Aerospace and Defense Manufacturing Case Studies
Industrial5 Case StudiesUpdated Sep 2026
These five cases come from the production and sustainment side of aircraft components and defense hardware. Each one starts with a part, a line, or a schedule that stopped behaving the way the drawings said it would. The evidence in every case is spread across documents that rarely get read together: design constraints in one file, operating data in another, tolerance stacks that were never recalculated, and maintenance records nobody connected to the failure. The organizations are fictional, but the decisions are the ones working teams make every month.
Case Study 1
Nose gear bracket crack and the fatigue question nobody asked
Calder Ridge Precision Components, a machining supplier of titanium brackets, fittings, and hydraulic hardware for transport aircraft.
The situation
Calder Ridge Precision Components machines titanium brackets and fittings for two transport aircraft programs. One bracket supports a nose gear actuator. Parts in the affected lot had logged about 2,400 flight hours on average when a line mechanic found a crack at the inner radius of a fillet during a scheduled inspection. The crack was roughly 4 millimeters long and had grown from a surface mark nobody had flagged. Every shipset had passed dimensional checks at receiving, and the material certifications were complete. The open question was whether the failure came from the part, the design, the way the fleet was flown, or an inspection interval set from earlier fleet data.
The challenge
The engineering team had to answer three things before the customer would accept any corrective action: whether the crack started from a manufacturing defect or from loading the design never anticipated, whether other brackets in service carried the same risk, and whether the inspection interval still fit the way the fleet was flown. The operating data showed heavier landing loads on a subset of routes, but the maintenance records were split across two systems that disagreed on serial numbers.
Analysis
Teams grounded in Aerospace and Defense Manufacturing usually start by separating what the fracture surface can prove from what the paperwork claims. Beach marks radiating from the fillet point to fatigue growth under repeated loading, not a single overload. The drawing radius met the stress analysis minimum, but that analysis used a load spectrum from an earlier mission profile. Operating data from the two busiest routes showed a larger share of cycles near the upper limit. The radius tolerance was plus or minus 0.2 millimeters, and the process had drifted toward the low end, so the failed part sat near the bottom of its band.
The maintenance records made the picture harder to read. Inspection intervals had been set from fleet averages built mostly on regional missions. Two inspection logs used different serial number formats, so about one entry in six could not be tied to a specific bracket. After the records were reconciled by hand, the cracked part showed more cycles than its neighbors. Corrective options were a larger fillet radius, shot peening on the existing geometry, or a tighter inspection interval for high-cycle aircraft. Each carried a different cost in redesign, requalification, and fleet downtime.
- Inspection dayMechanic flags crack at fillet inner radius on one bracket.
- First weekFractography points to fatigue growth, not a single overload.
- Second weekStress analysis found to use an outdated load spectrum.
- Within the monthRecords reconciled by hand; extra brackets identified for inspection.
- End of quarterCombined fillet and peening fix approved for production.
Actions taken
- Quarantine the remaining brackets from the affected lots and pull their certificates of conformance, then sort them by serial number so every installed part can be traced to its heat treatment and machining run.
- Rerun the fatigue analysis using the measured load spectrum from the two busiest routes, not the original mission profile, and report the margin at the lowest permitted fillet radius.
- Merge the two inspection databases under one serial number standard and require a matched record for every cycle count before the next interval review, so the fleet data can be trusted.
- Run a controlled fillet radius test on twelve production parts at the upper and lower tolerance limits, and pair the results with a peening trial before choosing a corrective path.
Results
The team chose a combined path. The fillet radius moved to the upper half of its tolerance band with a 0.1 millimeter increase in the minimum callout, and the shot peening step was added after the test parts showed better fatigue life than the unpeened control set. Fleet inspection intervals for high-cycle aircraft were shortened, and the reconciled records identified nineteen additional brackets that needed inspection. No further cracks were reported in the following inspection cycle, though requalification took the better part of two quarters and cost the customer a planned delivery window.
Questions for discussion
- Which of the three evidence sources in this case would you have trusted least at the start, and what would have changed your ranking?
- How would you decide whether a tighter inspection interval is a sound interim measure or a way of avoiding the design question?
- What records would your own team need to reconcile before a fleet-wide fatigue review could be trusted?
Case Study 2
Grinding cell output loss traced to dressing and gauges
Harrow Valley Machining, a plant that grinds steel actuator pins for two airframe programs and ships to assembly lines on a fixed weekly schedule.
The situation
Harrow Valley Machining runs a grinding cell that finishes steel actuator pins for two airframe programs. The cell is the constraint on the line. Over several months its output per shift dropped from about 340 pins to just under 260, even though operators and programmed cycle times had not changed. At the same time, scrap and rework on the outside diameter climbed from under 2 percent to nearly 7 percent. No lot had been rejected by a customer yet, but two shipments were held at receiving for diameter drift. The plant manager needed to know whether the loss came from the machine, the material, the inspection process, or the way the cell was scheduled.
The challenge
The pins carry a bearing journal tolerance of plus or minus 0.01 millimeters, and the surface finish callout limited how hard the team could push feed rates. The maintenance log showed the wheel dresser had been replaced late in the previous cycle, and coolant concentration readings lived on paper, not in the machine record. Nobody had matched gauge data against bar stock heat lots, so it was unclear whether the drift followed the material, the wheel, or the measurement.
Analysis
Mapping the cell before changing anything separated real problems from convenient explanations. The team traced the flow from bar stock receiving through grinding, gauging, and packing, and listed the variables at each step. Four families of variation stood out: hardness across heat lots, wheel condition after dressing, coolant condition, and gauge repeatability. Each left a different signature. Hardness shifts changed grinding force, wheel wear produced a slow diameter taper across a shift, and coolant problems showed up as finish scatter. Gauge repeatability had never been measured, and it accounted for more of the apparent drift than anyone expected.
The design constraints shaped which fixes were allowed. The journal tolerance and surface finish callout came from the actuator drawing, so changing feed rates alone could not fix the problem if the finish suffered. Shift logs showed output losses clustered in the first hour after a wheel dress, which pointed to dressing practice rather than the machine. The team weighed three responses: a tighter dress interval, an in-process gauge at the grinder, and a material certificate check on incoming heats. Each had a different cost in downtime, capital, and inspection labor, and only the first two could be tested within the existing schedule.
Actions taken
- Set a fixed wheel dressing interval based on ground part count rather than elapsed time, and log each dress with the operator, wheel serial, and first-part gauge reading.
- Install an in-process diameter gauge on the grinder with automatic compensation, and run it for two weeks alongside the bench gauges to measure repeatability before trusting either number.
- Match every bar stock heat lot to its hardness certificate and record grinding force by lot, so material drift can be separated from wheel and gauge effects in the next review.
- Set acceptance criteria before the trial: output at or above 320 pins per shift, scrap below 2.5 percent, and no diameter lot average outside the center third of the tolerance band over twenty shifts.
Results
The dress interval change alone recovered most of the lost output. Over the trial, output per shift returned to about 325 pins and scrap settled near 2.2 percent. The in-process gauge showed bench gauge repeatability was the weakest link, and the plant replaced two bench fixtures. Material hardness varied within certificate limits, so it was not the main driver, although two heat lots still needed closer watching. The acceptance criteria were met by the twenty-second shift of the trial, and the line returned to full schedule the following month.
Questions for discussion
- If the dressing interval fixed most of the loss, what evidence would convince you that the gauge repairs were still worth the spend?
- How would you set the first-hour sampling plan so that wheel condition and material effects could be separated without stopping the cell?
- Which acceptance criterion in this case would you tighten first, and what would it cost the schedule to do so?
Case Study 3
Pressure stand release held for a stale relief calculation
Ashgrove Test Systems, a designer and builder of ground test stands for hydraulic and pneumatic flight hardware.
The situation
Ashgrove Test Systems was asked to commission a new ground test stand that will pressurize hydraulic actuators to 3,000 psi using stored nitrogen. The stand includes a bank of accumulators, a remotely operated isolation valve, and a control PLC that interlocks the pressure sequence. The project was on track for commissioning later in the quarter, and the customer's quality lead wanted a formal safety and regulatory review before the first energized test. The review team included a process safety engineer, a controls lead, a quality representative, and an outside notified inspector. It was asked to say whether the stand could be released, and if not, what evidence was still missing.
The challenge
The review had to decide three things: whether the stored energy in the accumulator bank was adequately controlled, whether the documentation matched the stand as built, and whether the approval gates in the project plan had actually been passed. The accumulator supplier had changed during procurement, the pressure vessel data sheets were for the original design, and the interlock logic had been validated on a bench simulator rather than the installed hardware. Maintenance access points were not yet drawn on the layout.
Analysis
Stored energy drove most of the review. A 3,000 psi nitrogen bank can throw fittings and components across the test cell if a line fails, so the team worked down the control hierarchy. Substitution to lower pressure was rejected by the customer, so the controls had to carry the load: an isolation valve, relief protection, and a physical barrier between operators and the stand. The relief sizing calculation was the weak point. It assumed the original accumulator volume, and nobody could show that the revised bank had been carried into it, which left the governing pressure code check unfinished.
Documentation took most of the review time. The controls narrative, the piping and instrumentation diagram, and the hazard log had been written at different points, and the hazard log still listed the old supplier's accumulator. The interlock logic had been validated on a bench simulator, which proved the sequence but not the wiring or the response of the installed isolation valve. Under the project's own gate structure, the stand could not pass to energized testing until the design review, the controls validation, and the operator training record were each signed. Only the design review was signed. The remaining two gates were treated as formalities in the schedule.
- Hazard identification List stored energy paths and confirm each against the installed hardware.
- Control verification Test relief, isolation, and interlock trips on the stand as built.
- Documentation alignment Bring drawings, logs, and calculations to one revision.
- Gate sign-off Named approvers sign each gate before energized testing starts.
Actions taken
- Recalculate pressure relief sizing for the revised accumulator volume and the maximum credible flow, and have a second engineer check the calculation against the governing pressure code before the stand is energized.
- Reissue the hazard log and piping and instrumentation diagram under one revision, replacing the old accumulator data, and tie each hazard to a named control and a verification record.
- Run the interlock sequence on the installed stand with the isolation valve and pressure transmitters connected, using dry nitrogen at low pressure, and record each trip point against the expected value.
- Hold the controls validation and operator training gates as separate sign-offs with named approvers, and make energized testing conditional on all three gates in the project record.
Results
The review released the stand for low-pressure commissioning and held energized testing. Within the following weeks the relief calculation was redone with the revised volume, and it showed the original relief valve was undersized by a margin that would have mattered in a full-bank release. The valve was replaced, the hazard log was reissued with twenty-three traceable controls, and the interlock trip points passed on the installed hardware after two setpoint corrections. Energized testing started once all three gates were signed, about a month later than the original plan.
Questions for discussion
- Which of the three missing items would you have insisted on before any energized test, and which could have followed under a controlled restriction?
- How should a project record show that a supplier change has been carried through the hazard analysis and the calculations behind it?
- When a schedule and a gate conflict, who in your organization has the authority to hold the gate, and how is that authority recorded?
Case Study 4
Pocket change and billet slip on a wing fitting milestone
Northmere Aerostructures, a builder of wing spar fittings and landing gear brackets for a defense trainer program.
The situation
Northmere Aerostructures builds wing spar fittings and landing gear brackets for a defense trainer program. The milestone in question is first article delivery of a fitting set to final assembly, due at the end of the quarter. Two problems hit at once. A forging supplier lost a die and delivered titanium billets about three weeks late, and a design change to the pocket geometry was released after the machining programs had already been posted to the floor. The planning group estimated the milestone would slip by about five weeks if nothing was done. The program office wanted a recovery plan that protected the assembly date without reopening the fixture build.
The challenge
The recovery options had to respect three constraints: the qualified machining capacity on two five-axis centers, the inspection hold points that could not be skipped without requalification, and a fixed budget reserve of 180,000 dollars for schedule recovery. Operating data from the machining cells showed the five-axis centers were already loaded near ninety percent, and the maintenance records flagged one spindle with recurring vibration warnings. Every option also had to be checked against the revised drawing, which had not yet been fully checked for tolerance stack.
Analysis
Each recovery option was priced in three numbers: weeks gained, dollars spent, and risk added to later gates. Extended overtime on the five-axis centers could recover about two weeks for close to 60,000 dollars, but the spindle warnings made that time uncertain. A second forging source could recover the billet delay for about 95,000 dollars, but its first article needed most of a month, so the recovered time was not fully available. Resequencing the non-critical brackets freed the centers for about a week at almost no cost, though it pushed a second customer's shipment by the same margin.
The coordination gap was harder to price. The geometry change reached the machining programs through a planner who was not on the design distribution list, so two cells cut to the old revision for a shift before anyone noticed. The scrap was small, but the tolerance stack across the spar interface had not been recalculated for the new pocket. Any plan that added overtime without fixing release control would repeat the error at the next revision. The chosen path combined resequencing for two weeks with a conditional second source, and it reserved 40,000 dollars for the stack analysis and a controlled release step.
Actions taken
- Move the second forging source to a parallel first article track now, with billet samples and certificates due in the first week, so the conditional option can be released without a second delay.
- Run the tolerance stack analysis on the revised pocket geometry, including the spar interface and the fixture datum, and release the machining programs only after the stack is signed by design.
- Resequence the non-critical bracket orders to free the five-axis centers for two weeks, and notify the second customer in writing with the revised ship date and the reason for it.
- Restrict overtime to the spindle that passed its vibration check, and set a stop rule that ends extra shifts if that spindle logs another warning during the recovery window.
Results
The team kept the milestone date, though not in the way first proposed. Resequencing and the release control recovered the schedule within two weeks, and the conditional second source was never needed. Overtime ran for nine shifts on the healthier center and stopped when its own vibration check drifted, which is the stop rule working as intended. Total spend came to about 118,000 dollars against the 180,000 dollar reserve, and the second customer accepted a four-day shift in its shipment. The fitting set reached final assembly on the revised date, with the stack analysis signed before the new program was posted.
Questions for discussion
- Which of the three recovery options would you have ranked first on paper, and what would have changed that ranking once the spindle data arrived?
- How would you price the risk that a late design change adds to later gates, so it can sit beside weeks and dollars in the comparison?
- When a planner outside the design distribution sees a revision, what would make that gap visible before cutting starts?
Case Study 5
Autoclave energy and trim scrap at a composite panel shop
Corvell Composite Works, a shop that cures carbon fiber panels for wing skins and interior structures in two autoclaves.
The situation
Corvell Composite Works cures carbon fiber panels for wing skins and interior structures in two autoclaves. Energy is the largest utility line on the cure side, and the monthly energy cost for the autoclave hall had risen as the panel mix shifted toward thicker layups. Trim scrap from ply cutting was also high, roughly 22 percent of the prepreg issued to the cutting table. The sustainability lead and the operations manager were asked to reduce energy and waste per good panel without changing the void content or resin content specification, and to show the case in lifecycle terms rather than as a utility invoice.
The challenge
Three constraints framed the options. Void content had to stay under the specified limit. The laminate had to pass the ultrasonic scan. The cure cycle was qualified in the material specification, so any change to the ramp or hold needed requalification. The maintenance records showed one autoclave's vacuum pump was running on a longer duty cycle than its partner, which added energy use unrelated to the cure. The team also had no good record of trim scrap by layup type.
Analysis
Energy and scrap look like separate problems in the utility report, but they share a cause. The cure cycle was built around the thickest layup in the mix, so thin panels were held at temperature far longer than their resin needed. Energy per good panel, not per cycle, is the right denominator, because a cycle that produces a rejected panel spends its energy for nothing. The team built a model covering purchased electricity, nitrogen, scrap disposal, and the embodied energy of discarded prepreg. Scrap carried most of the embodied burden, so the cutting table mattered more than the oven.
Four options were compared on energy per good panel, scrap rate, void margin, and throughput. A shorter ramp was tempting, but the hold time was tied to void control, so it needed a full coupon program. Nesting software that rearranged ply patterns to reduce trim offered a larger effect on waste with no change to the cure. Out-of-autoclave prepreg would remove most of the energy, but it needed new tooling and a qualification the customer had not approved. A vacuum pump duty fix was cheap and stayed within existing controls.
Actions taken
- Install nesting software on the cutting table and run a six-week trial on the three highest-volume layups, measuring trim scrap per good panel against the current hand-nested baseline.
- Service the lagging vacuum pump and match the duty cycle of both autoclaves, then log energy per cycle for each machine so the difference is visible in the monthly review.
- Build a coupon program for a shorter hold on the thin layups, with void content and ultrasonic results recorded for every panel, and do not change the cure recipe until the data clears the specification.
- Set the monthly measure as kilowatt-hours and trim scrap per good panel, reported beside void margin and throughput, so no gain in energy can hide a drop in quality.
Results
The pump fix cut autoclave energy per good panel by about a fifth in the first full month, and the nesting trial brought trim scrap from roughly 22 percent to just under 14 percent on the trial layups. The shorter hold was not released across the board. Coupon results showed void content within specification at the reduced hold on most thin layups, but two layups stayed marginal, so the original recipe stays for them and the change goes ahead for the rest. Out-of-autoclave prepreg went into the next capital review as a separate decision.
Questions for discussion
- Which measure in this case would you distrust most as a single number, and what second measure would you pair with it?
- How would you weigh a large energy saving that requires requalification against a smaller saving that does not?
- What would your plant need to record for each autoclave cycle before a lifecycle comparison could be defended to a customer?
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.