error, once with the model error. You set both assumptions yourself. The random generator starts from a fixed seed, so every visit shows the same result.

Mean weekly routes
Spread of weekly demand
Forecast error today
Forecast error with model
Misplanning cost per misplanned route
Distribution of annual misplanning costs: 500 years per scenario
Median annual costs today
Median with model
Expensive year (one in ten)
today → with model
Model ahead
out of 100 years each

Calculation: weekly demand normally distributed around the mean, truncated at zero. The planning error per week is normally distributed around zero, and its mean absolute value equals the set forecast error (conversion factor 1.2533 to the spread, fixed value). Annual costs = sum of misplanned routes × misplanning cost over 52 weeks. 500 years per scenario, fixed random seed, independent random streams per scenario: the metric compares self-contained runs. Extreme weeks are only represented as far as the flank of the distribution allows. Because a consistently smaller error accumulates over 52 weeks, the model is ahead in almost every year at a clearly smaller error. At the initial values the metric therefore shows 100 out of 100. As both error assumptions move closer together, it drops.

Control

The forecast does not plan a single route automatically. How much capacity dispatch holds and what a misplanned route costs is your assumption at the slider.

The driver retention funnel

The existential topic of the industry, honestly calculated: two rates dampen multiplicatively, so many departures become few retained drivers. That a small number stands at the end is not a weakness of the calculation but its honesty.

Driver headcount
Turnover per year
Replacement cost per departure
Hit rate of the model
Success rate of the conversations
Cost per measure
From departure to retained driver: the funnel dampens twice
Departures per year
Retained drivers
per year
Measure costs
per year, on all identified cases
Net savings
per year

Calculation: departures = headcount × turnover. Identified = departures × hit rate. Retained = identified × success rate. Savings = retained × replacement cost. Measure costs = identified × cost per measure (on all identified cases, not only the retained ones). Net = savings − measure costs. False alarms among staying drivers are not included, so the actual costs lie above.

Control

The scoring is aggregated at team level, no person-level ranking. Whether and how a conversation is held is decided by fleet management.

What this page assumes: fully disclosed

Every number on this page follows arithmetically from the slider values. In addition:

  • Example values: all initial values are fictional: no customer data, no industry benchmarks and deliberately not the values of the sample carrier from the six modules.
  • Not additive: the four results must not be summed: simulation 01 measures driven empty kilometres, simulation 03 measures planning deviation in routes, and a misplanned route can cause an empty kilometre. The page therefore shows no grand total.
  • Separation from the modules: dwell time effects are calculated only by simulation 02. Tyres and breakdowns do not appear. Simulation 01 values empty kilometres at variable cost per km without a separate diesel calculation. A subcontractor premium is claimed nowhere.
  • Order volume: weekly demand and planning error normally distributed, 52 weeks, 500 simulated years, fixed random seed, independent random streams per scenario, conversion factor 1.2533 from the mean absolute error to the spread. Extreme weeks are only represented as far as the flank of the distribution allows.
  • Driver retention: measure costs run on all identified cases, the savings only on retained ones. False alarms among staying drivers are not included, so the actual costs lie above. The scoring stays aggregated at team level.
  • No shared sliders: the four simulations are deliberately decoupled so that no hidden conversion assumption (such as kilometres per route) arises.
  • Nominal amounts: without costs beyond those named, without discounting, without VAT.
Note on the context of this portfolio

This page is a basis for conversation, not an offer. All values are model calculations based on freely adjustable assumptions; no number is a commitment, a forecast for a specific carrier or an industry figure.

In a real project your data replaces the sliders: your telematics and dispatch data provide the empty-run ratio and waiting times, your order history the forecast error, your personnel data the turnover. The calculation paths stay the same.

All 6 modules: AI in logistics