Mining and Mineral Processing Engineering Case Studies
Industrial5 Case StudiesUpdated Sep 2026
Mineral processing engineers work where design assumptions meet ore that never quite matches the feasibility study. A crusher liner wears unevenly, a flotation bank starts short-circuiting, a tailings line plugs at a bend nobody inspected. The five cases here are drawn from fictional operations, but the pressures are familiar: tight tolerances, uneven ore, aging equipment, and records that disagree with each other. Each case follows the same thread of evidence, constraints, options, and outcomes. Read the numbers closely, because the analysis usually starts where the numbers stop agreeing.
Case Study 1
Cracked Trunnion Bearing Housing in a Copper Ball Mill
Sable Ridge Copper, a fictional concentrator treating about 28,000 tons of sulfide ore per day through one grinding line.
The situation
The primary ball mill at Sable Ridge Copper runs near 12 megawatts and grinds the bulk of the plant's copper sulfide ore. In the fourth month after a scheduled liner change, the discharge side trunnion bearing housing developed a crack that ran from a bolt hole into the outer web. Operators heard a knocking sound through the drive for two shifts before the mill was stopped. The housing had been cast to the original drawing, which called out ductile iron and a stress limit with a comfortable margin on paper. It had also been dropped from the critical equipment list during the last shutdown review, so its inspection interval had quietly drifted to match the liners.
The challenge
Ruth, the maintenance manager, needed three answers before the mill could restart: why the housing cracked, whether the matching housing on the feed side was at risk, and what would stop a repeat. Stores held one replacement casting, but its mill test report listed a heat number that did not match the failed part. The plant had already lost a little over 300 hours of grinding availability that quarter, and finance wanted a restart date that did not depend on hope.
Analysis
Teams grounded in Mining and Mineral Processing Engineering usually begin with the load path, not the crack. The trunnion housing carries the shell torque reaction and the bearing loads, and both swing with charge level and mill speed. The drive historian showed the mill had run above design fill for several weeks, and the bearing temperature trend was flatter than expected, which hinted the load was not shared evenly between housings. Wall thickness and fillet radius were inside drawing tolerance, but the drawing never called out a minimum fillet, and the foundry had cast a sharper transition at the bolt boss than the pattern intended.
The maintenance record told a more useful story than the mill test report. Dye penetrant checks had been done at commissioning and at the last overhaul, but ultrasonic readings covered the shell only, never the housing web. Two bolt retorques in the preceding months had closed as routine. The team compared three corrective options: a new casting with a heavier web and gussets, a fabricated steel housing with a verified weld procedure, or keeping the original design under crack growth monitoring. The fabricated option cost more up front but had the shortest lead time and gave the team full traceability, which the casting could not offer.
Actions taken
- Section the failed housing and map the fracture origin against the fillet radius, confirming whether the crack started at the bolt boss or at a casting inclusion.
- Pull six months of drive and bearing temperature logs and overlay them on recorded mill fill and throughput, so the team can see the true load history.
- Recalculate housing stress using measured fill and a fillet radius at the drawing minimum, then add the fillet callout and a casting hold point to the drawing.
- Add ultrasonic thickness readings on the housing web and bolt bosses to the overhaul scope, with a trigger that shortens the interval once any reading falls below the alarm value.
Results
The fabricated steel housing went in during the fifth week, and the mill was back at full charge within two shifts. Lost availability ended near 410 hours, against an estimate of 300 hours that had assumed a casting repair would pass. The opposite housing showed web thinning of about 6 percent at the next monitoring cycle, turning a feared emergency swap into a planned one. The drawing now carries a minimum fillet and a casting hold point, and the mismatched casting was retired. The main finding was that inspection had followed the shell instead of the load path.
Questions for discussion
- Which load records would have shown the sustained overfill sooner than the bearing temperature trend did?
- How should an inspection interval be set for a part that is not classed as critical but shares a load path with one that is?
- When a replacement casting cannot be traced to the original heat, what evidence is enough to return it to service?
Case Study 2
Falling Throughput and Rising Silica in a Pellet Feed Plant
Hollin Creek Iron, a fictional pellet feed concentrator producing about 4,200 tons of magnetite concentrate per day.
The situation
Over several weeks the magnetic separation circuit at Hollin Creek Iron delivered less concentrate per hour than the plan allowed, while silica in the product drifted upward. The plant runs a primary crusher, two stages of grinding, and a series of wet drum separators feeding a vacuum filter bank. Feed came from two pits that were being mined in alternating blocks, so the ore hardness and magnetite content changed from shift to shift. Customers had begun to flag silica in the pellet feed, and two shipments were held back for re-testing. Nobody on the operating team could say whether the falling output and the rising silica had one cause or several.
The challenge
Marcus, the plant manager, had a budget for one shutdown in the next quarter and no appetite for trial and error on the separators. He needed a ranked list of improvements, each with a measurable acceptance test, so the shutdown scope could be agreed with maintenance and quality by the end of the week. Throughput had dropped about 9 percent below the nameplate rate, and the silica exceedance was running at roughly one in every five lots.
Analysis
Start with a process map from feed to filter cake, marking every point where the material changes character. At Hollin Creek Iron the crusher discharge carried a wide size spread, the mill circuit ran at a fixed grind target that had been set for softer ore, and the drum separators were sized against a magnetite grade that the south pit no longer reliably delivered. Operating data showed that the throughput loss tracked feed hardness on the days the south pit was the main source. The silica tolerance was tight, and the separator matrix was the only place it could be recovered, so variation there mattered more than average throughput.
Maintenance records pointed to the screen ahead of the mill. Work orders showed it had been cleared by hand four times in the last period, and the deck blinded faster on wet days. Blinding sent oversize back to the mill and raised circulating load, which capped hourly throughput. Improvements were ranked by silica and throughput recovered per unit of shutdown time: a higher open area deck, a separator matrix change, and a filter cake moisture loop. Each had an acceptance test set in advance, such as undersize efficiency held above target for four weeks, and silica under the contract limit on at least nineteen of every twenty lots.
- Feed hardness Harder south pit blocks cut screen efficiency and raise mill load
- Screen blinding Wet fines blind the deck, sending oversize back to the mill
- Separator matrix Coarse matrix lets silica report to concentrate on the worst lots
- Filter cake moisture Moisture drift changes pellet behavior and triggers customer holds
Actions taken
- Install a screen deck with higher open area and a spray bar, then measure undersize efficiency on wet and dry days for four weeks before accepting the change.
- Replace the separator matrix on the south circuit with the finer grade and log field strength readings on each lot, so magnetic recovery can be compared against the old matrix.
- Add a filter cake moisture control loop that adjusts vacuum level from the pan sensor, with a high alarm set at the moisture limit the pellet plant specifies.
- Move feed hardness into the daily plan, with a blend rule that holds the south pit share below an agreed ceiling on days the hardness index runs high.
Results
The screen deck went in during the shutdown, and undersize efficiency held above target through the four weeks that followed. Hourly throughput climbed back to within about 2 percent of nameplate, with the blend rule deliberately trimming output on the hardest days. Silica in the concentrate stayed under the contract limit on 19 of 20 lots after the first two weeks, meeting the acceptance test, though the matrix change only partly closed the gap on the worst feed. The filter moisture loop sharply cut out-of-range cake readings.
Questions for discussion
- Which measurement would separate the effect of feed hardness from the effect of screen blinding on hourly output?
- How much shutdown time is a reasonable price for a silica improvement that only applies to the worst feed?
- If the acceptance test passes but the worst lots still fail, which criterion should govern the next change?
Case Study 3
Tailings Dam Raise Review with Missing Seismic Evidence
Brannock Valley Minerals, a fictional copper and molybdenum producer that stores all of its tailings in one valley facility.
The situation
The project team at Brannock Valley wanted to raise the tailings dam by one stage to extend storage by about two years of production. The current facility is an upstream-raised embankment built over several campaigns, and the proposed raise would use a downstream method with a compacted rockfill shoulder. Ahead of the design freeze, an internal review was asked to check the package against the site's tailings management standard and the regulator's submission checklist. The review team had the geotechnical investigation, the water balance, and the construction drawings, but the seismic hazard assessment was a draft that had not been updated since the earlier design.
The challenge
Sofia, the engineer of record for the dam, had three weeks before the approval gate. The review had to say whether the package could go forward as submitted, what evidence was missing, and which controls were already in place. Her concern was not only the seismic case but the lack of a documented operating limit for the pond level during an extreme rainfall event, plus an emergency response plan that still named a retired contact list.
Analysis
Start with which hazards the package addresses. Design constraints for a raise are the foundation, the pond and the spillway. The water balance showed the pond would run within about one meter of the crest during a design wet season, leaving less freeboard than the standard asked for, and existing pond data showed level swings the proposed limit ignored. Quality tolerances on the shoulder fill were written in terms of compaction, but the placement records held only moisture checks from the first two lifts. A quality tolerance without records is not evidence, so the review had to mark the shoulder specification as unsupported.
Approval gates are where missing evidence becomes a decision. The team separated the findings into three groups: items that block the design freeze, items that must close before construction, and items that belong in the operating manual. The seismic case blocked the freeze because the consequence of a shoulder failure is severe and the hazard basis was stale. The emergency response plan and the pond level limit could be closed before the first lift, provided the plant manager signed the updated documents. Documentation rules are strict: every control must point to a record, and every record to a named owner, or an auditor will treat the control as absent.
Actions taken
- Commission an independent seismic hazard update and hold the design freeze until the consequence class and design earthquake are documented and signed by the engineer of record.
- Set a written pond level limit with alarm and action thresholds, derived from the water balance and the operating data, and include it in the operating manual.
- Reconstruct the shoulder placement record from survey and test data, and where lifts lack density results, schedule test pits before the second lift is placed.
- Update the emergency response plan contact list and run a tabletop exercise with the site crew, then file the signed record with the regulator submission checklist.
Results
The design freeze moved back by about five weeks while the seismic update was completed. The updated hazard basis raised the design earthquake level, which changed the shoulder width and added a filter zone that was absent from the original drawings. The reconstructed placement records covered most lifts, and test pits were cut where they did not. Two lifts that failed the reconstructed density check were removed and rebuilt. The approval gate was passed with conditions tied to the pond limit and the response plan, both closed out before the first construction season began.
Questions for discussion
- Which hazard should have triggered a design hold in the earliest review, and who was responsible for raising it?
- When placement records are incomplete, how much testing is enough to keep a shoulder specification in use?
- How should a regulator submission handle a control that exists in practice but has no signed document behind it?
Case Study 4
Recovery Options for a Late Crusher Commissioning
Kestrel Mesa Mining, a fictional gold and silver operation building a secondary crushing plant for its heap leach feed.
The situation
Kestrel Mesa planned to commission a new secondary crushing and screening plant so that heap leach feed could reach its target size distribution before the rainy stretch. The plant was a modular design with a cone crusher, two screens, and a fine ore conveyor, procured as a package from a single fabricator. Fabrication of the main frame was about three weeks late when the milestone schedule was reviewed, and the conveyor drive had been revised twice after a clearance clash between the transfer chute and the existing stacking structure. The current crushing line was still operating, but it was producing oversize that the heap could not tolerate for long.
The challenge
Adrian, the project director, had to choose a recovery path within the week, because the fabricator's revised date and the conveyor redesign both pushed commissioning past the point where the heap could absorb oversize. Each option cost money, changed the risk of a second slip, and affected the operating budget. He wanted the trade-offs quantified before the steering meeting, not described.
Analysis
The slip had three roots, and they interacted. The fabricator was short on skilled welders, so the frame schedule had no slack. The clearance clash came from a design constraint: the transfer chute had been drawn to the original stacking structure, which had since been modified, and the revision moved the conveyor drive into a space the maintenance team needed for access. Operating data from the existing line showed oversize rising to a level the heap's permeability could tolerate for only a few weeks. The product quality tolerance on the heap feed was a fixed size distribution, so any recovery plan had to hold that target, not just the date.
The four options were compared on one basis: cost to recover the date, probability of a second slip, and effect on heap performance. Overtime at the fabricator and a second welding shift cost the most in spend but kept the design intact. Splitting the delivery, with the frame first and the conveyor after a redesign, was cheaper but added two commissioning interfaces. Accepting a later date and running the old crusher harder, which the maintenance records showed was already due for liner replacement, carried the lowest capital but the highest risk of an unplanned stop. Sourcing a second screen fabricator reduced single-source exposure but needed a fresh drawing review.
Actions taken
- Authorize overtime and a second welding shift at the fabricator for the main frame, with a daily progress check against the milestone and a penalty clause tied to the delivery date.
- Redraw the transfer chute and conveyor drive against the as-built stacking structure, using a laser scan, and issue the revised drawing before the frame ships.
- Schedule a liner change and bearing inspection on the existing crusher during the slip window, so the old line can carry oversize for the full period without a surprise stop.
- Qualify a second fabricator for the screens only, with a drawing review in the first week, so single-source exposure drops without restarting the whole package.
Results
The team chose overtime at the fabricator combined with the chute redraw, and held the existing line in reserve with an early liner change. Commissioning finished about 2 weeks later than the original milestone, against a projected slip of 5 weeks for the split delivery option. Direct recovery cost came in near the mid-range of the options compared, and the schedule risk score fell once the drawing was reissued. The old crusher ran through the window with one short stop for a bearing check. Heap feed stayed inside its size target for the whole period.
Questions for discussion
- Which assumption in the split delivery estimate would change the ranking of the options most if it proved wrong?
- How should the cost of a heap upset be weighed against overtime spend when the milestone is already late?
- When a redesign and a schedule recovery pull in opposite directions, who should hold the authority to decide?
Case Study 5
Cutting Specific Energy in a Gold Grinding Circuit
Larkspur Gold Mill, a fictional gold operation that grinds about 6,000 tons of ore per day ahead of cyanide leach and carbon recovery.
The situation
Larkspur Gold's grinding circuit had become the largest single draw on site power, and the energy bill had risen faster than throughput over the past months. The circuit runs a single ball mill in closed circuit with cyclones, fed by a primary crusher with a fixed setting. Water is recycled from the leach tailings thickener, and fresh water make-up has crept up as the thickener underflow density became harder to hold. The site had no formal target for specific energy, and the operating team tracked only mill power and tons milled. Management asked for a reduction in energy and fresh water without a drop in gold recovery or in the grind size needed by the leach.
The challenge
Tomasz, the process superintendent, had to bring a proposal to the operating committee that would hold P80 at the leach target and recovery above the agreed floor. Three options were on the table: a variable speed drive on the mill, a high-pressure grinding roll ahead of the ball mill, and a tighter thickener control loop with a water recovery upgrade. Each had a different capital cost, life cycle footprint, and effect on the grind size distribution.
Analysis
Measure before changing anything. The first step was to set a specific energy baseline in kilowatt hours per ton, from mill power and tons milled, and to tie it to P80 and gold recovery on the same time base. Operating data showed the mill drawing near its design power at a circulating load that was higher than the cyclone classification needed, so energy was going into rock that was already fine enough. The quality tolerance on P80 was narrow, and the leach could not accept a coarse tail, so any saving had to come from better classification or from breaking the ore more efficiently, not from simply grinding less.
The three options were compared on operating measures and on lifecycle cost, defined as capital, power, wear parts, and the embodied footprint of equipment over its planned life. The variable speed drive was the simplest change and held P80 steadiest, but it trimmed power only at the mill, while the classification records showed circulating load was the real energy sink. The high-pressure roll offered the largest saving, though it added a wear component whose rebuild cycle showed in the maintenance records and needed a heavier feed system. The thickener loop protected fresh water with the least new equipment, yet its benefit was capped by the water the mill itself consumed.
- 1Crushing and feedFixed primary setting feeds a coarse, variable size spread into the mill
- 2Ball mill grindingMill draws near design power, with most energy spent on already fine rock
- 3Cyclone classificationCirculating load sends fines back to the mill and carries the real energy cost
- 4Thickener water recoveryUnderflow density control sets how much fresh water make-up the circuit needs
- 5Leach and carbon recoveryRecovery and grind size limits set the classification target upstream
Actions taken
- Set a specific energy target in kilowatt hours per ton, with P80 and gold recovery as hard limits, and report all three on one daily sheet for the operating committee.
- Pilot a classification change on the cyclones, reducing circulating load by a set step, and log power draw, P80 and tail assays for four weeks before any capital decision.
- Install density and rake torque sensors on the leach thickener, and tie fresh water make-up to measured underflow density, with the maintenance team setting the alarm limits.
- Compare the high-pressure roll option against a full lifecycle cost that includes wear parts, rebuild intervals from the maintenance records, and the embodied footprint of the equipment.
Results
The classification pilot cut circulating load and lowered specific energy by a modest share in the first month, with P80 and gold recovery holding inside their limits. The thickener loop reduced fresh water make-up once underflow density held steady. The high-pressure roll was not bought because the lifecycle comparison showed its wear part rebuilds would erase much of the power saving at the site's ore hardness. The operating committee adopted specific energy as a standing measure and the daily sheet shows it beside gold recovery. What the team took away was that circulating load, not the mill, drove energy use.
Questions for discussion
- Which operating measure should govern the trade-off when lower energy and a tighter P80 point in different directions?
- How much weight should the embodied footprint of new equipment carry against a modest but certain operating saving?
- What would the lifecycle comparison look like if the ore hardness rose by the level the pit plan predicts for the later stages?
Learning Resources
Industrial CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.