Research, Development and Testing Services Case Studies
Professional5 Case StudiesUpdated Sep 2026
Research, development and testing work sits between an idea and a product someone will depend on, and the five cases here follow that gap. Each describes a fictional engagement like those a test laboratory, a prototype shop or an engineering group handles every season. The situations are untidy on purpose. Evidence arrives late, budgets are fixed, and the people who bear a decision's consequences are rarely the people who made it. Each analysis weighs objectives, evidence, operating limits and stakeholder impact. Read the reasoning as working notes from practitioners, and expect to disagree with some of the calls.
Case Study 1
Repeated seal leaks in food plant washdown service
Harrow Valley Fluid Systems, a mid-sized maker of industrial pump seals that runs its own rig and outsources only accelerated wear trials.
The situation
A polymer lip seal used in the washdown lines of two food processing customers began weeping fluid after roughly six weeks in service. Field returns reached 240 units across the first shipments, against a design target of fewer than 15 failures per 1,000 units in the first year. The seal had passed the standard bench endurance test with no visible wear. Engineers suspected the gap lay between the bench cycle and real conditions: frequent hot water rinses, caustic cleaning agents, and pressure spikes when valves closed too quickly. The product manager wanted a fix in the next production run, and the largest customer had asked for a written root-cause summary.
The challenge
The team had three problems at once. The mechanism was not confirmed: returned seals showed chemical attack on the lip, but some also showed compression set that could have started in manufacturing. A compound change meant a new material qualification of about eleven weeks, while a gland geometry change could be checked on the existing rig in less time but might miss the chemistry. Quality also had to decide whether to hold the current lot, which would cost the customer its line schedule.
Analysis
The business objective was narrower than the engineering question. Harrow Valley needed to keep both accounts, protect margin on a seal line already priced thinly, and avoid a recall that would cost more than the product line earned in a year. The evidence was uneven. The returned seals were a small and partly contaminated sample, and the bench data had been gathered at constant temperature. Teams grounded in Research, Development and Testing Services usually begin by separating what the failed parts prove from what the lab data merely failed to rule out, because those two kinds of evidence support different corrective actions and carry different weight when a customer asks for a written account.
The constraints narrowed the options quickly. A compound change needed fresh qualification against the customer's sanitation specification, and the elastomer supplier could not guarantee lot consistency for another six months. A gland redesign was cheaper and could be checked on the existing rig, but it would only help if pressure spikes were the main driver. Maintenance crews carried the cost of every leak, the sales lead carried the relationship risk, and finance wanted scrap exposure capped before approving any production change. The team scored each option against all three interests, not only the engineering metric.
Actions taken
- Pull every returned seal into a joint teardown with the customer's maintenance lead, photographing the lip and recording the cleaning cycles each unit saw before failure.
- Run a designed test on the rig that combines hot rinse cycles, caustic exposure and valve closure rates at three levels, so the failure can be reproduced rather than inferred.
- Approve the gland geometry change for the next production run, hold the current lot, and start compound qualification in parallel with a fixed decision date.
- Send the customer a written root-cause summary that separates confirmed mechanisms from open hypotheses, with the validation plan and a close date for each open item.
Results
The teardown showed lip damage concentrated on units that had seen the most frequent hot rinse cycles, which pointed to thermal cycling combined with pressure spikes. The rig reproduced the leak on the original gland geometry within a few runs and stopped leaking on the revised geometry under the combined profile. Field returns on the revised lot fell from 240 units in the first shipments to 9 in the following period. The compound change completed qualification later and became the second, safer material. The customer kept the account, though its maintenance team asked for a longer validation window on future changes.
Questions for discussion
- Which of the four actions would you sequence first, and what evidence would make you reverse that order?
- How should a small sample of returned parts be weighed when a customer expects a firm root cause quickly?
- When a fix changes both geometry and material, what does a clean validation result actually tell the team?
Case Study 2
Finish mill bottleneck and rising bracket rejects
Ridgeline Prototype Works, a contract development shop that machines and heat treats test articles for product engineering teams.
The situation
Over a few months, throughput on a bracket family used in a customer's endurance program fell from about 60 parts per shift to around 41, while the share rejected at final inspection rose from under 2 percent to roughly 9 percent. Customer test programs were waiting on finished brackets, and two deliveries were already late. The route runs through five stages: programming and setup, rough milling, heat treatment, finish milling and coordinate measurement. The shop had added a second heat treatment furnace earlier in the year, and output had not moved. Management needed the shop running at full rate again before the next program release.
The challenge
The bottleneck was not obvious. Operators blamed the furnace queue, programmers blamed fixture changeovers, and inspection reported that most rejects came from bore diameter drift. Each group had data that supported its view. The shop manager had to choose improvements that would restore throughput and cut rejects within the quarter, without adding headcount the budget could not carry. Two programs carried penalty clauses tied to delivery dates.
Analysis
The business objective had two parts that pulled against each other. Throughput mattered because test programs stalled when parts arrived late, and the shop was paid per completed program rather than per part. Quality mattered because every rejected bracket consumed machine time twice, once in making it and again in remaking it. The evidence was better than the arguments suggested. The shop logged queue times at each station and kept inspection results by bore. A practitioner starting from that position usually builds a simple flow map first, timing each stage and tagging each defect to the stage where it could first have been caught, before anyone proposes a fix.
The data showed the furnace was not the constraint on most days, although its queue was the most visible. Time-stamped logs placed the longest waits at finish milling, where setups varied by operator and fixture condition. Bore drift traced to a worn collet and to parts that left heat treatment before stress relief was logged complete. The shop could not run a third shift, and any fixture change needed the customer's sign-off on the drawing. Operators carried the schedule pressure, inspectors carried the blame for late rejects, and the customer cared mainly about parts arriving test-ready. Each improvement was ranked by the throughput and reject measures it would move.
- 1Programming and setupFixture changeovers vary by operator; setups logged at roughly forty minutes.
- 2Rough millingStable cycle time; rarely the source of late or rejected parts.
- 3Heat treatmentQueue builds here; stress relief completion was not always recorded.
- 4Finish millingLongest waits and widest setup variation; bore drift begins here.
- 5Coordinate measurementCatches rejects only after the machine time is already spent.
Actions taken
- Standardize finish mill setups with a locked fixture checklist and a fixed collet replacement interval, then time each setup for two weeks before accepting the change.
- Hold every batch after heat treatment until the stress relief cycle is logged complete, and reject any batch without that record before it reaches finish milling.
- Set an acceptance line of 55 parts per shift at under 4 percent rejects, sustained for four consecutive weeks, before the bottleneck is declared closed.
- Move bore inspection to an in-process gauge at finish milling so drift is caught on the second part rather than at final inspection, and log a first-article check for each setup.
Results
Setup time at finish milling fell by about a third during the first month, mostly because operators stopped searching for the collet and the right wrench. The stress relief hold removed most of the bore drift, and rejects at final inspection came down from roughly 9 percent to under 4 percent over six weeks. Throughput reached 54 parts per shift in the fifth week, just below the acceptance line, and cleared it in the seventh. One penalty clause was avoided. The second program slipped a few days because a fixture change needed the customer's approval, which was outside the shop's control.
Questions for discussion
- When the most visible queue is not the true constraint, what evidence would you need before moving work away from it?
- How should an acceptance line be set when throughput and rejects pull in opposite directions?
- Who should own the decision to add an in-process gauge, the shop floor, quality, or the customer's engineer?
Case Study 3
Approval gates for a high-voltage battery test bay
Brightwater Test Facility, an independent lab that runs abuse and environmental tests on battery modules for portable equipment makers.
The situation
Brightwater planned to commission a high-voltage test bay for abuse testing of battery modules, on behalf of a client launching a portable power product. Each bay could hold up to 450 volts DC with stored energy of about 30 kilowatt-hours. The design review found that vent duct sizing had been based on a generic gas release estimate rather than measured off-gas data for the client's chemistry. The arc flash study covered the main switchgear but not the portable cabling used inside the bay. Training records were incomplete for two technicians. The client's launch depended on qualification results within the coming weeks.
The challenge
The review had to decide whether the bay could be energized for qualification testing or whether that had to wait for evidence that did not yet exist. Missing items included the off-gas measurement, arc flash labels on the portable cables, and a signed hazard analysis from the building safety officer. Some controls, including a gas detection interlock, were specified in the design but not yet installed, and the reviewer could not treat specified controls as operating controls.
Analysis
The business objective was to start a credible test program without taking on a liability the lab could not carry. Brightwater sells results that clients submit to their own approval bodies, so a weak safety record or an unsupported method would damage the client's submission as well as the lab's standing. The evidence was partial but informative. The switchgear study was current and the stored energy calculation was sound. The vent sizing was the one document resting on assumption. Reviewers tend to sort findings into three groups: those that block energization, those that block qualification testing, and those that can close during operation, because each group carries different consequences for schedule and liability.
Constraints were practical. The vent duct could not be upsized without a contractor visit and a revision to the building file, which the facilities group estimated at about five weeks. Interim options existed, but each carried a cost: a temporary exhaust hood, a lower energy limit per test, or a restricted operator roster. Stakeholders had different stakes. Technicians would work beside the bay every day and had the most to lose from an unclear interlock. The client's compliance team needed data that would survive scrutiny. The building safety officer owned the sign-off and would not accept verbal assurance, so the review had to show which controls were installed, which were specified, and which were assumed.
- Before any installationHazard analysis and arc flash study checked against the actual room layout.
- In the first weekOff-gas measurement commissioned; vent sizing held until results arrive.
- Before first energizationGas interlock trip witnessed; building safety officer signs the hazard file.
- During low-energy checkoutRestricted roster and reduced voltage limit until competency records complete.
- Before full energy useDuct upsizing verified, labels corrected, full hazard file accepted by client.
Actions taken
- Hold energization until off-gas is measured on a representative module of the client's chemistry and the vent sizing is recalculated from that data.
- Commission the gas detection interlock and test it with a witnessed trip before any module is powered, and record the result in the hazard file.
- Allow low-energy checkout at a reduced voltage limit with a restricted operator roster, so technicians can finish training while the duct work proceeds.
- Add arc flash labels and a portable cable inspection to the approval checklist, and have the building safety officer sign the revised hazard analysis before qualification starts.
Results
The off-gas measurement showed a higher release rate than the generic estimate, which confirmed the duct needed upsizing. Qualification testing started about five weeks later than the client wanted, after the duct work and a witnessed interlock trip. Low-energy checkout gave technicians enough supervised practice to clear their competency assessments, and the arc flash labels were corrected in the same week. The client accepted the delay but asked for the full hazard file in its approval package. Two further review cycles were needed before the bay was signed off for full energy.
Questions for discussion
- Which of the three groups would you place the missing vent sizing in, and who should make that call?
- When a client's launch depends on test results, how much interim operation is acceptable before the hazard file is complete?
- What does a witnessed interlock trip establish that a written specification cannot?
Case Study 4
Pilot line acceptance slips on a late exchanger
Tallow Creek Engineering, a design and build group that delivers pilot processing lines for specialty chemical and food customers.
The situation
Tallow Creek is building a pilot processing line for a specialty chemicals customer that will run its own development trials on the equipment. The contract value is 1,250,000, with 15 percent of it held back until commissioning acceptance, which is due at the end of the quarter. Forty-two days ago the heat exchanger supplier reported that the shell would arrive late after a casting defect in the first batch. Around the same time, the controls group found that the outlet temperature loop could not hold setpoint within the customer's tolerance. Project spend stands at 412,000 against a budget of 380,000.
The challenge
The project manager needed a recovery that protected the acceptance date without creating a second failure. Options included expediting a shell from a second source, accepting partial commissioning on a temporary exchanger, or extending the date through a change order. Each moved cost, risk and retention timing differently, and the customer had already said it would not accept a line that ran but could not hold temperature.
Analysis
The business objectives were the retention payment, the customer relationship, and a margin already thinned by the first overrun. The evidence was firm on supply and weak on design. The supplier's revised ship date was in writing, but the controls fix rested on a simulation that had not been checked against the exchanger's actual response curve. A recovery built on one unverified path can fail on the other, so it helps to separate the steps that depend on a supplier from the steps that depend on an assumption before comparing options. Engineers who have run pilot systems tend to do this almost automatically, and it saves weeks of argument later.
Constraints were tight. The site crew could work extended hours for three weeks at most before fatigue rules applied, and the customer's own trial calendar was fixed. A temporary exchanger needed a pressure rating sign-off first. Expediting added 46,000, partial commissioning 26,000, and a change order 41,000 with six weeks of retention delay. Stakeholders read the trade-offs differently. The customer's process team wanted reliable data more than an early date, finance wanted the overrun capped, and the site crew would absorb the extra hours. The project manager had to propose a recovery all three could live with.
Actions taken
- Place an expedite order with the second exchanger source, accepting a higher unit price for a shipment that recovers about two of the three lost weeks.
- Have the controls group test the revised loop on the bench against the real exchanger response curve before committing any further installation labor.
- Offer the customer partial commissioning on the temporary exchanger, with a written outlet temperature limit and a change order covering the remaining scope.
- Circulate a weekly variance log of schedule and cost movements to the customer, finance and site leads on the same day, so no one is working from a different number.
Results
Partial commissioning was signed on the original date, and retention for that portion was released. Full acceptance followed about three weeks later, once the new exchanger arrived and the bench-tested loop was installed. The total overrun settled at 58,000 above the original budget, compared with roughly 78,000 under the expedite route. The customer ran its first trials inside the written limit, and its process team has since asked for a full-scope test, scheduled under a separate change order.
Questions for discussion
- Which recovery would you have recommended on the first day, and which assumption would have changed that view?
- How much weight should a supplier's written ship date carry when the controls fix is still unverified?
- When a customer accepts a limited scope, what obligations does the contractor keep for the rest of the work?
Case Study 5
Cutting chamber energy and offcut waste without drift
Greyfield Materials Testing, a commercial lab that runs fatigue, aging and climatic tests on polymer and composite samples.
The situation
Greyfield operates fourteen environmental chambers that hold samples at set temperature and humidity for campaigns lasting several weeks. Submeter data showed the chambers drew about 38,000 kilowatt-hours a month, and a large share went to holding idle chambers at setpoint between campaigns. Sample preparation added a steady stream of offcuts and single-use fixtures, roughly 1,600 kilograms of plastic waste per year. Clients compare results across campaigns, and the accredited method sets tight tolerances on chamber conditions. Management wanted visible reductions in energy and waste within the next budget cycle.
The challenge
Any change had to leave test results unchanged. A chamber that held conditions a little less tightly could invalidate a series, and the quality manager would not sign off a change that had not been validated against the accredited method. Savings also had to avoid relaxing tolerances, since clients would notice drift before they noticed a lower energy bill. The team had to find reductions that worked inside the existing method.
Analysis
The business goal was cost, but energy and waste reduction also mattered to two large clients who had asked for footprint reporting on their own samples. Evidence was good on energy because the chambers had submeters, and reasonable on waste, where the team had weighed offcuts by fixture type for a month. Lifecycle thinking mattered here. An option that cut chamber energy but shortened fixture life could leave total footprint unchanged, and a waste cut that raised rework had the same problem. Each option was compared on energy per valid test hour, waste per campaign and rework rate, not on any single figure.
The constraints were operational. A chamber could not be switched off without roughly two days of re-stabilization, and some campaigns ran for six weeks without interruption. Reusable fixtures needed cleaning validation, and cleaning used water and solvent, which created a waste stream of its own. Stakeholders included chamber technicians, who would handle new fixtures daily, the quality manager, who held the accreditation risk, and clients, who cared most about schedule and comparability. Options were judged against these limits rather than against an ideal lab that never existed.
- Metering Submeter data shows chamber draw by campaign and idle period.
- Standby scheduling Idle chambers held within tolerance until 24 hours before campaigns.
- Fixture trials Reusable fixtures tested with cleaning validation against single-use baseline.
- Quarterly reporting Energy per valid test hour and waste per campaign shared with clients.
Actions taken
- Move idle chambers to a standby setpoint inside the accredited tolerance band, and bring each one to full conditions about 24 hours before a campaign starts.
- Set up a cutting station with a labeled bin for each polymer grade, so clean offcuts can go back to a recycler without mixing with contaminated material.
- Trial reusable sample fixtures on one campaign with a cleaning validation run, and compare breakage and rework against the single-use baseline.
- Share energy per valid test hour and waste per campaign with clients each quarter, attaching the validation results so comparability stays visible.
Results
Standby scheduling reduced chamber energy by roughly a fifth over the following quarter, mostly from idle periods between campaigns. Offcut segregation raised the share sent for recycling to a clear majority of the stream. Fixture trials kept breakage within the single-use range for two of three polymers, while the third showed more rework and stayed on single-use fixtures. The quality manager signed the standby schedule once two campaigns showed no drift in control data. Some of the gain was offset by extra cleaning labor that the team had underestimated.
Questions for discussion
- Which measure would you trust most to judge a standby schedule, and which would you distrust?
- When a reduction in waste raises rework somewhere else, how far down the chain should the team look before deciding?
- What would a client need to see in a quarterly report before accepting a lower energy figure as comparable?
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.