Skip to content

Systems and records

Construction equipment dispatch data: available vs suitable machines

By SourceX Editorial · Reviewed by Noah Loul ·

10 sections

Short answer

Treat availability and suitability as two separate tests. Availability asks whether a machine is free, in service and able to arrive on time. Suitability asks whether it can do this job: rated capacity at the real working radius, ground and access, power line clearance, a qualified operator and a legal haul. Construction equipment dispatch data should store both verdicts for every candidate, with evidence and the reason a free machine was rejected.

Evaluation split

Construction equipment

This page covers construction equipment dispatch data: available vs suitable machines.Illustrative

Key takeaways

  • Availability and suitability are separate tests and should be stored as separate verdicts for every candidate machine.
  • Most dispatch systems log the assignment but not why free machines were rejected, which is the signal an AI model needs.
  • Crane capacity, ground, power line, operator and inspection rules give several suitability checks a federal basis.
  • The task specification has job requirement fields, candidate evaluation fields and four validation rules.
  • Outcome records graded as planned, confirmed or verified show whether the dispatch reasoning held on site.

How should a dispatch task separate an available machine from a suitable one?

Run two tests, in order, and record each one separately. Availability asks whether a machine is free on the requested dates, in service after its last inspection, and close enough to reach the site in time. Suitability asks whether that machine can do this particular job: lift or dig what is required at the real working position, fit the site, stand on the ground there, stay clear of power lines, run with a qualified operator and travel legally to the gate.

The two tests fail for different reasons and are owned by different people. Availability lives in the rental or fleet calendar, the shop's work orders and the telematics feed. Suitability lives in the job request, the lift plan, the site's ground and utility information and the operator roster. A dispatch record that stores only the final assignment hides both, and a model trained on it learns that the nearest idle machine is usually the answer.

Construction equipment dispatch data becomes useful for AI when each candidate machine carries both verdicts, the evidence behind them and the reason a free machine was passed over. That is the gap this page fills with a field-level task specification a dispatcher, rental desk or data team can adopt as written.

Why dispatch and telematics records rarely show suitability

Most systems log what was assigned, not why the alternatives were rejected. A rental counter or contractor ERP stores the contract line, the unit number, the delivery and pickup dates and the billing status. A telematics portal stores where the machine sits and how long it ran. Neither stores the conversation in which a dispatcher said the free excavator was too wide for the alley or the free crane could not reach the far corner of the roof.

The telematics side is better standardized than many teams assume. ISO/TS 15143-3 defines how machine status data, collected by a telematics provider from earth-moving and mobile road construction machines with location and time instrumentation, is passed from the provider's server to customer applications over the internet. That makes location and operating history comparable across mixed fleets, but it describes the machine's state, not its fit for a task.

Suitability evidence is scattered: load charts and lift plans in a shared drive, a general contractor's site logistics plan in email, utility locates in a ticket system, operator certifications in an HR or safety file, haul permits with the trucking vendor. The specification below pulls each of those into a field with a named source, so the decision can be rebuilt later without the dispatcher in the room.

Seven conditions that disqualify a machine that is free

A free machine should be rejected when any one of these conditions fails, and the record should say which. Several come straight from federal rules for cranes and highway weights, which is why they belong in the task as hard constraints rather than preferences.

The OSHA crane rules are the clearest example. Equipment must not be operated beyond its rated capacity, the employer must not require an operator to do so, and the load must be verified as within capacity by a method the rule lists. Ground must be firm, drained and graded enough to meet the manufacturer's support and level specifications before equipment is assembled or used, and the controlling entity must pass on known hazards under the set-up area such as voids, tanks and utilities. Where equipment could come within 20 feet of a power line up to 350 kV, the employer must follow one of the rule's clearance options and identify a work zone first.

Seven conditions that disqualify a machine that is free
ConditionQuestion the task asksEvidence recordBasis
Rated capacity at the working radiusDoes the load chart for this configuration cover load plus rigging at the planned radius and boom length?Load chart reference, configuration, planned radius, load weight and how it was determined29 CFR 1926.1417(o) for cranes
Ground supportWill the ground at the set-up position meet the manufacturer's support and level specifications?Site ground information, mats or cribbing plan, hazards disclosed by the controlling entity29 CFR 1926.1402
Power line proximityCould any part of the machine or load come within 20 feet of a line up to 350 kV?Line location and voltage source, chosen clearance option, work zone definition29 CFR 1926.1408
Operator qualificationIs a qualified or certified operator for this equipment type available for the same window?Operator id, certification type, issuing body, expiry29 CFR 1926.1427
Inspection stateIs the machine in service, with no open deficiency judged a safety hazard?Latest shift, monthly or annual inspection result and open deficiencies29 CFR 1926.1412
Transport to siteCan the machine and its trailer reach the site within legal weight limits or under a permit?Haul configuration, gross and axle weights, permit number and route23 CFR 658.17 on the Interstate
Access and attachmentsDoes the machine fit gates, overhead clearances and swing space, and does it carry the needed coupler or tool?Site access drawing or photo, machine dimensions, attachment listEditorial: site and manufacturer data

The task specification: job requirement fields

Start with the job, not the fleet. Every field below describes what the work needs, independent of which machine ends up doing it, so the same request can be scored against any candidate. Required means the record fails validation without it; conditional fields apply only when the job type triggers them.

At a glance

The task specification: job requirement fields

The task specification: job requirement fields
FieldValues or formatRequiredWhy it matters
job_id and request_timeStable id; UTC timestamp of the requestRequiredAnchors every later decision to what was known at request time
work_typeControlled list: lift, excavate, compact, trench, demolish, otherRequiredSelects which suitability checks apply
work_windowStart and end in UTC plus site time zoneRequiredAvailability is tested against this window, not the whole day
duty_specLoad weight and rigging weight with method; or volume, depth and materialRequiredCapacity is relative to the job, never a machine-wide number
working_geometrySet-up position, radius, height or reach, swing sectorConditional: lifts and reach workLoad charts are read at a radius and configuration
site_constraintsAccess width, overhead limits, ground notes, disclosed underground hazardsRequiredCaptures the conditions that reject machines that would otherwise fit
power_line_checkNearest line distance, voltage source, clearance option chosenConditional: any line within reachTurns a safety rule into a recorded decision
operator_requirementEquipment type, certification needed, operated or bare rentalRequiredA machine without a qualified operator is not suitable for the window
service_termsContract or rental line id, delivery window, who supplies the operatorRequiredDefines what counts as on time and who carried which obligation
approverRole and id of the person who accepted the requirementRequiredSeparates a dispatcher's guess from an approved requirement

The task specification: candidate evaluation and decision fields

Then score every candidate the dispatcher considered, including the ones rejected in seconds. One row per candidate per job keeps the availability and suitability verdicts apart and makes the rejection reason a field instead of a memory.

  • Validation rule: an assigned candidate must have available_status free, a passing capacity_check where the work type requires one, a passing site_fit_check and an operator_match.
  • Validation rule: every rejected candidate carries at least one reason code that points to a failed field, so 'closest but unsuitable' cases are countable.
  • Validation rule: decided_at falls between request_time and the start of the work window; anything else is a reconstruction and is flagged as such.
  • Validation rule: requirement fields are frozen at decision time; later edits create a new version rather than overwrite the one the dispatcher saw.
The task specification: candidate evaluation and decision fields
FieldValues or formatTestFailure example
candidate_unit_idFleet or rental unit id, make, model, configurationBothUnit id with no configuration, so the load chart cannot be chosen
available_statusfree, booked, in_shop, out_of_service, in_transitAvailabilityMarked free while a shop work order is still open
inspection_refLatest inspection id, date and open deficienciesAvailabilityNo inspection on record for the shift
eta_to_siteEstimated arrival in UTC with haul plan referenceAvailabilityArrival after the work window opens
capacity_checkpass, fail or not_applicable, with chart reference and margin notedSuitabilityPass recorded without the radius used
site_fit_checkpass or fail per access, ground and clearance itemSuitabilityGround check skipped because the site was visited last month
operator_matchOperator id or none; certification type matchedSuitabilityOperator certified for a different equipment type
decisionassigned, rejected, held, substituted; reason code; free-text noteBothRejected with no reason code
decided_by and decided_atRole, id and UTC timestampBothDecision time later than the machine's departure

Illustrative case: the closest free crane fails at the working radius

This example is illustrative; the machines, distances and timings are invented to show how the fields fill in. A mechanical contractor requests a crane to set rooftop units on a three-story building, with the set-up area on a side street and the heaviest unit sitting near the far edge of the roof.

Three cranes are candidates. Crane A is free and parked closest, about 20 minutes away. Its load chart covers the unit plus rigging at the near edge of the roof but not at the radius needed for the far unit, so capacity_check fails. Crane B is free and its chart covers the radius, but the only operator free that morning is certified for a different equipment type, so operator_match fails. Crane C has the capacity and a matched operator, but it is booked until mid-morning on another job.

The dispatcher's first instinct is to move Crane A closer to the building so the radius shrinks. The site constraints field records that the general contractor disclosed a utility vault under that part of the street, so the alternative set-up position fails the ground check. The exception is documented, not argued over the radio.

The decision record assigns Crane C, shifts the work window by four hours with the customer's approval in the service terms, and stores two rejections with reason codes: capacity at radius for A, operator type for B. A model trained on hundreds of such rows sees that the nearest free crane lost, and why.

  • Crane A: available yes, suitable no; reason code capacity_at_radius; alternative position rejected for ground_hazard.
  • Crane B: available yes, suitable no; reason code operator_type.
  • Crane C: available after a four-hour shift, suitable yes; assigned with the customer's approval of the new window.

Which record proves the assignment held up on site?

The assignment is only half a training example; the outcome tells a model whether the reasoning held. Add an outcome record that links back to the job and the assigned candidate and states what happened on site, who confirmed it and from which document.

Useful outcome fields include actual arrival and release times, whether the planned set-up position was used, any substitution or second mobilization, and the delay cause if the window slipped. Contract conditions decide whether an obligation was met, so the outcome should reference the same service terms line: an operated rental that arrived inside the delivery window met its obligation even if the lift was delayed by weather, while a bare rental that arrived without the agreed attachment did not.

Grade each record's evidence maturity so a buyer can filter. A planned value comes from the request; a confirmed value comes from a delivery ticket, a signed field report or a telematics start time; a verified value has been reconciled against a second source. Training on planned values alone teaches a model the plan, not the job.

Keep failed assignments too. A machine that was assigned, arrived and then could not work, because the ground was softer than reported or an overhead line was closer than drawn, is the most instructive row in the set. Record it as a suitability failure discovered on site, link the replacement assignment to it, and note which requirement field was wrong so the error is traced to its source rather than blamed on the machine.

Ranking, verdict or evaluation set: which task are you exporting?

The same records support three different tasks, and each needs a slightly different cut. Decide which one you are building before exporting, because the split rules and the label differ.

A ranking task gives a model the job requirement and the full candidate list and asks it to order the candidates or pick one. The label is the assigned machine, and the most informative rows are the ones where the assigned machine was not the nearest free one. A classification task gives the model one job and one candidate and asks for the verdict and the failed condition, using the reason codes as labels. An evaluation set holds back a sample of hard cases, such as ground hazards disclosed late or operator gaps, so a dispatch assistant can be tested on the decisions people found difficult.

Split by time and by site, not at random. A random split puts the same job's candidates on both sides and lets a model memorize a site rather than learn a rule. Keep requirement versions as they stood at decision time, and drop or flag rows reconstructed after the fact, since they describe what someone remembered rather than what the dispatcher saw.

Watch the base rate. In many fleets the nearest free machine is the right choice most days, so a dataset of routine assignments teaches little. Oversampling rejected candidates, with their reason codes intact, is what turns a log of assignments into a lesson about suitability.

Where the specification ends and a lift planner or counsel takes over

The specification records decisions; it does not make engineering or legal ones. Lift plans, critical lift reviews and ground bearing assessments stay with qualified people, and the dataset should store their conclusions and references rather than recompute them.

Narrow the scope when a fleet mixes cranes, which carry detailed federal rules, with earth-moving equipment, where the suitability checks are mostly site and manufacturer data. Starting with one work type gives a cleaner set of reason codes. Note too that the crane operator rule changed in 2018 so that certification may be by equipment type, or by type and capacity, and state plans may differ, so the operator field should store the certification's own wording.

Bring in counsel before records leave the company. Rental and dispatch histories name customers, job sites and operators, and machine data may be governed by OEM or telematics contracts. Who may use those records is a contract question, separate from whether names are removed.

Licensing equipment dispatch history built this way

A contractor or rental business that already keeps candidate-level dispatch decisions holds something most fleets do not: the reasons a free machine was wrong for the job. Candidate-level dispatch history is the kind of record SourceX helps rental houses and contractors license, taking care of qualification, the rights review, contracting, preparation and delivery.

On volume, SourceX's general guide is that buyers want dispatch records consistent over two or more years and numbering in the tens of thousands in one system; specialized lifting or hauling work can qualify with fewer, judged deal by deal. No job record is handed over before the business approves scope and terms and signs. Fit is checked first in a short assessment of the company that needs no fleet data.

Frequently asked questions

What is construction equipment dispatch data?

It is the record of which machine was sent to which job, when and with which operator, plus the supporting evidence: the job requirements, the candidates considered, inspection state, haul plans and outcomes. For AI use, the valuable part is the candidate-level reasoning, especially why a free machine was rejected.

Is an available machine the same as a suitable machine?

No. Available means free, in service and able to arrive in time. Suitable means able to do this job at this site with a qualified operator and a legal haul. A machine can pass the first test and fail the second, which is exactly the case most dispatch logs fail to record.

Can telematics data alone tell a model which machine fits a job?

Not on its own. Telematics status data, standardized for mixed fleets by ISO/TS 15143-3, shows where a machine is and how it has run. Fit depends on the job's load, geometry, ground, power lines and operator, which live in other records.

Which rejection reasons should a dispatch dataset capture?

At minimum capacity at the working radius, ground or set-up hazards, power line clearance, operator qualification, inspection state, transport limits and site access or attachment fit. Each should be a reason code tied to the field that failed, with a free-text note for the dispatcher's detail.

Who should approve the job requirements in the record?

The person accountable for the work, usually a project manager, superintendent or lift planner, depending on the job. Recording the approver separates an approved requirement from a dispatcher's estimate, which matters when the outcome is later compared with the plan.

Sources

  • ISO/TS 15143-3:2020 defines a communication schema for machine status data passed from a telematics provider's server to customer applications, for earth-moving and road construction machinery with location and time instrumentation. Source
  • OSHA 29 CFR 1926.1417(o) prohibits operating equipment beyond its rated capacity and requires the operator to verify the load is within rated capacity. Source
  • OSHA 29 CFR 1926.1402 requires ground that is firm, drained and graded to meet the manufacturer's specifications, and requires the controlling entity to inform users of known hazards such as voids, tanks and utilities. Source
  • OSHA 29 CFR 1926.1408 requires a work zone and one of three options when equipment could come within 20 feet of a power line up to 350 kV. Source
  • OSHA 29 CFR 1926.1427, as amended in 2018, allows crane operator certification by equipment type, or by type and capacity. Source
  • OSHA 29 CFR 1926.1412 requires shift inspections and removal from service when a deficiency is determined to be a safety hazard, until it is corrected. Source
  • 23 CFR 658.17 sets an 80,000-pound maximum gross vehicle weight on the Interstate except where the bridge formula requires less, and allows special permits for nondivisible loads. Source

See if your company qualifies

A short company assessment. No data uploads are needed.

See if your company qualifies