Update on Proposed ASTM D4929

Procedure C - XRF
Presented at the Crude Oil Quality Association meeting, St. Louis, MO June 8, 2017
By Leslie Johnson, XOS Applications Engineer/ASTM WK57076 Technical Contact
 ASTM D4929 Organic Chlorides in Crude
Methodology:
A crude oil sample is distilled to 400F to collect the naphtha
fraction.
Inorganic chlorides are removed by water wash
Chloride content of the naphtha fraction is determined by one of
two methods:
Procedure A, sodium biphenyl reduction and potentiometry
Procedure B, combustion and microcoulometry
Chloride content of the crude is determined by multiplying the
naphtha chloride concentration by the naphtha fraction

What if there was a Procedure C, using XRF?

6/8/17 COQA meeting 2

 ASTM D4929 Procedure C XRF Timeline

June 2014: June 2015: July 2016: April 2017:

Present MWDXRF pilot Present revised ILS study Procedure C/D ILS D02.03 (17-02)
study results at ASTM outline at ASTM meeting (10 finished ballot closes
D02.03 XRF TG meeting labs each performing MWD, Five negative votes
Obtain permission for a new WD, and MED/EDXRF) on Procedure C
work item for D4929 XRF TG vote 34-1-6 in favor
Procedure C - XRF of revised study

June Dec June Dec July Dec April June

2014 2014 2015 2015 2016 2016 2017 2017

Dec 2014: Dec 2015: Dec 2016: June 2017:

Present XRF ILS outline at WDXRF dropped from Present ILS Working to
ASTM meeting (30 labs total: Procedure C study study results at adjudicate
10 MWD, 10 WD, and 10 Procedure D CIC, running ASTM meeting Procedure C
MED/EDXRF) concurrently using same ILS Obtain negatives at
samples permission to ASTM meeting
move to ballot

6/8/17 COQA meeting 3

 D4929 Procedure C ILS Design
ILS design was presented to the ASTM D02.03 XRF task group for
approval (ASTM WK57076, ILS 1122):

Distillation and wash procedure retained as written in the method

No revisions to Procedures A or B

Procedure C includes multiple XRF technologies: MWDXRF,

MEDXRF, EDXRF
Original design included WDXRF as well, but it was dropped in Dec 2015

ILS designed to replicate existing 1987 Procedure A/B ILS as

much as possible
10 crude oil samples in blind duplicate per original D4929 design
Thank you Dennis Sutton for identifying crude oil equivalents
Thank you Marathon for supplying WTI, LLS, Mars bulk crude samples
0-12 mg/kg doped organic chloride per original design

6/8/17 COQA Meeting 4

 D4929 Procedure C ILS Design (cont.)
Up to 10 end user participants
June 2015: reduced from the originally proposed 30 participant design (10
MWD, 10 WD, 10 MED/EDXRF labs) due to lack of participants and funding for
the ILS study

Each lab ran one distillation per sample aliquot (10 samples in
blind duplicate = 20 distillations)

XRF analysis to be performed on each naphtha fraction using

MWDXRF, WDXRF, MED/EDXRF analysis
WDXRF dropped in Dec 2015 due to lack of available participants or demo
analyzers

Note: participants did not run Procedures A and B for bias.

Why? Financial and participant constraints.
This information was discussed with and approved by the ASTM D02.03 XRF task
group prior to the start of the ILS.

6/8/17 COQA Meeting 5

 D4929 Procedure C ILS Design (cont.)
The ASTM Statistician will determine whether Procedure C
precision should be pooled or separated for each technology
based on the ILS data.

Results will be balloted as D4929 Procedure C unless otherwise

advised by ASTM Statistician to ballot as a stand alone method.

This ILS design was presented to the D02.03 XRF task group at the
June 2015 Fort Lauderdale meeting, and it was approved by a vote.

ILS design changes/additions after XRF task group vote:

Dec 2015: WDXRF dropped in due to lack of available participants and demo
analyzers
Dec 2015: Procedure D - CIC analysis added as a concurrent ILS (WK53424,
ILS 1329). Naphtha samples were analyzed by XRF and CIC when possible.

6/8/17 COQA meeting 6

 D4929 Procedure C XRF Participants
I would like to thank the following D4929 XRF Method C
participants:

Amspec Corpus Christi

Inspectorate New Orleans
Intertek Chicago
Intertek Los Angeles
Intertek New Orleans
Intertek San Francisco
Marathon Catlettsburg
SGS Deer Park
Shell Global Solutions Houston

6/8/17 COQA meeting 7

 D4929 XRF Participant Breakdown
Lab Code
MWDXRF MEDXRF EDXRF
(randomized)
1 X X X

2 X X X

3 X X X

4 X X* X

5 X X X

6 X X X

7 X X

8 X X

9 X X

10

Total participants 9 9 (10) 6

*Submitted results from two different MEDXRF systems
6/8/17 COQA Meeting 8
 D4929 ILS Sample Set Comparison
D4929 Procedures A/B 1987 ILS: D4929 Procedure 2016 ILS:
Doped Doped
ID Crude Dopant Cl ID Crude Dopant Cl
(mg/kg) (mg/kg)
1,1,1- 1,1,2-
A Liberty 1 A LLS 1
trichloroethane trichloroethane
1,2- 1,2-
B Butte 12 B Mars 12
dichloropropane dichloropropane
Butte/ 1,1,1- 1,1,2-
C 10 C Mars/LLS 10
Liberty trichloroethane trichloroethane
1,2- 1,2-
D Butte 3 D Mars 3
dichloropropane dichloropropane
E WTI blank 0 E WTI blank 0
1,2- 1,2-
F Liberty 1 F LLS 1
dichloropropane dichloropropane
o- o-
G Butte 5 G Mars 5
dichlorobenzene dichlorobenzene
H Liberty blank 0 H LLS blank 0
methylene methylene
I WTI/Butte 6 I WTI/Mars 6
chloride chloride
1,1,1,- 1,1,2,-
J WTI 5 J WTI 5
trichlorethane trichlorethane
6/8/17 COQA Meeting 9
 D4929 Procedure C ILS Statistics
Statistical analysis was performed by ASTM Statistician Alex Lau,
D02 CS94 chair:
Extreme results were identified and excluded using the GESD Many-Outlier
Detection technique as per ASTM D7915.
The pre-screened data was then analyzed in accordance with ASTM D6300 to
obtain the precision.
A D6708 assessment was performed to determine relative bias between XRF
techniques.

Recommendations:
Precision for EDXRF needs to be listed separately.
Notwithstanding the similarity between MWDXRF and MEDXRF precision,
these also should be listed separately due to the observed bias.
The observed relative bias between technologies should also be listed.

6/8/17 COQA meeting 10

 Statistical Findings
Precision Equations, Method C XRF
Repeatability r, mg/kgA Reproducibility R, mg/kgA
MWDXRF r = 0.643 * X0.44 R = 1.235 * X0.44
MEDXRF r = 0.591 * X0.44 R = 1.500 * X0.44
EDXRF r = 0.934 (X +0.4)0.48 RB = 2.000 (X +0.4)0.48
A
Where X = mg/kg chloride
B
The degrees of freedom associated with the reproducibility estimate from this round robin study are
29. Since the minimum requirement of 30 (in accordance with Practice D6300) is not met, users are
cautioned that the actual reproducibility may be significantly different than these estimates.

Between-Method Bias Outcome

Method X Method Y Suggested Bias CorrectionA
MEDXRF MWDXRF = X - 0.658
EDXRF MEDXRF = X + 0.519
EDXRF MWDXRF No bias correction
A
Where:
= predicted Y-method value
X = single X-method measured results

6/8/17 COQA Meeting 11

 Repeatability Comparison: A, B, C

Note: data is from two different studies, 1987 Procedure A and B study and 2016 Procedure C study.

6/8/17 COQA Meeting 12

 Reproducibility Comparison: A, B, C

Note: data is from two different studies, 1987 Procedure A and B study and 2016 Procedure C study.

6/8/17 COQA Meeting 13

 Low concentration samples appear to be dominated by naturally
occurring chlorides (100+% recovery)
High concentration samples show that not all of the volatile
compounds will distill from a complex crude oil under the
conditions of this test method

6/8/17 COQA Meeting 14

 D02.03 (17-02) Subcommittee Ballot
D4929 Procedure C results were authorized at the December 2016
meeting to go to subcommittee ballot.

D02.03 (17-02) Item 1: D4929 Procedure C XRF subcommittee

ballot results:
45 affirmative
5 negatives + 1 non-subcommittee member negative
110 abstain
3 comments

6/5/2017 15
 Summary of Negative Voter Rationale
1. Relative bias between procedure C vs. procedures A and B was
not determined.
2. This should be balloted as a separate stand alone test method
rather than as a part of D4929.
3. Procedure C has poor precision and should not be added to
D4929, particularly EDXRF.

Other negative voter concerns:

Scope of the proposed procedures is well above what is normally encountered.
The distillation step was not performed in duplicate.
The recommended calibration range is too high (0-100 mg/kg).
Each XRF method should have its own procedure (e.g. C, D, E).
All XRF equipment manufacturers should be named in the method (instead of
naming one sole source manufacturer).
No XRF equipment manufacturers should be named in the method.
A reference to the concurrent CIC ILS should have been included in the ballot
rationale.

6/8/17 COQA Meeting 16

 Next Steps
Working to adjudicate negative votes
Next ASTM D02 meeting is June 26-29th in Boston, MA
A ballot item must pass subcommittee and main committee ballots
before approval

Questions?

Thank you for your time.

Leslie Johnson | Applications Engineer

O 518.880.1500 ext 402 | E ljohnson@xos.com
XOS | www.xos.com

6/8/17 COQA Meeting 17