Electrical Power Systems Engineering Case Studies
Technical5 Case StudiesUpdated Sep 2026
These five case studies come from electrical power work, each built around a situation a practicing engineer could meet on a live project. The scenarios cover a failed transformer, a stalled assembly line, a battery installation waiting for energization, a substation milestone under schedule pressure, and a hospital campus trying to cut losses. Each case draws on the same four kinds of evidence: design constraints, operating data, quality tolerances, and maintenance records. The organizations and figures are fictional. They are meant to show how those sources interact, and why a sound decision usually depends on reading them together rather than one at a time.
Case Study 1
Transformer Winding Failure at a Peak Load Substation
Harrowgate Utility Cooperative, a rural distribution utility serving about 38,000 customers across three counties
The situation
The Pell Creek substation feeds roughly 9,400 customers through two 25 MVA transformers that were installed when the station was built. In the middle of a heat wave, transformer T2 tripped on differential protection in the early evening, with load running near 91 percent of nameplate. Crews found a burned lead connection at the low voltage bushing and a scorched section of winding near the tap changer. The unit had been in service for a little over eleven years and had passed its routine tests on schedule. Nobody had flagged it as a priority, though the dissolved gas record told a less comfortable story.
The challenge
The utility needed to know whether the failure came from a manufacturing defect, an overloaded duty cycle, or a maintenance gap before deciding whether to rewind, replace, or redistribute load. The sister unit showed similar gas readings. A replacement transformer carried a lead time of several months, so the decision also determined how much of the service territory would depend on emergency switching for the rest of the season.
Analysis
Start with the design constraints. The nameplate rating assumed an average ambient of 30 degrees C, but the station had recorded hot-hour ambients well above that for three consecutive summers. Loading data from the supervisory system showed T2 spent long stretches above 90 percent during those afternoons, and the cooling stage logs show that only two of the three fan banks ran during several of those events. Combined with a hot spot calculation the vendor had never updated, the thermal margin was thinner than the original paperwork implied. The tolerance question matters too. The insulation paper specification allowed a moisture content that the oil test trends suggested was being approached from below.
Teams grounded in Electrical Power Systems Engineering usually check maintenance records against the physical evidence before drawing conclusions, because a clean test calendar can hide a slow trend. In this case the dissolved gas analysis had moved from a caution level into a warning band over two sampling intervals, yet the work order was closed with a note that the trend was within limits. Oil dielectric results were marginal but never out of specification. Taken together, the records point toward accelerated thermal aging of the paper, a process that is cumulative and rarely visible in any single reading. The defect hypothesis stayed open, since the winding wire lot certificate had never been compared against a retained sample.
- Hot spot calculation Recheck winding temperature using actual fan banks and ambient records.
- Gas trend review Track rate of change in dissolved gases, not only absolute values.
- Cooling capacity check Confirm every fan bank runs during afternoon peak loading.
- Maintenance closure Close work orders only after the trend is explained and documented.
Actions taken
- Pull the full gas-in-oil history for both units and set alarm thresholds on rate of change rather than absolute values, with a written escalation path for each band.
- Recalculate winding hot spot temperature using the actual cooling configuration and current ambient records, then compare the result against the insulation life assumed in the asset register.
- Send the failed lead section and a retained sample of the wire lot to an independent metallurgical laboratory, requesting a root-cause report before any rewind contract is signed.
- Restore full fan bank operation on T1 and add a load-shedding rule for afternoons above a defined temperature, reviewed each season by operations and engineering together.
Results
Forensic work confirmed a localized connection overheating combined with paper degradation consistent with long running at elevated hot spot temperatures. The laboratory found no defect in the wire. T2 was rewound on a contract that returned the unit to service after the next season. Rate-of-change alarms were adopted for both units, and the maintenance procedure now requires a review whenever gas trends cross a caution band, not only a warning band. Two cooling banks were repaired on T1, and the afternoon dispatch rule reduced measured hot spot excursions in the following summer. The team still plans a replacement study for the older unit.
Questions for discussion
- Which design assumptions in the original nameplate rating would you test first, and what data would you need to test them?
- How should a utility weigh a slow gas-in-oil trend against a closed work order that technically meets every limit?
- If the laboratory had found a manufacturing defect in the wire lot, how would the response change for the sister unit?
Case Study 2
Busbar Line Throughput Drop and Torque Defects
Corran Switchgear Works, a mid-sized manufacturer of low voltage panelboards and busbar assemblies for commercial buildings
The situation
Corran Switchgear Works assembles panelboards and main busbar sections on two lines in a plant of about 310 workers across two shifts. The busbar line, which drills, torques, and inspects copper bar assemblies before insulation sleeving, had run at 62 finished assemblies per shift for most of the previous year. Over the last several weeks output fell to 48 per shift, and final dielectric test rejects rose from under 1 percent to about 4 percent. Supervisors first blamed staffing, then a new supplier of insulating sleeves, but neither explanation matched the timing of the drop.
The challenge
The plant manager wanted throughput recovered before a large export order entered its final build window, yet the quality team would not accept more escapes to the test bench. The line had no single bottleneck on paper. Every station reported its own cycle time, and each owner argued that the problem sat upstream or downstream. The team had about three weeks to find a fix that held without adding a third shift.
Analysis
The process map shows seven stations, and the sleeve insertion step is the only point where the bar is handled twice. The torque specification calls for a value within a narrow band on each bolt of every joint, and the drilling fixture clearance was never rechecked against the drawing tolerance after a tooling change. Station logs record cycle time for every assembly, which lets the team separate true slowdowns from waiting. The torque station's average time was steady, but its spread widened sharply, and the widest spread lined up with the hour after changeover. Variation, not average speed, was driving the lost output.
Experienced line engineers often read maintenance logs before they look at operator practice, since a worn fixture or drifting tool can produce defects that look like training problems. The torque driver calibration records were current, but the work order history showed that two drivers had been swapped between stations without updating the station tag. Calibration checks were scheduled only in the morning, while the complaint pattern peaked after lunch. Those gaps matter because a driver that reads correctly on a bench can still deliver a wide spread of values under production vibration and hose wear. The sleeve supplier change was real, but its effect on the data was small.
Actions taken
- Verify torque driver calibration at each station after every changeover, and tag each driver to its station so the record follows the tool rather than the bench.
- Measure fixture clearance against the drawing tolerance with a coordinate check, and replace any locating pins that show measurable wear.
- Run a two-week control chart on torque values and cycle time, with a defined limit that stops the line rather than leaving a note in the log.
- Set acceptance criteria of at least 58 assemblies per shift, first-pass dielectric yield of 99 percent or higher, and no more than two torque excursions per shift across three consecutive weeks.
Results
Calibration moved to the start of each shift and to the first assembly after every changeover, and the fixture pins were replaced within the first week. Torque spread narrowed within two weeks, and throughput rose to 57 assemblies per shift by the end of the third week. First-pass dielectric yield recovered to just above 98 percent. The 58 target was not reached until the control chart had run for a full month. The sleeve supplier remains under review, because one lot shows a small difference in cut length that operators can feel but the gauges do not always flag.
Questions for discussion
- Which measurement would you take first to separate a fixture problem from a tool problem, and why?
- When a control chart signals a shift but the average has not moved, how should the line respond?
- What tradeoff would you accept between a faster line and a tighter first-pass yield target?
Case Study 3
Battery Storage Review Before First Energization
Lindqvist Energy Partners, an integrator that designs and commissions behind-the-meter battery systems for commercial and campus customers
The situation
A 4.2 MWh lithium iron phosphate battery system was being built into the rear of a university research building, connected through a new switchgear lineup to a 2,500 kVA service. Five weeks before the planned energization, the owner's engineer asked for a formal review. The package held single line diagrams, a cabinet layout, and the vendor's product data sheet, but the arc flash labeling study and the short circuit study were both still marked draft. Fire protection drawings showed a gas-based suppression system in a room that also housed the inverters. The authority having jurisdiction had not yet seen the package.
The challenge
The project team needed to decide whether energization could go ahead on the planned date, what conditions to attach if it could, and which gaps had to close before the inspector arrived. Some missing items were paperwork. Others, such as available fault current at the new switchgear, changed equipment ratings and could alter labels, protective device settings, and the personal protective equipment that operators would need.
Analysis
The most important missing evidence was the short circuit study. Without verified fault current at the switchgear bus, the interrupting ratings of the breakers and the arc flash incident energy values cannot be confirmed. The inverter output had been modeled with a nominal contribution, but the data sheet gives a range that shifts with state of charge and grid voltage. Tolerance stacking matters here. A modest rise in fault current can push a device past its rating, and a rating that holds on paper can still fail if the relay settings were based on an earlier design. The review treated the unfinished study as a hazard in its own right until it was reconciled with the as-built equipment.
Reviewers who work in this area treat documentation as part of the energization gate rather than a closeout item, because the inspector, the utility, and the insurer each ask for different evidence at different points. The electrical code requires arc flash labels on equipment likely to be examined under energized conditions, and the fire code governs both the battery room separation and the choice of suppression agent. The review recommended splitting the gate into three approvals covering mechanical and structural work, electrical work with the verified study, and fire and emergency response. A single sign-off would have hidden the fact that only the first was complete.
Actions taken
- Hold the energization date as conditional until the short circuit study is finalized and checked against the nameplate values of the installed switchgear.
- Recalculate arc flash incident energy and regenerate labels, then confirm that protective device settings match the updated study before the first closing operation.
- Require the fire protection engineer to reconcile suppression agent and room separation with the inverter heat load, and file the revised drawings with the authority having jurisdiction.
- Split approval into three signed gates covering mechanical, electrical, and fire response, and do not authorize energization until all three are recorded.
Results
The owner accepted a conditional schedule. The short circuit study was finalized within two weeks and showed fault current about 12 percent above the draft assumption. That moved two breakers closer to their ratings and required a revised arc flash category on the switchgear. Labels were replaced before inspection. The fire engineer changed the suppression approach for the inverter room, which added a cost the owner had not planned for. Energization went ahead nearly three weeks after the original date, and the inspector's first visit closed with two minor comments only.
Questions for discussion
- What minimum set of documents should a reviewer require before allowing a first closing operation on a new battery installation?
- How would you handle a short circuit study that changes a breaker rating after the equipment has already been delivered?
- Which review area in this case would you expect to cause the most disagreement among the owner, installer, and inspector, and why?
Case Study 4
Substation Upgrade Slips on a Delayed Breaker Shipment
Tamsin Ridge Electric Cooperative, a distribution cooperative serving about 21,500 members across a mix of farm and light industrial load
The situation
The cooperative planned a rebuild of its Dunmore Road substation, replacing two 69 kV breakers, updating protection relays, and adding a feeder position for a new processing plant. The milestone was energization of the new bay before the plant's first load test. The breakers had been ordered against a delivery window that the supplier later moved by nearly two months, citing a casting shortage at a foundry. The relay vendor delivered protection panels on time, but its settings file was built on an older one-line diagram that did not show the new feeder.
The challenge
The construction manager needed a recovery plan that could hold the load test date, which was tied to a customer contract with a penalty clause. Options ranged from paying for expedited breakers from a second supplier, to routing the new feeder through a temporary tie, to moving the outage window and accepting a later test. Each option carried a different cost, a different risk to the existing feeders, and a different dependence on protection settings that were already behind.
Analysis
The schedule impact is easiest to understand through the protection work. The new bay's relays cannot be tested against final breaker timing until the breakers arrive, and the mechanism specification allows an operating time tolerance that affects coordination margin with the upstream device. Operating data from the existing breakers showed trip times near the upper end of their band during the hottest weeks, so the design margin is narrower than the drawings suggest. Any settings change made to accommodate the new feeder has to preserve that margin. The delay therefore changed what could be verified and when, not just how long the team had to wait.
Project teams that handle this kind of delay usually quantify each recovery path in the same currency before ranking them, since a cheaper option that pushes the outage into a peak-load period can cost more than it saves. Outage records showed that the existing feeders had a maintenance window with limited switching flexibility, and an unplanned interruption was estimated at about $185,000 in crew overtime and member relations. Expedited breakers from a second supplier added roughly $640,000 but preserved the test date. A temporary feeder tie added about $210,000 and brought in a maintenance concern about the spare cable. Moving the outage lowered direct cost but triggered the contract penalty.
- 1Breaker shipment slipsFoundry casting shortage moves delivery back by nearly two months.
- 2Relay settings stalledSettings file was built on an outdated one-line diagram.
- 3Test window at riskContract penalty applies if the load test misses its date.
- 4Recovery option chosenExpedited breakers selected after comparing cost and risk.
- 5Revised commissioning dateBay energized with interim settings and a later settings review.
Actions taken
- Ask the relay engineer to rebuild the settings file from the current one-line diagram, including the new feeder, and check coordination margins against measured breaker timing.
- Qualify the second breaker supplier with a witnessed factory test and a written delivery guarantee, accepting a higher unit price in exchange for schedule protection.
- Hold a weekly three-party review of drawings, settings, and procurement so that any revision reaches the relay vendor within one week of being issued.
- Hold the outage in the agreed low-demand window and confirm switching steps with operations, so the existing feeders keep serving members without an unplanned interruption.
Results
The cooperative accepted the expedited breaker option. The second supplier delivered in the sixth week of the recovery period and passed the witnessed test, and the bay was energized with interim settings that kept coordination margins inside limits. The load test ran three weeks after the original date, and the customer accepted a reduced penalty once the cooperative showed the revised schedule. Total added cost came to about $640,000 over the budgeted rebuild. Final relay settings were issued only after the feeder was fully loaded, which required a second site visit and a small additional expense.
Questions for discussion
- How would you decide between paying a premium for schedule protection and accepting a later test date when both options carry penalties?
- What evidence would convince you that interim relay settings are safe to use before the final design is verified?
- Which document control practice would have caught the outdated one-line diagram earliest in this project?
Case Study 5
Cutting Distribution Losses at a Hospital Campus
Mercer Valley Health System, which runs four hospitals and a regional laboratory network with about 6,800 staff
The situation
The system's largest campus draws about 14 MW at peak, and its electrical distribution dates from several expansions. Two older dry-type transformers and a long run of aging feeder cable account for a meaningful share of losses, which the facilities group measured against metered energy at the main. The campus runs two standby generators and a UPS that supports imaging and surgery, and those critical loads must stay powered through any work. The sustainability office asked for a plan that reduces electrical losses and the waste from retired equipment by a set percentage.
The challenge
Any change had to leave critical loads untouched during replacement, preserve existing redundancy, and keep protection coordination intact. Options included replacing transformers with low-loss units, upsizing feeder conductors, adding power factor correction, and changing UPS operating modes. Each option had a different lifecycle footprint, and clinical staff had firm views on which changes were acceptable while patients were in care.
Analysis
Measured losses rose sharply in the afternoon, and the metering shows that the older transformers run at a fairly low average load for most of the day, which is where no-load losses dominate. A low-loss replacement reduces that constant loss, but the payoff depends on the duty cycle, so the decision cannot rest on nameplate efficiency alone. The feeder conductors were sized for an earlier load, and voltage drop at the far end of the surgical wing sits close to the tolerance the imaging vendor specifies. Upsizing the conductors would cut resistive loss, but it would also require a shutdown that the redundancy rules do not allow without a standby plan.
Sustainability reviewers compare options over the full equipment life, because a transformer that saves energy but must be replaced sooner can lose its advantage in material and disposal. The maintenance history shows the older units needed oil sampling and gasket replacement at intervals that added labor and produced waste fluid. A dry-type replacement with a lower loss rating removes that maintenance stream, and the retired units can go to metals recovery under the waste contract. Power factor correction is inexpensive and reversible, but it moves little energy. UPS eco mode reduces losses only if its transfer time still falls inside the tolerance written for the imaging equipment.
- First monthMetering installed on feeders and transformers to set the baseline.
- Second monthDuty cycle analysis and vendor transfer time testing begin.
- Within the next quarterLow-loss dry-type replacement on the older transformer bank.
- Following quarterFeeder conductor upgrade during a shutdown with generators online.
- Next budget cycleSecond transformer and UPS mode decision reviewed with clinicians.
Actions taken
- Install revenue-grade metering on both transformers and the surgical wing feeder, and collect at least one full duty cycle before committing to any replacement.
- Replace the older transformer with a low-loss dry-type unit, selected on loaded losses across the measured duty cycle rather than on nameplate efficiency.
- Ask the imaging equipment vendor to confirm UPS transfer time in eco mode, and adopt the mode only if the figure stays inside the tolerance in the equipment specification.
- Schedule conductor upgrades for a shutdown that keeps standby generation online, and record the recovered oil and metal in the lifecycle report as diverted waste.
Results
Metering showed that the older transformer bank accounted for about 18 percent of total campus distribution loss at typical afternoon load, which justified replacing it first. The new dry-type unit lowered measured losses on that bank, and the maintenance log shows no further oil sampling for it. Power factor correction went in during the same window at low cost. UPS eco mode was not adopted because its transfer time sat too close to the imaging tolerance. Conductor work is scheduled for the next cycle, and the sustainability office reports that retired oil and metal were diverted from landfill.
Questions for discussion
- How would you set the baseline period so that one unusually hot week does not distort the loss comparison?
- When a lower-energy option reduces redundancy margin, how should the decision be documented, and who should sign it?
- Which lifecycle boundary would you use for a transformer decision, and what would you deliberately leave out?
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.