DieselNet - Emission Standards Database for Heavy-Duty Engines

DieselNet is the reference layer every emissions conversation eventually cites: an information service on engines, fuels and clean transportation technology, maintained continuously since 1997, whose Emission Standards section organizes the world's engine emission regulation summaries into one browsable structure. Pick a region, pick an engine category, read the limits by model year.

The regional spread is genuinely global. North America covers United States federal and California programs alongside Canada and Mexico. Europe spans the EU collective and the national deviations that matter - Germany, Norway, Sweden, Switzerland, Turkey - plus the Russia/EAEU framework. Asia runs through China, India, Japan, Korea, Singapore and Thailand; Australia and the major South American markets (Argentina, Brazil, Chile, Colombia, Peru) carry their own summaries; and an International tier holds IMO marine rules, UIC locomotive requirements, the Gothenburg Protocol and World Bank guidelines.

Within any jurisdiction the categories stack the way engineers actually meet them: cars and light trucks, heavy-duty truck and bus engines, nonroad diesel engines with their in-service conformity provisions, nonroad spark-ignition engines, stationary engines, marine engines and locomotives, then greenhouse-gas and fuel-economy rules, low-emission zones, periodic technical inspections and occupational-health limits. Limit tables sit at model-year grain - the US heavy-duty page alone tabulates CO, HC, HC+NOx, NOx and PM from 1974 onward. Test-cycle and fuel-regulation libraries, a technology paper archive, calculators and a glossary round out the corpus. Get a sample of this dataset cut to the jurisdictions you sell into.

One regulation, one engine category, one model year, one pollutant per column - the opening row of the US heavy-duty highway compression-ignition table:

table  : US Heavy-Duty Highway Compression-Ignition Engines

year   : 1974
co     : 40        hc+nox : 16
hc     : -         nox    : -        pm : -

Read the dashes as carefully as the numbers. The 1974 federal standard capped carbon monoxide at 40 and wrote its hydrocarbon-plus-NOx requirement as a single composite of 16, leaving the separate hc, nox and pm cells empty on that row - so the row is simultaneously a limit sheet and a snapshot of how the standard itself was constructed in that era. As the table advances year by year, composites split into separate columns and particulate limits arrive, which turns the full table into a readable history of regulatory architecture rather than just a list of numbers. Request a sample and the rows come back filtered to your markets and engine categories, every cell populated or dashed exactly as shown.

Six fields carry every limit row, and all six were verified against the published limit-table headers rather than inferred from descriptions. Five hold the pollutant limits themselves - CO, HC, the HC+NOx composite, NOx and PM - and one anchors the row in time. The identifying stack is deliberately spare, which is what makes the tables comparable across decades: a 1974 row and a current-model-year row occupy the same grid, and a difference in values always means a difference in the standard, never in the schema.

Where a standard writes its requirement as a composite, the HC+NOx column carries the number and the separate columns dash out - and vice versa once regulators began splitting the pollutants. Everything the corpus documents beyond this grid folds under [additional fields on request](#field_dictionary-footnote): the GHG and fuel-economy entries, the low-emission-zone and inspection-regime rows, the nonroad in-service provisions and the test-cycle library that pairs every numeric limit with the cycle used to certify it.

Geography - worldwide by design. The summaries resolve to roughly twenty-five distinct jurisdictions across five inhabited continents, from the two programs that dominate compliance calendars (US federal and California, EU) down to single-market frameworks such as Singapore and Peru, with the international tier - IMO marine, UIC locomotive, Gothenburg Protocol, World Bank guidelines - covering the regimes that cross borders. For anyone certifying or pricing a machine into several markets, the value is that all of them follow one organizational logic.

Temporal - historical through current, at model-year resolution. The US heavy-duty tables open at 1974 and run unbroken to the present, and the reference has been maintained continuously since 1997, so a compliance timeline and a history of the regulation both fall out of the same rows.

Granularity - one row per regulation, engine category and model year, one cell per pollutant. That is the level at which engineering decisions happen: whether a 2027 engine needs a different aftertreatment package than a 2024 one is a question answered by adjacent rows, not by averaging.

API, files, or your warehouse. Daily, weekly, or hourly.

You choose the channel and the cadence; the field dictionary above travels unchanged through all three. Standards text never reaches your pipeline as markup to parse - rows arrive flattened, one observation per pollutant-year-category, ready to join against your own engine or product master. Cadence changes are a settings conversation, not a re-integration project, and a sample cut to your jurisdictions comes first either way.

• Product planning and homologation - engine and equipment builders lay target-market limit tables side by side per model year to decide what aftertreatment hardware a platform needs before tooling commits.
• Regulatory horizon scanning - compliance teams track when each jurisdiction's next tier bites, converting effective dates into engineering deadlines and retrofit budgets.
• Market-entry prioritization - manufacturers expanding across borders compare stringency by market to sequence launches where certified hardware already complies.
• Emissions inventory and air-quality modeling - researchers feed fleet-age distributions against per-year limits to estimate what a regional fleet actually emits, the input layer beneath inventory models.
• Competitive intelligence - suppliers of aftertreatment, sensors and fuels map which upcoming standards create demand for their components, and where.
• Citation-grade reference work - consultants, litigators and academics anchor claims to a specific limit, year and jurisdiction instead of a secondhand summary.

Competitive intelligence and product teams get the regulatory clock their competitors are watching, organized per market and model year; see competitive intel product teams use cases. Market researchers and consultants get the demand-side timeline - every tightening standard is a procurement event for someone - anchored to citable limits; see market researchers use cases. Journalists, academics and students get fifty years of US heavy-duty limits as a continuous, attributable series; see journalists academics use cases. Data scientists and ML engineers get a tidy six-field grid that joins cleanly against fleet, registration and production datasets to model fleet-wide emissions trajectories. Across all of them the constant is the same: one schema spanning every jurisdiction makes cross-market questions answerable without re-reading twenty regulatory websites.

Three things worth knowing upfront. First, the corpus is a standards reference, not a measurement set - rows carry the legal limits, not test results or fleet averages, so pair it with an inventory or registration dataset when you need what engines actually emit in the field.

Second, the core grid is intentionally narrow. Six columns cover the criteria-pollutant limits that define the classic heavy-duty tables, while GHG and fuel-economy entries, low-emission-zone rules, inspection regimes and the test-cycle library exist as documented extensions - they fold under additional fields on request, so name them when you scope your sample rather than expecting them in every delivery.

Third, granularity is per regulation, category and model year. If your question lives below that line - serial-level production data or per-engine certificates - that is a different dataset entirely, and we will say so during the sample conversation instead of overselling this one.