What 1,783 US Government Audits of Auto-Parts, Forging, Foundry and Machine-Shop Plants Found: Top Measures, Payback, Take-Up
How do you cut cost in an auto-parts, forging or machine-shop plant? In the US DOE ITAC database (release of 19 Oct 2026), 1,783 such plants received 13,897 recommendations. The median measure paid back in 0.79 years, yet owners implemented 46.0% of them. Fixing compressed-air leaks came up in 59% of plants and was done 75% of the time.
Last checked: 20 Oct 2026
US DOE data, US plants, fiscal years 1982–2026; dollar values stay in US dollars and are shown mainly as shares, ratios and paybacks.
By the Mama Editorial Team. Every figure is our own count of the public database, with the filter listed; the rows are in a CSV.
Verified against
- US DOE Industrial Training and Assessment Centers (ITAC) database, file
ITAC_Database_20261019.xlsx(sheets ASSESS and RECC), download page, which lists 23,234 assessments and 170,633 recommendations, generated 19 Oct 2026. Checked 20 Oct 2026. - Licence: Creative Commons Attribution 4.0, per the data.gov record published by the DOE Office of Energy Efficiency & Renewable Energy. Checked 20 Oct 2026.
- Measure wording and group names: ITAC recommendation code (ARC) index. Checked 20 Oct 2026.
- Visit and follow-up method: ITAC assessment protocol. Checked 20 Oct 2026.
- Our filter: NAICS 3363 (motor vehicle parts), 3321 (forging and stamping), 3315 (foundries), 3327 (machine shops, turned products, screws, nuts and bolts). Where NAICS is blank in older records: SIC 3714, SIC 3462/3463/3465/3466/3469, SIC 332x/336x foundries, SIC 3451/3452. SIC 3599 (mixed industrial machinery) is left out. "Implemented %" = status I ÷ (I + N); 874 recommendations with other or blank status are left out of that ratio.
What the program is
ITAC sends university teams of engineering students and faculty to smaller US manufacturers free of charge, under the DOE Office of Manufacturing and Energy Supply Chains (ITAC site, checked 20 Oct 2026).
The team usually spends one day on site, then phones the plant 6–9 months later and logs which measures it carried out. That follow-up is where the "implemented" figures on this page come from. The same cut for textile mills is in what US audits of textile plants found.
The four trades side by side
Per trade, ITAC database, release of 19 Oct 2026, our count.
| Trade (filter) | Plants (count) | Median employees (count) | Recommendations (count) | Median payback (years) | Implemented (%) |
|---|---|---|---|---|---|
| Auto parts (NAICS 3363 / SIC 3714) | 658 | 230 | 5,168 | 0.74 | 46.5 |
| Forging and stamping (3321 / 346x) | 304 | 115 | 2,487 | 0.79 | 48.7 |
| Foundries (3315 / 332x, 336x) | 575 | 110 | 4,371 | 0.82 | 45.4 |
| Machine shops and fasteners (3327 / 3451, 3452) | 246 | 100 | 1,871 | 0.78 | 42.2 |
| All four | 1,783 | 147 | 13,897 | 0.79 | 46.0 |
Across all four trades, 36.7% of plants had between 200 and 1,000 employees, the size band of a mid-size Indian auto-component unit (ITAC database, our count).
The measures that came up most, with payback and take-up
The 12 measures recommended in the most of 1,783 plants, ITAC database release of 19 Oct 2026, our count.
| Measure (ITAC wording, shortened) | ARC code | Plants where recommended (% of 1,783) | Median simple payback (years) | Implemented (%) |
|---|---|---|---|---|
| Higher-efficiency lamps or ballasts | 2.7142 | 66.6 | 1.68 | 55 |
| Fix compressed-air leaks | 2.4236 | 59.1 | 0.20 | 75 |
| Compressor air intake from the coolest place | 2.4221 | 32.5 | 0.48 | 48 |
| Lower compressed-air pressure to the minimum | 2.4231 | 30.3 | 0.05 | 48 |
| Occupancy sensors | 2.7135 | 28.2 | 1.00 | 37 |
| Energy-efficient belts and drives | 2.4111 | 23.5 | 0.79 | 54 |
| Most efficient type of electric motor | 2.4133 | 22.1 | 1.76 | 59 |
| Recover waste heat from equipment | 2.2437 | 19.7 | 0.60 | 29 |
| Variable-frequency drives | 2.4146 | 16.7 | 2.01 | 23 |
| Insulate bare equipment | 2.2511 | 15.4 | 0.54 | 41 |
| Cut compressed-air use | 2.4239 | 14.1 | 0.32 | 45 |
| Optimise plant power factor | 2.3212 | 12.3 | 1.67 | 31 |
Compressed air dominates this sample. 1,476 of the 1,783 plants (82.8%) received at least one compressed-air recommendation (ARC 2.42xx), and these made up 20.4% of all recommendations (ITAC database, our count).
Lowering air pressure was the fastest money on the list: a median payback of 0.05 years, under three weeks, at a median cost of $140. Owners still did it only 48% of the time (ITAC database, our count).
Heat recovery and drives get postponed. Waste-heat recovery (0.60 years) and variable-frequency drives (2.01 years) were implemented 29% and 23% of the time (ITAC database, our count).
What these plants get more often than the rest of US industry
Compared with the other 21,227 ITAC plants that had recommendations, the four metal trades were 1.29 times as likely to be told to fix compressed-air leaks (59.1% vs 46.0%) and 1.36 times as likely to be told to recover waste heat (19.7% vs 14.5%), per the ITAC database (our count).
The trade split shows where heat matters. Preheating with flue-gas waste heat (2.2414) was recommended in 17.9% of foundries and 12.5% of forging and stamping plants, against 6.1% of auto-parts and 4.9% of machine-shop plants (ITAC database, our count).
Machine shops stand out on air pressure: 41.5% of them were told to lower compressed-air pressure, against 25.7% of foundries (ITAC database, our count).
Recycling casting sand (3.5221) appeared only in foundries, in 4.3% of them, with a median payback of 1.33 years (ITAC database, our count).
Production measures: rare, but bigger per measure
ITAC codes starting with 4 are "direct productivity enhancements", grouped as manufacturing, purchasing, inventory, labour optimisation, space utilisation, reduction of downtime, management practices and administrative savings (ITAC ARC index, checked 20 Oct 2026).
Only 365 of the 1,783 plants (20.5%) received any productivity recommendation, and these made up 4.2% of all recommendations (ITAC database, our count).
Per measure they were larger. The median productivity recommendation was worth $21,840 a year, against $3,196 for the median energy recommendation (nominal US dollars of the audit year, ITAC database, our count).
| Productivity measure (ITAC wording) | ARC code | Plants (count) | Median simple payback (years) | Implemented (%) |
|---|---|---|---|---|
| Replace equipment with a more suitable substitute | 4.652 | 45 | 0.96 | 36 |
| Add equipment or operators at a production bottleneck | 4.111 | 37 | 0.78 | 28 |
| Begin predictive / preventive maintenance | 4.611 | 31 | 0.30 | 65 |
| Install equipment to move product | 4.445 | 27 | 0.91 | 30 |
| Re-arrange layout to reduce labour costs | 4.513 | 20 | 0.25 | 48 |
| Train operators for maximum operating efficiency | 4.431 | 20 | 0.31 | 50 |
| Standard procedures to improve internal yields | 4.122 | 17 | 0.27 | 44 |
Counts are small: read each row as a direction. The CSV lists 82 measures with the share of plants in each of the four trades.
Waste and water measures (codes 3.x) had the lowest take-up, 36.8%; energy measures 46.8%; productivity 42.9% (ITAC database, our count).
Payback does not decide alone
Across all 13,897 recommendations, measures paying back in under six months were implemented 55.8% of the time, and those over three years 34.0% (ITAC database, our count).
Take-up also fell over time: 52.5% of measures were implemented in audits before 2000, against 41.8% from 2010 on, across 725 plants audited since 2010 (ITAC database, our count).
The energy bill test
For the median plant, the energy measures identified were worth 10.0% of its annual energy bill; the energy measures actually implemented were worth 2.7% of it (ITAC database, our count of 1,783 plants, electricity and fuel costs as each plant reported them).
Assume your forging or machining unit pays ₹4 crore a year for power and fuel. The US pattern would point at about ₹40 lakh a year of energy measures, and the US take-up pattern would capture about ₹11 lakh of it. The ₹4 crore is an assumption; swap in your own bill. This applies a US ratio to an assumed number and is not a forecast for your plant.
What changes in an Indian plant
- Walk the compressed-air line first. It came up in more than four of five US plants in these trades, and pressure and leak fixes paid back in weeks.
- Lighting may already be done. If your sheds are on LED, skip the top row.
- Furnaces decide the forging and foundry list. Flue-gas preheat and waste-heat recovery were recommended far more often there than in machine shops.
- Production measures need a measured baseline. Bottleneck, maintenance and layout measures were the rarest, and nobody can price a bottleneck without knowing how many hours each machine stood idle.
Where Mama fits
The cost side of an auto-component plant is set out in spending less on scrap, sorting, tooling and energy, the energy a stopped-but-running machine burns is covered in idle machines, lights and compressors, and the idle machine energy cost calculator estimates it. Mama reads a factory's existing cameras and tells the owner each morning which machines stood idle and for how long, the baseline the rare productivity measures above depend on. Leave an email or phone in the form below.
FAQ
Can an Indian forge or machine shop get an ITAC audit?
No. The ITAC protocol covers US manufacturers only (itac.university, checked 20 Oct 2026). An Indian unit can walk the same list itself; BEE's ADEETIE refunds an investment-grade energy audit in listed forging and foundry clusters.
Which compressed-air fix pays back fastest in these trades?
Lowering pressure to the minimum: a median payback of 0.05 years, yet done only 48% of the time. Leak repair, recommended in 1,053 of the 1,783 plants, paid back in a median 0.20 years and was done 75% of the time (ITAC database, Oct 2026 release, our count).
How can a machine shop reduce costs?
In 246 US machine-shop and fastener plants, the most frequent recommendations were lighting (69.1% of plants), compressed-air leaks (61.4%) and lowering compressed-air pressure (41.5%), per the ITAC database (our count). Take-up there was the lowest of the four trades, 42.2%.