Mining Equipment Benchmarking Studies: Which KPIs Enable Fair Comparisons?

A haul truck can report higher tonnes per hour because it travels a shorter route, carries easier material, or excludes queuing from its operating time. A crusher can appear more efficient because it produces a coarser product. When those figures enter an equipment evaluation without adjustment, the apparent winner may be the wrong machine for the intended duty.

Fair mining equipment performance benchmarking studies need a linked set of KPIs: quality-adjusted productivity, physical availability, utilization, failure frequency and repair duration, energy intensity, and lifecycle cost per accepted unit of output. These metrics become comparable only when time definitions, operating conditions, measurement boundaries, and output requirements are aligned. Safety and compliance should remain acceptance gates, not disadvantages that higher production can offset.

Define the comparison before selecting the KPIs

Start with the decision the study must support. Selecting a truck for an existing haul circuit is different from comparing complete truck-and-loader systems. The first assessment may hold loading infrastructure constant; the second must capture loader matching, dispatch, road capacity, and downstream congestion. Mixing these boundaries makes an equipment ranking difficult to interpret.

Write a comparison statement that identifies the equipment class, required task, output specification, operating envelope, and measurement period. For hauling, this includes payload limits, route distance, elevation change, road condition, and dump arrangements. For crushing, it includes feed characteristics, reduction ratio, product size distribution, and acceptable product quality.

Separate three questions that often become conflated:

• Capability: What can the machine deliver under controlled, matched conditions?
• Site performance: What does it deliver within the actual production system?
• Economic suitability: What does that output cost over the planned ownership period?

A controlled trial can establish capability but cannot, by itself, establish long-term reliability. Conversely, fleet records may describe realistic operating performance while concealing differences in equipment age, maintenance support, and assignment difficulty.

A KPI set that exposes trade-offs

No single metric provides a defensible overall ranking. Use a compact scorecard in which every numerator and denominator has a documented definition. The following metrics form a practical starting point; additional measures should reflect the equipment’s specific task.

These are working definitions, not universal conventions. A supplier’s “availability” or “MTTR” may use a different basis. Preserve the original reported value, then calculate a harmonized value from underlying records where possible. If the records cannot support harmonization, identify the metric as non-comparable rather than silently treating it as equivalent.

Productivity: specify what counts as useful output

Tonnes per operating hour describes production rate while the machine is operating. Tonnes per scheduled hour captures the combined effect of availability, utilization, and operating productivity. Report both when evaluating practical delivery capacity; otherwise a fast machine with frequent stoppages may look stronger than it is.

Under mutually consistent time definitions, scheduled-hour productivity can be decomposed as:

Output ÷ scheduled time = availability × utilization of available time × output per operating hour.

The relationship only holds when the time states reconcile and use the same study boundary. “Operating time” must not mean engine-on time in one dataset and productive-cycle time in another.

Hauling and excavation

For haul trucks, pair tonnes per hour with actual payload, cycle-time components, and payload compliance. Record loading, loaded travel, dumping, empty return, and queuing separately. Overloading should not be normalized into a performance advantage: it changes component stress and may breach operating limits.

For excavators, distinguish mass output from loose-volume output. Material density, fragmentation, bucket fill, digging resistance, and swing angle affect the result. A conversion based on nominal bucket capacity alone does not establish comparable production.

Crushing and processing

Crusher throughput needs a matching product specification. Report feed size distribution, hardness or relevant comminution characteristics, moisture, and product size. Higher throughput obtained by relaxing the required product size is not an equivalent result.

In processing equipment, accepted output may require quality or recovery criteria rather than raw feed tonnes. Where equipment affects recovery, report throughput and recovery separately as well as the resulting accepted output. This prevents increased feed rate from concealing greater losses.

Availability and reliability need a common event clock

A machine awaiting a replacement component can be classified as maintenance downtime, logistics delay, or standby in different reporting systems. Those choices materially change availability and repair metrics without changing the physical event.

Build mutually exclusive time states covering scheduled time, maintenance downtime, available-but-not-operating time, and operating time. Within downtime, distinguish planned work, corrective work, diagnosis, parts waiting, and other delays. Retain total downtime even when reporting narrower maintenance measures.

MTBF should use operating exposure, not an unexplained calendar interval. Define a qualifying failure before collecting data: for example, an event that prevents a required function and requires corrective intervention. Record repeat faults, partial functional losses, and software resets consistently. Excluding inconvenient event types from only one fleet biases the comparison.

Mean repair time needs equal care. Active hands-on repair time measures a different problem from total elapsed restoration time. The former helps assess maintainability; the latter captures the production consequences of diagnosis, access, staffing, and spare-parts support. Report both when the distinction affects equipment selection.

Means can conceal operationally important extremes. Include failure counts, exposure hours, downtime distributions, and major-event descriptions. A short trial with no observed failures does not demonstrate infinite MTBF or proven durability; it provides limited evidence over the observed exposure.

Normalize duty without erasing the operating reality

The strongest comparison uses matched assignments or controlled crossover trials, where machines perform comparable work across similar periods. When that is impractical, stratify records into comparable duty groups before calculating adjusted results.

• Material: density, fragmentation, abrasiveness, moisture, and relevant strength or grindability characteristics.
• Route and resistance: loaded and empty distance, elevation profile, rolling resistance, road condition, and traffic constraints.
• Machine condition: accumulated hours, configuration, attachment, tires, maintenance status, and control settings.
• Operating context: shift, operator or automation mode, ambient conditions, loading match, and production demand.

For haulage, tonne-kilometres can supplement energy per tonne, but distance alone does not capture gradients, empty returns, or acceleration losses. An energy comparison based only on tonne-kilometres may still favor a less demanding route.

Statistical adjustment can help when datasets contain overlapping conditions. Use a model whose inputs have physical relevance, disclose its assumptions, and show observed results alongside adjusted results. Do not extrapolate a machine into duty conditions absent from its dataset and present that estimate as measured performance.

Keep two views: matched-duty performance for equipment comparison and actual-site performance for operational planning. Normalization should remove assignment bias, not conceal poor integration with the intended mine.

Energy and cost boundaries can reverse the ranking

Energy intensity should use the same accepted-output denominator as productivity. Include idling and auxiliaries if they fall inside the declared operating boundary. For electrically powered equipment, distinguish onboard energy consumption from electricity measured at the charging or supply point; conversion and charging losses make these different quantities.

Diesel litres per tonne and electricity kilowatt-hours per tonne are useful operational measures, but they are not directly interchangeable. A cross-technology comparison needs a disclosed energy conversion basis. Cost and greenhouse-gas comparisons additionally require explicit fuel prices, electricity tariffs, and emissions factors. Tailpipe-only accounting should not be compared with broader energy-supply accounting.

Lifecycle costing should capture acquisition, installation or enabling infrastructure, energy, maintenance labor, consumables, major replacements, and disposal or residual value. Allocate shared infrastructure transparently. For battery-electric fleets, replacement assumptions and charging capacity may affect both cost and achievable output.

Use a common economic horizon and discounting convention. Dividing discounted costs by undiscounted tonnes produces a different metric from a consistently discounted unit-cost calculation; label the convention and use it across alternatives.

Separate direct equipment costs from modeled production-loss consequences. Otherwise downtime can be charged twice: once through reduced output in the denominator and again through an inadequately defined lost-production charge. Test sensitivity to utilization, energy prices, component life, and residual value rather than presenting one cost estimate as certain.

Evidence quality determines whether the ranking is usable

Before calculating results, reconcile telemetry, dispatch, maintenance work orders, payload records, and energy measurements. Check clock synchronization, missing intervals, sensor calibration, duplicate events, and inconsistent asset identifiers. Explain exclusions and report how much exposure or output they remove.

A useful study reports distributions and uncertainty, not just fleet averages. Unequal sample sizes, repeated observations from the same machine, and clustered shift conditions can make apparent differences less convincing than they appear. Resampling or statistical methods should respect those dependencies rather than treating every cycle as independent evidence.

Standards support measurement discipline but do not automatically make operating results comparable. Identify the applicable measurement method, edition, and scope. ISO 50015, for example, addresses measurement and verification of energy performance; it does not establish a universal mining-equipment ranking. Equipment safety standards and jurisdiction-specific mining requirements must be checked separately for applicability.

The evaluation record should connect each reported KPI to its data source, formula, exclusions, and normalization method. Supplier summaries without this traceability can inform initial screening, but should not carry the same evidential weight as auditable operating records.

Turn the scorecard into a defensible selection

Apply mandatory constraints first: required output quality, permissible payload, operating-envelope compatibility, and applicable safety requirements. Rank only the alternatives that satisfy those constraints. Then compare scheduled-hour output and lifecycle unit cost, using reliability and energy metrics to explain the differences.

Where rankings change under plausible assumptions, specify the evidence needed to resolve the decision. That may be a matched-route trial, longer failure exposure, verified product-size measurements, or a maintenance-cost review. Define acceptance criteria before testing. A technically sound benchmark may conclude that two machines cannot yet be distinguished reliably; that is more useful than a precise-looking ranking built from incompatible data.