Civil, Structural and Construction Engineering Case Studies
Technical5 Case StudiesUpdated Sep 2026
These five cases come from routine practice in civil, structural and construction work: a bridge bracket that cracked under light use, a precast line losing output and quality together, a basement excavation held for missing evidence, a steel milestone that slipped after a late design change, and a concrete operation cutting waste. Each one is told from the position of the person who had to decide. Every analysis leans on the same four records, design constraints, operating data, quality tolerances, and maintenance history, because those usually decide what can be fixed.
Case Study 1
Cracked bearing bracket on a pedestrian crossing
Corvane Structural Group, a consultancy of about 140 engineers and technicians that designs and inspects bridges and parking structures for municipal clients.
The situation
A regional transit agency commissioned a steel pedestrian crossing over a six-lane arterial road. The span was designed as a simply supported truss with elastomeric bearings at each abutment and a welded steel bracket that carried the bearing seat. Corvane's design passed independent review with no open comments. Eleven months after opening, a maintenance technician found a crack running about 60 millimeters along the toe of a fillet weld on the north bracket. The crossing had been closed for two days when the bracket was found, and the agency asked Corvane to determine why a detail that met code had cracked in service.
The challenge
The fracture surface pointed to fatigue, not overload, but the crossing carried far less pedestrian load than the design basis assumed. Operating counts from the agency's own sensors showed daily crossings well under the design figure. Shop drawings showed a weld access detail that had been revised late, and the inspection records for the fabricator were incomplete. The team had to decide whether the cracking reflected a design flaw, a fabrication deviation, or a maintenance gap, because each answer led to a different repair and a different liability position.
Analysis
Teams grounded in Civil, Structural and Construction Engineering usually start by separating the failure mode from the failure location. The crack sat in the heat-affected zone of a fillet weld, a classic site for fatigue initiation under repeated small stresses. The bracket was sized for static pedestrian load and wind, but nobody had checked stress cycles from footfall-induced vibration. Quality tolerances in the fabrication drawings allowed a weld reinforcement profile that was acceptable for static strength but not for a detail with a sharp toe. Each of those is a design question, and each leaves a trace in the drawings and the shop records.
The corrective options depend on what the records actually show. Sensor logs gave a usable crossing count and a rough acceleration record, enough to estimate cycles at the bracket against the fatigue category of the detail. The fabricator's file had no weld profile measurements for the north bracket, so the team could not say whether the flaw was a deviation or a design gap. Maintenance records helped more than expected: a technician had logged a loose bolt at the same location two months earlier. The realistic repairs were a gusset retrofit with a smoother transition or a machined replacement collar. A damper was cheapest but treated the symptom.
- 1Load and cycle reviewMeasured crossing counts and footfall data replaced the design basis estimate.
- 2Detail geometry checkToe profile of fillet weld checked against drawing and weld procedure.
- 3Fabrication record reviewInspection file lacked weld profile measurements for the north bracket.
- 4Maintenance log reviewA loose bolt logged at the same location two months earlier.
- 5Corrective design optionGusset retrofit with smoother transition, or machined replacement collar.
Actions taken
- Pull the fabricator's weld procedure and inspection records for every bracket, then measure the toe profile on the remaining three crossings before any of them reopens.
- Run a fatigue check on the bracket using measured crossing counts and the detail category, and report the stress range separately for footfall vibration and wind.
- Install the gusset retrofit with a smoothed transition and a full penetration weld, with ultrasonic testing on every new weld before the crossing reopens to the public.
- Add a fatigue-rated inspection interval to the maintenance plan and require technicians to log any loose fasteners at bracket locations, with a closeout check at the next quarterly review.
Results
The gusset retrofit went in during the third week, and ultrasonic testing found one porosity indication that was ground out and re-welded. Measured toe profiles on the other crossings were within tolerance, pointing to one bad detail rather than a batch problem. The fatigue estimate, built from measured counts, put the detail far below its design basis, so the weld was the weak point rather than the load. The agency reopened the crossing after six days, and bracket inspections now sit at each quarterly review. Corvane now requires a documented toe profile check on every bracket weld.
Questions for discussion
- Which of the three explanations, design, fabrication, or maintenance, would you test first, and what single measurement would rule out the most?
- How should a designer document cyclic loading assumptions when the owner's usage data is incomplete?
- When fabrication records are thin, what evidence is strong enough to support a repair decision?
Case Study 2
Wall panel line losing output and quality together
Brackenridge Precast, a plant of about 210 workers across two shifts that casts insulated wall panels and stair units for commercial contractors.
The situation
Brackenridge Precast runs a single casting line for insulated wall panels, with a shuttering table, a rebar cage station, a pour station, a steam-curing bay, and a finishing and inspection area. Over one quarter the line fell from an average of 38 finished panels per shift to 29, while rejected panels rose from about 2 percent of output to nearly 7 percent. Most rejects were surface cracking at the corners and embed plates set outside the drawing tolerance. The plant manager, Priya, wanted a recovery plan that did not add a second shift, and the quality lead, Tomas, wanted the rejects traced to their source before anything changed on the floor.
The challenge
Throughput and quality were moving together, which suggested a shared cause, but the line had changed in several ways during the same weeks. A new admixture supplier had been qualified, the steam curing profile had been shortened to free the bay, and the rebar cage station had a new operator. Operating data from the controllers was patchy. The team needed to find which variable explained the cracking and the embed errors, and how much of the lost output came from waiting on curing versus rework at finishing.
Analysis
The process map came from the travel tickets, not from memory. Each panel carries a ticket that records the time at each station, so the team could see that the steam-curing bay, not the pour station, held the most queued panels. Design constraints set the edges of the problem: the drawings allowed plus or minus 3 millimeters on embed plates and a minimum cover of 25 millimeters at the corners, which left little room for cage drift. Variation came from three places: the cage station's tie spacing, the admixture batch, and the shuttering table's level. Each one drives a different defect.
The maintenance log for the shuttering table showed a level check skipped for two weeks before the first embed rejects appeared, which fit the drawing tolerance failures. Controller data for the steam bay showed the ramp rate had been cut and the hold temperature was not reached on several cycles, which explained much of the corner cracking as concrete cured too quickly. The admixture batch test records were in order, so the supplier change was a weaker suspect than it first seemed. The team ranked each fix by rejected panels removed per shift and by how easily the change could be checked against a measurable limit.
- Start of quarterOutput averages 38 panels per shift with rejects near 2 percent.
- Week oneLevel check and curing alarm go in, and tie gauge catches bad cages.
- Week twoCorner cracking falls after the steam-curing profile is restored.
- Fifth weekThroughput reaches 33 panels per shift, still below the 35 target.
- Sixth weekRejects settle at 2.6 percent, under the 3 percent limit.
Actions taken
- Restore the daily shuttering table level check as a signed step on the line sheet, with a limit of 2 millimeters over the table length and a stop rule if it fails.
- Reinstate the original steam-curing ramp and hold profile, and set an alarm on the controller when the hold temperature misses target for more than ten minutes.
- Add a cage tie spacing gauge at the rebar station and reject any cage with ties outside 150 millimeters on center before it reaches the pour table.
- Set the acceptance test for the next four weeks at 35 finished panels per shift with rejects under 3 percent, and review the schedule only after both targets hold.
Results
The level check and the curing alarm were in place by the end of the first week. Corner cracking fell within two weeks, and embed rejects dropped after the tie gauge caught two bad cages. Throughput recovered more slowly. The line reached 33 panels per shift in the fifth week because the restored curing profile added time to each cycle, and it stayed below the 35 target. Rejects settled at 2.6 percent, under the 3 percent limit. Priya accepted the lower output, and the reject trend became the main measure at the weekly review.
Questions for discussion
- Which of the three variation sources would you measure first, and what data would show it is the dominant one?
- How should a plant decide between recovering throughput with a second shift and holding the curing profile that protects quality?
- What acceptance limit would you set for reject rate when the line is still learning a new process?
Case Study 3
Deep basement excavation next to an occupied building
Ardenne Site Engineering, a firm of about 75 designers and site engineers that prepares temporary works and excavation support packages for contractors.
The situation
Ardenne prepared the temporary works review for a four-level basement dug beside a six-story office block. The contractor proposed a secant pile wall with two rows of steel props, excavating in lifts of about 2.5 meters. The design depth was 9.2 meters, and the water table sat about 2.1 meters below the existing ground. The client's engineer, Marisol, asked for a formal review before the first lift, because the adjacent building had a shallow raft foundation and no recorded settlement history. The package arrived with a risk register, two boreholes, and a method statement that did not yet state how props would be removed.
The challenge
Three hazards were live at once. The face of an excavation of this depth in layered fill can fail without warning, the shallow raft next door could move as groundwater was drawn down, and perched water above the clay could flood the base during wet weather. The register rated each hazard but gave no evidence for the ratings. Most of the controls were described as monitoring or as to be confirmed, and nothing in the file showed who would stop work if a reading crossed a limit.
Analysis
The review began with the standards that govern the work rather than the drawings. The excavation and temporary works were checked against the local building code and the consensus guidance for excavation support, and the review noted that the method statement did not reference the permit for an excavation of this depth. Risk controls were assessed by asking whether each had a trigger, a responsible person, and a documented response. Props had no stated preload or removal sequence, and the monitoring plan set alert levels only as a percentage of an unmeasured baseline. A control without a trigger is a description, not a control.
Operating data was the weakest part of the package. Inclinometer baseline readings covered only two weeks, too short to separate seasonal groundwater change from the start of excavation. There were no settlement readings for the adjacent building, so any movement after the first lift would have no reference. Maintenance records for the dewatering pumps and hydraulic prop jacks were absent, which mattered because both would be running continuously through the wet season. The review therefore set approval gates rather than a single yes or no: each lift needed a signed monitoring check, a verified prop preload record, and a pump service entry before the next lift could start.
Actions taken
- Require a baseline of at least four weeks of inclinometer and settlement readings on the adjacent building before any excavation below the existing ground level.
- Add a prop removal sequence to the method statement, with a preload check at each level and a named engineer who signs before any prop is released.
- Set three-tier alert, action, and stop levels for wall movement, each tied to the measured baseline and to a named person who can halt work.
- Obtain dewatering pump service records and a standby power plan, and confirm that the perched water monitoring well is installed before the wet season begins.
Results
Ardenne issued a conditional approval, with the first lift held for eleven days while the contractor installed two additional monitoring points and filed the pump records. The four-week baseline was not complete, so the review accepted a three-week baseline plus daily readings for the first lift with alert levels to be recalculated. The prop removal sequence was added and signed by the contractor's engineer. The standby power plan revealed that the pumps had no backup supply, so a generator was added at the entrance. Early readings on the adjacent building stayed inside the alert band.
Questions for discussion
- Which controls in this package would you have refused outright, and which would you have accepted with a condition?
- How should alert levels be set when the baseline record is shorter than the seasonal change in groundwater?
- What evidence would you need from a contractor before accepting a prop removal sequence written by someone else?
Case Study 4
Steel erection slips after a stiffener design change
Hollin Ridge Contracting, a general contractor of about 320 staff running mid-rise medical and office projects for private owners.
The situation
Hollin Ridge is the general contractor on a four-story medical office building with a structural steel frame over a podium of cast-in-place concrete. Steel erection was scheduled to finish by the end of the eighth week, which fixed the owner's enclosure date. In the third week, the engineer of record revised 14 beam-to-column connections to add stiffeners after a coordination review with the mechanical team. The fabricator had to re-issue shop drawings and remake 22 connection plates. Ines, the project manager, learned the steel would arrive six working days late and the enclosure date carried a daily liquidated damages rate.
The challenge
Recovery options all cost money, and they traded against each other. Overtime at the fabricator could recover some days, but the fabricator's shop was already running two shifts. Expediting the plates by air freight shortened the gap but added cost per plate. Resequencing so mechanical crews started behind the steel would protect the enclosure date but cost crane time. The team needed to quantify each option against the penalty exposure and the cost of an unfinished enclosure.
Analysis
The slip itself was small, but its cause was a design change inside a coordination loop, which is where recovery plans usually fail. The stiffeners added weld length at connections already close to the limits in the fabricator's weld procedure, and the new detail needed a fresh check against qualified welder ranges. Quality tolerances on the erection were tight: column plumb within a set limit, and bolt pretension checked on a sample of connections. The schedule had float on the enclosure path of only four working days, so any recovery that did not start immediately would leave the milestone exposed.
Three recovery plans were costed against the damages clause, at $6,000 for each working day past the enclosure date. Fabricator overtime recovered about three of the six days for roughly $18,500, leaving $18,000 in damages and a total near $36,500. Expedited freight on the plates recovered four days for $31,000, with $12,000 still exposed, a total near $43,000. Overtime combined with resequencing, with mechanical rough-in moved behind the steel, recovered the remaining three days at about $9,400 in added crane standby, for a total near $27,900 and no damages. Resequencing carried a quality cost: rough-in at the stiffened bays had less access and penetration tolerances were tight.
- Design change Stiffeners added to 14 connections after a coordination review.
- Fabricator rework Shop drawings reissued and 22 connection plates remade in the shop.
- Erection queue Late plates hold crane lifts and push mechanical rough-in back.
- Enclosure exposure Six working days late, with damages running at a daily rate.
Actions taken
- Ask the fabricator for a written overtime commitment with named shifts and a plate delivery schedule, and tie the next progress payment to it.
- Freeze the 22 revised connection plates in a single approved shop drawing set, and reject any further stiffener change without a written cost and schedule impact.
- Resequence the mechanical rough-in to start behind the steel on each floor, with a crane booking log so standby time is measured against a daily rate.
- Hold a joint review with the mechanical and structural foremen each morning for the first two weeks of recovery, and log every access conflict at the stiffened bays.
Results
The combined plan was adopted in the third week after the revision. The fabricator delivered the plates across three shifts, and steel erection finished one day ahead of the enclosure date. Crane standby came to $10,100, about $700 above the estimate, because two lifts waited on a stiffener that arrived late. Rough-in at the stiffened bays needed two extra access reviews, and the sleeve penetrations in those bays passed inspection on the second attempt. The owner paid no damages, and the project manager reported the recovery cost as a single variance line rather than a schedule penalty.
Questions for discussion
- At what daily damages rate would you have chosen freight over overtime, and what assumptions would you check first?
- How should a contractor price the quality risk of compressing trades into a stiffened area?
- When a design change starts a schedule slip, who should own the cost and schedule estimate, and when?
Case Study 5
Cutting concrete waste and energy without losing strength
Dunmore Gray Ready Mix, a supplier of about 95 staff running two batch plants and a small precast yard for bridge and foundation contractors.
The situation
Dunmore Gray runs two batch plants that supply ready-mixed concrete to bridge decks, foundations, and slab work across a metro area. Its energy use per cubic meter had risen as the fleet aged, and return concrete from unused loads was being dumped into a settling pit at roughly 4 percent of production. The operations director, Kenji, asked for a plan that cut energy, waste, and embodied carbon without giving up the strength and durability classes that the bridge clients specified. Two design constraints framed the work: every mix had to meet its specified compressive strength at the test age, and the plants could not change their slump tolerance band without a retest.
The challenge
The options ranged from plant housekeeping to replacing part of the cement with supplementary materials not yet used in their mixes. Each option promised a different share of the 4 percent waste and a different energy effect, but the measured data was uneven. Mixer drum energy had been logged for only one plant, and the cement content on several mixes was rounded to the nearest five kilograms, which made lifecycle comparisons rough. Options had to be provable on the plant floor within one test cycle.
Analysis
Start with what cannot move. The strength class and durability exposure set the minimum cement content and the water-cement ratio, so the cement reduction was bounded by the specification, not by what the plant wanted. Supplementary materials can cut embodied carbon per cubic meter, but they change early strength gain and setting time, which affects stripping and the pour schedule. Tolerances on slump and air content already limit what a mix can tolerate, so a substitution that moves one property usually moves another. Lifecycle comparison has to include transport, because a material that is cheaper per ton can cost more once it travels across the metro area.
The operating data gave the team a baseline it could trust for two things: energy per cubic meter from the plant meters, and waste from the settling pit log. Maintenance records told a different story. The drum drive on the older mixer had needed frequent bearing service, and its energy draw climbed before each failure, so the energy saving from new equipment was partly a reliability saving. The team compared options by carbon per cubic meter delivered, waste diverted, and the change in unplanned downtime, and capped any rise in rejected loads at 0.5 percent. A housekeeping change that cut idle time was cheap but small.
Actions taken
- Fit the two oldest drum drives with energy meters and bearing condition logs, so that efficiency gains can be separated from reliability gains in the baseline.
- Install a return concrete reuse system that holds returned loads in a dedicated tank and feeds them back into low-strength mixes, starting with two trial batches per week.
- Tighten batch tolerances on cement and water to one kilogram on cement, then measure whether the rounding change alone reduces overdosing and waste.
- Run a 90-day trial of one supplementary blend on a single mix family, with strength cylinders tested at seven and twenty-eight days before any wider release.
Results
Energy meters showed the older drum drive used about 11 percent more power per cubic meter, and its bearing logs linked that gap to frequent service. Tighter cement batching cut overdosing on 6 percent of loads. Return concrete reuse diverted about 60 percent of the settling pit volume during the trial. The supplementary blend met strength at seven and twenty-eight days but gained strength more slowly early, so it was limited to mixes without an early stripping requirement. Waste fell to about 2 percent of production, and rejected loads rose by 0.2 percentage points, inside the cap.
Questions for discussion
- Which measure would you weight most heavily when two options trade carbon savings against early strength, and why?
- How would you decide whether a bearing failure history belongs in the energy baseline or in a separate reliability case?
- What cost of a rejected load would justify a slightly higher waste rate in exchange for a larger carbon reduction?
Learning Resources
Technical 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.