Energy Management and Audit Case Studies
Industrial5 Case StudiesUpdated Sep 2026
Energy management in an industrial plant depends on whether the numbers agree. Meter exports, invoices, certificates, and contract terms describe the same consumption from different angles and rarely match on the first pass. The five cases below come from fictional plants where energy teams answered questions from controllers, operations, and corporate reporting. Each follows one decision: reconcile a gap, revise a forecast, escalate a warning sign, fund a project, or close an audit finding. The analysis returns to financial records, risk exposure, control evidence, and decision thresholds, which is where energy programs win or lose credibility.
Case Study 1
Tracing a utility cost gap across three ledgers
Harrow Creek Castings, an iron foundry with two electric arc furnaces and a four-person energy team reporting to the plant controller
The situation
Harrow Creek's energy cost ledger for the trailing twelve months showed $2,418,600 in electricity and natural gas charges. When the controller compared that figure with the supplier invoices on file, the ledger came out about $141,300 higher. The energy management system held a third set of numbers, interval meter exports for each furnace and the main service entrance, and those totals did not match the invoices either. Sofia, the energy analyst, was asked to explain the gap before the quarterly close, when the cost per ton figure would go into the pricing model for the next bid cycle.
The challenge
The gap was not a single error. Some months matched within a few hundred dollars, while others were off by amounts that looked random until the bills were lined up against the meter reads. Several invoices had been scanned without their read dates, and two were estimated bills that were never trued up in the ledger. Sofia had to decide which source to treat as authoritative, whether the variance was timing or substance, and how much of it had already reached the cost model.
Analysis
From a financial records standpoint, the discrepancy matters because energy cost feeds cost per casting, which sets minimum pricing on repeat orders. Sofia traced the $141,300 into three components. Two estimated bills, posted in two separate months, were replaced by actual reads but the original accruals were never reversed, which double counted roughly $58,200. A demand charge credit of about $43,900 was booked in the period after the utility issued it. The remaining balance traced to a tariff change applied from the wrong effective date. The tariff error matters most, since it repeats every billing cycle until someone corrects the rate table.
Teams grounded in Energy Management and Audit tend to treat the reconciliation as a control design problem as much as a cleanup task. Harrow Creek's controller set three thresholds: under 0.5 percent of monthly spend the analyst may clear the item with a written note, between 0.5 and 2 percent the controller reviews source documents, and above 2 percent posting is held until the variance is explained. Control evidence now includes a monthly three-way match of meter export, invoice, and ledger entry, plus a required read date on every scanned bill. Thresholds decide who acts and when.
Actions taken
- Reverse every estimated bill accrual automatically when the actual read posts, and add a report that lists open accruals older than one billing cycle.
- Load the utility tariff schedule into the ledger with a dated rate table, and require the controller to sign off before any rate change takes effect in the system.
- Add the read date and service account number as mandatory fields on the invoice intake form, so scanned bills without them cannot be filed.
- Run the three-way match each month on the closing schedule and log every item above threshold in a shared exceptions register with an owner and a target clearance date.
Results
Within the next close cycle, the team cleared $119,700 of the $141,300 gap. The double counted accruals and demand credit timing were corrected through one adjusting entry with a memo attached. The tariff table was rebuilt, and the cost per casting figure used in pricing moved enough to change two open quotes. The remaining $21,600 traced to a meter that reset its interval logs after replacement, and the vendor's correction letter is on file. Since then, no variance above threshold has gone unexplained.
Questions for discussion
- Which of the three sources should the controller treat as authoritative when the meter export and the invoice disagree, and what evidence would justify overriding either one?
- How would the thresholds change if the plant bought power through a contract with index pricing rather than a fixed rate?
- What is the minimum documentation a reviewer needs before clearing a variance below the lowest threshold without a second look?
Case Study 2
Rebuilding an energy budget after gas prices move
Tallow Bend Packaging, a corrugated box plant with two gas-fired dryers, a compressed air system, and a shift schedule that changes with customer orders
The situation
Tallow Bend set its annual energy budget early in the year using a natural gas price assumption and a production forecast of 14,200,000 square feet of board. Two quarters in, gas costs ran about 19 percent above the planned run rate, and output was tracking closer to 11,800,000 square feet because a large customer moved volume to another plant. Ruth, the plant finance lead, was asked to produce a revised forecast for the remaining quarters before the budget review. Corporate wanted one number. The operations manager wanted a range, since the dryer schedule depended on orders that had not yet been confirmed.
The challenge
Gas price, volume, and energy intensity all moved at once, and the sub-metering only showed usage at the dryer level. Ruth could not tell from the data whether higher cost per unit came from price, from lower throughput spreading fixed loads, or from dryers running longer to hit moisture targets on a heavier board grade. Any forecast she produced had to separate those effects and hold up when the budget committee asked which assumption carried the most weight.
Analysis
The records gave a starting point but not a clean one. Gas invoices for the period totaled $486,200 against a plan of $408,100, a gap of $78,100. Ruth estimated roughly $52,000 of that came from price, using the contract index that had moved since the plan was set, and the rest from intensity, since fewer units were spreading the dryers' standby load across less output. Risk exposure sat mostly in the contract. The index reset quarterly, so each price move reached the forecast with a lag, and a further rise would widen the gap before the next review.
Ruth built three scenarios instead of one. The base case used the revised order book and the current index forward curve. The downside added a 10 percent volume drop and a price move equal to the largest quarterly change in the contract history. She tested assumptions one at a time, and dryer hours proved the most sensitive input, since a 3 percent change in run time moved the annual total more than the index did. The committee agreed on two thresholds: a forecast more than 8 percent over budget triggers load shifting options, and more than 15 percent requires a supplemental approval.
- Refresh inputs Pull invoices, index prices, and dryer run hours for the closed months
- Test assumptions Vary one driver at a time and rank sensitivity before combining effects
- Compare to thresholds Check the scenario range against the 8 and 15 percent triggers
- Recommend response Present the lever that fits the threshold crossed, with its cost and lead time
Actions taken
- Rebuild the gas forecast from metered dryer hours instead of the plan's flat intensity factor, and tie each hour block to the board grade that drove it.
- Add a contract index reset date to the monthly close checklist so price changes flow into the forecast within the same reporting cycle rather than one period later.
- Set up a standing scenario template with the three cases, the assumption owners, and the threshold triggers, so the next revision starts from the same structure.
- Bring the load shifting option to operations for a trial on one dryer, with hours and moisture results logged so the savings claim can be checked before it is booked.
Results
The revised forecast came in 11 percent above budget in the base case and 17 percent above in the downside. The downside crossed the supplemental approval line, while the base case called for load shifting options. Operations trialed load shifting on one dryer for the rest of the period, recovering about a third of the intensity gap. The committee approved a smaller supplemental request tied to the trial results, and the scenario template is now used in each monthly review.
Questions for discussion
- Which assumption in this forecast would you least want to defend in front of the committee, and what evidence would make it easier to defend?
- How should the thresholds be adjusted if the contract index were replaced with a fixed price for the coming period?
- When a load shifting trial shows savings on one dryer, what should be checked before the result is applied to the whole plant?
Case Study 3
Renewable certificate purchases flagged on the reporting register
Greystone Valve and Pump, a valve assembly plant with heat treatment furnaces and a large compressed air network, which buys renewable energy certificates to support customer sustainability claims
The situation
Greystone bought renewable energy certificates through a broker to cover a share of its annual electricity use, and the purchase was cited in the customer sustainability report. In the third week of the quarter, the broker's confirmation listed certificates tied to a generator ID and a meter account that did not appear in the plant's registry. The volume was modest, about 2,400 megawatt hours, but it represented roughly a fifth of the claimed coverage for the period. Marcus, the compliance lead, had to decide whether the claim could stand, what the plant owed the broker, and who else needed to know before the report went out.
The challenge
Three warning signs appeared together. The meter account on the certificate was not on the plant's list of registered loads, the serial numbers did not match the format used in the prior two deliveries, and the delivery arrived late in the period with a note that the generator was under review. None of these alone proved misstatement. Together, they meant the draft could not carry the claim as written, and the sales team was already quoting the figure to two customers.
Analysis
The exposure has two layers. The direct cost is the purchase price, around $31,800 for the contracted volume, which is small beside the plant's energy spend. The larger risk is what the claim supports. A sustainability statement that overstates coverage can lead to customer questions, contractual remedies in supply agreements, and reputational damage that is hard to size. Marcus framed the exposure around what a customer would find if they asked for the certificate trail: if the serial numbers could not be traced to the plant's meter, the claim had no support regardless of the amount.
Greystone's policy already required a verification step, but it was built for volume, not for identity. The procurement file held the broker's invoice and a summary sheet, and the registry match was done by eye. Marcus set escalation rules that depend on both exposure and documentation. Mismatched meter identity always goes to the compliance committee, a serial number gap above one percent of volume holds the claim, and a clean trail proceeds with a standard note. The same warning sign looks minor on a small purchase and serious once it is quoted externally.
Actions taken
- Add the meter account and generator ID to the certificate intake form, and reject any delivery where either field is missing or does not match the plant's registry.
- Require the broker to supply the full serial trail in the registry format before the purchase is approved, with the generator's current status attached.
- Hold the sustainability claim in the draft report until the registry match is complete, and notify the sales team in writing so no further figures go to customers.
- Review the broker contract in full for delivery warranties and remedies, and confirm with legal whether supply agreements contain sustainability representations.
Results
The committee reviewed the registry match in the same week. The meter account turned out to be a sub-meter added during a line expansion and never entered in the registry, so the certificates were valid. The serial gap came from a formatting problem in the broker's export. The report went out with the trail attached and a registry update noted. Two customer quotes had used a figure containing a unit conversion error, and both customers received corrected letters.
Questions for discussion
- Where should the line sit between a registry formatting issue and a documentation gap that justifies holding a claim?
- How would the escalation rules change if the purchase had been made directly from a generator rather than through a broker?
- What checks would you add to the sales process so that figures from the sustainability report reach customers only after compliance review?
Case Study 4
Heat recovery funding request tested against downside scenarios
Marrow Glass Works, a container glass plant with two furnaces and a cullet handling line, seeking capital for a waste heat recovery project
The situation
Marrow Glass asked the capital committee for $1,860,000 to install a waste heat recovery system on its larger furnace, with the stated aim of generating steam for process needs and offsetting natural gas use. The vendor's proposal projected annual savings of $412,000 and a payback a little under five years. Felix, the energy engineer who prepared the request, had the vendor model, two months of furnace exhaust readings, and the plant's gas invoices. The committee wanted to know whether the savings would hold if the furnace ran at lower pull rates or if gas prices fell.
The challenge
The vendor model assumed the furnace would run near its design pull rate for most of the year, and the exhaust temperature it used came from a short measurement window. Marrow's own production plan showed two planned reduced-pull periods. A second project was also competing for the same capital headroom, so this request had to show that it held up under a downside, not only that it cleared the hurdle rate in the base case.
Analysis
The financial case rests on three numbers: the energy baseline, the recovery rate, and the gas price. Felix rebuilt the baseline from twelve months of invoices. The vendor's figure ran about 6 percent above metered gas use, which cut annual savings to $387,000. The recovery rate was the least certain input, since the exhaust sampling covered a narrow temperature range. Felix treated recovery as a range rather than a point estimate and carried the lower bound into the downside case.
Felix ran four scenarios: the base case at vendor assumptions, a reduced-pull case with 15 percent fewer operating hours, a gas price fall of 20 percent, and a combined case. The combined case extended payback to about six years and pushed net present value close to zero at the company's required return. Suitability mattered as much as returns. A project that pays back only at design pull locks the plant into a schedule it may not keep. The approval threshold was a payback under five years in the combined case, with net present value above zero, and a vendor guarantee covering at least 85 percent of modeled recovery.
Actions taken
- Re-base the savings case on metered gas use for the trailing twelve months, and require the vendor to restate the projection using that baseline before the next review.
- Extend the exhaust sampling across the full range of pull rates in the production plan, with at least one measurement window in each reduced-pull period.
- Negotiate a performance guarantee covering at least 85 percent of the modeled recovery, with a remedy schedule tied to measured steam output over the first full year.
- Approve the request as conditional, releasing a first tranche for engineering and holding the balance until the guarantee and restated baseline are signed.
Results
The committee approved the project as conditional. The vendor restated its projection on metered gas use, and the guarantee rose to 88 percent of modeled recovery. Engineering funds of $142,000 were released at once, and the rest of the capital was held. Under the revised figures the base case paid back in about four and a half years, and the combined case ran just past five years, so final release was tied to a production plan that avoided two reduced-pull periods.
Questions for discussion
- Which of the four scenarios would you weight most heavily if the plant's production plan were uncertain for the next two years?
- What evidence would justify a lower performance guarantee than the one the committee requested, and what would it need to cover?
- How should a project that clears the hurdle rate in the base case but fails in the combined case be presented to a capital committee?
Case Study 5
Metering control gaps ranked before the energy audit closes
Quillan Forge Co., a closed-die forging plant with four induction furnaces and a three-shift schedule, whose energy audit covers both electricity and natural gas
The situation
Quillan's internal energy audit reviewed the metering and reporting controls that feed its energy performance indicator, which the plant reports to a corporate sustainability group and uses in supplier scorecards. The audit found eleven control weaknesses across meter calibration, data retention, reconciliation, and access to the energy management system. Heather, the internal audit lead, had two weeks to prepare findings for site leadership before the remediation plan was set. Several findings touched the same numbers, so the plan had to show which weaknesses actually changed the reported figures.
The challenge
Management had a long list and limited staff. The energy team of three could not fix everything at once, the maintenance group had a backlog of meter work, and IT owned the historian server that stored the interval data. Heather needed a ranking that finance and operations would both accept, owners who could be held to dates, and closure tests an outside reviewer could repeat without relying on the energy team's word.
Analysis
Materiality here is measured against the indicator, not the energy bill. Heather ranked findings by how much each could move the reported energy intensity and whether the error ran in one direction. Two findings scored highest. An uncalibrated sub-meter on the second furnace had been reading low by a margin the audit estimated near 4 percent, which overstated the reported improvement. A gap in historian retention meant interval data for part of the prior period could not be re-pulled. Lower-ranked items, such as a shared login on a reporting screen, weakened control but did not move the reported figure.
Each finding received one accountable owner, named by role, and a closure test that required evidence from outside the team doing the fix. For the furnace sub-meter, the test was a calibration certificate issued after the repair and a week of side-by-side reads against a reference meter within the tolerance set in the audit standard. For retention, the test was a successful re-pull of a sampled month from the historian into the audit folder. A finding closed only when the test passed and internal audit had reviewed the evidence, never when the fix was reported as done.
- 1Rank findingsScore each finding by effect on reported intensity and direction of error
- 2Assign ownersName one accountable role per finding, with a due date agreed with site leadership
- 3Define closure testsSpecify the evidence an outside reviewer can repeat without the owner
- 4Execute and documentOwner completes the fix and files evidence in the audit folder
- 5Independent reviewInternal audit re-performs the test before the finding is closed
Actions taken
- Rank all eleven findings in a single register that shows the effect on reported intensity, the direction of error, and the estimated size where it can be measured.
- Assign each finding one owner by role, and set due dates with site leadership so that no item depends on a shared team commitment.
- Write the closure test before the fix starts, so the owner cannot shape the test to fit the repair, and store test scripts in the audit folder.
- Hold a monthly closure review where internal audit re-performs a sample of closed items, and reopen any finding whose evidence cannot be repeated.
Results
Seven of the eleven findings closed by the end of the following quarter. The sub-meter calibration confirmed a reading error near 4 percent, and the prior period's intensity figure was restated, which trimmed the reported improvement. The historian re-pull succeeded for the sampled month but exposed an earlier month that could not be recovered, which was disclosed as a limitation. Two findings were reopened after re-performance failed, and the last two are scheduled for the next maintenance window.
Questions for discussion
- Which ranking criterion would change the order of the top two findings if the indicator were used for supplier scorecards instead of corporate reporting?
- How should an owner respond when a closure test is sound in principle but the historian cannot produce evidence for part of the period?
- What is the right number of re-performed items each month before internal audit can say the closure process is working?
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.