OREDA (Offshore and Onshore Reliability Data) is a Joint Industry Project that has been collecting equipment-failure data across the petroleum industry since 1981. The 6th edition of the OREDA Handbook (2015), prepared by SINTEF and NTNU and distributed by DNV, contains failure rates, failure modes, and repair times across 25 equipment classes spread over a Topside volume and a Subsea volume. OREDA also produced ISO 14224 (the data-collection format the rest of the industry uses) and ISO 20815 (production assurance and reliability management).
If your FMECA quotes a failure rate and the source line is empty, you are guessing. OREDA exists so you do not have to. The harder question, after access, is whether the OREDA population resembles your population enough to use the numbers as is.
What OREDA is
OREDA is three things at once, and the confusion between them is the source of most misunderstandings in industry:
- A Joint Industry Project (JIP) with seven to eleven oil and gas operators as members, running continuously since 1981. The members fund the work, contribute their own data, and elect a Steering Committee that sets direction.
- A reliability databank, populated by member-supplied operating, failure, and maintenance records from offshore and onshore production installations. Geographic coverage spans the North Sea, Gulf of Mexico, West of Shetland, Angola, the Adriatic, the Caspian, and other producing regions.
- A forum for the development of reliability methods and the production of international standards. ISO 14224 and ISO 20815 both came out of OREDA work and remain anchored in OREDA practice.
Most engineers first meet OREDA as a thick blue Handbook on a senior reliability engineer's desk. That is one delivery channel. The data also lives in OREDA@Cloud, a member-and-subscriber-only platform hosted on DNV's Veracity portal, where the same population is queryable interactively.
OREDA in one sentence
The longest-running, most widely cited reliability-data programme in the petroleum industry, owned by a closed group of operators, prepared by SINTEF and NTNU, and distributed by DNV.
A 35-year history
The project started in 1981 in cooperation with the Norwegian Petroleum Directorate (now the Norwegian Offshore Directorate). It has progressed through twelve project phases since then, with handbook releases at most major phase boundaries:
| Year | Edition | Notes |
|---|---|---|
| 1984 | 1st | Offshore reliability data, first published handbook. |
| 1992 | 2nd | Expanded equipment coverage, refined taxonomy. |
| 1997 | 3rd | Failure-mode and -mechanism data added. |
| 2002 | 4th | Methodology aligned with the early drafts of what became ISO 14224. |
| 2009 | 5th | Two-volume split into Topside and Subsea. |
| 2015 | 6th | Current published handbook. Onshore data formally added to the population. |
The phasing matters because it is the timeline against which the standards lineage runs. ISO 14224's first edition appeared in 1999 (and was revised in 2006 and 2016) and tracks closely to the OREDA data structure that hardened during phases 4 and 5. OREDA was where the format was proven; ISO 14224 is the format made portable.
Why OREDA matters
Three reasons reliability engineers cite OREDA in deliverables, even when site-specific data exists:
- Industry recognition. Regulators, operators, classification societies, and EPC contractors all know OREDA. A failure rate sourced "OREDA 2015 Vol 1, page 134" survives review meetings that an internally collected number does not.
- Population size. OREDA aggregates failure events across decades and dozens of installations. A site collecting its own data on a single fleet of pumps will run out of statistical power long before OREDA does.
- No real alternatives. For upstream petroleum specifically, there is no comparable peer-reviewed, industry-funded, standards-aligned dataset. Process-industry data exists in pockets (PERD, FMD-91 for electronics, etc.), but for offshore and subsea production equipment OREDA stands alone.
Site data versus OREDA. Site-specific reliability data, when it exists in usable form, is almost always more representative of your asset than OREDA. The catch is the qualifier. Most CMMS-based attempts to extract failure rates locally produce numbers with no statistical power because the population is too small or the failure-mode coding is inconsistent. OREDA is where you go because the rigorous local effort almost never gets done.
Who runs it
The JIP is governed by its member companies, which have included over the years (membership rotates) Eni, BP, Petrobras, Gassco, Engie, Equinor (formerly Statoil), Total, and others. Members supply operational data into the central database under confidentiality agreements; the published handbook anonymises and aggregates the contributions so no operator can be back-identified from any failure record.
The roles split as follows:
- OREDA JIP issues the data and the handbook. Steering Committee chairs and Project Manager rotate on a multi-year cycle.
- SINTEF and NTNU (Norway) prepare the handbook content. SINTEF's reliability group has been the technical partner for most of the project's history.
- DNV distributes the handbook and runs the OREDA@Cloud platform on its Veracity portal.
- SATODEV developed the OREDA software platform.
The data structure
OREDA data is organised into three categories. Anyone who has read ISO 14224 will recognise the shape, because this is where ISO 14224 came from.
Inventory data
Identifies what each equipment unit is, how it is configured, and how it is operated. Five sub-fields:
- Classification data. Equipment class, type, sub-type, operating mode.
- Identification data. Tag number, manufacturer, model, serial number, installation site.
- Specification data. Design parameters: power, capacity, pressure rating, temperature, fluid handled.
- Maintenance data. Planned-maintenance regime applicable to this unit.
- Operation data. Operating profile, environment, surveillance period in calendar hours.
Failure event data
Records what failed, how, and what happened next:
- Identification. Which equipment unit, which subunit, which maintainable item.
- Failure event data. Failure mode, failure mechanism, failure cause, severity, the date and the operating state at the time of failure.
- Remarks. Free-text context, used by analysts to disambiguate edge cases when the data is later mined.
Maintenance event data
Records the response to a failure (corrective) or a scheduled task (preventive):
- Identification. Linked to the failure event when corrective.
- Maintenance event data. Activity performed, time taken, downtime caused, dates.
- Maintenance resources. Crew size, disciplines involved, special equipment used.
- Remarks. Same role as for failure events.
Why the structure matters
This is the same trio of categories ISO 14224:2016 specifies for any equipment-reliability collection programme. If your CMMS work-order types, failure-mode codes, or operating-state codes do not map back to this structure, your data is not OREDA-compatible and will not aggregate against the OREDA population. Building the mapping is most of the work in a credible reliability programme.
System hierarchy
OREDA hangs everything off a three-level system hierarchy. A failure record cannot be created until the failed item is placed at the right level.
The hierarchy is mirrored exactly in ISO 14224. It is also the reason why where you draw the equipment boundary matters for the failure rate you compute. A failure rate for "compressor" that includes the driver is a different number from one that does not. OREDA publishes equipment-boundary diagrams for every class precisely to remove that ambiguity.
Equipment boundaries
The boundary defines which subunits and maintainable items belong to the equipment unit. OREDA boundary diagrams (compressor, gas turbine, control system, and so on) call out:
- The driver (electric motor, gas turbine, steam turbine): included or excluded?
- The lubrication subunit: separate equipment unit, or part of the parent?
- The instrument and control loop: at what point does the loop stop being "compressor instrumentation" and start being "control system"?
- Power supply: main feed only, or step-down transformer included?
Two analyses that disagree on boundary assumptions will produce different failure rates for the same physical asset. When you cite an OREDA number in your own work, cite the boundary too.
Failure definitions
OREDA defines a failure as "the termination or the degradation of the ability of an item to perform its required function(s)." That language matches IEC 60050-191 and ISO 14224. The detail is in what counts and what does not.
What counts as a failure
- Complete failure of the item.
- Failure of part of the item that causes the item to be unavailable for corrective action.
- Failure discovered during inspection, testing, or preventive maintenance that requires repair.
- Failure on safety devices, control devices, or monitoring devices.
What does not count
- Unavailability due to preventive or planned maintenance.
- Shutdown of the item due to external conditions, or where no physical failure condition of the item is revealed. A shutdown is not a failure unless there is a recorded maintenance activity attached to it.
The "spurious shutdown" trap. Trip-and-restart events on rotating equipment often look like failures in a CMMS work-order log but do not qualify as OREDA failures unless the maintenance team actually intervened. Many programmes inflate their failure rates by counting every nuisance trip. OREDA does not.
Severity classes
Every recorded failure carries a severity class, scored at the equipment-unit level (not at the subunit or maintainable-item level, and not considering anything outside the equipment boundary).
| Class | Effect on equipment-unit function | Typical example |
|---|---|---|
| Critical | Immediate and complete loss of function. | Compressor trip with no restart possible. |
| Degraded | Function is not lost but is compromised. The item still performs, below specification. | Pump delivering reduced flow or head. |
| Incipient | An imperfection in the state or condition of the item. No immediate effect on function. | Vibration trend climbing toward an alarm threshold. |
The split matters when you query OREDA for a failure rate. The headline rate is usually all severities combined. Critical-only rates are typically much lower and are the right number to use in a safety-related risk analysis. Degraded and incipient rates feed condition-monitoring intervals and P-F-interval estimates.
What's in the handbook
For each equipment class the handbook publishes a consistent block of tables. The exact column layout has evolved across editions; the categories have not.
- Failure rates. Mean (constant lambda), with confidence intervals from the underlying chi-squared estimate. Per-mode breakdowns when the population supports it.
- Failure mode distribution. Percentage breakdown of how the recorded failures distribute across the standard mode codes for that class.
- Repair times. Active repair time and total downtime, mean and percentiles.
- Maintainable item versus failure mode. Cross-tabulation showing which maintainable items contribute which failure modes. Useful for FMECA scoping.
- Failure mechanism versus failure mode. Cross-tabulation showing the underlying cause for each observed mode. Useful for root-cause work and for designing condition-monitoring strategies.
A note on confidence intervals
OREDA reports uncertainty alongside the point estimates. A failure rate of 4.5 per million calendar hours with a 90% confidence interval of [3.1, 6.4] is doing more work than the bare 4.5. When the confidence interval is wide, the population was small or the failure mode was rare, and you should hesitate before quoting the mean as a deterministic input. When the interval is narrow, the data is doing what you wanted it to do.
Equipment classes covered
The 6th-edition handbook covers 25 equipment classes across the Topside and Subsea volumes. The OREDA database itself extends beyond that to additional classes that are accessible via OREDA@Cloud but not necessarily printed in the handbook.
Topside (Volume 1)
Rotating: pumps, compressors, gas turbines, steam turbines, electric generators, electric motors, combustion engines. Static: heat exchangers, vessels, fired heaters, piping, valves, storage tanks. Electrical: power transformers, frequency converters, UPS, switchgear. Safety and control: fire and gas detection, control logic.
Subsea (Volume 2)
Subsea production systems, wellhead and Christmas trees, manifolds, control modules, jumpers, flowlines, power and signal cables.
Additional classes in the database
Beyond the printed handbook, the OREDA database also contains data on:
- Cranes
- Fire-water systems
- Frequency converters
- HVAC systems
- Loading arms
- Nozzles
- Subsea power cables
- Switchgear
- Swivels
- Turrets
- Wellhead and X-mas tree (dry)
- Winches
- Subsea control system
- Dry-tree riser
- Electrical power distribution
- Subsea pumps
- Subsea vessels
- Common components
Applications
OREDA data shows up in six main analytical contexts:
Availability studies
Production-availability estimates, design optimisation, equipment-selection comparisons. The same MTTF and downtime numbers feed both. Common deliverable: a Reliability, Availability, and Maintainability (RAM) study during front-end engineering.
Risk analysis
Estimating probabilities of critical events, computing survival times for safety-critical items, sizing risk-based inspection (RBI) intervals. OREDA failure rates feed into the demand rates and on-demand probability terms used in IEC 61508 and IEC 61511 functional-safety calculations.
Maintenance planning and optimisation
Reliability-Centred Maintenance (RCM) per SAE JA1011 uses failure rates to score consequence likelihood and to size condition-monitoring intervals. Spare-parts requirements come from MTTF and lead-time math. Cross-tabulating OREDA mechanism-versus-mode data with your own equipment list reveals where your designs are weak relative to the industry population.
Operations
Setting condition-monitoring thresholds and trend-monitoring windows. The mode distribution tells you which failure mechanisms matter for the equipment in front of you. For new installations without local history, OREDA is the only way to set sensible alarm bands on day one.
Benchmarking
Comparing your fleet's measured failure rate against the OREDA population reveals whether you are running better, the same, or worse than the rest of the industry on a per-class basis. Benchmarking is a use case the JIP encourages because it is what justifies operator membership.
Life-cycle cost
LCC analyses consume failure rates and repair-time distributions to calculate the maintenance cost component of total cost of ownership. Without OREDA, the maintenance term in an LCC model is essentially guesswork until late in the life cycle.
Relationship to ISO 14224 and ISO 20815
The two standards that cite OREDA explicitly:
ISO 14224, Reliability and maintenance data collection
ISO 14224:2016 specifies the format in which the petroleum and natural gas industries should collect and exchange reliability data. The taxonomy in Annex A, the failure-mode codes in Annex B, the data structure for inventory, failure events, and maintenance events: all of it is the OREDA structure made into an international standard. If you are setting up a CMMS for an offshore asset, the question is not "should I use ISO 14224?" but "where do I get the manpower to enforce it?"
ISO 20815, Production assurance and reliability management
ISO 20815:2018 covers production-assurance management throughout the life cycle of an upstream installation. It specifies how reliability requirements are flowed down from production targets to equipment specifications. OREDA failure-rate data is the assumed input where local data is unavailable.
The takeaway. If you use ISO 14224 for your CMMS taxonomy and ISO 20815 for your reliability programme, you are running a system that is structurally compatible with OREDA. You can compare your data to OREDA's, and OREDA can absorb your data if you ever choose to contribute. That portability is OREDA's lasting contribution to the industry.
Access
OREDA is not free. The data was paid for by the operator members who funded thirty-five years of collection effort, and it is licensed accordingly.
The handbook
Print and PDF copies of the 6th-edition handbook are sold through DNV. Two volumes (Topside and Subsea) at industry-standard reference-book prices. Suitable for an individual reliability engineer, a small consulting practice, or a single project's needs.
OREDA@Cloud
Subscription access to the live database via DNV's Veracity portal. Search, filter, and export functionality across the full population, beyond what is printed in the handbook. Priced at the operator-corporate tier; appropriate for organisations doing repeated reliability analyses across multiple projects.
JIP membership
Full membership in the OREDA JIP requires operator status (you must own and operate offshore or onshore production assets) and the willingness to contribute your operational data into the central database. Members get the deepest access plus a vote on Steering Committee decisions.
How Bluestream uses OREDA
OREDA appears at four points in the Bluestream Maintenance Development Toolbox:
- FMECA tool. When the user lands on a recognised OREDA equipment class, the tool primes the per-mode failure-rate field with the appropriate OREDA reference. The user enters the actual lambda from their own OREDA access; the tool does not redistribute the data.
- Concept Builder. The library of Generic Maintenance Concepts is structured to ISO 14224, which means it is OREDA-compatible. Each concept is keyed to an OREDA equipment class.
- Work Instruction Generator. Failure-mode codes on every work step are OREDA-aligned (PLU, BRD, ELP, FTS, and the rest). When the technician closes the work order with the correct code, that closure data feeds back into a future reliability calculation that can be benchmarked against OREDA.
- Criticality classification. The consequence categories of NORSOK Z-008:2024 are independent of OREDA, but the survival-time and demand-frequency inputs that feed the matrix typically come from OREDA-style analyses.
Common pitfalls
Mistakes that show up in real reliability deliverables:
- Ignoring the equipment boundary. Citing an OREDA pump failure rate for a pump unit while including the driver in your own analysis. The OREDA boundary may have excluded the motor; your number is now apples to oranges.
- Mixing severity classes silently. Quoting an "all-severity" rate where the analysis context requires "critical only" inflates the calculated risk by an order of magnitude or more.
- Using a wide confidence interval as if it were narrow. A point estimate from a population of twelve failure events deserves caveats. Using it as a deterministic input is overconfidence dressed as data.
- Stale data on rapidly evolving equipment. The 6th-edition handbook is from 2015. Equipment classes that have changed materially since then (subsea control systems, certain frequency-converter topologies) may need a hedge or a literature supplement.
- Treating OREDA as a replacement for site data. OREDA is a useful prior, not a destination. The mature programme uses OREDA to bootstrap, then converges on site data once the local population is large enough to be informative.
References
- OREDA project home page. DNV-hosted overview of the JIP, member companies, and OREDA@Cloud access.
- OREDA handbook page. Purchasing information for the printed Topside and Subsea volumes.
- ISO 14224:2016. Petroleum, petrochemical and natural gas industries: collection and exchange of reliability and maintenance data for equipment.
- ISO 20815:2018. Petroleum, petrochemical and natural gas industries: production assurance and reliability management.
- SINTEF project page. The technical preparation partner for the handbook.
- Eisinger S. and Clavé N., Offshore & Onshore Reliability Data (OREDA) Collection: OREDA JIP Status, 4th ISO seminar on international standardization in the reliability technology and cost area, Statoil, Houston, 4 May 2018. The primary source used to verify the historical and structural claims on this page.