🍎 Determination of constituents (moisture, fat, protein, starch, crude fibre) and technological parameters (e.g. digestibility) in animal feeding stuffs, cereals and milled cereal products by near infrared spectrometry (NIR, 770–2500 nm) — reflectance or transmittance measurement, conversion of the spectrum to content by a calibration model developed on representative samples and validated locally on an independent set of at least 20 samples (bias, slope, SEP, RMSEP), instrument stability checks and control charts of NIR–reference differences; PN-EN ISO 12099:2017-10

Food chemistry PN-EN ISO 12099

Overview

WHAT THE STANDARD COVERS. We read the text of the standard in the iTeh sample of ISO 12099:2017 (second edition, English text) — PN-EN ISO 12099:2017-10 adopts EN ISO 12099:2017 and ISO 12099:2017 as identical (PKN search): contents, foreword, introduction, Clauses 1–6 and Clause 7 up to 7.3 — the sample ends on page 6. We do not read the rest of the statistics (7.3–7.6: bias, RMSEP, SEP, slope), sampling (8), procedure (9), checking instrument stability (10), running performance check of calibration (11), precision and accuracy (12) or Annexes A (guidelines for specific NIR standards), B (examples of outliers and control charts) and C (supplementary terms); we know the titles from the contents.

ACCORDING TO THE TEXT OF ISO 12099:2017. The standard was prepared by ISO/TC 34 “Food products”, Subcommittee SC 10 “Animal feeding stuffs”. The second edition replaces the first (ISO 12099:2010). Introduction: the document is based on ISO 21543 | IDF 201 (milk and milk products). Scope (1): guidelines for the determination by near infrared spectroscopy of constituents — moisture, fat, protein, starch, crude fibre — and parameters such as digestibility in animal feeding stuffs, cereals and milled cereal products. No normative references (2). Definitions (3): an NIR instrument predicts constituent contents and technological parameters through relationships to absorptions in the near infrared; feeding stuffs include raw materials, fodder, meat and bone meal, mixed feed and pet food; constituent content is the mass fraction determined by an appropriate, standardized or validated chemical method (e.g. moisture, fat, protein, crude fibre, NDF, ADF); NIR methods can be developed for other parameters and sample types if the procedure of the standard is followed, and units are those of the reference methods. Principle (4): spectral data in the near infrared are transformed into contents by calibration models developed on representative samples. Apparatus (5): reflectance or transmittance instruments over 770–2500 nm (12 900–4000 cm⁻¹) or parts of it, dispersive, interferometric or non-thermal (diodes, lasers), with diagnostics of noise, reproducibility and wavelength accuracy; a sufficiently large sample volume or surface; a mill — changes in grinding conditions can affect the result, e.g. by heating and loss of water. Calibration and validation (6): the instrument is calibrated before use; no general calibration procedure can be given; the validation set should cover composition, seasonal, geographic and genetic effects, processing, storage, temperature and differences between instruments — at least 20 samples; reference methods in statistical control, with known precision, where possible traceable to SI (certified reference materials); outliers: a sample is not an outlier if it is within the working range and the spectral variation of the calibration (e.g. Mahalanobis distance) and its residual is below the limit; spectral (x) outliers are checked first — such samples are removed; for y-outliers the measurement and reference are repeated, and a confirmed sample is not removed; equations are validated locally on an independent set of at least 20 samples for each sample type, constituent, temperature and other factors; the calibration is valid only for the range in which it was validated; samples with a residual above ±3 SEP are outliers; a model that does not give acceptable statistics should not be used (what is acceptable is decided by the parties); regression yref = a + b × yNIRS; a bias significantly different from zero is removed by adjusting the constant term; a slope significantly different from 1 means skewness — slope adjustment is generally not recommended, it is better to expand the calibration set and repeat validation on a new set; after a change of conditions (sample population, presentation, temperature, another instrument) validation is not valid without additional calibration — e.g. a calibration on grass silages from one area may not give the same accuracy for another; transfer between instruments may require bias or slope adjustment or standardization; the calibration is checked after replacement or repair of a major part of the instrument. Statistics (7): validation set independent of the calibration set (in a plant — new batches, in agriculture — a new crop or location), carefully analysed by the reference method, at least 20 samples, the same NIR protocol as in routine (one or two measurements); residual ei = yi − ŷi, so predictions that are too high give a negative bias; a reference–NIR plot shows correlation, bias, slope and outliers (Figure 1); bias is the mean difference, caused e.g. by new sample types, instrument drift, drift in wet chemistry, changes in the process and in sample preparation.

STATUS (as of 03.10.2026). The PKN search engine (query “PN-ISO 12099”, 03.10.2026) lists PN-EN ISO 12099:2017-10 in Polish and English versions, “Introduces: EN ISO 12099:2017 [IDT], ISO 12099:2017 [IDT]”, without a withdrawal note; the results also include PN-EN ISO 12099:2010 and PN-EN 12099:2002 (a different standard). There is no “PN-ISO 12099” in the results.

HOW MANY LABORATORIES AND IN WHAT FORM (copy of the accreditation scope database, data up to 17.09.2026, read 03.10.2026). The number 12099 appears in 10 records at 4 laboratories, but at AB 237 it is PN-EN 12099:2002 (polyethylene — volatile content, gravimetric method), i.e. a different standard; this NIR standard is held by AB 822 (6 records), AB 856 (2) and AB 1616 (1). In the current PCA documents (read 03.10.2026; AB 822 issue no. 13 of 23.07.2025, AB 1616 issue no. 13 of 14.10.2025, AB 856 issue no. 18 of 21.11.2025) the number appears at these laboratories and additionally at AB 1029 (issue 23 of 04.05.2026), whose item is not in the database copy; in every document the number of page markers equals the number of PDF pages, without repetitions. There are three written forms: “PN-ISO 12099:2017-10 z zastosowaniem analizatora NIR” (AB 822 — no such standard exists in the PKN catalogue), “PN EN ISO 12099:2017” without hyphens (AB 856, marked with a triangle) and “PN-EN ISO 12099” without year (AB 1616); AB 1029 has “PN-EN ISO 12099:2017-10” with procedure PB-36-00. None of these items is suspended. The “Laboratories” tab splits along these forms: query “PN-ISO 12099” — 1 laboratory (AB 822), “PN-EN ISO 12099” — 1 (AB 1616), “PN EN ISO 12099” — 1 (AB 856), “ISO 12099” — 0 — checked by tab queries on the labcoda.pl production server on 03.10.2026.

WHAT THE LABORATORIES TEST (items in PCA). AB 822: maize silage — dry matter (25.0–44.0) g/100 g, crude ash (1.1–1.8), total protein (2.0–4.0), starch (6.0–16.0), crude fat (1.0–2.0), crude fibre (5.0–9.0) g/100 g. AB 856: feed materials (wheat, barley, maize, soybean meal) and compound feeds for poultry and pigs — moisture 7.00–14.0 %, crude ash 1.50–6.70 %, total protein 6.00–46.0 %, crude fat 1.20–24.0 %, crude fibre 2.20–12.6 %; feeds for laying hens — narrower ranges (e.g. protein 17.2–18.4 %); reflectance NIRS as a predictive method. AB 1029: poultry feed — crude protein 13.7–23.0 %, crude fat 3.6–7.8 %, crude ash 2.1–10.8 %, crude fibre 2.1–6.9 %, moisture/dry matter 87.5–90.5 %. AB 1616: agricultural products — content of constituents by NIR, without ranges in the database copy.

WHERE A SEEMINGLY CORRECT RESULT IS EASY TO GET. Validity range: a calibration is valid only for the sample types, range and temperature of the validation (6.4.1) — the ranges in PCA (e.g. maize silage at AB 822, layer feeds at AB 856) are exactly such limits, and a sample outside them gives a number without support in the validation. Population: the same content but a different population (region, variety, processing) may give a different accuracy (6.5). Sign of the bias: the residual is reference minus NIR, so overestimating predictions give a negative bias (7.2) — confusing the sign in the correction doubles the error. Slope: slope adjustment is not recommended unless the calibration is applied to new samples or instruments; first look for high-leverage outliers (6.4.3). Outliers: y-outliers confirmed by repetition must not be removed from the validation statistics (6.3). Instrument: after replacement or repair of the optics or detector the calibration must be checked (6.5). Grinding: heating in the mill may drive off water (5.2). Designation: “PN-ISO 12099” does not exist in the PKN catalogue, and forms without hyphens or year scatter the search — in the tab each of them gives one laboratory.

WHAT WE DO NOT GIVE. We do not quote the formulae for RMSEP, SEP and slope, significance limits, sample handling, instrument stability checks, control charts, precision data or the annexes — they lie outside the sample. We do not summarize the reference methods (moisture, protein, fat, fibre, starch) in this entry. We do not give acceptable SEP values or requirements for feeds.

Method principle

The reflectance or transmittance spectrum of the sample is measured in the near infrared (770–2500 nm or part of it). A calibration model, developed on representative samples analysed by reference methods, converts the spectrum into constituent content or parameter value. The model is used only after local validation on an independent set of at least 20 samples (bias, slope, prediction error) and only within the range in which it was validated; in routine use, instrument and calibration stability are checked with control samples and comparison with the reference method.

Applications

• Moisture, fat, protein, starch, crude fibre and digestibility in animal feeding stuffs, cereals and milled cereal products (1)
• Calibration, local validation and transfer of calibrations between instruments (6)
• Running control of NIR analysers in feed laboratories (10, 11 — outside the sample)
• Maize silage, feed materials, compound feeds for poultry and pigs — items of 4 laboratories in PCA

Key parameters

Standard

Standard number

PN-EN ISO 12099:2017-10

Title (PL)

Pasze, ziarno zbóż i produkty przemiału — Wytyczne stosowania spektrometrii bliskiej podczerwieni

Title (EN)

Animal feeding stuffs, cereals and milled cereal products — Guidelines for the application of near infrared spectrometry

Step-by-step procedure

1. Calibration

   Model on representative samples with reference values (4, 6.1, 6.2). PN-EN ISO 12099:2017-10 — no withdrawal note (PKN, 03.10.2026).

2. Local validation

   Independent set ≥ 20 samples; plots, bias, slope, SEP; outliers > ±3 SEP (6.4, 7).

3. Adjustments

   Bias — constant term; slope — preferably expansion of the set and new validation (6.4.2–6.4.4).

4. Routine measurement

   Same protocol as in validation, sample within the calibration range (7.1; Clause 9 outside the sample).

5. Running control

   Control sample, diagnostics, control charts of NIR–reference differences (10, 11 — outside the sample; titles from the contents).

Required equipment and apparatus

Reagents, media and consumables

Health and safety (OHS)

• The part of the standard read contains no separate safety requirements; usual rules for working with a mill apply when grinding samples (5.2)

Frequently asked questions