Quality Management and Process Excellence Case Studies
Professional5 Case StudiesUpdated Sep 2026
Quality management is often taught as a toolkit, but the cases here show how those tools get used under pressure. A part fails in service. A line falls behind and defects climb at the same time. A regulator asks for a record nobody filed. A delivery date slips. Each case follows one site through its problem and asks what the evidence supported, what the constraints allowed, and who absorbed the consequences. Treat them as working notes rather than model answers. The numbers belong to the fictional sites, and the trade-offs are the part worth arguing about.
Case Study 1
Cracked manifold ports in a packaging machine line
Brindle Hollow Fluid Works, a mid-sized maker of hydraulic valve manifolds with two plants and a small product engineering group.
The situation
Brindle Hollow Fluid Works builds valve manifolds for packaging and material handling machines. Its flagship block, a cast aluminum manifold with three threaded ports, had shipped in steady volume with a low warranty rate. In the second quarter, the customer's maintenance group began returning units with fatigue cracks at the port closest to the pressure relief valve. Dana, the product engineer, pulled 42 returned units and the records for the lot they came from. Early reading suggested the cracks started at the thread root after a pressure spike, not a material defect. The machine builder had recently raised its peak cycle pressure by a modest margin to speed up a press.
The challenge
The question was not only why the ports cracked but whether the fix could be made without requalifying the whole product line. Tomas, the operations lead, wanted the plant running at its current pace, since two customers had orders scheduled for the next month. Cost was a concern too: a forging tool change would take weeks and a redesign would need new pressure testing. Dana had 42 units, but only nine had complete service histories, which made it hard to separate pressure exposure from installation error.
Analysis
The business objective was narrower than the engineering question. Brindle Hollow needed to protect its relationship with the packaging customer, keep the second plant's allocation intact, and avoid a recall that would reach other builders using the same block. Forty-two returns sounds substantial, but nine complete histories carry most of the weight, because they show operating pressure and installation torque. Dana sorted the returns by fracture origin, then matched each group to the pressure log the customer had supplied. Teams grounded in Quality Management and Process Excellence tend to treat this sorting as the first real analysis, since a fracture surface often says more than a report written afterward.
The corrective options were constrained differently. A thicker boss at the thread root reduced stress but changed the envelope, so the mounting footprint and two fixtures needed rechecking. Shot peening could be done within days and needed no new tooling, yet it did not touch the pressure spike. A forged block was the most robust choice but added weeks of tooling and a new validation run. Tomas weighed these against the order calendar, while the customer's engineer cared whether the revised part would fit existing machines without a retrofit. Each option pushed cost onto a different group, so the choice was also about who carried the risk.
- Gather returns Sort 42 units by fracture origin before drawing any conclusion.
- Test pressure Match peak cycle pressure to each complete service history.
- Bench validate Compare peened and unpeened blocks at measured peak pressure.
- Review design Reopen the boss geometry only if bench data points to it.
Actions taken
- Hold the 42 returns in quarantine and photograph each fracture surface with a fixed scale, so the origin groups can be reviewed by someone outside the original sorting.
- Ask the customer for pressure logs on all nine complete histories and request the peak values, not averages, so the spike can be tested directly against the thread root stress estimate.
- Run a short bench fatigue test on peened and unpeened blocks at the customer's measured peak pressure, with the result due before the scheduled order release.
- Approve peening as interim containment for promised shipments, and open a controlled boss redesign only if bench data points to the pressure spike as the driver.
Results
Peened blocks in the bench test ran past the customer's target cycle count with no crack initiation at the root in the sample. The interim containment let the plant meet its scheduled orders, and the customer accepted the lot with a pressure limit added to its maintenance sheet. Tomas kept the plant at its planned pace, and no forging tool was ordered. Two returned units had been installed with torque well above the drawing value, which the customer corrected at its own sites. The boss redesign stayed on file as the next step if field returns rise.
Questions for discussion
- Which evidence would have changed the containment decision if it had arrived one week later, and how should that delay have been handled?
- When a customer's maintenance practice contributes to failure, where should the boundary of corrective action sit?
- How would you weigh a fast containment that leaves the root cause open against a slower redesign that closes it?
Case Study 2
Trim cell bottleneck and rising flash defects
Calder Ridge Molding, a contract injection molder making plastic housings for kitchen appliances, with three presses feeding one shared trim and inspection cell.
The situation
Calder Ridge Molding makes plastic housings for kitchen appliances, using three injection presses that feed a shared cell of manual trim and inspection stations. Over several weeks, weekly good output from the cell fell while the share of parts rejected for flash and short shots rose. The presses were not the constraint: their cycle times were steady and their shot weights held tolerance. Michael, the cell supervisor, saw carts stacking up in front of trim, and the operators reported that a few parts per hour came back to them with flash that the molding setup should have prevented. Nobody on the floor had mapped the process end to end since the cell was reconfigured.
The challenge
The business problem was a missed delivery commitment on the largest appliance account, along with rising rework cost. The obvious fix was a fourth trim operator, but the shift had no spare labor and the deburr fixture was shared with a second product. Ayesha, the quality manager, needed output gains without added headcount, and results she could check against the customer's rejection limits within a month.
Analysis
The first analysis is about which measurements can be trusted. The cell had counts of good parts at final inspection, but no record of where a part was rejected in the process, so the team could not tell whether flash came from the press, the carts, or the trim operators handling parts. Ayesha asked for a two-week tally at each station, with each reject tagged by defect and by the shift that molded it. Sorting like this belongs at the front of any review, since a bottleneck that looks like a press problem is often a handling problem one station downstream.
With the tally in, the sources were clearer than expected. Flash rejects clustered on parts from one cavity that had been reground last month, and short shots came mostly after material changes. Ayesha ranked four candidate changes by output recovered and by how fast each could be checked against the customer's limits. A trim fixture change was the largest gain but shared a tool with a second product, so it had to fit between runs. Cart staging was cheap and did not touch molding, which made it the right first step.
Actions taken
- Replace the shift tally sheet with a tagged reject log at each station, and require the operator to record the defect and the press cavity on every rejected part.
- Stage carts in a fixed bay with a first in, first out rule and cap work in process at trim to twelve carts, so queues stay visible.
- Pull the reground cavity from production until its flash rate returns to baseline, and log the first hour after each material change as a separate run for two weeks.
- Schedule the trim fixture change into the gap between runs of the second product, with a go or no go check against the customer's rejection limit after the first full shift.
Results
Cart staging and the reject log went live in the second week. Queue length at trim fell from a daily peak of about forty carts to under fifteen, and operators stopped inheriting unmarked flash from upstream. Flash rejects from the reground cavity returned to baseline once it was pulled. The fixture change passed the customer's limit on its first full shift, and weekly good output rose by about eleven percent. Ayesha sent the tally data to the account team, which settled the question of whether the presses were at fault.
Questions for discussion
- If the trim fixture change had been the only option that could be scheduled, how would the team have communicated the delay to the account?
- What would a tagged reject log need to capture to separate a press problem from a handling problem on the first day?
- Which measurable acceptance criteria would you set for the fixture change before it is installed, and who should be able to reject it?
Case Study 3
Blending line review before commissioning approval
Tessmore Industrial Coatings, a powder coatings maker adding an automated blending and bagging line at its main plant.
The situation
Tessmore Industrial Coatings makes powder coatings for metal furniture and appliance customers, with one plant that is adding an automated blending and bagging line. The project was six weeks from its commissioning gate when the internal review group began its safety and documentation check. The line includes a high speed blender, a bag filler, and a dust collection unit tied to interlocks on the blender lid. The project engineer had completed a hazard analysis, but it treated dust as a housekeeping issue, and the interlock test records were not yet signed. Peter, the health and safety lead, had asked for the review to confirm that each control was in place rather than planned.
The challenge
The plant's business objective was a commissioning date that protected a supply agreement for a furniture customer, and the team had already bought the blender. Delay had a real cost, but a line that started with an unvalidated dust interlock could create an explosion hazard and an uninsured loss. Peter could not sign off on evidence that did not exist yet, and the project manager, Mei, wanted to know which gaps were blockers and which could be closed during commissioning.
Analysis
The review had to separate two kinds of gaps. Some were missing evidence, such as the unsigned interlock test and the dust layer measurements. Others were design gaps, such as a bag filler guard that opened without stopping the auger, which no document addressed. Mei's schedule pressure was real, but it did not change what counted as proof. The team treated a control as in place only when a test record showed the interlock tripping the blender under a simulated fault. A precise list of unproven controls is often the most useful output of a review.
The hazard analysis had to be rebuilt around controls, not around dust. Combustible dust risk depends on the dust layer, ignition sources and ventilation, and each has a testable control. The team ranked the controls by how much they reduced exposure if they failed: a verified interlock and a grounded collector ranked above a housekeeping schedule, which was useful but could slip. The approval gates were set by the plant's documented quality system, so the review checked that each gate had an owner, a required record, and a defined stop condition. Peter also wanted the operators' procedure, still a draft, checked against each guard reset.
Actions taken
- Hold commissioning until the blender lid interlock is tested under a simulated fault, and file the signed record against the hazard analysis control it verifies.
- Rebuild the dust hazard analysis around the dust layer, ignition sources and ventilation, listing a test or inspection record for each control before the gate opens.
- Allow bag filling to start in a supervised trial only after the guard interlock is verified, with the auger stop time measured and written into the procedure.
- Finish the operators' cleaning and guard reset procedure, walk it through with the night shift, and record their objections as open items with named owners.
Results
The commissioning gate moved by eleven days. The interlock test failed on its first run because a limit switch had been wired to the wrong input, which would have left the blender running with the lid open. After rewiring and a second signed test, the record went into the file. The dust layer survey found a ledge above the filler that had been missed, and a cleaning task was added to the schedule with an owner. Mei told the plant board that the delay bought evidence, not a failure of the project.
Questions for discussion
- Which control on the list would you have verified first, and what would you have accepted as proof short of a full test?
- How should a review handle an approval gate whose owner has changed since the hazard analysis was written?
- When operators' procedures are still in draft at commissioning, what conditions should limit the first production runs?
Case Study 4
Control cabinet delay and three recovery options
Marrow Creek Controls, an automation integrator that builds control panels and commissions process lines for beverage and dairy clients.
The situation
Marrow Creek Controls builds control panels and commissions automated lines for beverage and dairy processors. Its current project is a filling line for a regional client, with a factory acceptance test scheduled for the start of the eleventh week and a site installation booked for the following month. The control cabinet depends on variable frequency drives from a single supplier, and the drives were quoted at eight weeks. Two weeks after the order, the supplier moved the delivery date out by five weeks after reallocating components. Meanwhile, the panel designer had revised the cabinet layout to fit a relocated operator panel, which added a rework of wiring ducts already fabricated.
The challenge
The milestone was a fixed date in the client's production plan, and a late acceptance test would push site installation into the client's seasonal shutdown window, when a missed slot meant waiting for the next one. Sanjay, the project director, had to choose between paying for expedited parts and freight, reworking the sequence of cabinet builds, or offering a partial acceptance test that covered the filler logic without the drives. Each option carried a different cost and a different risk to the client relationship.
Analysis
The evidence that mattered was the supplier's revised allocation letter, the wiring duct rework estimate, and the labor hours already booked on the cabinet. Sanjay asked for the critical path in days, not percentages, because the client's shutdown window was a hard boundary. The revised drive date left 19 working days of float before the acceptance test would have to slip, and the layout change had consumed about six of them in engineering review. Each option has to be quantified against the same float and the same penalty clause, since otherwise a cheap-looking recovery can hide a larger cost in a later milestone.
The three options were not equally reversible. Expedited freight cost a fixed premium, and the team estimated it at about 4,800 in added freight and handling, but it recovered about seven working days if the supplier's partial shipment arrived as promised. Resequencing the cabinet builds cost nothing in parts but pushed two technicians off a second project, which put that client's delivery at risk. A partial acceptance test kept the date but left the drive interface untested until site, which meant a likely punch list during installation. Sanjay weighed these against the penalty clause and the client's stated preference for a sound start over an early one.
- 1Supplier confirms partialWritten quantities and carrier window fix the earliest usable date.
- 2Freeze cabinet layoutSingle approved duct change protects the float the plan relies on.
- 3Expedite critical drivesFreight premium limited to the module that sits on the critical path.
- 4Provisional filler acceptanceClient engineer signs tests that run without drives, pending site checks.
- 5Site installation in windowInstallation lands in the shutdown window with a short punch list.
Actions taken
- Confirm the partial drive shipment in writing with the supplier, including quantities and the carrier's delivery window, before approving any freight premium.
- Freeze the cabinet layout now and approve the duct rework as a single change, so engineering review stops consuming float that the recovery plan depends on.
- Agree with the client's engineer on which filler logic tests can run without drives, and write those results into the acceptance record as provisional.
- Keep the two technicians on the second project and add a fixed overtime budget for cabinet builds, capped at a level the finance lead approves in advance.
Results
The team chose a combined path: expedited freight for critical drives, a fixed overtime budget, and a provisional filler test. The total premium stayed under 6,000, which was below the penalty exposure for a two week slip. The factory acceptance test ran in the twelfth week, one week after the original target, and the client's engineer signed the provisional filler results. Site installation moved into the shutdown window with a small punch list that closed in the first week on site. Sanjay noted the client kept its payment milestone, which was the real stake.
Questions for discussion
- How would the comparison change if the client had been willing to accept a later date in exchange for a fully tested drive interface?
- What evidence should a project director ask for before accepting a supplier's revised date as a planning fact?
- Where does a partial acceptance test stop being a recovery tool and start transferring risk to the client's installation team?
Case Study 5
Drying oven energy and pulp scrap reduction
Lowmoor Packaging, a producer of molded pulp trays and inserts for produce and bakery customers.
The situation
Lowmoor Packaging molds pulp trays and inserts for produce and bakery customers. Its largest plant runs drying ovens that consume most of the site's energy, and its scrap rate on the forming line had crept up over several quarters. Leadership set two goals for the coming budget cycle: lower energy use per good tray and lower fiber waste, without changing the strength that customers measured in drop and stack tests. Ruth, the sustainability and operations manager, had a list of ideas from the floor, including heat recovery from the oven exhaust, a finer forming mesh, and a lighter tray design that used less fiber. Each idea promised a different mix of savings and risk.
The challenge
The difficulty was that the cheapest gains could damage performance. A lighter tray saved fiber on every unit, but a drop failure at a produce customer's distribution center would cost more than a season of savings. Heat recovery required a new heat exchanger and a shutdown of the oven line for several days, which the plant could not spare during peak orders. Ruth needed one comparison that put energy, scrap, strength and lifecycle effects on the same page.
Analysis
Measures had to be stated per good tray, not per batch, since a scrap cut that also reduced good output would look like an energy win on paper. Ruth set three measures: kilowatt hours per good tray, fiber lost as a share of pulp fed, and strength retained against the customer's drop and stack specifications. Lifecycle effects used one tray delivered to the customer as the functional unit, so a lighter design got credit only for fiber and energy it actually removed. The baseline is usually the first deliverable, so the team pulled eight weeks of meter and scale data before modeling.
The options separated cleanly once they were scored on the same basis. Heat recovery cut energy per good tray the most, but its payback depended on shutdown cost and the peak season schedule. A finer forming mesh reduced fiber loss with almost no capital, yet it slowed the line and raised rejects on thin walls. The lighter tray design saved the most fiber per unit, but it needed a strength retest with the produce customer, whose distribution team had the final say. Ruth's group paired the mesh change with a tighter drying profile and sent heat recovery and the lighter design to separate gates.
- Weeks one to eightBaseline per good tray set from meter and scale data.
- Weeks nine and tenFiner forming mesh trialed on one shift with strength sampling.
- Week elevenDrying profile tightened in steps, held at the third setting.
- Week twelveHeat recovery reviewed at its own gate with peak season payback.
- After customer retestLighter tray design considered only once strength data is in.
Actions taken
- Set the baseline per good tray for energy, fiber loss and strength over eight weeks, and publish the calculation sheet so every option is scored from the same numbers.
- Trial the finer forming mesh on one shift for two weeks, with strength samples pulled every two hours and a stop rule if thin wall rejects exceed the baseline.
- Tighten the drying profile in small steps, logging kilowatt hours per good tray after each step, and keep the change only where strength holds at the customer's test levels.
- Send the heat recovery exchanger and the lighter tray design to separate approval gates, each with its own payback, shutdown plan, and customer retest requirement.
Results
The mesh trial ran for two weeks on one shift and cut fiber loss, with thin wall rejects staying within the baseline band. The drying profile step-down lowered kilowatt hours per good tray by about nine percent, and strength held through the third step but not the fourth, so the team kept the third setting. Heat recovery went to its own gate, and the lighter design waited for the produce customer's retest. Ruth reported the combined changes to site leadership with the full calculation sheet attached.
Questions for discussion
- What would change in the ranking of the options if the energy meters had reported per batch rather than per good tray?
- How much weight should a sustainability target carry against a customer's strength specification when the two conflict?
- What evidence would justify reopening the lighter tray design after its retest has been deferred?
Learning Resources
Professional 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.