Purchased, Net and Onsite Electricity in Manufacturing Data

Unmarked manufacturing electrical conduits and cabinet beside a blank electricity-flow worksheet, with onsite equipment and rooftop solar visible through windows.
Official records and their applicability
Statistic or methodSource and boundary
Establishment-count tableTable 4.4: national counts, not electricity volumes.
Consumption table catalogueMECS data catalogue: separate physical-unit and Btu presentations.
Collection unit and sampleCensus methodology: manufacturing establishments and a nationally representative sample.

These records establish definitions and scope. They do not establish a supplier’s electricity price, capacity or efficiency.

How far apart are purchased electricity and electricity demand?

Purchases covered 86.5% of the electricity U.S. manufacturers used in 2022, and far less in the three largest onsite generators, paper at 56.3%, petroleum and coal products at 72.7% and chemicals at 78.1%, which together produced 86.4% of manufacturing’s onsite electricity.

Zinit Research · EIA data, 2022 survey

Paper plants bought 56.3% of the electricity they used in 2022, the lowest share of any subsector; across manufacturing, purchases were 86.5% of net demand

Purchased electricity as a percent of net demand for electricity, by manufacturing subsector (NAICS), 2022. Net demand is purchases plus transfers in plus all onsite generation, less what was sold or transferred offsite, so it is the electricity the plants used. EIA released these tables in August 2025. Choose a subsector in the check to highlight it.

Paper56.3%
Petroleum and coal products72.7%
Chemicals78.1%
Food91.7%
Apparel94.1%
Textile product mills94.6%
Primary metals95.6%
Computer and electronic products95.8%
Wood products95.8%
Printing and related support96.1%
Transportation equipment97.4%
Textile mills97.6%
Machinery98.0%
Miscellaneous98.5%
Beverage and tobacco products98.5%
Plastics and rubber products98.6%
Leather and allied products98.8%
Nonmetallic mineral products99.1%
Electrical equipment, appliances, and components99.9%
Fabricated metal products100.1%
Furniture and related productswithheld by EIA–
All manufacturing86.5%

Which electricity figure is a supplier quoting?

The fields start with a subsector’s 2022 national totals in million kWh. Replace them with one plant’s records, all in one unit and for one period, and add the figure the supplier gave you.

EIA’s 2022 figures for paper, million kWh: net demand 96,642 (Table 11.1); net electricity 49,711 (Table 1.1). EIA rounds each component, so a total rebuilt from them can differ by a unit or two.

Electricity demand
96,642
Net electricity
49,711
Purchased share
56.3%
Quoted figure matches
Purchases

These flows give an electricity demand of 96,642 and a net electricity of 49,711, in the unit you entered, so purchases cover 56.3% of the electricity used and 42,258 more came from transfers and onsite generation, net of what the plant sold. A quoted 54,384 matches purchases, so it describes what the plant bought from outside and leaves out the rest of what it used, which matters if the supplier presents it as consumption.

From EIA’s definitions: demand = purchases + transfers in + all onsite generation − sold or sent offsite and net electricity = purchases + transfers in + solar, wind, hydro and geothermal − sold or sent offsite. A match is within 1% of the measure, and purchased share is purchases ÷ demand.

Source: U.S. Energy Information Administration, 2022 Manufacturing Energy Consumption Survey, Table 11.1 (components of net demand) and Table 11.3 (components of onsite generation), released August 2025, and Table 1.1 (net electricity), released March 2026. EIA withholds purchases or net demand for furniture and related products. Chart: Zinit Research. Data and method

Which electricity question does the comparison answer?

A procurement comparison should begin with a written decision: buying an offsite electricity supply, understanding electricity used inside a plant, or comparing a defined energy input. Put that sentence above the worksheet. Ask the energy owner to name the relevant meter boundary and reporting period before selecting a public reference. A familiar column heading is insufficient when one site generates power and another buys it.

EIA defines electricity demand using purchases, inbound transfers and total onsite generation, less electricity sold or transferred away. Its net-electricity definition instead includes noncombustible renewable generation and excludes onsite generation from combustible energy. The first-use explanation avoids counting energy transformed from other energy inputs twice. Keep these definitions attached to the chosen measure.

Why can an electricity column still be the wrong denominator?

MECS Table 4.4 reports establishment counts. Its electricity footnote includes electricity purchased or transferred in, but that does not turn a count into an electricity quantity. A count of establishments using a source cannot substitute for the volume used by a plant or for the price paid under a supply agreement. Confirm both the column definition and the table’s stated unit.

The official catalogue separates net-electricity tables presented in physical units from those presented in Btu. Preserve the selected table and unit beside each extracted field. For a buyer’s own worksheet, retain the units supplied by the meter or invoice; any conversion should carry an explicit method and responsible reviewer. Keep a volume reconciliation separate from a monetary bid comparison.

For example, a proposed plant comparison may arrive with a utility invoice, a generator operating log and a company transfer statement. Treat those as different evidence objects. Ask whether each covers the same location and period, whether the transfer is already inside another record, and whether a generation value is electricity output or fuel input. Keep the comparison on hold where those answers remain unavailable. This is an authored review scenario, with no invented volumes or claim that its balance has been verified.

Counting establishments ranks the subsectors differently from measuring their electricity

Fabricated metal products had 20.7% of the 184,124 manufacturing establishments that bought or received electricity in 2022 and used 5.2% of that electricity, while chemicals had 5.5% of the establishments and used 20.4%. Table 4.4 and Table 4.2 define electricity the same way, as purchased or transferred in, so the gap is the difference between a count and a quantity. The unit needs the same care: Table 4.2 reports that electricity as 2,769 trillion Btu, which at EIA’s 3,412 Btu per kilowatthour is 811,547 million kWh, and Table 11.1 reports purchases plus transfers in of 811,472 million kWh, the same quantity within rounding.

Share of establishments using purchased or transferred electricity, and share of that electricity, by subsector, 2022

Chemicals5.5%20.4%
Primary metals1.5%12.3%
Food9.9%9.9%
Plastics and rubber products4.6%8.7%
Nonmetallic mineral products6.7%7.6%
Paper1.7%6.7%
Petroleum and coal products1.0%6.5%
Transportation equipment4.1%5.7%
Fabricated metal products20.7%5.2%
Computer and electronic products3.6%3.4%
Machinery7.6%2.7%
Wood products5.2%2.6%
Beverage and tobacco products3.3%1.8%
Electrical equipment, appliances, and components1.8%1.4%
Miscellaneous7.1%1.2%
Printing and related support6.7%1.1%
Textile mills0.6%1.1%
Textile product mills2.0%0.4%
Apparel1.8%0.1%
Leather and allied products0.3%0.0%
Furniture and related productselectricity withheld4.4%withheld

Source: EIA MECS 2022, Table 4.4 (establishment counts, released December 2025) and Table 4.2 (offsite-produced fuel consumption, trillion Btu, released March 2026). Shares are of the manufacturing totals. EIA withholds the electricity quantity or count for furniture and related products. Chart: Zinit Research. Data and method

What belongs in the plant electricity-flow worksheet?

An authored electricity-flow reconciliation record
RecordEvidence and boundaryIf unresolved
Decision and selected measureBuyer question, named definition, accountable energy ownerHold the denominator choice
Site and reporting periodMeter or establishment boundary; start and end datesDo not combine unlike periods or locations
Purchases and inbound transfersInvoices, meter records and company transfer documents kept distinctRequest the missing flow record; do not assume no transfers
Onsite generationElectricity-output record; separately identified fuel or generation typeKeep unclassified generation outside an approved comparison
Sales and outbound transfersExport meter and transfer documents; overlap checked against other recordsLeave the outflow unconfirmed
Units and conversionsOriginal unit, conversion method and reviewerPreserve the original field; do not mix unlike units
Reconciliation and exceptionsRecord overlaps, missing evidence and unresolved differencesKeep an open exception with an owner
Commercial comparisonRelevant agreement, charge scope and volume boundaryKeep statistical reconciliation separate from price approval

Authored evidence workflow, not an official MECS reporting form, a certified balance, or a validated financial model. Completing a field does not verify the fact entered into it.

Use the worksheet as a handoff between procurement, the plant energy owner and finance. Give each supplied record a source, coverage period and reviewer. Retain the original value beside any normalized value so a later reviewer can reconstruct the choice. A difference should produce an exception to investigate rather than a balancing adjustment created only to make the totals agree. The worksheet’s useful output can be a clearly described hold.

Before applying the selected definition, review how a company transfer is represented in invoices and local meter records. Ask the record owner whether it is separately metered, centrally purchased, or already included in another reported flow. Keep the answer and its evidence together. Procurement can collect the records and document the commercial question; an energy-accounting owner should confirm the physical boundary and reconciliation.

Most onsite generation burns fuel, and net electricity leaves that generation out

Manufacturers generated 151,366 million kWh onsite in 2022, and 98.5% of it came from fuel burned in cogeneration or other generators; solar, wind, hydro and geothermal supplied 1.5%. Net electricity keeps only that last part, so in paper, where onsite generation equalled 48.9% of the electricity the plants used, net electricity was 51.4% of electricity demand, compared with 84.0% across manufacturing. Of the subsectors charted, computer and electronic products had the largest renewable share, 27.6% of its 1,237 million kWh, so two plants with the same generation volume can sit on different sides of the net-electricity line, and the worksheet records the generation type beside each volume.

Onsite generation by type, 2022, share of each subsector’s total (total in million kWh in brackets)

Chemicals (60,426)99.6%
Paper (47,281)99.3%
Petroleum and coal products (23,016)99.0%
Food (7,372)98.8%
Primary metals (4,920)99.7%
Wood products (2,271)97.8%
Computer and electronic products (1,237)72.4%
Transportation equipment (1,152)86.5%

Source: EIA MECS 2022, Table 11.3, components of onsite generation, and Table 1.1, net electricity. Generation from fuels is total onsite generation less noncombustible renewables. The value shown is the share from fuels. The eight subsectors that generated at least 1,000 million kWh, together 97.6% of all onsite generation, are shown. Chart: Zinit Research. Data and method

How should the source vintage and population be recorded?

The MECS release schedule identifies different release dates for consumption, expenditure, count and hand-suppressed tables. The tables describe their named survey year. A later release date does not make the observations a current plant measurement. Record observation vintage, release date and access date separately, and retain the exact table edition used in the comparison.

Census describes manufacturing establishments as the collection unit and a nationally representative sample as the source of questionnaire data. The July 2026 overview also announces movement toward annual delivery. That statement is not a newly observed annual dataset. A representative manufacturing survey supplies context for a manufacturing question; it does not establish a particular supplier’s current operating conditions.

A buyer reviewing a temporary energy arrangement at a data center should keep that facility’s own evidence in view. This article analyzes a manufacturing-survey accounting boundary. It does not apply a manufacturing estimate to a data center, validate a reported fuel-delivery arrangement, or establish the output of generating equipment. The transferable step is to name the evidence boundary before comparing records; any sector-specific assessment remains separate.

When should procurement leave the comparison open?

Leave the denominator unresolved when a material flow has no supporting record, when the period cannot be matched, or when the reviewer cannot tell what an equipment log measures. Name the missing evidence and the person responsible for obtaining it. Set the next review around the decision that needs the answer, rather than treating a completed spreadsheet as authority to approve a sourcing conclusion.

Use the supplier-risk assessment guide for the separate supplier-evidence process. Put continuing ownership and follow-up into the supplier-relationship management workflow. Further methods are available in the Journal. These links organize subsequent work; they do not fill missing electricity records.

Where does this data come from?

What the numbers are
The components of net demand for electricity (Table 11.1) and of onsite generation (Table 11.3), in million kilowatthours, for the 21 manufacturing subsectors in EIA’s 2022 Manufacturing Energy Consumption Survey, with EIA’s published net electricity (Table 1.1, physical units), the count of establishments using purchased or transferred electricity (Table 4.4) and the quantity they consumed (Table 4.2, trillion Btu). The figures describe the 2022 survey year at subsector level and never a named plant. Tables 11.1 and 11.3 were released in August 2025, Table 4.4 in December 2025, and Tables 1.1 and 4.2 in March 2026. EIA withholds some cells (Q: relative standard error above 50%; D: to avoid disclosing an establishment) and marks estimates under 0.5 with *; withheld cells are blank here. In the check, a cell EIA marks as under 0.5 starts at 0.
Source files
Definitions used
Electricity demand is estimated as the sum of electricity purchases, transfers in, and total onsite generation minus the quantities of electricity sold or transferred offsite (EIA MECS Terminology; Table 11.1 calls it net demand for electricity). Net electricity is the sum of purchased electricity, transfers in, and generation from noncombustible renewable resources minus the quantities sold and transferred offsite, so it excludes electricity generated or cogenerated onsite from combustible energy sources (EIA MECS Terminology). Table 11.3’s renewable column excludes wood and other biomass; its noncombustible sources are solar, wind, hydropower and geothermal power.
How the page stays current
The survey tables change only when EIA revises a table or fields the next survey, which it plans for 2027; the page is rebuilt from whichever edition is retained, and the release month of each table is printed with it.
Download the data

One ZIP holds the CSV file below and a README that lists the sources and the citation.

  • eia-mecs-2022-electricity-balance-by-subsector.csv

Leave your work email and company, and the download starts as soon as you send the form.

We respect your privacy. No spam. Your data is never sold.

Related Zinit Research
Cite this page

Zinit Research. “Purchased, Net and Onsite Electricity in Manufacturing Data.” Data: U.S. Energy Information Administration, 2022 Manufacturing Energy Consumption Survey, Tables 11.1, 11.3, 1.1, 4.4 and 4.2. https://zinit.com/journal/research/manufacturing-electricity-accounting-boundaries, data retrieved October 22, 2026.

How this page is produced
Zinit Research downloads the source files listed above, keeps the originals, and computes every chart, table and figure in the data sections from them; an automated check rebuilds each number from the files before the page is published. The article text follows the Editorial Policy.

Sources

  1. MECS Terminology (undated living reference; accessed 2026) — U.S. Energy Information Administration, 2026. Foundational evidence (official report): Citation in b1, b6; supports only the stated definition, unit, vintage or methodology boundary.
  2. 2022 MECS Table 4.4: establishment counts, released December 2025 — U.S. Energy Information Administration, 2025. Contextual evidence (official report): Citation in b3, b8; supports only the stated definition, unit, vintage or methodology boundary.
  3. 2022 MECS data catalogue and release schedule, 2025–2026 releases — U.S. Energy Information Administration, 2026. Contextual evidence (official report): Citation in b3, b9, b16; supports only the stated definition, unit, vintage or methodology boundary.
  4. Manufacturing Energy Consumption Survey (MECS) Methodology — U.S. Census Bureau, 2026. Contextual evidence (official report): Citation in b3, b17; supports only the stated definition, unit, vintage or methodology boundary.

Global Procurement Brief

Procurement news, briefed

The market moves, supplier signals, and cost levers that matter — curated by the team behind this Journal. Daily or weekly, your call.

We respect your privacy. No spam. Your data is never sold.

Related procurement resources

Request a demo
Request a demo