Engineering, Architecture and Construction Consulting Case Studies
Professional5 Case StudiesUpdated Sep 2026
Consulting in engineering, architecture and construction often begins after something has gone wrong, or after a plan has been drawn but not tested. The cases below cover a failed compressor housing, a fabrication line losing output to rework, a rooftop retrofit with missing evidence, a late switchgear delivery, and a precast yard cutting energy use. Each draws on four sources: design constraints in the basis of design, operating data from logs, quality tolerances that define acceptable work, and maintenance records showing what was done to the asset. Those sources rarely agree, so the work is largely deciding which to trust.
Case Study 1
Repeated compressor housing cracks in a cold-storage plant
Brackwater Refrigeration Systems, a mid-sized firm that designs and services ammonia cold-storage plants for regional food distributors.
The situation
Brackwater was retained by a regional food distributor whose cold-storage facility had lost its second screw compressor housing in under two years. The original design was sized for a continuous load within the vendor's rated envelope. The basis of design documented oil injection rates, suction superheat targets, and a vibration limit on the mounting frame. Operating logs showed the compressor cycling far more often than the design assumed, because a control change made during commissioning widened the suction pressure band. Maintenance records held three oil changes logged late and one filter replaced with a non-specified part. The owner wanted a cause, a fix, and a recommendation on whether to replace the whole unit.
The challenge
The housing cracks sat at a fillet near the discharge port, and the pattern looked like fatigue, but fatigue alone did not explain why the first failure came so early. The vendor blamed liquid slugging during defrost. The maintenance lead blamed a contaminated oil batch. Brackwater had to decide which evidence to trust, and the data had gaps. The logger was blank for nine days during the second failure period, and oil samples were taken only at quarterly service visits, so no trend could be drawn from them.
Analysis
The first move was to treat the basis of design as a contract rather than background reading. Every later judgment depends on knowing which numbers the designer promised and under what conditions. In this plant, the suction pressure band and the oil return rate were the two values that mattered, and both had drifted from the envelope after commissioning. Overlaying the operating log on that envelope showed the compressor spending a large share of its hours near the low edge of the oil return window. That did not prove the cause, but it changed the question from why the housing cracked to which operating states the housing had actually seen.
Maintenance records were the weakest source and the most revealing one. The late oil changes clustered in the months when the site ran with two technicians out, and the non-specified filter had a coarser rating than the original. Brackwater set these records against the metallurgy report, which placed the crack origin at a weld toe with striations consistent with cyclic loading. Corrective options were then sorted by what each would remove. A revised control band addressed cycling, a fixed-hour oil interval addressed return rate, and a larger fillet radius addressed the stress concentration. None of the three alone justified replacing the whole unit, which was the point the owner most needed to hear.
- At commissioningSuction band widened to reduce short-cycling complaints
- Eleven months laterFirst housing crack found at discharge fillet
- Two months after repairSecond crack; logger blank for nine days
- Following quarterControl band restored and fillet redesign approved
- Replacement cycleEnlarged-radius housing installed with fixed-hour oil service
Actions taken
- Restore the suction pressure band to the original design envelope and log the change with the commissioning engineer's signature, so control setpoints match the basis of design again.
- Replace calendar oil changes with fixed run-hour intervals, and require oil analysis at every service so return-rate trends appear in the monthly report.
- Specify a larger fillet radius at the discharge port on any replacement housing, and request a fatigue assessment from the vendor before accepting the revised drawing.
- Install a continuous logger with a backup channel and a gap alarm, so that a nine-day hole in the trend data cannot recur during a future investigation.
Results
The revised control band cut compressor cycling within the first full month, and the logger showed suction pressure holding inside the design envelope for most of the following quarter. The owner accepted a replacement housing with the enlarged fillet, installed in the third week after the drawing was approved, and no new cracks appeared at the next inspection. The original frame was kept, which avoided a larger capital request. The defrost theory was tested against the recorded liquid levels, which stayed within limits, and the vendor's claim was set aside. Fixed-hour oil intervals were adopted as standard across the owner's other sites.
Questions for discussion
- If the replacement budget had been capped, which of the three corrective options would you have funded first, and what evidence would have changed that order?
- How should a consultant weigh a maintenance log with known gaps against a logger that has its own blind spots?
- What would the investigation have looked like if the metallurgy report had come back inconclusive?
Case Study 2
Falling throughput and rising rework on a fabrication line
Tallis Ridge Metalworks, a contract fabricator of structural steel connection plates and bracket assemblies for commercial building projects.
The situation
Tallis Ridge Metalworks fabricates bolted connection plates and bracket assemblies for mid-rise building projects. The line runs cutting, punching, press brake forming, weld preparation and final inspection on two shifts. Over two quarters, output per shift fell from 412 pieces to 317, while rework rose on the final inspection report. The client's shop drawings set a hole position tolerance of half a millimeter and a bend angle tolerance of one degree. The shift log recorded parts started, completed and returned, and the maintenance file showed the press brake backgauge recalibrated twice in the period, along with a punch set swapped without a documented first-article check.
The challenge
Management assumed the press brake operators were slowing down. Dario, the shift supervisor, suspected the steel coil lots. Rework ran at 9 percent on the second shift and under 3 percent on the first, which pointed to something specific to one shift rather than the whole line. The team had three weeks before a client audit and could not stop production for a long study.
Analysis
Start with a map of where parts wait and where they are measured, because the flow chart alone hides the constraint. Here the backgauge station limited throughput, sitting between a punching step and a bending sequence that ran at a different tempo. Consultants trained in Engineering, Architecture and Construction Consulting usually separate common-cause variation, which the process produces on any shift, from special-cause variation, which traces to a specific change. The two shifts differed less in operator speed than in how often they changed punch sets. The second shift changed sets more often, and each changeover had skipped the first-article check, so defects entered the line at a predictable moment.
Tolerances then decided what to fix first. The hole position limit of half a millimeter was tight enough that a backgauge drift of a few tenths would push parts out of spec, which explained the concentration of rework at the bend stage on one shift. Improvements were ranked by how much rework each would remove per hour of effort. A locked changeover check ranked first because it was cheap and aimed at the defect that clustered. Backgauge verification with a reference block ranked second. Coil lot sampling ranked last, since the evidence for it was weakest. Each action was given an acceptance criterion before work began: rework under 3 percent on both shifts for two consecutive weeks.
Actions taken
- Write a standard punch set changeover sequence that ends with a first-article check on hole position within half a millimeter, and do not allow the line to restart until it passes.
- Verify the backgauge against a reference block at the start of each shift and log the reading, so recalibration drift shows up in the shift record rather than in rework counts.
- Build a daily sheet that records each rework reason by station and shift, reviewed at the morning meeting, so the supervisor can see which shift carries a defect and act that day.
- Set the acceptance criterion at rework below 3 percent on both shifts for two consecutive weeks, and keep coil lot sampling as a low-rate secondary check.
Results
Within the first week under the new changeover standard, second shift rework fell from 9 percent to under 4 percent. Over the following two weeks both shifts held below 3 percent, which met the acceptance criterion. Output per shift recovered to about 390 pieces, short of the earlier 412, because operators now spend several minutes on each changeover. The team judged the trade worthwhile, since fewer rejected brackets saved more time than the checks consumed. The coil lot question stayed open at a low sampling rate. The client audit went ahead on schedule, and the first-article records were accepted as evidence of process control.
Questions for discussion
- If the second shift had changed punch sets no more often than the first, what other evidence would you look for to find the source of its rework?
- Where should a line place the boundary between a hard stop that protects quality and a looser check that protects output?
- How would you set an acceptance criterion for a defect that only appears at final inspection?
Case Study 3
Hazard review of a rooftop retrofit over occupied space
Pellwick and Strand Building Engineers, a consultancy that reviews mechanical, electrical and structural packages for owners and general contractors on medical and commercial projects.
The situation
The owner of a four-story medical office building asked for a review of a rooftop retrofit ahead of permit. The package added a heat recovery unit on a new curb above an occupied clinic floor, with a new exhaust stack and a new electrical feed. The design constraints were roof live load capacity, setback distances from operable windows and air intakes, and clearance from the parapet. The operating data included occupancy counts on the clinic floor and exhaust readings from the building. Maintenance records covered the last roof inspection and two past incidents of crews working near the parapet without a plan. The drawings were marked issued for construction, but the structural sheet for the curb had no stamp.
The challenge
The review turned up three hazards and two gaps in evidence. The added curb load appeared to exceed the allowance for one bay, with no calculation on file. The exhaust discharge sat inside the setback from an operable window. The lockout plan referred to a disconnect not yet installed. Roof access had no fall protection plan, and the only anchor certificate on file belonged to an earlier project on a different building.
Analysis
The review ranked each finding by severity. The first control question was whether a hazard could be removed. Relocating the exhaust and choosing a lower-load bay for the curb removed two exposures without added equipment. Where elimination was impractical, engineering controls came next: reinforced joists, a guardrail at the roof edge, and a labeled isolation point. Administrative controls such as a crew briefing and a written access procedure sat below those. The building code, the local fall protection rule and the energy code's setback each set a floor, and the review checked against all of them.
Documentation was where the package fell short, and the gaps were easy to miss because the drawings looked complete. A stamped structural calculation for the curb was absent, so no reviewer could confirm the load. The anchor certificate was out of date for the building, and the lockout plan cited a device without a tag number. Approval gates should close in order: structural sign-off before the permit, a fall protection and access plan before mobilization, and owner acceptance of isolation and energization before power is applied. The review file listed each hazard, its control, its evidence and the signature that closed it.
- Identify hazards List roof load, exhaust, access and isolation hazards from drawings
- Verify evidence Request calculations, certificates and existing roof records
- Set controls Apply hierarchy: relocate, guard, isolate, then train crews
- Gate approval Hold release until structural and safety sign-offs are filed
Actions taken
- Hold the permit application until a structural engineer stamps the calculation for the added curb, including snow and equipment service loads on the existing deck.
- Move the exhaust discharge to a height and distance that clears the operable windows, and dimension the revised setback on the drawings so the reviewer can check it.
- Require a fall protection plan and a current anchor certificate for this roof, since the only certificate on file belongs to a different building and an earlier project.
- Tie the lockout plan to the installed disconnect by tag number, and have the electrical contractor walk the isolation sequence with the owner before energization.
Results
The structural engineer found the added curb pushed one bay above its allowable load, and the design team reinforced the joist section before fabrication. The exhaust was moved, and the revised drawings cleared the owner's safety gate in the third week of review. The access plan was accepted once the contractor supplied a new anchor certificate. Energization slipped about two weeks, which the owner accepted because the alternative was an unmanaged hazard above an occupied floor. The closed review file listed each hazard, its control, its evidence and its signature.
Questions for discussion
- Which of these hazards would you have ranked highest if the structural calculation had arrived on time but the exhaust setback had not been checked?
- How much weight should a consultant give to a certificate from an earlier project when the current building differs?
- What is the minimum evidence an owner should see before approving energization of new rooftop equipment?
Case Study 4
Delayed switchgear delivery threatens a clinic wing handover
Ostrander Field Engineering, a project controls group that advises owners and contractors on healthcare and laboratory construction schedules and budgets.
The situation
A clinic wing expansion carried a handover milestone at the end of the eighth month, when the owner needed to install imaging equipment. The new switchgear was a custom assembly whose bus rating had been revised after a load study. When revised shop drawings were released, manufacturer lead time had grown and delivery now sat eleven weeks past the milestone. Coordination then found the revised busway ran into a steel beam already erected. The construction budget stood at 4,200,000 dollars with 180,000 dollars of contingency. Operating data from the existing building showed breaker trips on older gear at afternoon peaks, which is why the owner asked for the higher rating.
The challenge
The team had two weeks to choose a recovery plan before the contractor locked its sequence. Three options were on the table: pay a premium to expedite the original gear, redesign around a stock unit with rework to the busway and beam, or energize in phases with a temporary generator. Each moved cost and schedule differently. None of the cost figures had been checked against the revised load, and the owner's delay cost had never been written down.
Analysis
The trade-offs only became clear once each option was priced in the same units. The owner's model put each week of delay at 22,000 dollars, so eleven weeks of slip carried 242,000 dollars of exposure. Expediting the original gear cost a 186,000 dollar premium and was expected to recover eight weeks, leaving a three-week residual slip. The stock unit redesign cost about 58,000 dollars in rework, recovered seven weeks, and left four weeks outstanding. The generator option cost 64,000 dollars per month and protected the imaging suite only, so the rest of the wing still waited on the gear.
The expedite option looked fastest, but it rested on a verbal date from the manufacturer, which is weak evidence for a premium of that size. The stock unit needed its own checks: switchgear pads had to align within three millimeters, and the revised busway needed clearance at the beam confirmed by a field survey before fabrication. Redesign also meant reissuing arc flash labels and updating the single line diagram. Ranking the options meant weighing dollars of delay against confidence in the date. The team chose the option with the lowest exposure per week of residual slip, and set a condition that any future expedite come with a written vendor commitment.
Actions taken
- Obtain a written delivery commitment from the manufacturer for the expedited unit, including a penalty clause, before paying the premium, because a verbal date does not move the schedule.
- Commission a short electrical and structural scope for the stock unit, limiting busway changes to two runs and confirming beam clearance with a field survey.
- Recover float by resequencing finishes and overhead rough-in in the imaging suite, so that room can take equipment before the full wing is energized.
- Rebuild the cost model with the revised load study and reissue it as a single table, so the owner decides on figures that match the current design.
Results
The owner chose the stock unit redesign with resequenced finishes. Rework cost 58,000 dollars, and the team recovered seven of the eleven weeks. A single month of generator support covered the imaging suite at 64,000 dollars, so the combined draw on contingency came to 122,000 dollars out of 180,000 dollars. The imaging suite opened four weeks after the original milestone. The expedite premium was not paid, since the manufacturer would not put a delivery date in writing. The remaining work held within the reduced contingency.
Questions for discussion
- How would the choice change if the manufacturer had offered a written date with a penalty clause for the expedited unit?
- What would you need to confirm about the owner's delay cost before accepting a slip with no recovery spending?
- When a redesign introduces rework to finished structure, how should the consultant test whether the fix is acceptable?
Case Study 5
Reducing curing energy and form waste at a precast yard
Corrin Vale Precast, a manufacturer of architectural concrete panels and structural elements that supplies regional building projects from one yard.
The situation
Corrin Vale wanted to lower steam use and material waste across its panel casting beds over the next two quarters. Panels cure under steam before stripping, and forms are plywood with release agent applied before each pour. The design constraints were a stripping strength that the structural engineer had set in the mix specification, and a dimensional tolerance of plus or minus three millimeters with a required finish on exposed faces. Operating data included steam consumed per casting cycle, rejected panels per quarter, and the number of uses per plywood sheet. Maintenance records showed steam trap failures on two beds and a boiler service that had slipped by three weeks.
The challenge
The target was a 15 percent cut in steam and waste without any drop in strength or finish. Several ideas were on the table: insulated curing enclosures, a mix with less cement and more supplementary material, longer form life with a different release agent, and reclaiming rinse water from the yard. Each would affect a different measure, and each carried some risk to the panels that left the yard. Testing had to happen on live production.
Analysis
A lifecycle view changes the ranking. Cement is the largest embodied input in a concrete panel, so a mix change can outweigh any energy measure, if strength holds at stripping and in service. Steam is an operating cost that recurs every cycle, which makes curing enclosures attractive on a daily basis. Form life and rinse water are smaller in absolute terms but easy to measure. Comparing options meant putting each on the same scale: energy per cycle, rejects per quarter, and risk to structural elements. A large steam saving that causes rejects scores lower than a small one with no change to finish.
Performance guardrails were set before any trial. Stripping strength was checked with cubes cured beside every batch, and no mix went into structural elements until its cubes passed for three consecutive weeks. Finish was judged against a visual standard agreed with the architect's representative. Scores reflected both saving and risk. The curing enclosure ranked highest because its saving was daily and its risk was limited to sealing checks on each bed. The mix change scored lower because its benefit depended on a test result that had not yet been earned.
Actions taken
- Pilot insulated curing enclosures on one casting bed for ten weeks, logging steam per cycle and stripping strength on every batch to confirm the saving holds.
- Trial the revised mix on non-structural panels first, with cube tests at stripping age and a finish inspection, before any structural element uses it.
- Change the release agent and track uses per plywood sheet, replacing a sheet only when its surface record shows finish loss rather than at a fixed count.
- Reclaim rinse water through a settling tank and filter, and log the volume reused each week so the water saving is measured rather than assumed.
Results
The enclosure pilot cut steam per cycle by 19 percent on the test bed, and stripping strength stayed above the required value on every batch. The mix trial passed on non-structural panels, and structural use was deferred to a second round. Rejected panels fell from 46 to 31 in the quarter after the changes, although that count also reflected a more careful pour sequence introduced at the same time. Form life rose, with fewer sheets scrapped at the planned count. The combined measures met the 15 percent target for steam, and rejects dropped by about a third.
Questions for discussion
- How would you weigh a measure that saves steam every day against a mix change whose benefit depends on a test that has not yet passed?
- What evidence would a structural reviewer need before accepting a lower cement mix for load-bearing panels?
- Which option would you drop first if the enclosure pilot showed only half the expected saving?
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.