Power Generation and Energy Systems Case Studies
Industrial5 Case StudiesUpdated Sep 2026
Power generation and energy systems carry a cost in downtime, emissions, and safety exposure whenever a plant decision goes wrong. The five cases below come from fictional operators and contractors, each drawn from situations that practitioners meet regularly. They cover a turbine component that fails early, a boiler line losing output as defects climb, a commissioning review with missing hazard evidence, a transformer delay that slips a substation milestone, and a steam efficiency program with competing options. Read each one as a decision record. Ask what evidence existed, what limited the response, and who absorbed the result.
Case Study 1
Turbine Blade Cracking Under Cyclic Startup Loads
Cinder Bend Generation, an independent operator of a 410 megawatt combined-cycle plant that also supplies steam to a nearby processor.
The situation
Cinder Bend Generation runs a 410 megawatt combined-cycle plant that supplies a regional grid operator and, under a separate contract, a food processor that buys steam for its drying lines. During a planned outage, inspectors found cracks at the root fillet of four first-stage blades on the second gas turbine. That unit had logged 14,000 operating hours against a published design life of 25,000. Dispatch had asked for far more starts than the original model assumed, averaging three a week since the regional wind fleet grew. Two cracked blades went to a metallurgy lab, and the outage ran nine days longer while replacement sets were sourced.
The challenge
The plant manager needed a root cause that would hold up with the manufacturer, the grid operator, and the insurer, and a decision on whether to restart the other three units. The vendor pointed to excessive starts and suggested a dispatch limit. Engineering suspected a casting porosity issue in one heat lot, but only two blades were available for metallography, and the original certification records were incomplete. Restarting at full load within the week would cut into a capacity payment window worth a large share of monthly margin.
Analysis
The evidence was strong on where the cracks started and weak on why. Fractography showed fatigue beginning at the fillet, which fits thermal cycling, but the same fracture surface could come from a design stress concentration or from a subsurface inclusion. The business goal was narrower than the physics. The plant needed a defensible cause quickly enough to restore the running envelope and keep the grid contract. Blaming starts alone would let dispatch carry on unchanged. Blaming material would question the whole heat lot, and the records did not link every installed blade to its lot.
Operational constraints shaped the option set more than the metallurgy did. The outage window was fixed by the grid operator's maintenance calendar, and new blade sets had a long lead time, so the plant could not simply swap in an untested design. The food processor depended on steam, which meant any restart plan had to protect that supply first. Corrective options included a revised fillet radius with a finite element check, a tighter start-count limit with a blade inspection interval, and a coating change. The team preferred a combined approach: a dispatch cap for the next season, a borescope schedule, and a redesign review that would outlast the outage.
- At outage startRoutine borescope inspection scheduled on the second gas turbine.
- On first inspection dayFour first-stage blades show cracking at the root fillet.
- Within two daysTwo blades removed for fractography and metallography.
- Later in outageStart cap and borescope interval agreed before return to service.
- At closeoutUnit returns nine days late with remaining blades inspected.
Actions taken
- Sample the remaining blade sets from every installed heat lot and pull receiving inspection records, so the plant can say which units share the suspect casting history.
- Cap starts on the second unit at the level the original model assumed, and log each start with its thermal state so dispatch and maintenance share one count.
- Add a borescope inspection of the fillet region at a fixed interval in operating hours, with acceptance limits agreed with the manufacturer before the next outage.
- Commission a finite element and fatigue review of the fillet geometry, and hold redesign approval until the grid operator and food processor have reviewed the restart plan.
Results
The unit returned to service under a start cap and a borescope interval, and the remaining blades were inspected before closeout. Lab work tied the initiation to a combination of thermal cycling and a surface finish defect on two of the four blades, which pointed to both a dispatch problem and a receiving gap. The other three units continued with restricted starts during the finite element review. Steam supply to the food processor held throughout. The insurer accepted the findings, and the plant adopted a rule that every blade certificate must match a physical lot number.
Questions for discussion
- How would the root-cause conclusion change if the receiving inspection records had matched every blade to its heat lot?
- What evidence would you have required before accepting a dispatch cap as the corrective action, and who should have approved it?
- How should the team weigh a longer redesign review against the cost of restricting starts for another season?
Case Study 2
Weld Station Constraint Driving Boiler Tube Rejects
Larkin Hollow Thermal Works, a fabricator of finned boiler tube assemblies with two forming lines and one shared welding cell.
The situation
Larkin Hollow Thermal Works fabricates finned boiler tube assemblies for power plant customers, running two forming lines that feed one shared welding cell and a hydrostatic test bay. Over the last stretch of the quarter, weekly output fell from 2,460 assemblies to 1,990 while the rejection rate at hydrostatic test rose from 1.2 percent to 3.8 percent. Nobody had changed the forming tools or the filler material, and the scrap log showed the new defects clustering on the second shift. The welding cell had been added as a short-term capacity fix and was now running close to full time.
The challenge
The operations director wanted output back above 2,400 assemblies a week before the next customer shipment cycle. The quality manager wanted the rejection rate under 1.5 percent before any new capacity was approved. Maintenance believed the weld torches were the source because torch tips were replaced on a timed schedule that had slipped. Production planning had only three weeks of partially verified data, and the shift supervisors disagreed about whether faster welding or slower welding was driving the defects.
Analysis
The process map showed the welding cell as the constraint, with forming upstream producing ahead of it and test downstream waiting for it. That alone did not explain the rejects, since rejects clustered on the second shift and not on the slowest station. Weld parameter logs covered about two-thirds of the shifts, and the torch replacement record had gaps that matched the defect spikes. Teams grounded in Power Generation and Energy Systems usually separate three sources at this point: fixture repeatability, consumable condition, and operator method. The goal was recovered volume without a third line, so each source needed a cost to correct as well as an effect on defects.
Constraints were tight. Customer schedules were fixed for two shipment cycles, the welding cell could not stop for full recalibration, and any fixture change needed a documented first-article check. Each stakeholder watched a different number: operations counted assemblies per shift, quality counted rejects per lot, and customers counted late deliveries. A torch replacement schedule keyed to weld count was the cheapest change and could be tested within one shift. Fixture tolerance work would take two weeks. The team ranked options by expected effect per unit of disruption and set acceptance criteria first: rejects below 1.5 percent for four consecutive weeks, and output at or above 2,300 assemblies a week.
Actions taken
- Move torch tip replacement from a calendar interval to a weld count, and log each change against the shift that ran it.
- Run a two-shift trial of the revised torch interval, comparing reject rates against the prior four weeks before any fixture work is approved.
- Commission the fixture tolerance study with a toolmaker and set a target of repeatable positioning within the tolerance used by the weld program, verified on sampled parts.
- Brief customer account managers on the staged recovery plan, with a weekly note showing reject rate and output against the two acceptance criteria.
Results
The torch interval change alone brought the rejection rate from 3.8 percent to 2.1 percent within the trial window, and the second shift's defect cluster largely disappeared. Output recovered to about 2,240 assemblies a week, short of the 2,300 target, because the fixture study ran longer than planned and the team did not want to restart the cell for a partial fix. The rejection rate was still above the 1.5 percent criterion at the end of the quarter, so the fixture work continued. Two customers accepted a revised delivery schedule without penalty, and escalations fell.
Questions for discussion
- Which evidence would have been needed to rank the torch interval change above the fixture study if the second shift had not shown the defect cluster?
- How should the team have balanced the operations target for output against the quality criterion during the period when both were unmet?
- What would a customer-facing delivery commitment look like if the fixture study had been scheduled before the torch change instead of after?
Case Study 3
Missing Hazard Evidence Before Biomass Plant Commissioning
Pellham Ridge Energy, a developer and owner-operator of a 62 megawatt biomass cogeneration plant under commissioning.
The situation
Pellham Ridge Energy is commissioning a 62 megawatt biomass cogeneration plant that will supply steam to a paper mill next door and export surplus power to the grid. The project review was scheduled ahead of the first fuel delivery. Reviewers found that the combustion safety case rested on a hazard study written before the fuel handling design changed to a finer chip size. The fire protection drawings showed a deluge system over the fuel conveyor, but no functional test appeared in the commissioning plan. Of 46 monitored dust points, 9 still lacked calibration records, and the sign-off sheet for the boiler interlocks carried one signature where two were required.
The challenge
The project director wanted an approval to load fuel within the month so the mill's steam contract would start on time, and the lender's technical adviser had flagged the schedule. The safety lead argued that the review could not be signed off until the hazard study was reissued and the deluge test was witnessed. Operations wanted to start a commissioning trial on the fuel conveyor alone to save time. Nobody could say whether the regulator would treat the trial as a separate permitted activity or as part of commissioning.
Analysis
The hazard study set the basis for every control in the plant, and its scope was drawn for a fuel the project no longer used. A finer chip size changes dust generation and the conveyor's ignition behavior, so the old conclusions cannot be assumed to carry over. The fire protection narrative matches the drawings, but two documents built on the same outdated assumption agree without verifying anything. The missing functional test is the clearest gap. A drawing shows intent, while a witnessed test shows that the deluge valve opens and water reaches the conveyor. A delayed first fuel load costs less than a finding after startup.
Controls and approval gates need to be read together. The plant relied on interlocks, a deluge system, and housekeeping procedures for the fuel conveyor. A deluge valve is only as good as the detector feeding it, and 9 of the 46 dust points lacked calibration records, so the detector layer was unverified. The applicable combustible dust and boiler standards call for documented hazard analysis and verified interlocks before fuel introduction. Approval gates should run in sequence: reissued hazard study, witnessed interlock and deluge tests, calibration closeout, then permission to load fuel. Stakeholders carried different exposure. The mill's steam contract, the lender's covenants, and the crew's time during a trial all depended on that order holding.
- Hazard study Reissue against the finer chip specification, independently reviewed.
- Functional tests Witness deluge and interlock tests with recorded valve travel times.
- Calibration closeout All nine outstanding dust points calibrated and filed.
- Sign-off complete Both interlock signatures in place and regulator position recorded.
Actions taken
- Reissue the combustion hazard study against the finer chip specification, and have the reissued study reviewed by someone who did not write the original.
- Witness the deluge and interlock functional tests on the fuel conveyor, recording valve travel time and the detector trip point for each test.
- Close out calibration for the nine outstanding dust points and file the records before any fuel enters the conveyor, with no exceptions for trial runs.
- Get written confirmation from the regulator on whether a conveyor-only trial counts as a permitted activity, and keep the interlock sign-off open until both signatures are in place.
Results
The project director accepted a three-week slip in first fuel load and the review closed with a conditional approval. The hazard study was reissued, and the witnessed deluge test found a valve travel time longer than the design basis, which led to a nozzle and piping adjustment before loading. Four of the nine uncalibrated dust points failed initial checks and were replaced. The steam contract with the mill was amended to start in stages, and the lender's adviser accepted the revised schedule. The team now treats any change to fuel specification as a trigger for reviewing hazard studies.
Questions for discussion
- Which of the review findings would you have treated as a hold point, and which as a condition to close after loading?
- How should the review team handle a document that is internally consistent but was written against an outdated design basis?
- What written record would convince the lender's adviser that a conveyor-only trial was controlled, and who should sign it?
Case Study 4
Transformer Delay Pushes Substation Energization Milestone
Vellard Transmission Works, an engineering and construction contractor delivering substation upgrades for regional utilities, with about 140 field staff.
The situation
Vellard Transmission Works is upgrading a 138 kilovolt substation for a regional utility, with a new transformer, a switchgear lineup, and a protection scheme that must energize in a fixed utility outage window. The transformer was ordered early with a delivery date that held through the design freeze. In the fourth month, factory tests found a winding insulation issue, and the replacement unit's delivery moved out by 11 weeks. The relay vendor also shipped firmware that did not match the approved settings file, adding 4 days of rework on site. The contingency reserve stood at 1,800,000 dollars, and the milestone was now 9 weeks past its window.
The challenge
The client's project manager wanted energization inside the next outage window, which meant a temporary transformer or a schedule compression that the contractor's own planners rated as high risk. The procurement lead had a second manufacturer's unit that could arrive sooner but needed a design change to the mounting pad and oil containment. Relay rework could happen on site or at the vendor's plant, with different cost and commissioning risk. Finance wanted the cost of each option stated against the liquidated damages clause before any choice was made.
Analysis
The business objective was energization inside the window at a cost the reserve could absorb. The evidence was clearer than usual. The insulation failure was documented in the factory report and the replacement lead time came from a written delivery commitment. Liquidated damages were set at 45,000 dollars per week up to a cap of 300,000 dollars so the 9-week exposure was capped. The rental quote was 390,000 dollars before transport and hookup and parallel commissioning overtime was estimated at 60,000 dollars for a recovery of six weeks. Those figures set the trade-off and the team kept them on one sheet so no option was judged on schedule alone.
Constraints pulled in different directions. The utility could not move its outage window without a new regulatory notice, and the second manufacturer needed two weeks of design review first. Each party paid for a different outcome. The client wanted reliable outage timing, procurement wanted bounded exposure, and the commissioning crew wanted a test schedule it could actually staff. The rental protected the window but moved risk into crew hours and cash. Accepting the slip kept cash intact and handed the client another season on a weakened network. The team recommended parallel commissioning first, with the rental held as a fallback if the transformer date moved again.
Actions taken
- Start parallel commissioning now by running relay settings verification and transformer bay preparation at the same time, with a daily reconciliation of open items.
- Hold the rental quote as a fallback with a decision date tied to the transformer's factory retest result, so the cost is committed only if the date slips again.
- Ask the relay vendor to cover the firmware rework at its own facility, and write the settings file check into the acceptance test so mismatches surface before shipment.
- Send the client a written trade-off sheet showing damages at the cap, rental cost, and parallel overtime, and ask for a decision before the next utility planning meeting.
Results
The team ran parallel commissioning and recovered six of the nine weeks, while the transformer was retested at the factory. The retest passed on the second attempt, so the rental was never ordered. Energization landed three weeks past the original window, and damages came to 135,000 dollars against 475,000 dollars for the rental and transport. Overtime cost 60,000 dollars. The client accepted the recovery plan, and the relay vendor absorbed the firmware rework without a change order.
Questions for discussion
- How would the comparison change if the liquidated damages cap had been set at a higher figure with no weekly rate?
- What evidence would justify ordering the rental before the transformer retest result was known?
- Which stakeholder should carry the residual risk of a three-week slip, and how should that be recorded?
Case Study 5
Steam Efficiency Options Compared on Lifecycle Cost
Stillwater Hollow Board Company, a paperboard mill that runs its own boilers and a 48 megawatt backpressure turbine.
The situation
Stillwater Hollow Board Company runs a paperboard machine that depends on steam from its own boilers and a 48 megawatt backpressure turbine. Steam losses across the distribution network had grown as the site added two dryer sections, and the energy team estimated that condensate return sat at 38 percent of generated steam flow, well below what the original design assumed. A corporate target asked the mill to cut fuel use per ton of board by 8 percent within the next budget cycle, while keeping the machine running at rated speed and holding paper quality within specification.
The challenge
The mill had three candidate projects: a condensate recovery loop with new pumps and tanks, a flash steam unit feeding a low-pressure header, and a control upgrade that tightened dryer steam setpoints. Each promised savings, but they differed in capital cost, risk to machine uptime, and carbon effect. The dryer setpoint change had only a short trial behind it. The controller wanted a lifecycle comparison ready for the next capital meeting, and the quality manager refused any change to dryer profiles without a trial.
Analysis
The mill had five weeks of steam flow data across both boilers, and it showed condensate losses concentrated in two dryer headers. That concentration set a practical ceiling on what each option could recover. The business objective was an 8 percent fuel reduction per ton without losing uptime or paper quality, so a project that saved energy on paper but cost a day of machine time was not a success. The team modeled each option over its full equipment life, with capital, maintenance, fuel, and the carbon effect of the fuel mix, and checked each result against measured flows rather than vendor estimates.
Operating limits narrowed the field. The machine could not take an unplanned stop longer than one shift, and the turbine's output depends on steam flow to the dryers. Each option was scored on energy and operational effect, not only on lifecycle cost. The condensate loop recovered the most fuel but needed a planned shutdown for tie-ins. The flash steam unit reduced turbine output whenever the low-pressure header ran full. The setpoint upgrade was cheap and fast, but it carried the highest quality risk. The controller valued payback, the quality manager valued profile stability, and the maintenance lead valued predictable spares.
Actions taken
- Stage the setpoint change on one dryer header first, with paper quality samples taken before and after each step, and keep the old profile loaded for fast rollback.
- Schedule the condensate loop tie-ins into the next planned shutdown, and pre-fabricate the pump skid and tank connections so the shutdown stays inside one shift.
- Set the flash steam unit's operating rule so the low-pressure header never forces turbine output down, and log every hour the rule overrides normal flow.
- Track fuel per ton, condensate return percentage, and turbine output monthly against the baseline of 38 percent return, and report the lifecycle payback to the capital committee.
Results
The mill chose the condensate recovery loop as the main project and the setpoint change as a staged trial. Condensate return rose from 38 percent to 61 percent within the first full season, which cut fuel use per ton by about 3.4 percent. The setpoint trial held paper quality across both grades after two adjustments, and the flash steam unit was deferred to a later budget cycle. Turbine output held at the planned level throughout. The 8 percent target was not met in the first season, but the lifecycle comparison now shows a payback that the capital committee accepted.
Questions for discussion
- How would the ranking change if the lifecycle model had used a stricter carbon accounting basis for the fuel mix?
- Which measure should decide between a project with the largest fuel saving and one with the lowest quality risk?
- What monitoring would you require before expanding the setpoint change from one header to the whole dryer section?
Learning Resources
Industrial CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.