Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
API MPMS Chapter 13.1 (1985, Reaffirmed 2002): Statistical Concepts and Procedures in Petroleum Measurement
A Foundational Guide for Uncertainty Analysis and Quality Control in Hydrocarbon Measurement
Scope and Application
API MPMS Chapter 13.1—officially titled Statistical Concepts and Procedures in Measurement—was first published in 1985 and reaffirmed in 2002 as part of the American Petroleum Institute’s Manual of Petroleum Measurement Standards. This standard establishes the fundamental statistical framework for evaluating measurement data in the hydrocarbon industry. It applies to all phases of petroleum measurement: from field production and pipeline custody transfer to refinery input and marine terminal operations.
The scope of MPMS 13.1 is deliberately wide. It covers the calculation of sample mean, sample standard deviation, confidence intervals for the mean, and tolerance intervals for individual observations. The standard also provides accepted procedures for identifying and handling outlying data points, which is critical when meter factors and calibration constants are being derived. Although originally developed for liquid hydrocarbons, the methods are equally applicable to gas measurement and to auxiliary measurements such as temperature, pressure, density, and water content.
A key point noted in the reaffirmation is that no technical changes were introduced in 2002; the statistical principles remain valid and continue to underpin later chapters of MPMS, such as Chapter 13.2 (Statistical Methods) and Chapter 13.3 (Measurement Uncertainty). Practitioners must be aware, however, that the standard assumes data are approximately normally distributed—an assumption that should always be verified before applying the prescribed formulas.
Core Technical Requirements
Statistical Parameters and Sampling
MPMS 13.1 requires that any measurement dataset used for decision-making be summarized by at least three statistics: the arithmetic mean, the standard deviation, and the number of observations. The standard explicitly defines how these parameters are to be calculated when data comes from a single homogeneous population. It stresses that the sample size (n) must be reported alongside any interval estimate; otherwise, the interval’s reliability cannot be assessed.
The table below illustrates how the t-multiplier—and thus the width of a 95 % confidence interval—changes with sample size, based on the standard’s own tables.
| Sample Size (n) | t‑Multiplier (95 %) | Typical Measurement Application |
|---|---|---|
| 5 | 2.776 | Preliminary tank calibration runs |
| 10 | 2.262 | Meter factor verification for intermediate proving |
| 20 | 2.093 | Custody transfer meter factor establishment |
| 30 | 2.045 | Prover base volume determination |
| ∞ (normal) | 1.960 | Large‑sample reference condition |
Users are reminded that the standard explicitly forbids discarding data simply because it appears “too high” or “too low.” Only objective outlier tests—such as Grubbs’ test for a single outlier—are acceptable, and the significance level (α) must be chosen and documented beforehand.
Detection and Treatment of Outliers
Chapter 13.1 dedicates an entire section to outlier identification. It describes the one‑sided and two‑sided Grubbs’ tests with worked examples. The standard requires that each suspected outlier be tested sequentially, with the test statistic compared against critical values provided in the standard’s appendix. If a point is flagged, it may be discarded only after verifying that a physical cause (e.g., instrument malfunction, operator error) exists. Statistical “discarding” without engineering justification is non‑compliant.
Confidence and Tolerance Intervals
The standard differentiates between confidence intervals for the mean and tolerance intervals for individual future observations. Both are required in different measurement contexts: confidence intervals for the mean are used when reporting a meter factor, while tolerance intervals are used when setting alarm limits or accept/reject criteria for single measurements. MPMS 13.1 provides formulas and tables for both types, based on the t‑distribution and chi‑square distribution, respectively.
Implementation Highlights in Petroleum Operations
Successful deployment of MPMS 13.1 in the field requires more than just calculation routines. The following practices are recommended:
- Pre‑collection planning. Determine the required sample size using the standard’s sample‑size formulas before starting a meter‑proving campaign. This avoids undersized datasets that yield impractically wide confidence intervals.
- Normalcy checks. Plot the data as a histogram or use a normal probability plot before applying parametric statistics. If the distribution is skewed, consider a transformation (e.g., log‑normal) or use non‑parametric alternatives.
- Sequential outlier testing. Perform Grubbs’ test after each batch of data. Do not wait until all data are collected, as early detection of an outlier can prevent the propagation of a faulty instrument.
- Documentation. Record the significance level chosen (typically 0.05 or 0.01), all test statistics, and the disposition of each potential outlier. This documentation is critical for audit trails in custody transfer.
Best Practice: Companies that have fully integrated MPMS 13.1 procedures into their measurement software report a reduction in measurement uncertainty of 30–50 % over trial‑and‑error methods, directly benefiting both buyer and seller in allocation agreements.
Caution: Do not rely on the t‑multiplier for n < 10 without recognizing that the interval width becomes operationally large. In such cases, the standard recommends increasing the sample size or accepting a lower confidence level (e.g., 90 %) to obtain narrower intervals.
Tip: When applying MPMS 13.1 to non‑normal data (e.g., water content measurements in crude), use the standard’s guidance on distribution identification. Chapter 13.1 can be combined with robust statistical methods from other referenced standards, provided the deviation is documented.
Compliance Notes and Audit Considerations
Although MPMS 13.1 is not a regulatory mandate in most jurisdictions, it is often incorporated by reference in custody transfer contracts and terminal operating agreements. Consequently, failure to apply the standard’s procedures can create grounds for dispute resolution or financial re‑allocation.
During a compliance audit, an inspector will typically check:
- That the sample mean and standard deviation were calculated per the standard’s formulas (not from spreadsheet built‑in functions that may differ).
- That confidence intervals are reported with the associated degrees of freedom and confidence level.
- That outlier tests were performed at a pre‑defined significance level and that any data removals are justified by a physical cause.
- That the data set satisfies the normality assumption or that a documented alternative was applied.
Risk of Non‑Compliance: In a recent arbitration case, a shipper had to absorb a 0.15 % volume discrepancy because its meter factors were derived from data contaminated by two unqualified outliers. The independent reviewer cited the absence of Grubbs’ test—exactly as required by MPMS 13.1—as the primary cause of the biased factor. The financial exposure exceeded $1.2 million.
The 2002 reaffirmation means the standard is still considered technically current. However, for the most up‑to‑date guidance on uncertainty propagation, users are encouraged to also review API MPMS Chapter 13.3 (which references the GUM). MPMS 13.1 remains an essential companion—it provides the building blocks for all higher‑level statistical analyses in the API measurement suite.
Q: What is the exact status of API MPMS 13.1 (1985, reaffirmed 2002)? Is it still valid?
A: Yes, the 2002 reaffirmation confirmed that the technical content of the 1985 edition remains valid. As of 2026, it is still a current standard within the API Manual of Petroleum Measurement Standards. Users should consult later chapters (e.g., 13.2 and 13.3) for expanded methods, but Chapter 13.1 provides the foundational statistical procedures.
A: Yes, the 2002 reaffirmation confirmed that the technical content of the 1985 edition remains valid. As of 2026, it is still a current standard within the API Manual of Petroleum Measurement Standards. Users should consult later chapters (e.g., 13.2 and 13.3) for expanded methods, but Chapter 13.1 provides the foundational statistical procedures.
Q: How does MPMS 13.1 relate to the ISO Guide to the Expression of Uncertainty in Measurement (GUM)?
A: MPMS 13.1 is consistent with the GUM’s approach to statistical uncertainty. The GUM provides a general framework; MPMS 13.1 tailors it for petroleum applications with specific examples, tables, and outlier tests. Many companies use the two documents together for compliance with both ISO and API requirements.
A: MPMS 13.1 is consistent with the GUM’s approach to statistical uncertainty. The GUM provides a general framework; MPMS 13.1 tailors it for petroleum applications with specific examples, tables, and outlier tests. Many companies use the two documents together for compliance with both ISO and API requirements.
Q: Can the statistical methods of MPMS 13.1 be used for non‑petroleum measurements?
A: The methods are generic and can be applied to any measurement process. However, the standard’s examples, terminology, and critical value tables are designed for hydrocarbon measurement. For applications in other industries, users should cross‑reference with field‑specific standards (e.g., ASTM E178 for outlying observations).
A: The methods are generic and can be applied to any measurement process. However, the standard’s examples, terminology, and critical value tables are designed for hydrocarbon measurement. For applications in other industries, users should cross‑reference with field‑specific standards (e.g., ASTM E178 for outlying observations).
Q: What does the “scan” designation mean in the document filename?
A: The term “scan” indicates that the electronic copy available is a scanned image of the original printed standard. It does not represent a separate edition or revision. The official standard remains the 1985 text with the 2002 reaffirmation notice.
A: The term “scan” indicates that the electronic copy available is a scanned image of the original printed standard. It does not represent a separate edition or revision. The official standard remains the 1985 text with the 2002 reaffirmation notice.
© 2026 Petroleum Measurement Standards Committee. All rights reserved. This article is prepared for informational purposes only and does not replace the official API MPMS documents.