Closing out mine findings before a fixed window opens

A mine shutdown is not a refinery turnaround

In a refinery, a turnaround takes a unit down while the rest of the site keeps running, and the economics are the margin on that unit. In a single-train concentrator, stopping the SAG mill stops the entire plant. Every hour of the window is priced at full plant throughput, and the length of the window is set by the longest task on the critical path, which is almost always the mill reline rather than the inspection programme that happens alongside it.

That changes what a deficiency record is for. In refining, a finding competes for a slot within a scope that will be worked somewhere across a multi-week window, and a week of float is normal. In mining, a finding either fits inside a reline that is already fully loaded on day one, or it does not get worked at all for another twelve to twenty-four months. The decision is therefore binary and it has to be made early, on lead time as much as on severity.

Practically, the module has to hold two things that generic maintenance systems do not. The first is a shutdown event object that findings attach to, so scheduling is relative to a plan that moves. The second is a lead-time model per finding covering access, parts, procedure qualification and specialist crew. A finding assigned to a shutdown without those four attributes settled is not scope. It is a wish with an asset number attached.

Findings that exist only when the mill stops

A large grinding mill hides most of its defects behind its own contents and liners. Fatigue cracks radiating from shell liner bolt holes are found by magnetic particle only when the liners are out. Girth gear tooth root cracking is examined by inching the mill through positions, each position a separate access event with its own isolation and its own permit. Trunnion journals and head-to-shell weld regions need the charge out and the interior clear. Discharge cones, feed chutes and grate assemblies follow the same rule for the same reason.

The rest of the plant is similar in character but different in access. Gyratory crusher top shells and spider arms crack under load and are examinable only when the crusher is empty and locked out. Apron feeder frames, transfer chute weldments and thickener rakes are all confined space work with atmospheric testing and rescue provision. Stacker and reclaimer boom weldments need rope access or a large mobile crane that has to be booked well ahead. Conveyor pulley end disc weld cracking is only found with the belt off and the pulley accessible.

Because these are once-a-cycle opportunities, a missed examination is not a deferred task. It is a blind spot lasting until the next window on a component whose failure stops the whole plant. That is why the right pre-shutdown question is never simply what is overdue. It is what will become inaccessible again the moment we start up, and what are we choosing not to look at.

Aging against a shutdown calendar, not a date calendar

The most common failure in mining deficiency data is an ordinary date. An engineer writes re-inspect in twelve months against a crack that can only be examined during a reline, on a mill that relines every eighteen. The item goes amber, then red, then someone extends it because the alternative is an unplanned stop, and after two cycles nobody can say whether the original interval had a technical basis or was simply the default the form offered.

The fix is to make the shutdown event the unit of scheduling. A finding is due at a named shutdown, with the calendar date derived from the current plan rather than typed by a person. When the shutdown moves, and it does move, because liner wear does not respect the plan, every dependent finding moves with it. The items that cannot move surface immediately as a short list requiring an engineering decision, rather than sitting silently in overrun.

That short list is the real deliverable. For each item that genuinely cannot wait, you need a documented option: run to the window with a fitness-for-service assessment and a monitoring regime, run at reduced duty with a defined limit, or take an unplanned stop and accept the production loss. Recording which option was chosen, by whom and on what technical basis is what turns a deferral into a defensible engineering decision instead of an accumulating and undocumented risk.

MSHA, and what a timely manner is measured against

MSHA's metal and nonmetal standards state the obligation plainly. Under 30 CFR 56.14100 for surface operations, and its Part 57 counterpart underground, defects on equipment that affect safety must be corrected in a timely manner to prevent the creation of a hazard, and where continued operation would be hazardous the equipment is removed from service until the defect is corrected. There is no numeric deadline anywhere in the rule.

The absence of a number is the point, and it is routinely misread as leniency. In practice, timely is judged against your own documented intentions and your own history with similar findings. If the record shows that a particular class of crack indication is normally repaired at the next available stop, and one instance was carried through two cycles with no engineering justification recorded, the inconsistency is the exposure, not the interval itself. A deficiency system that captures the classification, the intended treatment and every deviation from it is doing regulatory work rather than administrative work.

North of the border the requirements for shafts and hoisting become considerably more explicit. Ontario's mining regulation under the Occupational Health and Safety Act requires documented testing and professional engineering involvement for hoisting plant, including nondestructive examination of ropes and conveyance components at defined intervals. Those intervals are calendar-driven and do not bend around a reline that slipped six weeks, so the module has to run statutory and event-driven schedules side by side without quietly collapsing them into one field.

Access is the scope

In a mining shutdown the examination is rarely the constraint. Reaching the surface is. Confined space entry into a mill, a bin or a transfer chute requires atmospheric testing, an attendant, rescue provision and an isolation certificate, and each of those has an owner who is not the inspector. Rope access on a headframe or a stacker boom requires technicians who are certified for rope access and qualified in the examination method, which is a much smaller labour pool than either qualification produces alone. Scaffold has a build time measured in shifts and a footprint that will conflict with the reline crane.

So the readiness report a planner actually needs is not a list of findings sorted by severity. It is a matrix with a row per finding and columns for asset, examination method, access method, isolation required, crew type, duration, prerequisite parts and the date by which access must be ordered. Built four to six weeks before the window, it converts arguments about priority, which nobody wins, into arguments about scaffold and crane time, which are at least resolvable by people with calendars.

Field conditions matter as much as planning. Many sites are remote, operating on fly-in fly-out rosters, with intermittent connectivity in the plant and none underground. Capture has to work fully offline, with the asset register and open findings cached locally, and it has to synchronize without creating duplicates when a technician who worked two shifts underground finally gets signal in the crib room. Anything less produces notebooks, and notebooks produce the backlog the system was bought to eliminate.

Pressure equipment on a mine site follows different rules

Mining sites carry serious pressure equipment that sits well outside the refining inspection culture. Pressure oxidation autoclaves in gold and nickel processing are brick-lined, frequently titanium-clad, operating hot and chloride-bearing, and they are opened rarely. Acid plants carry converters, heat exchangers and acid coolers. Alumina refineries run digestion trains at high temperature and high causticity. Very little of this is governed by API 510 as a matter of law; most of it falls under the jurisdiction's boiler and pressure vessel regime, with repairs and alterations run through the National Board Inspection Code.

That distinction matters for deficiency tracking in a specific way. An owner-user programme in refining can lean on API 510 to set intervals and acceptance criteria and be judged against a code everyone recognizes. On a mine site the interval is usually set by the site's own written programme, informed by a damage mechanism review, and audited by a jurisdictional inspector who is looking for consistency with what you wrote rather than compliance with a code you never formally adopted. Your document trail therefore has to include the technical basis for every interval you set, not just the interval.

Autoclave scope decisions deserve a workflow of their own. Removing brick lining to examine the shell is expensive and adds weeks to the critical path; leaving it in place means accepting a partial examination and living with the gap. Recording that trade-off explicitly, meaning what was examined, what was not, why, and what compensating measure covers the unexamined area, is the single most valuable record an autoclave file can hold. It is also, reliably, the record that turns out to be missing when a leak appears three years later.

Discovery work and the decision rights that go with it

In mining, the ratio of discovered work to planned work is high, because genuine access really is rare. The reline crew pulls liners and finds cracking that nobody could have seen from outside. The crusher comes apart and the spider arm tells a different story than the last inspection did. Examination scope grows sharply in the first forty-eight hours of a window that was built with no slack. Everything then depends on who is allowed to decide, and how quickly they can be reached.

The module should carry pre-authorized dispositions to handle this. For defined finding classes on defined assets, the treatment and the approver are agreed before the shutdown starts, so a technician's finding at three in the morning routes to a named person with a decision rule attached rather than into a queue that will be read after breakfast. Anything falling outside those classes escalates by design. This is the practical difference between a shutdown that absorbs discovery and one that overruns and then gets analysed for months afterward.

It should also close the loop once the plant is running again. Comparing planned scope against discovered scope, broken down by asset and by damage mechanism, is how the next window gets planned on reality rather than on the last plan. Sites that never measure that ratio build every shutdown from the previous shutdown's intentions, and the overrun repeats with impressive consistency.

Evaluating the module before the next window

Test it against a real shutdown rather than a demonstration dataset. Take last cycle's findings and ask the vendor to produce, as of four weeks before a hypothetical window, a full readiness pack: open findings by asset, each carrying method, access, crew, duration, prerequisite parts and an order-by date, plus the separate list of statutory items whose intervals fall due before the window regardless of what the plan says. If what comes back is a filtered task list, the vendor has not understood the problem and no amount of configuration will fix that.

Then move the shutdown. Push the window six weeks and watch what the system does. Findings tied to the event should follow it automatically. Statutory items should not move at all. Anything that now breaches an interval should surface as an exception naming the specific rule it breaches. That one test separates event-aware systems from systems with a date field more reliably than any feature comparison a procurement team can assemble.

Finally, check the field end honestly. Ask to see capture working with the network switched off, on a rugged tablet, operated with gloves, in a workflow that a technician on a fourteen-day roster will actually use at the end of a twelve-hour shift. The best deficiency data model in the industry is worth nothing if the person who found the crack writes it on the back of an isolation permit and hands it to a supervisor who is already walking to the bus.

Request a consultation