INDUSTRY BRIEFING · BATTERY-ELECTRIC HAULAGE
Development of a Dispatching Model for Battery-Trolley Haulage Systems
Cutting the two biggest costs in electrified mine haulage — charger congestion and premature battery replacement — inside a single dispatch decision.
Mohammad Mirzehi
·
PhD Candidate, The University of Queensland
·
Supervisors: Prof. Mehmet Kizil · Prof. Peter Knights · Dr Micah Nehring
·
INDUSTRY BRIEFING · BATTERY-ELECTRIC HAULAGE
Development of a Dispatching Model for Battery-Trolley Haulage Systems
Cutting the two biggest costs in electrified mine haulage — charger congestion and premature battery replacement — inside a single dispatch decision.
Mohammad Mirzehi
·
PhD Candidate, The University of Queensland
·
Supervisors: Prof. Mehmet Kizil · Prof. Peter Knights · Dr Micah Nehring
·
01 · WHO I AM
A practitioner's view — not just a researcher
Four years working across open-pit mine sites before bringing that operational lens to electrified haulage as a PhD candidate at The University of Queensland.
Short- to long-term planning — building mine plans from shift schedules through to life-of-mine strategy.
Haulage operations management — running truck–shovel systems against live production targets on site.
Dispatch data analysis — turning raw dispatch output into decisions operators can act on.
Productivity & cost optimization — pushing throughput and operating cost in the same direction, not against each other.
Contents
01
02
The shift to electrification
Our project
03
Modelling battery degradation
04
Dispatching the fleet
01
The shift to electrification
01 · CONTEXT
Electrification has moved from aspiration to necessity
OEMs are shipping commercial battery-electric haul trucks, carbon pricing is tightening in key mining regions, and ESG commitments are hardening into procurement decisions. Haulage — not processing — is where most surface mines will make or break their decarbonization targets, which is why the diesel-to-battery transition is now a board-level question rather than a pilot-project curiosity.
40–60%
of surface-mining energy use is haulage
#1
Scope-1 emitter at surface mines
3–5 yr
battery pack life under mining duty, before replacement
The hardware transition is already underway. The open question — and the one this project answers — is how a site actually dispatches an electrified fleet once the trucks arrive: how it schedules charging, swapping and trolley assist without either starving production or burning through battery packs faster than budgeted.
01 · THE PROBLEM (1/2)
Electrification trades a fuel bill for a replacement bill
Battery packs under intensive mining duty cycles fail years before the truck does — replacement is an ongoing expense that decides whether electrification pencils out.
Harsh operating environment
High ambient and cell temperatures accelerate ageing — heat compounds every other stress factor on the pack.
Demanding duty cycles
Deep discharges dominate wear: depth of discharge is the strongest and most nonlinear driver of degradation under mining duty.
01 · THE PROBLEM (2/2)
Today's tools weren't built for electrified haulage
Commercial dispatch software is inflexible — it can't jointly represent chargers, swap bays, trolley sections, state-of-charge limits and infrastructure capacity in one framework.
Battery life is analysed separately from operations — degradation is evaluated after the fact, and charging is treated as fixed rules rather than decision variables the dispatcher can tune.
The result: decisions that look optimal in the moment can quietly wear the battery faster, raising long-term replacement cost while the schedule reports a productive shift.
02
Our project
02 · OUR PROJECT
Simultaneously optimizing production and battery life
We simultaneously schedule charger bays, swap stations, trolley sections and trip sequencing across the whole fleet, respecting the physical infrastructure actually on site. In parallel, a battery-degradation model quantifies how each dispatch and charging decision affects battery life — and that cost is fed directly into the dispatch objective, so production and pack longevity are optimized together, not in separate steps.
This directly attacks the two biggest cost drivers in electrified mine haulage: productivity loss from charger congestion, and premature battery replacement from poorly managed duty cycles. Both are treated as one decision, solved once, rather than two separate reviews after the shift.
02 · OUR PROJECT
One model that unites dispatch, battery life & charging
01 · Dispatch across every charging mode
Stationary charging, battery swapping and dynamic (trolley) charging — allocation, routing and trip sequencing in a single framework.
02 · Degradation cost inside the decision
Model and evaluate battery ageing, then feed its cost into the objective so wear is weighed in real time, not reviewed afterwards.
03 · Optimized charging & infrastructure
Size and place chargers, swap bays and trolley sections to match the fleet — infrastructure becomes a design variable, not a fixed input.
03
Modelling battery degradation
03 · MODELLING BATTERY DEGRADATION
Rainflow: counting the cycles that age the pack
Borrowed from metal-fatigue analysis: treat the state-of-charge trace like a stress history, decompose it into depth-tagged cycles, and weight damage by depth across the entire trajectory — not instant by instant.
1
SoC trace
The input signal, from real or simulated operating data.
→
2
Cycle extraction
Turning points identified and decomposed into depth-tagged charge/discharge cycles.
→
3
Depth-weighted life loss
Cycles weighted by depth of discharge across the whole trajectory, not a single instant.
→
4
Degradation cost
Translated into a cost per shift, ready to feed the dispatch objective.
03 · MODELLING BATTERY DEGRADATION
Comprehensive by design — and cross-validated
Rainflow core
Depth-weighted cycle damage across the full state-of-charge trajectory — the whole history, not a snapshot.
Blind spots covered
Paired with an empirical multi-factor model adding temperature and C-rate — the stresses rainflow alone can miss.
Data-driven refinement
Real and simulated SoC data calibrate the method to site conditions and improve its accuracy.
Running the physics-based and data-driven methods side by side gives a degradation estimate we can validate — not a single unchecked number.
04
Dispatching the fleet
04 · DISPATCHING THE FLEET
A physics-informed MILP for the whole fleet
Every truck's trip, source assignment and charging choice is a decision variable — not a fixed schedule imposed in advance.
Trucks compete for loading and unloading just as they compete for charging stations and trolley sections, so congestion in either system is captured by the same model.
Physical constraints — charger counts, bay capacity, trolley coverage, state-of-charge limits — bound every feasible solution, so the schedule is one the site can actually run, not a theoretical optimum that ignores infrastructure.
Formulated as a mixed-integer linear program (MILP), solved on a rolling horizon so dispatch decisions can be refreshed as conditions on site change.
04 · DISPATCHING THE FLEET
Balancing today's tonnes against tomorrow's battery
MAXIMIZE
short-term production
−
MINIMIZE
long-term battery replacement cost
=
MILP OBJECTIVE
solved once, every dispatch cycle
Degradation results enter the objective as a cost term, so the optimizer weighs the trade-off automatically — instead of treating battery wear as something reviewed after the shift.
04 · DISPATCHING THE FLEET
Every charging mode, one framework
Stationary charging
Trucks charge at fixed bays during queuing or idle time — the baseline strategy most fleets already use.
Battery swapping
A depleted pack is exchanged for a charged one at a swap station, trading downtime for deeper, more damaging cycles.
Dynamic (trolley) charging
Trucks draw power from trolley lines on loaded ramps, cutting depth of discharge without stopping to charge.
04 · DISPATCHING THE FLEET
Infrastructure becomes a design variable
Charger and swap-bay counts, placement and trolley-section coverage are sized to the fleet, not assumed fixed in advance.
The model captures competition for these shared resources alongside loading and unloading — the same congestion that limits productivity today.
This turns infrastructure planning into an output of the model rather than an input assumption — useful for sizing a new electrified fleet or expanding an existing one.
The dispatch logic is being cross-checked in HaulSim, an independent discrete-event simulator, alongside the MILP solver itself, to validate behaviour before it is proposed for a real site.
INDUSTRY BRIEFING · BATTERY-ELECTRIC HAULAGE
Thank you
A haul-truck battery runs out of life, not just charge — we're building the dispatch model that plans for both.
Deeper material available on request: full research proposal · battery-degradation conference deck · fleet-simulation movie.
Mohammad Mirzehi
·
m.mirzehi@uq.edu.au
·
The University of Queensland