CEN/TR 16885:2015

SLOVENSKI STANDARD
01-december-2015
7HNRþLQDIWQLSURL]YRGL3UHLVNRYDQMHSUHVNXVQHPHWRGH]DPHUMHQMHRNVLGDFLMVNH
VWDELOQRVWLGL]ODLQ)$0(GL]HOVNHPHãDQLFHVNLVOLQVNLPãWHYLORPSRVWDUDQMX
Liquid petroleum products - Investigation on test method for measurement of the
oxidation stability of diesel and diesel/FAME blends by Acid Number after ageing
Flüssige Mineralöl-Erzeugnisse - Bericht über die Bestimmung der Oxidationsstabilität
von Diesel und Diesel/FAME-Mischungen durch Bestimmung der Säurezahl nach
Verälterung
Produits pétroliers liquides - Recherche de la détermination de la stabilité à l'oxydation
du gazole et des mélanges gazole/EMAG par l'indice d'acide après vieillissement
Ta slovenski standard je istoveten z: CEN/TR 16885:2015
ICS:
75.160.20 7HNRþDJRULYD Liquid fuels
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

CEN/TR 16885
TECHNICAL REPORT
RAPPORT TECHNIQUE
September 2015
TECHNISCHER BERICHT
ICS 75.160.20
English Version
Liquid petroleum products - Investigation on test method
for measurement of the oxidation stability of diesel and
diesel/FAME blends by Acid Number after ageing
Produits pétroliers liquides - Recherche de la Flüssige Mineralöl-Erzeugnisse - Bericht über die
détermination de la stabilité à l'oxydation du gazole et Bestimmung der Oxidationsstabilität von Diesel und
des mélanges gazole/EMAG par l'indice d'acide après Diesel/FAME-Mischungen durch Bestimmung der
vieillissement Säurezahl nach Verälterung

This Technical Report was approved by CEN on 17 August 2015. It has been drawn up by the Technical Committee CEN/TC 19.

CEN members are the national standards bodies of Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia,
Finland, Former Yugoslav Republic of Macedonia, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania,
Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey and
United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION

EUROPÄISCHES KOMITEE FÜR NORMUNG

CEN-CENELEC Management Centre: Avenue Marnix 17, B-1000 Brussels
© 2015 CEN All rights of exploitation in any form and by any means reserved Ref. No. CEN/TR 16885:2015 E
worldwide for CEN national Members.

Contents Page
European foreword . 3
1 Scope . 4
2 Context and creation of a dedicated subgroup . 4
3 Participants in the work . 5
4 Meetings of the subgroup „Acid No.“ . 5
5 Main steps of the work item study . 6
5.1 Creation of the NWI . 6
5.2 Test method used . 6
5.3 First Round Robin Test . 7
5.4 Improvement of the test method . 7
5.5 Pass/fail methodology . 8
5.6 Second Round Robin Test . 9
6 Conclusions . 11
7 Acid number determination method available for lab use . 11
8 Acknowledgements . 11
Annex A (informative) Test method transcription . 12
Annex B (normative) Round Robin Results . 20
B.1 October 2010 results . 20
B.2 2012/2013 RRT . 22
Annex C (normative) Pass-/Fail discriminant analysis . 25
Bibliography . 31

European foreword
This document (CEN/TR 16885:2015) has been prepared by Technical Committee CEN/TC 19 “Gaseous and
liquid fuels, lubricants and related products of petroleum, synthetic and biological origin”, the secretariat of
which is held by NEN.
Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. CEN [and/or CENELEC] shall not be held responsible for identifying any or all such patent rights.
1 Scope
This Technical Report describes the investigation into the development of a standard test method to
determine oxidation stability of diesel fuel and fatty acid methyl ester (FAME) blends in diesel by the use of
determining the acid number after ageing at elevated temperature. It provides conclusions following this
work that have been discussed by CEN. The result thereof is that no European Standard has been developed.
2 Context and creation of a dedicated subgroup
In case of poor diesel or biodiesel quality, ageing of the fuel in the fuel system under high pressure and
temperature (recirculation of fuel, high injector temperature, long storage in the vehicle fuel tank) may
cause various car problems due to the formation of acidity through oxidation (i.e. deposit of sediments,
deposit of lacquer, corrosion, lube oil deterioration).
Acidity of the fuel is therefore considered as a relevant parameter to evaluate oxidation stability of the Diesel
fuel. Test methods based on the measurement of the acid number (AN) after an ageing step were studied. An
ageing test temperature of 115 °C which is significantly higher than the test temperature of 95 °C applied in
EN ISO 12205 [1] has been chosen because it better discriminates fuel’s oxidation stability. Additionally, it is
closer to the temperature range prevailing in fuel systems of current and future engine technologies (i.e.
common rail systems).
Customer complaints related to fuel degradation linked to oxidation stability in France are shown in
Figure 1.
Figure 1 — Customer complaints linked to fuel degradation in France
A test method based on the change of the acid no. of a fuel during ageing, Delta AN, was evaluated in
CEN/TC 19/JWG1 'FAME Test methods' in 2008. In the Delta AN method, the fuel is aged at 115 °C for 16 h
by passing a stream of oxygen through the fuel using the oxidation cell of EN ISO 12205. The acid number of
the fuel before ageing is subtracted from the acid number of the aged fuel. The results of Round Robin tests
made on the Delta AN method led to the conclusion that the Delta AN test method, although discriminative,
exhibits a precision not enough robust ; this test method needed some analytical improvements. A draft
report about the test results applying the Delta AN method performed in 2008 was presented to
CEN/TC 19/JWG1 in January, 2011.
Further work concerning the improvement of the Delta AN test method was carried out in France in 2009. A
new test method based on the measurement of the acid number of the fuel after ageing was developed.
Based on the results of a cross check test, it was decided at the JWG1 meeting on September 4, 2009, that
additional work would be necessary concerning the robustness and precision of the new method. As such
work being not covered by the CEN/TC 19 mandate to JWG1, it was proposed that experts continue the
improvement work and issue a proposal for a NWI to WG 24.
Based on the results of the work of the French experts the continuation of the work was accepted by WG24
in March 2010. JWG1 started the work, creating a dedicated subgroup for this preliminary new work item
(PNWI).
3 Participants in the work
Several European experts were active within this project, represented by one or more member(s)
participating in the meetings. The memberships are listed in Table 1.
Table 1 — Members of the Subgroup „Acid No.“
Company Country Members
PSA France P. Jestin
S. Duperrier; P. Manuelli; P. Pestiaux; A.
TOTAL France
Vincent; A. Gandubert
SHELL M. Schmidt
Germany
Deutsche BP Germany W. Strojek
Neste Oil M. Kuronen
Finland
IFPEN L. Pidol
France
OMV Austria W. Koliander
ADM Germany J. Groos; J. Fischer
ASG Germany T. Wilharm
Metrohm Switzerland C. Haider; U. Loyall
SGS Germany, France M. Kulikowski; D. Juillet
4 Meetings of the subgroup „Acid No.“
The members of the group have been working on the assessment of the oxidation stability of diesel and
diesel/FAME blends by determination of the acid value after ageing from beginning of 2010 to mid-2014.
The meetings are listed in Table 2. This work have been reported and discussed within JWG1 at each session.
The main orientations and action plans have systematically been validated by JWG1.
Table 2 — Meetings of the Subgroup „Acid No.“
Meeting Date and location
Meeting 1 April 27, 2010
Conference call
Meeting 2 July 07, 2010
Meeting 3 January 14, 2011
Meeting 4 May 24, 2011
PSA Peugeot Citroën – Paris
Call conference July 25, 2011
Conference call
Meeting 5 September 02, 2011
IFPEN – Rueil
Meeting 6 March 22, 2012
PSA Peugeot Citroën – La Garenne Colombes
Meeting 7 November 13, 2013
TOTAL – Paris La Défense
5 Main steps of the work item study
5.1 Creation of the NWI
The first meeting of the group took place in April, 2010. The scope was presented to the members: the
objective was to improve the precision of the new acid number test method applicable to diesel fuels from
B0 to B10. In that context, some adjustments were made on the test method protocol and it was decided to
run first a cross-check test. Necessary improvements based on the outcome of the study should be
implemented to the method. A Round Robin test should finally be conducted in order to develop the
precision of the method.
5.2 Test method used
The method used has been developed to be applicable to diesel fuels from B0 to B10. The main analytical
parameters are listed hereafter and the full description of the test method is given in Annex A.
— Sample amount: (10 ± 0,2) g;
— Heating bath temperature: (115 ± 0,2) °C ;
— Oxygen rate: (1 ± 0,1) L/h;
— Running time for fuel oxidation: 16 h ± 5 min;
— Maximal time between the end of oxidation step and the AN measurement: 4 h.

The fuel was aged either in an oil bath or an heating bath as applied in the Rancimat equipment
5.3 First Round Robin Test
A RT was run in October, 2010 to assess the precision of the proposed new AN method on both colorimetric
and potentiometric determination of the AN. Nine samples were used for the RRT: 3 B0, 4 B7 and 2 B10.
Samples were representative for the European Market, some containing cetane improver (content between
100 and 1000 ppm), CFPP additives and/or lubricity additives. Thirteen labs out of fourteen participants
have returned their results on time: ten labs have performed colorimetric determination (oil bath and
[2]
Rancimat bath according to EN 15751 ) and eleven labs have performed potentiometric determination (oil
bath and Rancimat bath according to EN 15751). The results of this RRT are given in Annex B.
The RRT results led to the following comments:
— Even if there was a discrimination between “good” and “bad” products, results were worse than
expected, in particular for the potentiometric version. When the dispersion of results with the
potentiometric method was discussed, all participants agreed that experimental parameters were
perhaps not optimized and that it was necessary to work on it (electrode system, solvent, dynamic
titration, etc.).
— “Home-made” diesels, meaning diesels formulated by blending “good” and “bad” B0 or B7 in order to
reach certain AN target, seemed to have a strange behaviour. Even if the formulated products seemed to
be homogeneous, the results obtained by the labs were really different and the statistical distribution of
results indicated strong issues.
— There were some difficulties of being more precise on very good samples (AN <0,1 mg KOH/g). For non
acidic samples, the resulting precision is poor due to the precision of colorimetric titration (in test
method ISO 6618 [3] the reproducibility is 0,04 mg KOH/g for samples with AN <0,1 mg KOH/g).
— No impact of Rancimat bath compared to oil bath was observed, no bias was observed.
Thus this RRT pointed out that the method could not be used in the current state to be submitted for
standardization. It was decided to continue the work to understand potentiometric results, to identify what
could have an influence on the results dispersion and thus improving the method (work on experimental
parameters, propose a few tests to assess the new parameters, …).
In parallel, the group members have decided to ask CEN/TC 19/WG 36 (statisticians) how a pass/fail test
could be established, as this method could be considered as such.
All the details about this RRT are available in the internal document “Round Robin Study Report 2010-831”
of CEN/TC 19.
5.4 Improvement of the test method
In order to improve the potentiometric titration test method, the participants of the first RRT were asked for
detailed information of settings and conditions of their instruments. While there was no significant
difference on the equipment (brand of device, software, electrode system, analytical parameters), the way of
detection of the equivalent point was not the same for all participants. Indeed, the determination of the
equivalent point can be automatically or manually done and some labs used the point corresponding to the
pH 11 aqueous buffer. In parallel, several tests were performed by TOTAL to estimate the impact of various
analytical parameters. Based on these results, some improvements were found to optimize the titration step:
a) Set all titration program parameters as proposed;
b) Use the colorimetric solvent and add indicator solution (to follow the solution colour change, especially
for blank titration);
c) Perform a manual (re)check on equivalent point for each titration;
d) Do NOT use the point corresponding to the pH 11 aqueous buffer (to define the KOH sample volume).
At that stage of the study, the group decided organizing a new RRT with a well-defined measurement
parameters for this RRT in order to minimize the variations from one lab to another. Nevertheless, all the
experts agreed on the fact that the critical part of the test lies in the ageing step more than in the acid
number determination.
5.5 Pass/fail methodology
The results obtained during the first RRT showed that it would be very difficult to propose a method with
“classical” precision according to EN ISO 4259 [4] (r and R versus acid number). Indeed, the acid numbers
measured on the RRT samples were not evenly distributed. The AN were either low or high and the samples
formed two separate populations that can be considered as “good” samples and “bad” samples. By
consequence, the members of the Subgroup proposed to use the preliminary tests on 44 B0 to B10 samples,
conducted in August 2011, for the development of a Pass/Fail method. This model, developed in close
cooperation with WG 36 experts, is based on General Discriminant Analysis (GDA). GDA applies the methods
of the general linear model to the discriminant function analysis problem. It is a strong tool for detecting the
variables that allow to discriminate between different groups, and for classifying samples into different
groups with an accuracy better than chance.
In the two-group case, discriminant function analysis can be thought of as a special kind of multiple
regression. If we code the two groups in the analysis as P (pass) and F (fail) and use that variable as the
dependent variable in a multiple regression analysis, we would then get results that are analogous to those
we would obtain via Discriminant Analysis. In general, in the two-group case a linear formula of the type:
Group = a + b *x + b *x + . + b *x (1)
1 1 2 2 m m
where:
a is the constant
b - b are regression coefficients
1 m
The interpretation of the results of a two-group problem is straightforward. Those variables with the largest
(standardized) regression coefficients are the ones that contribute most to the prediction of group
membership. Another major purpose to which discriminant analysis is applied is the issue of predictive
classification of cases. Once a model has been finalized and the discriminant functions have been derived, we
can predict to which group a particular sample belongs. The classification functions can be used to
determine to which group each case most likely belongs. There are as many classification functions as there
are groups. The classification functions can be used to directly compute classification scores for some new
observations. Once the classification scores for a case are calculated it is easy to decide how to classify the
case: in general the case is classified as belonging to the group for which it has the highest classification
score.
Another important item is the probability that a new sample will make the predicted choice. Those
probabilities are called posterior probabilities and are defined as the probability, based on the knowledge of
the values of other variables that the respective case belongs to a particular group. Posterior probabilities
can be used to evaluate the risk of a bad classification. In the case of the pass/fail two group classification
with less than 0,95 or 0,99 probability should be disregarded. Like in regression models, a model needs to be
validated on new samples not used for the model fitting.
In order to determine the feasibility of a pass/fail methodology for the determination of AN after ageing on
Bx, the group members selected a set of 44 samples, from B0 to B10 (14*B0, 2*B5, 14*B7, 1*B8 and 13*B10),
the preliminary test was run in August 2011. All the samples were analyzed by one laboratory (TOTAL). The
results were processed by applying the General Discriminant Analysis leading to the classification of each
sample as Pass or Fail. Figure 2 shows the results of the preliminary tests.
Figure 2 — Results of preliminary pass-/fail evaluation
Those preliminary tests were satisfying so the group decided to run a full Round Robin Test in agreement
with CEN/TC 19/JWG1. Details are given in Annex C.
5.6 Second Round Robin Test
The RRT was performed by using the new AN method with AN measurement by potentiometric titration on
19 diesel blends (Bx). Those samples were either taken directly from European filling stations and refineries
or formulated by blending B0 with FAME. In order to encourage labs’ participation, it was proposed running
the RRT in two parts in order to spread the workload for participants. The approach was agreed on by JWG1.
Part 1 was launched in December 2012 and part 2 in February 2013. Seven laboratories out of eleven have
provided full sets of results.
a) EN ISO 4259 approach
The evaluation of the data confirmed the result of the 2010 RRT: it was impossible achieving a precision
which would have been acceptable for a standardized test method. Repeatability and reproducibility

according to EN ISO 4259 [4] were not sufficient, the 2R criteria being not fulfilled for most of the samples
(Annex B).
b) Pass-/Fail model
The model was improved by processing data of the Round Robin and the data of the preliminary using a AN
threshold of 1,0. Details of the data evaluation (including the characterization of the sample aspect after
ageing) are shown in Annex C. In contrary to the classical approach, processing the data by using the
Pass/Fail model lead to robust classification functions and allowed the group to confirm the performance of
the model on the new AN method. The classification functions (“ax + b” type) are the following ones:
PASS FAIL
-0,480 096 -11,942 8
4  ,584      223       31   ,2  89   8
a
The classification of an unknown sample can be executed according to the following protocol:
— Measure AN after ageing;
— Calculate Pass and Fail criteria;
— The highest criteria gives the classification.
EXAMPLE
AN = 0,3 mgKOH/g Pass criterion = 0,894 270 9 Fail criterion = -2,555 86
Sample is classified as PASS.
AN = 0,8 mgKOH/g Pass criterion = 3,186 382 4 Fail criterion = 13,089 04
Sample is classified as FAIL.
Based on the evaluation of the results according to the described protocol the pass-/fail methodology seems
to be robust and can distinguish between “good” and “bad” fuels; this was also confirmed by
CEN/TC 19/WG 36 experts. The method can therefore be regarded as “validated” as a Pass/Fail method to
determine the oxidation stability of diesel fuels which were experimentally covered by the discriminant
analysis. A safe application of this method to fuels of unknown origin is not possible.
The best configuration was a discriminant analysis with a AN threshold of 1,0 (23 samples). Details of the
construction of the pass/fail model are shown in Annex C.
All the details of this RRT are available in the internal CEN/TC 19 document “Round Robin Study Report
2013-460”.
6 Conclusions
In spite of all efforts and valuable work for improving the measuring part, the method still shows a number
of "defects" which would be very difficult to handle:
— the investigation if/if not real samples from the field would follow the same Pass/Fail population as the
ones used in test method development,
— the question if there would be sufficient separation between the Pass and Fail sections in order to avoid
errors in the pass/fail declaration,
— the potential need for a grey zone, where a pass/fail decision cannot be made with sufficient safety.
As it seems to be impossible to overcome these issues within a reasonable time the group concluded that for
the time being there is no progress to foresee. Based on this information CEN/TC 19/JWG1 decided to apply
to CEN/TC 19 to cancel the PNWI for the development of this method and to disband the subgroup. The new
ageing method will not be submitted to CEN as EN method.
7 Acid number determination method available for lab use
The new AN method is described in Annex A and laboratories are free to use it to evaluate the PASS/FAIL
criteria on their Bx samples, from B0 to B10.
8 Acknowledgements
Thanks to all the experts who have taken part to the work on the AN method during several years.
Thanks to the RRTs participants for their valuable activity in carrying out the analytical measurements.
Thanks to the companies for having supplied several batches of fuels and FAME.
Thanks to T. Feuerhelm, DIN FAM, and P. Pestiaux, TOTAL, for their comprehensive statistical evaluations.
Annex A
(informative)
Test method transcription
This annex is a copy of the test method for the measurement of the oxidation stability of Bx by Acid Number
after ageing at 115 °C using potentiometric titration
1 Scope
This test method covers the measurement of
...