Industrial Safety and Process Engineering Case Studies
Industrial5 Case StudiesUpdated Sep 2026
These five case studies are written for engineers, reliability staff, and safety leads who work on plants, production lines, and utilities. Each one follows a fictional organization through a decision where the evidence is incomplete and the schedule is unforgiving. The shared thread is the data behind every judgment: design constraints, operating logs, quality tolerances, and maintenance records. Read the situation and challenge before the analysis, then decide what you would do. The diagrams summarize the options each team weighed. The questions suit group discussion, and most have more than one defensible answer.
Case Study 1
Shaft fracture in a caustic cleaning pump
Tillman Dairy Processing, a fictional cheese and cream plant with a central clean-in-place system
The situation
Tillman Dairy Processing runs a cheese and cream plant with two packaging halls and a central clean-in-place system. The CIP supply pump is a close-coupled centrifugal unit rated for hot caustic solution. In the ninth month of operation, the pump shaft fractured at the sleeve shoulder during a routine wash cycle. The unit was pulled from the loop, and caustic reached the drain pit before the level alarm tripped. No one was hurt, but the line stayed down for four days and the shift lost most of a week's output. The team had the maintenance log, the original equipment specification, and seven months of temperature and flow trend data from the historian.
The challenge
The maintenance manager needed to know whether this was a material defect, a sizing error, or a procedure problem. The supplier wanted to close the matter as a warranty claim. The plant engineer had noticed that caustic supply temperature ran about 10 degrees above the design basis during summer, and that change had never been signed off. The spare pump on the shelf was the same model, and the next cheese run was scheduled for the following week. Restart depended on the answer.
Analysis
The first question was whether the design basis still matched the duty. The original specification assumed a caustic solution at a set temperature and concentration, with torque margin for start-up transients. The historian showed supply temperature drifting upward through the warmer months, and several wash cycles starting with the return valve partly closed, which pushed the pump toward shutoff. Neither condition alone should have broken a shaft, but together they raised cyclic bending stress at the sleeve shoulder. Teams grounded in Industrial Safety and Process Engineering usually start here, because the trend data shows what the plant actually did, which is often different from what the procedure says.
The second question was whether the part met its tolerances. The fracture showed a fatigue origin at a fillet machined near the low end of the drawing radius, and the supplier's hardness report sat close to the lower limit. The sleeve had been replaced twice in the past twelve months, each time fitted by feel rather than measured, and one replacement came from a different batch. The notes did not prove a defect, but they meant the failure could not be blamed on design or material alone. Any fix should be tested before a redesign is approved, since a larger shaft may only move the crack.
- Setpoint drift Supply temperature rose above design basis through warmer months.
- Start-up stress Partly closed return valve pushed pump toward shutoff region.
- Fillet fatigue Low-radius fillet and low hardness set up crack initiation.
- Undocumented change No management of change record existed for the setpoint shift.
Actions taken
- Pull the full operating envelope from the pump datasheet and overlay the historian data, counting hours outside the rated temperature and suction conditions over the last seven months.
- Section the fractured shaft at the fillet and send it for fractography and hardness testing, with the supplier witnessing the cut so the results cannot be disputed later.
- Replace the sleeve fit method with a measured interference check and a fitting record signed by a second technician, and quarantine remaining parts from the suspect batch until testing is complete.
- Add a start-up interlock that holds the wash cycle until the return valve is confirmed open, and require a documented management of change before the supply temperature setpoint is altered.
Results
Testing confirmed a fatigue crack starting at the fillet, with hardness near the lower specification limit. The sleeves from the suspect batch were quarantined, and two other units from that batch turned up with similar fit records. The team did not redesign the shaft. Instead the interlock and measured fit check went in, and the supply setpoint was returned to the design basis through a formal management of change. The supplier agreed to a tighter fillet note on the drawing, and the rebuilt pump has run without recurrence since.
Questions for discussion
- Which of the three hypotheses would you test first, and what result would make you drop it?
- How should a plant decide whether an undocumented setpoint drift needs a formal change review?
- What evidence would you require before accepting a supplier's claim of a manufacturing defect?
Case Study 2
Flash defects and a starved inspection cell
Keel Point Molding, a fictional injection molder making housings for small motor assemblies
The situation
Keel Point Molding runs eleven injection presses for housings used in small motor assemblies. Every part passes through a trim and inspection cell before packing, and that cell is the plant's constraint. Over the last quarter, good parts per hour through the cell fell by about a fifth, while rejection at final inspection rose from under 2 percent to about 5 percent. Operators reported that the cell was starved in some hours and blocked in others. The team had shift output logs, press cycle data, gauge results for two critical bore dimensions, and a maintenance record showing three presses with off-pattern mold changes.
The challenge
The plant manager wanted throughput back before the customer's next release. Quality wanted the rejection rate under the agreed limit. The cell lead proposed adding a second inspector, a cost the budget did not carry. Three ideas were on the table: a faster trim tool, a tighter setup routine at the press, and a gauge-based sorting step at each press. Nobody agreed on which problem came first, or whether they were the same problem.
Analysis
Mapping the process first changed the picture. The cell had ample nominal capacity, but its real pace depended on how parts arrived. Press logs showed flash defects clustering after mold changes and during the first hour of each run, before mold temperatures settled. Those bursts flooded the cell and forced operators to inspect in batches, so the bottleneck was being fed by variation upstream. That makes it as much a flow problem as a quality problem. That is a different fix from adding inspectors.
The gauge data told a second story. The two critical dimensions had tolerances of plus or minus a few hundredths, and the drift on the bore diameter tracked with gradual wear on one press's core pins. The maintenance log showed those pins had been adjusted twice without a new baseline measurement. Off-pattern mold changes had also been done by a crew working from memory rather than the setup sheet, which explains the spread in first-hour scrap. A sorting gauge at the press would catch some of this, but it would not remove the cause. Restoring setup discipline comes first.
Actions taken
- Record a verified baseline on each press after every mold change, using the two critical dimensions and a first-article check, and hold release until the baseline is logged.
- Rebuild the setup sheet as a step-by-step document with mold temperature and warm-up time, and require the crew lead to sign it before each off-pattern change.
- Stagger the start of the affected presses so that first-hour parts arrive at a rate the cell can inspect without batching, and track cell idle time by hour.
- Trial a faster trim tool on one cell station for two weeks, accepting it only if it cuts cycle time by at least 8 percent without raising the flash escape rate above 0.5 percent.
Results
Within the quarter, first-hour scrap on the two worst presses fell after verified baselines and the rebuilt setup sheet were in use. Cell idle time in the first hour of each run dropped, and flow through the cell settled. The final rejection rate returned to just under the agreed limit, though not to the level the team had hoped for. The faster trim tool failed its acceptance test because the flash escape rate went above 0.5 percent. The second inspector was not hired, and the sorting gauge was deferred. Output recovered to within a few percent of the earlier baseline.
Questions for discussion
- If the flash defects did not follow the first-hour pattern, which upstream variable would you measure next?
- How much idle time at the cell would justify a second inspector instead of process changes?
- What acceptance criteria would you set before removing the sorting gauge from the press?
Case Study 3
Ammonia loop reviewed before the design gate
Larkfield Cold Storage, a fictional operator of three frozen-food warehouses in one region
The situation
Larkfield Cold Storage is adding an ammonia refrigeration loop to its newest warehouse, which will hold frozen goods for two regional grocers. The design package includes a high-pressure receiver, two evaporative condensers on the roof, and a machine room that shares a wall with the loading dock. The project is at its design review gate, and the construction schedule assumes a permit decision within a few weeks. The package contains a process flow diagram, an early piping diagram, a draft hazard analysis, and an operating procedure outline that operations staff have not yet read.
The challenge
The review team had to decide whether the design could move into detailed engineering, which would lock in equipment and pipe routing. Three things looked wrong. The machine room had a single door onto the loading dock, the relief header discharged near an air intake, and the hazard analysis had not considered loss of power to the condensers. The project manager argued these were detail items for later. Leadership wanted an answer by the end of the week.
Analysis
The reviewer started with the documentation, since missing evidence is easier to find than hidden hazards. The relief header routing appeared on an early piping and instrumentation diagram with no discharge calculation, and the machine room ventilation had not been sized against the refrigerant inventory. Relief routing and room ventilation are foundational. Changing them after equipment is ordered means reopening vendor agreements and possibly the building permit. Each control should be checked against the applicable pressure vessel and refrigeration requirements, with the calculation in hand. At a gate like this, open questions become expensive quickly.
The second concern was hazards that appeared nowhere on the drawings. The machine room has one door, and it opens onto the loading dock, so an operator leaving during a leak would pass the area where trucks idle. The hazard analysis did not consider loss of condenser fans or power, which would raise head pressure toward the relief settings. Maintenance records from the current site were useful here, since two earlier leaks had traced back to valve packing outside the inspection schedule. Listed controls had no verification step, test date, or owner.
Actions taken
- Require a relief sizing and discharge routing calculation, stamped by an engineer, before the piping isometrics are released, and move discharge away from the air intake.
- Add a second egress path from the machine room, or justify a single path with a documented escape plan, and document the travel distance against the building's exit requirements.
- Extend the hazard analysis to cover loss of power, loss of condenser fans, and high head pressure, with each scenario tied to a listed control, a test, and a named owner.
- Hold the design gate open until operations staff have reviewed the draft procedure and the valve packing inspection schedule is written into the maintenance plan.
Results
The project team accepted the relief calculation and moved the discharge to the roof ridge, which took about three weeks and added some piping cost. The egress concern was resolved by relocating one dock door and adding a second exit from the machine room. The design gate was held for one extra review cycle, and the hazard analysis was reissued with each scenario tied to a control and an owner. Operations approved the draft procedure after edits. The permit application went in later than planned, but it came back without conditions that would have forced equipment changes after the order.
Questions for discussion
- Which of the open items would you require closed before the design gate, and which could move to detailed engineering?
- How would you test whether a single egress path is acceptable for a machine room of this size?
- What evidence would convince you that a hazard analysis covers loss of utilities adequately?
Case Study 4
Recovering a skid milestone after three overlapping delays
Pennock Process Skids, a fictional fabricator of modular membrane filtration units for small utilities
The situation
Pennock Process Skids fabricates modular membrane filtration units for small municipal water systems. The current job, for a utility serving a few thousand residents, has a contractual delivery milestone at the end of the quarter with a daily delay charge of 4,000 dollars. The contract value is 386,000 dollars, with 18,000 held as contingency. Three things went wrong in the same month. The actuated valve package arrived about nine weeks late, a piping revision was issued after the control panel had been built to the prior drawing, and the customer's site access was postponed by a separate civil contract. The planning lead had the purchase log, the drawing revision register, and the weekly progress reports.
The challenge
Pennock could not meet the delivery date with parts in hand. The operations director wanted the milestone protected, the finance lead wanted spending held near contingency, and the customer asked whether partial delivery was possible. Three recovery options were on the table: expedite the valves at a premium, rework the panel to the new piping drawing in-house, or ship the skid early and leave the valve package for a later field visit. Each carried a different cost and a different risk to commissioning.
Analysis
The first step was separating the three delays, because each had a different recovery curve. The valve delay was a supply problem with a known lead time, and expediting could recover perhaps four weeks at a premium of roughly 22,000 dollars. The piping revision was a design coordination failure, and the panel rework carried the largest uncertainty because the control logic depended on signal counts that had moved. The site access delay was outside Pennock's control. Pricing any fix before confirming the critical path wastes money, because a premium on a non-critical item buys nothing.
Some steps could not be compressed at all. Factory acceptance testing on the panel required a full loop check against the revised drawing, and the valve package had to pass seat leakage tolerances that the customer's specification set tightly. Pennock's own maintenance records for the same valve model showed two stuck-position faults in earlier skids, both traced to actuator lubrication not done before shipment. Rushing the valves might save a week, but a field failure would cost far more than the premium. The plan that held up kept the test sequence intact.
- 1Valve supply delayActuated valves arrived nine weeks after the planned date.
- 2Piping revision lagPanel built to the prior drawing before the change was issued.
- 3Site access postponedCivil contract delay pushed the earliest installation date.
- 4Expedite and reworkPremium valves and in-house panel rework recovered the schedule.
- 5Split delivery agreedFrame and piping shipped on time, commissioning fixed later.
Actions taken
- Expedite the valve package under a premium order that recovers roughly four weeks, with the supplier's written commitment and a pre-shipment actuator lubrication check added to the release.
- Rework the control panel to the revised piping drawing in-house, freezing signal counts first and running the loop check on the bench before any wiring is installed in the enclosure.
- Offer the customer a split delivery that ships the skid frame and piping on the milestone date, with a written schedule for the valve package and the field commissioning visit.
- Set a weekly reconciliation of purchase orders, drawing revisions, and site access dates, with any change to the critical path reported to finance within two working days.
Results
The team chose in-house panel rework and expedited valves. Together they cost about 31,000 dollars against 18,000 of contingency, and the difference came out of job margin. The skid frame and piping shipped on the milestone date, and the customer accepted the split delivery once commissioning dates were fixed in writing. The loop check on the revised panel found one signal mismatch, corrected on the bench. Commissioning started about two weeks later than the original plan, and the delay charge was waived for the partial delivery. The job closed with a thinner margin than bid.
Questions for discussion
- Which of the three delays would you price first, and how would you confirm it sat on the critical path?
- At what premium on expedited parts would you stop chasing supply and propose a split delivery?
- What change control rule would have caught the piping revision before the panel was built?
Case Study 5
Cutting cure oven energy without losing adhesion
Redbrook Pressed Metals, a fictional contract manufacturer of coated steel brackets
The situation
Redbrook Pressed Metals stamps and coats steel brackets for two agricultural equipment customers. Its powder coating line uses a gas-fired cure oven that still runs at the setpoint chosen when the line was first commissioned. Energy per coated part sits above the plant average, and rejects from coating defects, mostly orange peel and under-cure, consume about 6 percent of coated output. A corporate target asks for a cut in energy per part and scrap within the next budget cycle. Available records include hourly gas meter logs, part cure temperature readings, adhesion test results, powder usage, and maintenance logs for the burner and exhaust fan.
The challenge
Any change must keep adhesion and salt-spray results inside customer specification, and the plant cannot lose throughput during peak season. The options were a lower oven setpoint with a revalidated cure profile, a heat recovery exchanger on the exhaust, a low-cure powder, and tuning the burner. Each carried a different capital cost, a different effect on powder supply, and a different risk to the customer's sign-off.
Analysis
The starting point was the cure profile, since the oven's setpoint sat above the powder supplier's recommended window. Hourly gas logs showed the oven running at full setpoint even during idle periods, and cure records showed parts at the cold end of the oven reaching temperature late, which explained most of the under-cure rejects. A lower setpoint would reduce energy per part, but only if the cure window still reached the adhesion requirement at the coldest part location. The cheapest energy is energy the process does not need.
The remaining options needed a comparison on lifecycle and operational terms. Heat recovery on the exhaust would capture a meaningful share of waste heat and was the largest capital item, but the exhaust fan's maintenance log showed repeated bearing replacements, and recovery equipment adds pressure drop that the fan may not handle. A low-cure powder reduces energy per part and cuts cure time, but it required customer approval and carried a different chemistry footprint that needed lifecycle review. Burner tuning was cheap and reversible, and it was the first step to take because it reduced gas use without changing the process window.
- First weekBurner tuned and gas use logged against combustion analysis.
- Third weekTrial batch run at lower setpoint with coldest-location sampling.
- Sixth weekAdhesion and salt-spray results reviewed with customer quality.
- Eighth weekLower setpoint adopted; under-cure rejects tracked daily.
- End of quarterExhaust fan rebuilt; heat recovery sized for next budget cycle.
Actions taken
- Tune the burner on the cure oven against a combustion analysis, and log gas use per part before and after for four weeks to confirm the saving.
- Run a revalidated cure profile at a lower setpoint on a trial batch, testing adhesion and salt-spray at the coldest part location before any production part is run at the new setting.
- Replace the exhaust fan bearings on the planned schedule, then size a heat recovery exchanger against the fan's measured pressure drop before ordering any equipment.
- Hold the low-cure powder option for a lifecycle comparison covering embodied energy, powder supply, and customer approval time, and decide only after the other changes are measured.
Results
Burner tuning cut gas use per part by roughly eight percent within the first month and needed no customer approval. The revalidated lower setpoint passed adhesion and salt-spray tests on the trial batch and was adopted after sampling at the coldest location. Under-cure rejects fell, and overall scrap dropped to the lowest level the plant had recorded in two quarters. The exhaust heat recovery was deferred until the fan was rebuilt, and the low-cure powder remained under review. Throughput held through the peak season.
Questions for discussion
- Which measurement would you need before accepting a lower oven setpoint as a permanent change?
- How would you weigh a reversible, low-capital measure against a larger change with better lifecycle numbers?
- What would make you delay the low-cure powder decision even if its energy case looks strong?
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.