Construction digital twin: haul cycles, truck counts and cost per m³
- A construction digital twin reproduces haul cycles from real site data so you can test fleet decisions before applying them.
- A 3D model describes the geometry of the ground, while a digital twin describes how the fleet works and what each m³ costs.
- The match factor compares truck arrival rate with the loader's loading rate; above 1, trucks queue at the loader.
- Past the balance point, each extra truck adds hourly cost without adding output, so cost per m³ goes back up.
- A twin is only reliable once it has been calibrated against measured cycle times, idle time and payloads.
A construction digital twin lets you test fleet decisions before you make them on site: how many trucks to assign to an excavator, what speed limit to set on the haul road, whether engines should be shut off in the queue. It reproduces the haul cycle (load, haul, dump, return) from real GPS positions and engine data, then calculates output in m³/h and cost per m³ for each scenario.
What is a construction digital twin, in practical terms?
ISO/IEC 30173 defines a digital twin as a "digital representation of a target entity with data connections that enable convergence between the physical and digital states at an appropriate rate of synchronisation" (Designing Buildings Wiki, on ISO/IEC 30173). For an earthmoving contractor, two ideas matter: the twin is fed by data from the site, and it has to stay in step with what actually happens there.
Digital twin or 3D model: what is the difference?
The same article sums up the ISO position: a digital twin is about exchanging relevant information, and its representation can be graphical (a 3D model) or conceptual (a diagram), depending on the subject. A 3D terrain model describes the volumes to be cut. A haul cycle twin describes how the fleet will move them, how fast and at what cost.
| Criterion | 3D model (DTM, BIM model) | Haul cycle digital twin |
|---|---|---|
| Question answered | What to build, what volumes | How and how fast to move them |
| Inputs | Surveys, drawings | GPS, engine hours, idle time, fuel, payloads |
| Updates | With each new survey | Continuously, cycle by cycle |
| Outputs | Geometry, quantities | m³/h, queue at loader, excavator utilization, cost per m³ |
| Typical use | Design, quantity takeoff, grade control | Fleet sizing, testing site rules |
| Field check | Comparison with a new survey | Comparison of simulated and measured cycle times |
What makes up a haul truck's cycle time?
A cycle starts when the truck arrives at the loading area and ends when it gets back there empty. Mining literature breaks it into eight terms (Mohtasham, Journal of Mining and Environment, 2025):
- waiting at the loader (the queue);
- manoeuvring into position at the loader;
- loading;
- loaded travel to the dump point (distance divided by speed);
- waiting at the dump;
- manoeuvring at the dump;
- dumping;
- empty return.
Durations vary widely between sites. On an urban earthwork job in Montreal tracked by GPS, a sample cycle took 33.75 min, including 7.75 min loading, 10 min hauling and 8 min returning (Alshibani, Buildings, 2018). Haul and return times depend on road length and the posted speed. Loading sets the maximum pace.
What is the truck-shovel match factor?
The match factor (MF) compares the rate at which trucks arrive with the rate at which the loading unit can serve them. In the formulation used by Burt (PhD thesis, Curtin University, 2008) and attributed to Morgan and Peterson (1968):
MF = (number of trucks × time to load one truck) ÷ (number of loaders × average truck cycle time excluding waiting)
- MF < 1: the excavator waits for trucks, and hauling limits output.
- MF = 1: theoretical balance, neither excavator nor trucks wait.
- MF > 1: trucks queue, and the excavator limits output.
Burt adds two useful caveats. An MF of 1 gives maximum output, which is not necessarily the lowest-cost point. And a theoretical MF of 1.0 does not guarantee an actual MF of 1.0, because trucks tend to bunch up on the haul road.
Why does output plateau when you add more trucks?
Here is a worked example. The values are hypothetical and only there for the arithmetic:
- one excavator, 3 min to load a truck, so at most 20 loads/h;
- 12 m³ trucks, 15 min cycle excluding waiting;
- hourly costs: excavator 900 MAD/h, truck 400 MAD/h.
4 trucks. MF = 4 × 3 ÷ 15 = 0.8. The fleet makes 16 trips/h, or 192 m³/h. The excavator works 48 min out of 60 (80%). Cost: 2,500 MAD/h, about 13.0 MAD/m³.
5 trucks. MF = 1.0. The excavator reaches 20 loads/h, or 240 m³/h, for 2,900 MAD/h: about 12.1 MAD/m³.
6 trucks. MF = 1.2. The excavator still loads 20 trucks an hour, so output stays at 240 m³/h. The cycle stretches to 18 min, including 3 min in the queue per truck. Cost rises to 3,300 MAD/h, or 13.75 MAD/m³.
The sixth truck adds 400 MAD/h and zero extra cubic metres. This deterministic calculation is still optimistic. On a real site, breakdowns, uneven loading and bunching create waiting even with an MF below 1. A discrete event simulation study (Kostyuk et al., AusIMM, 2014) shows that with even small process variation, the fleet that maximizes productivity and value has, on average, some trucks queuing at the shovel, and that this optimal fleet grows as variation and unscheduled downtime increase. A queueing theory analysis (Kaboli et al., Journal of Mining and Environment, 2020) also shows that the fleet size that minimizes cost per unit of production also minimizes emissions per unit of production.
Which what-if scenarios can you test with a digital twin?
Number of trucks
Vary the fleet and read off m³/h, queue length and cost per m³. The right number meets the production target at the lowest unit cost, given the variability measured on your site.
Haul road speed limit
Suppose the road in the example is 2 km long. The loaded trip takes 4 min at 30 km/h and 6 min at 20 km/h, and the same goes for the return, so the cycle grows from 15 to 19 min. With 5 trucks, MF drops to 5 × 3 ÷ 19 = 0.79 and the excavator starts waiting again. A lower limit, set for safety or because of road conditions, therefore changes the optimal truck count. The twin puts a number on that effect before the sign goes up.
Engine-off while queuing
In the 6-truck example, the fleet accumulates 20 trips × 3 min = 60 min of queuing per hour, which is one engine hour of idling per working hour. Argonne National Laboratory data published by the U.S. Department of Energy put idle consumption at 0.44 gal/h for a diesel medium heavy truck (U.S. DOE), about 1.7 L/h. Caterpillar states that most construction equipment burns an average of one gallon an hour at idle (For Construction Pros), about 3.8 L/h. Over a 10-hour shift, that is 17 to 38 L of diesel with no m³ moved, plus engine hours that count toward maintenance. For more on this, see our article on reducing diesel consumption in a construction fleet.
What data does a digital twin need?
A credible twin is built on measurements taken on your own site:
- Time-stamped GPS positions for trucks and excavators, with geofences around loading and dumping areas. This is the method used in the Montreal case: a point within a zone's radius marks loading or dumping, and everything else is hauling or returning (Alshibani, 2018).
- Engine data in ISO 15143-3 (AEMP) format: cumulative operating hours, idle hours, fuel used, locations, load counts and cumulative payload totals (Proemion ISO 15143-3 documentation).
- Actual payload per trip, where available. In a 1:50 scale model test, productivity estimated from GPS alone showed a 6.05% mean absolute error, against 2.20% with payload sensors, the gap coming from the assumption that trucks are always full (Ibrahim and Moselhi, ITcon, 2014).
- Actual hourly cost for each machine, in MAD or CAD.
- Haul road geometry: distances, grades, speed limits.
GeoChantier is being designed to collect much of this data from GPS units or manufacturer telematics. The platform is still under development.
How do you calibrate a twin against field data?
Calibration follows a simple loop:
- Measure each cycle component over several shifts and keep the full distribution (minimum, median, maximum).
- Feed those distributions into the simulation, with the current fleet and haul road.
- Compare simulated output, cycle time and queue length with values measured over the same period.
- Adjust the parameters that are off, then repeat after every change: new haul road, new face, new machine.
Step 3 matters. In the Montreal case, the GPS-based estimate and a probabilistic simulation model differed by −26% to +20% depending on fleet size, between 2 and 14 trucks (Alshibani, 2018). An uncalibrated model can lead to the wrong fleet decision.
In GeoChantier, this kind of analysis is planned to feed costed recommendations: reassign a machine, shorten a loading queue, cut idling.
Where GeoChantier stands
GeoChantier plans a haul-cycle digital twin, but it is not available on this site yet. To follow the pilot programme, tell us about your fleet or write to contact@canalytics-group.com.
Frequently asked questions
What is a digital twin used for on an earthmoving project?
It lets you test fleet decisions before applying them on site: trucks per excavator, haul road speed limit, engine-off while queuing. Fed with GPS positions and engine data, it simulates the haul cycles and reports, for each scenario, output in m³/h, the queue at the loader, excavator utilization and cost per m³ in MAD or CAD.
What is the difference between a digital twin and a 3D model?
A 3D model describes geometry: existing ground, design surface, volumes to cut. A digital twin has data connections to the real site that keep it in step with what is happening, as defined in ISO/IEC 30173. For earthmoving, it reproduces truck and excavator cycles and calculates the resulting output and cost.
How do I calculate how many trucks an excavator needs?
Use the match factor: number of trucks times the time to load one truck, divided by number of loaders times the average truck cycle time excluding waiting. A value of 1 is the theoretical balance. In practice, breakdowns and truck bunching shift the optimum, which is why a simulation calibrated on your own measurements is useful.
What data do you need to build a fleet digital twin?
You need time-stamped GPS positions for each machine, geofences around loading and dumping areas, engine data (operating hours, idle hours, fuel used), ideally payload per trip, haul road geometry and the actual hourly cost of each machine. ISO 15143-3 standardizes much of the telematics data that manufacturers expose.
How do you check that a digital twin is reliable?
Compare its results with field measurements over the same period: cycle time by component, output in m³/h and queue length. Adjust the parameters until the gaps are acceptable, then recalibrate after every significant change on site, such as a new haul road, a new working face or a new machine joining the fleet.
Sources
- An Optimisation Approach to Materials Handling in Surface Mines (C. N. Burt, PhD thesis, Curtin University, 2008)
- Development of a Match Factor for Optimizing Haulage Fleet Efficiency in Open-Pit Mining Operations (Journal of Mining and Environment, 2025)
- Is Truck Queuing Productive? (Kostyuk et al., AusIMM, 2014)
- Optimal Earthmoving Fleet Size for Minimising Emissions and Cost (Journal of Mining and Environment, 2020)
- The ISO answer to what is a digital twin (Designing Buildings Wiki, ISO/IEC 30173)
- ISO-15143-3 (AEMP) documentation (Proemion)
- Automation of Measuring Actual Productivity of Earthwork in Urban Area, a Case Study from Montreal (Buildings, 2018)
- Automated productivity assessment of earthmoving operations (ITcon, 2014)
- Fuel Consumption at Idle for Selected Gasoline and Diesel Vehicles (U.S. Department of Energy)
- Caterpillar: save a gallon an hour on the jobsite by reducing idle time (For Construction Pros)
Join the GeoChantier pilot programme
GeoChantier is under development. Tell us about your fleet so we can prepare the pilot.
Book a demo