EVALUATION OF IGNITABILITY METHODS (LIQUIDS)
By
M. Umana, W. Gutknecht, C. Salmons, R. Chapman, R. Handy,
and E. Pellizzari
Research Triangle Institute
Post Office Box 12194
Research Triangle Park, North Carolina 27709
Contract Number 68-03-3099
Project Officer
James J. Lichtenberg
Physical and Chemical Methods Branch
Environmental Monitoring Systems Laboratory
Cincinnati, Ohio 45219
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45219
-------
DISCLAIMER
This report was prepared under contract to an agency of the United
States Government. Neither the U.S. Government nor any of its employees,
contractors, subcontractors, or their employees makes any warranty,
expressed or implied, or assumes any legal liability or responsibility
for any third party's use or the results of such use of any information,
apparatus, product, or process disclosed in this report, or represents
that its use by such third party would not infringe on privately owned
rights.
Publications of the data in this document does not signify that the
contents necessarily reflect the joint or separate views and policies of
each sponsoring agency. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
ii
-------
CONTENTS
Page
Abstract iv
List of Abbreviations and Symbols v
List of Tables vi
Acknowledgements vii
1. Introduction 1
2. Conclusions 3
3. Recommendations 4
4. Materials and Methods 5
5. Experimental Procedures 11
6. Results and Discussion 14
7. References 33
Appendix A. Method 1010 35
Appendix B. Method 1020 39
Appendix C. Community Bureau of Reference (BCR) Information. . 48
Appendix D. Quality Assurance Report 66
iii
-------
ABSTRACT
The purpose of this research was to evaluate the ignitability
Methods 1010 (Pensky-Martens) and 1020 (Setaflash) as described by OSW
Manual SW846 (1). This effort was designed to provide information on
accuracy and precision of the two methods. During Phase I of this task,
six standards and simple mixtures were tested. In addition, during
Phase II, twelve actual wastes were tested. The results are contained
in this Final Report.
The results of Phase I determined that both methods are applicable
to characterize the ignitability of liquid wastes. During Phase I, no
significant interferences were identified. During Phase II experiments;
however, water-containing wastes could not be tested using Method 1020.
No direct comparison of Phase I results and data published in literature
reports was made due to the uncertainty associated with the previously
published data. However, during Phase II a search of the Chemical
Abstract Data Base provided more reliable information than previously
published data for |>-xylene flash point.
Based on standards and simple mixtures results, it can be concluded
that no significant difference exists between the accuracy and precision
of the two methods. The results of actual waste experiments however,
provided information that showed significant differences between the
methods. The Setaflash method, when applicable, was determined to be
more accurate and more precise than the Pensky-Martens. The Pensky-
Martens method is not applicable for wastes which have flash points
below 13°C (55°F). The Setaflash method is not applicable to complex
mixtures with substantial amounts of water and high surface tension.
IV
-------
Abbreviations
LIST OF ABBREVIATIONS AND SYMBOLS
ASTM
BCR
EMSL-CI
EPA
OSW
RCRA
RSD
RTI
SD
V
W.A.
American Standards for Testing and Materials
Community Bureau of Reference
Environmental Monitoring and Support Laboratory-Cincinnati
Environmental Protection Agency
Office of Solid Waste
Resource Conservation and Recovery Act
Relative Standard Deviation
Research Triangle Institute
Standard Deviation
Variance, Square of Standard Deviation
Work Assignment
Symbols
°F
°C
mm
mL
X
H20
A
N
Degrees Fahrenheit
Degrees Centigrade
Millimeters
Milliliters
Mean Value
Water
Percentage
Statistical Student t-test
Approximately
Greater than
Less than
Difference
Number of Measurements
v
-------
TABLES
Table Page
1 Reagents Used as Standards During Phase 1 6
2 Wastes Proposed During Phase I to Be Used
During Phase II 7
3 Twelve New Hazardous Wastes Chosen for
Replicate Flash Point Tests 8
4 Closed-Cup Triplicate Analyses of Standard
Compounds, Phase I 15
5 Literature Values for Flash Points 19
6 Waste Samples Acquired During Phase I and
Determined Inadequate for This Study 20
7 Mixture of Waste Samples and Used Motor Oil
or CC1, 22
8 Triplicate Results for Flash Point Determination
of Flammable Waste, Phase II 23
9 Flash Point Mean Values and Standard Deviation,
Phase 1 26
10 Test for Difference in Precision, Phase 1 27
11 Test for Difference in Means, Phase 1 28
12 Flash Point Mean Values and Standard Deviation,
Phase II 30
13 Test for Difference in Precision, Phase II 31
14 Test for Difference in Means, Phase II 32
VI
-------
ACKNOWLEDGEMENTS
The authors wish to thank Mr. Jack Pfaff of the Environmental
Protection Agency (Cincinnati, OH) for his assistance during the course
of this study. Also, we wish to acknowledge the helpful discussions
with Research Triangle Institute technical staff, Dr. L. Sheldon and
Mr. L. Michael and the cooperation of Ms. M. Dempsey.
vii
-------
SECTION 1
INTRODUCTION
Disposal of solid wastes at landfill sites presents several potential
hazards apart from intrinsic toxicity of the disposed materials. The ignita-
bility of waste substances is of concern because of the imminent danger of
uncontrolled burning, and also because of the probable toxicity of the combus-
tion products.
A waste has been defined by RCRA (40 CFR 261.21) as hazardous if any of
the following ignitability characteristics are present:
1. "It is a liquid, other than an aqueous solution containing less than 24%
alcohol by volume, and has a flash point less than 60°C (140°F), as
determined by a Pensky-Martens Closed-Cup Tester, using the test method
specified in ASTM Standard D-93-79 or D-93-80, or a Setaflash Closed-Cup
Tester, using the test method specified in ASTM Standard D-3278-78, or as
determined by an equivalent test method approved by the Administrator
under the procedures set forth in §260.20 and 260.21.
2. It is not a liquid and is capable, under standard temperature and pres-
sure, of causing fire through friction, absorption of moisture, or spon-
taneous chemical changes and, when ignited, burns so vigorously and
persistently that it creates a hazard.
3. It is an ignitable compressed gas as defined in 49 CFR 173.300 and as
determined by the test methods described in that regulation or equivalent
test methods approved by the Administrator under §260.20 and 260.21.
4. It is an oxidizer as defined in 49 CFR 173.151 (1)."
Specific ignitability tests have been developed and refined under several
previous work assignments to this contract, "Development and Evaluation of
Test Procedures for Ignitability Criteria for Hazardous Waste" (2).
As stated above, one characteristic for assessing a waste as hazardous is
if it exhibits the property of ignitability. Liquids exhibit this
characteristic by having a flash point less than 60°C as determined by using
either the Pensky-Martens Closed-Cup tester or a Setaflash Closed-Cup tester.
The directions for using these testers are set down in Method 1010 (Pensky-
Martens) and Method 1020 (Setaflash) in the OSW publication (1) (see Appendices
A and B).
In order to identify wastes which are subject to this hazard, this work
assignment, "Evaluation of Ignitability Methods (Liquids)," was undertaken.
The purpose of this task was to evaluate Ignitability Methods 1010 and 1020 as
described in OSW Manual SW846 (see Appendices A and B) using standards during
Phase I and waste samples during Phase II and to provide precision and accuracy
data for these methods.
-------
Some refinements of the procedures were necessary to perform the evalua-
tion. However, since the goal of this project was to evaluate existing methods
rather than improve them, only small modifications, if any, were made.
Since the equipment is designed for such measurements, all of the experi-
ments measured temperatures in degrees Fahrenheit (°F). When data was reported
in degrees Centigrade (°C) the data were calculated and approximated to the
nearest decimal unit.
-------
SECTION 2
CONCLUSIONS
Comparison and evaluation of Methods 1010 and 1020 during Phase I and
Phase II gave the following conclusions:
a. Based on the testing of standards and simple mixtures used in Phase
I, both Methods 1010 and 1020 were found to be applicable to char-
acterize the ignitability of liquids.
b. During Phase I, no significant interferences were found in the
measurement of ignitability of standards and simple liquid mixtures.
c. A comparison of open-cup (9) and closed-cup experiments shows that
open-cup values are approximately 7°C (20°F) higher than than those
obtained in closed-cup experiments.
d. No statistically significant difference was found between the results
obtained during Phase I using standards and simple mixtures for the
two methods under consideration.
e. From the results of Phase II experiments with actual waste samples,
it was also determined that the Pensky-Martens method is not prac-
tical to use with wastes of low flash points (13°C or 55°F) and the
Setaflash method is not applicable to complex aqueous mixtures which
have high surface tension.
f. Phase II experiments with actual waste samples also established that
flash point measurements made with the Setaflash method were more
precise than the Pensky-Martens method. Also, the Setaflash was
found to be more accurate based on tests with p_-xylene. Finally, it
was concluded that the Setaflash system is easier to use and requires
much less sample than the Pensky-Martens system.
-------
SECTION 3
RECOMMENDATIONS
From the results presented in this report, the authors recommend:
a. The use of Method 1010 (Pensky-Martens) to test the flash point of
ignitable wastes, despite the fact that the method does not easily
allow measurement of flash points below room temperature. The
Pensky-Martens system can identify those samples that flash at or
below room temperature and this is enough to label them hazardous.
Although the Setaflash method is more accurate and more precise, the
Pensky-Martens method has a wider range of applicability to the
needs of EPA.
b. The use of Method 1020 (Setaflash) when a determination of low
temperature flash points is required.
c. Further research, using synthetic, well characterized, complex
mixtures to confirm the tentative conclusions obtained during this
study.
-------
SECTION 4
MATERIALS AND METHODS
STANDARDS AND SIMPLE MIXTURES
During Phase I, six standards and simple mixtures were selected to be
used as a preliminary calibration test because in some cases they could be
compared with published data (1,16) thereby provide information on the behavior
of our experimental set-up. Table I shows the origin of the samples selected
for use in Phase I. The reagent water used for this phase was deionized in
the laboratory. All other reagent grade chemicals were obtained from commer-
cial sources and used as received.
WASTE SAMPLES
During Phase I, waste samples were collected for use during Phase II
(Table 2). The first six samples (2258-14-01-1 to -3 and 2258-14-02-1 to -3)
listed in Table 2 were obtained by mixing waste samples already on hand at
RTI. These samples had been used for previous ignitability tests under this
same contract. The remaining samples in Table 2 were collected specifically
for use in Phase II and had not been previously analyzed. Only 12 of the 14
samples in Table 2 were required for the experiments. The extra two wastes
were acquired in case some of the unknown samples did not have a measurable
flash point. The last two wastes (2258-14-04-1 and -2) listed in Table 2 were
obtained from the RTI Safety Department. One is a mixture of organic solvents
and the other is a mixture of chlorinated organic solvents originating in our
laboratories.
ACQUISITION OF A NEW SET OF WASTE SAMPLES
As explained in Section 6 of this report, the samples shown in Table 2
were not suitable for use in Phase II and it became necessary during to acquire
a new set of twelve ignitable waste samples with known flash points from local
industry. Enviro-Chem Waste Management Services, Inc. (Gary, NC) agreed to
supply RTI with twelve ignitable wastes which had been already tested for
flash point using Pensky-Martens closed-cup method. A complete list of the
new waste samples is shown in Table 3. The samples were obtained and stored
in glass containers.
METHOD 1010
Method 1010 uses the Pensky-Martens closed-cup tester to determine the
above room temperature flash point of fuel oils, lubricating oils, liquids
with suspended solids, liquids that tend to form a surface film under test
conditions, two phase liquids and other flammable liquids.
The sample was heated at a slow, constant rate with continual stirring.
A small flame was directed into the cup at regular intervals with simultaneous
interruption of stirring. The flash point is the lowest temperature at which
application of the test flame ignites the vapor above the sample.
The Pensky-Martens closed flash point tester was acquired from Fisher
Scientific Company and meets specifications of Method ASTM D-93.
-------
TABLE 1. REAGENTS USED AS STANDARDS DURING PHASE I
Sample Origin
£-Xylene Fisher Scientific Co., Lot No. 740150
n-Butanol Fisher Scientific Co., Not. No. 700404
Cyclohexanone Fisher Scientific Co., Lot No. 704956
Ethanol (15%):Water (85%) Fisher Scientific Co., Lot No. 735682,
Laboratory deionized
n-Nonane (5%):n-Decane (95%) Fisher Scientific Co., Lot No. 735248,
Lot No. 740338
n-Nonane (20%):n-Tetradecane Fisher Scientific Co., Lot No. 735248,
(80%) Lot No. 732916
-------
TABLE 2. WASTES PROPOSED DURING PHASE I TO BE
USED DURING PHASE II3
RTI Code Number
Description
2258-14-01-1 Clear, pink/orange liquid containing gasoline,
mineral spirits, paint, paint lacquer, alcohol, ace-
tone, and xylene. Expected flash point: 21°C
(•v70°F). Source: Triangle Resources, Inc.
2258-14-01-2 Brownish liquid with suspended solids containing
isopropanol, water, glycerine, phenol, alcohol,
acetone, xylene, methyl ethyl ketone, and butyl
cellosolve. Expected flash point: ~21-27°C
(•v70-80°F). Source: Triangle Resources, Inc.
2258-14-01-3 Clear, pink/yellow liquid containing hexane, iso-
octane, petroleum ether, ethyl ether, acetone,
nethanol, chloroform, water, dioxane, methylene
chloride, and tetrahydrofuran. Expected flash
point: 21°C (~700F). Source: Triangle Resources,
Inc.
2258-14-01-4 Clear pinkish liquid containing methanol, ethyl
ether, chloroform, water, tetrahydrofuran, dioxane,
and methylene chloride. Expected flash point:
<21°C (<70°F). Source: Triangle Resources, Inc.
2258-14-02-1 Dark liquid, oil based containing cellulose
nitrate lacquer, water, other unknowns. Expected
flash point: ~21°C (^70°F). Source: Pope Air
Force Base.
2258-14-02-2 Dark liquid containing denatured alcohol and other
unknowns. Expected flashpoint: >21°C (>70°F).
Source: Pope Air Force Base.
2258-14-03-1 Yellow liquid containing paint thinner. Expected
flash point: unknown. Source: Triangle
Resources, Inc.
2258-14-03-2 Dark, Cloudy liquid containing 1,1,1-trichloroethane
and other unknowns. Expected flash point: unknown.
Source: Triangle Resources, Inc.
2258-14-05-1 Two layer liquid: Top, cloudy brownish yellow;
Bottom, oily dark clear. Expected flash point:
Unknown. Source: local industry.
2258-14-05-2 Clear black homogeneous liquid. Expected flash
point: unknown. Source: local industry.
2258-14-05-3 Pale greenish cloudy liquid; two layers. Expected
flash point: unknown. Source: local industry.
2258-14-05-4 Pale brown clear liquid. Expected flash point:
Unknown. Source: local industry.
2258-14-04-1 Waste flammable liquid solvents. Expected
flash point: Unknown. Source: RTI.
2258-14-04-2 Waste flammable chlorinated solvents.
Expected flash point: Unknown. Source:
RTI.
*These wastes were later determined to be inadequate for this study.
-------
TABLE 3. TWELVE NEW HAZARDOUS WASTES CHOSEN FOR REPLICATE FLASH POINT TESTS
Flash Point
Waste Description Range, C°a
(°F)
2258-1A-06-1 "A 4-16-84 CHIP BIN WASTE." Two-phase liquid. 55-57
The top phase is dark brown and the bottom phase (131-135)
is a light brown milky liquid. Content unknown.
Source: Enviro-Chem. Amount: ^500 mL.
2258-14-06-2 "B DIRTY TANK 4-16-84." Two phases. Bottom 53-57
phase is a light green milky liquid, and the (128-134)
top phase is a dark green liquid. Contents
unknown. Source: Enviro-Chem. Amount: ViOO mL.
2258-14-06-3 "C EGA 0090 4-16-84." Light green milky liquid. 63-66
Contents unknown. Source: Enviro-Chem. (146-150)
Amount: ~500 mL.
2258-14-06-5 Two-phase liquid. Fuel oil is upper phase and 81-87
ground water is lower phase. Source: (177-188)
Enviro-Chem. Amount: VjOO mL.
2258-14-07-4 Yellow aqueous solution with orange solids 58-63
floating on top. Source: Enviro-Chem (136-144)
(#110-14-2). Amount: ^500 mL.
2258-14-07-9 Orange turbid liquid. Approximately 1% v/v 50-56
solids in the form of orange flakes. Source: (122-133)
Enviro-Chem (#110-54-1). Amount: ^500 mL.
2258-14-07-10 Two-phase mixture. Upper phase is a black 60-66
liquid and lower phase is a gray turbid liquid. (140-150)
Source: Enviro-Chem (#110-27). Amount: ^400 mL.
2258-14-07-11 Clear aqueous mixture containg a surfactant 58-64
and white/gray particulates. Source: (137-148)
Enviro-Chem #110-42). Amount: ^400 mL.
2258-14-07-12 Amber liquid containing gray particulates. 56-58
Source: Enviro-Chem #110-33). Amount: (132-137)
•v2 L.
2258-14-08-1 Two-phase mixture. Approximately 95% of 10-16
2258-14-07-3 (brown aqueous solution with light (50-60)
brown suspended solids; Source: Enviro-Chem
(#110-48-1) as upper phase and ^5% of 2258-
14-03-1 (yellow liquid containing paint thinner;
Source: Triangle Resources, Inc.) as lower
phase. Amount: ^400 mL.
(continued)
-------
TABLE 3 (continued)
Flash Point
Waste Description Range, C°a
2258-14-08-2 Mixture of approximately 15% 2258-14-02-2 40-48
and 85% 2258-14-07-1 which retained the (105-118)
appearance of an orange turbid liquid with
VL% v/v solids in the form of orange flakes.
Amount: ^-500 ml.
2258-14-08-3 Two-phase mixture of ^95% 2258-14-07-5 (brown ^36
turbid aqueous solution with black sediments, (^97)
Source: Enviro-Chem, #110-23-1) as lower phase
and of -^5% 2258-14-01-2 (brownish liquid with
suspended solids; Source: Triangle Resources,
Inc. as upper phase. Amount: ^400 mL.
As determined in our laboratory using Setaflash closed-cup method. Flame
was enlarged to ^-3/4 in height; flame application time was increased from
^2.5 sec to M.O sec; cup was shaken after sample injection.
-------
METHOD 1020
Method 1020 makes use of the Setaflash Closed Tester to determine the
flash point of homogeneous liquids such as fuel oils, paints, enamels,
lacquers, varnishes, and related products and their components that have flash
points between 0° and 110°C and a viscosity lower than 150 stokes at 25°C.
Tests at higher or lower temperatures are also possible.
The procedures may be used to determine whether a material will or will
not flash at a specified temperature or to determine the finite temperature at
which a material will flash.
The Setaflash closed tester, model 01SF, was acquired from ERDCO Engi-
neering. It was tested and the temperature control was calibrated for easy
use.
10
-------
SECTION 5
EXPERIMENTAL PROCEDURES
SETAFLASH EXPERIMENTS
The standards and simple mixtures were tested in triplicate, as required
in the Work Plan, using the Setaflash equipment (see Figure 1), following the
procedure specified by Method 1020. Using a 2 ml syringe, the sample (2 ml)
was introduced through a tight port into the closed test chamber, which had
been previously heated to within 3°C below the expected flash point. No
stirring is provided. After allowing approximately one minute for temperature
equilibration, a small flame (4 mm in diameter) was directed into the cup
through a small window. Careful observation was made as to whether the sample
flashed or not. If there was no flash, the temperature was sequentially
increased by 0.5°C until a flash was observed. A repeat determination was
then performed using a fresh sample. The data were obtained in degrees
Fahrenheit.
PENSKY-MARTENS EXPERIMENTS
Standard samples shown in Table 1 were analyzed in triplicate for flash
point by the Pensky-Martens closed-cup tester (see Figure 2) in accordance
with the procedure given in ASTM D-93. In this procedure, the sample cup was
cleaned, dried, and filled to the mark with sample. Approximately 50 mL of
sample are required. The cup was then placed in the heater, and the cup lid
was lowered and locked into place. The thermometer was inserted. The test
flame was lit and adjusted to 4 mm in diameter. The heater was turned on and
set so that the temperature of the solution rose 5-6°C (9 to 11°F) per minute.
The stirrer was turned on. Every 1°C (^-2°F), the shutter was opened, and the
test flame was lowered into the vapor space of the cup. The sample was not
stirred while the flame was lowered into the cup. The flash point was the
temperature at which the test flame application caused a distinct flash in the
interior of the cup. The data were determined in °F.
11
-------
Pinch Valve
On/Off Valve
Flame Height Adjuster
Thermometer
l.p.g. Recharge
Temperature
Control Knobs
On-Off Switch
4-mm Diameter
Flame Gauge
Testing
hamber
Window
Sample Port
Figure 1. Setaflash tester.
-------
Thermometer
Testing
Chamber
Temperature
Control
On-Off Switch
Figure 2. Pensky-Martens tester.
13
-------
SECTION 6
RESULTS AND DISCUSSION
CORRECTION FOR BAROMETRIC PRESSURE
When the barometric pressure differs from 760 mraHg, a correction must be
made for an accurate flash point temperature. The equation used for this
purpose was obtained from Methods 1010 and 1020 procedures:
Calculated flashpoint = C + 0.03 (760-P)
or
Calculated flash point = F + 0.06 (760-P)
where:
C,F = observed flash point in °C or °F.
P = barometric pressure in mmHg.
A daily reading of the barometric pressure was made at our laboratories to
provide a correction to the flash point.
LITERATURE REVIEW
A comprehensive literature review was beyond the scope of this task, a
brief survey of the published literature did provide some scattered and not
well characterized experimental values for flash points. Therefore, a direct
comparison with our results is not appropriate. The brief literature search
also revealed some methods developed for calculating flash points, some of
these literature values are shown in Table 4.
James C. Johnston in "Estimating Flash Points for Organic Aqueous Solutions"
(5) provides a simple, but not very accurate, method for calculating flash
points. His method is based on the lower explosive limit of the organic
component in air and its vapor pressure at a particular temperature. This
method for calculating does not specify the method used for experimentally
determining the flash point. The agreement seen between the calculated and
our experimental flash points shown in Table 3 is roughly equivalent to that
seen in Johnston's article.
John M. Lenoir, in "Predict Flash Points Accurately" (7), provides a
method for calculating flash points of mixtures in the Pensky-Martens and Tag
closed-cup testers. This calculation involves the summation of the products
of the mole fraction, the molecular weight, and the equilibrium ratio for each
component in the mixture. For the Pensky-Martens closed-cup tester, this
summation is equal to 1.03 at the temperature at which the mixture will flash,
14
-------
TABLE 4. LITERATURE VALUES FOR FLASH POINTS
Compound
n-Butanol
Flash Point
°C (°F) Conditions
35 (95) Closed-cup experiment,
Reference
3
p_-Xylene
Cyclohexanone
Ethanol/Water
25 (77)
27 + 0.6
(81 + 1.5)
26 + 0.6
(78.8 + 1.2)
42 (107)
63 (146)
49 (120)
36 (97)
n-Nonane/
n-Tetradecane
52 (125)
61 (141)
equipment not mentioned,
barometric pressure cor-
relation not mentioned.
No experimental condi-
tions specified.
Closed-cup experiments,
corrected for barometric
pressure difference, for
both Pensky-Martens and
Setaflash.
Closed-cup Setaflash,
barometric correction included.
Closed-cup experiment,
no equipment specified,
no barometric pressure
correction mentioned.
Calculated for open-cup
experimental conditions
and 14.6% ethanol. No
barometric pressure
correction mentioned.
No experimental conditions
specified, 10% ethanol
mixture, no barometric
pressure correction
mentioned.
No experimental condi-
tions specified, 20%
ethanol mixture, no
barometric pressure
correction mentioned.
Experimental value for
Pensky-Martens open-cup,
15% ethanol.
Closed-cup experiment
using Pensky-Martens
equipment. No baro-
metric pressure correc-
tion mentioned, 20%
nonane.
4
1
16
3
(continued)
15
-------
TABLE 4 (continued)
Compound
Flash Point
°C (°F)
Conditions
Reference
n-Nonane/
n-Tetradecane
n-Nonane/
n-Decane
57 (134) Calculated value for
Pensky-Martens closed-
cup equipment. No
barometric pressure
correction mentioned,
20% nonane.
72 (161) Experimental value
Pensky-Martens open-cup
20% nonane.
46 (115) Calculated value for
closed-cup TAG equipment,
5% nonane.
55 (131) Experimental value for
Pensky-Martens open-cup,
20% nonane.
-------
and for the Tag closed-cup tester, the summation is equal to 1.30 at the flash
fio1]^ A comparison of our experimental value for nonane/tetradecane mixture
62-63 C (143-146°F) seems to indicate that our experiments provide reliable
values.
Wilbur A. Affens and George W. McLaren in "Flammability Properties of
Hydrocarbon Solutions in Air" (8) give a method for calculating the flash
point of mixtures based on the Antoine-type equation:
log p =
m
t+a
+ b
where p is vapor pressure (atm.), t is temperature (°C), and a,b, and ra are
constants for a particular hydrocarbon. The equation for this method is:
X
•
"m . (0 . -6 )1
i i m 1
e.e
L i m J
X.10
i
*
(1642-6..)
(1642-6 )
m
= 1
.th
X. is the mole fraction of the i component in the liquid mixture;
.th
m. is the constant for the i component from the Antoine-type equation;
0. is the flash point of the i component plus 230 (°C), and
0 is the flash point of the mixture plus 230 (°C).
Flash points previously determined with the Pensky-Martens open-cup
tester have been used in the 0. term. This calculation has been done for
several mixtures, and the results are presented in an RTI report for a previous
EPA contract (9).
A literature review using a computerized search of Chemical Abstracts
Data Base was performed during Phase II, covering 1972 to present. The
purpose of this review was to determine if there were any reliable data
available for flash points with which to compare our data and determine
accuracy. The relevant articles obtained are listed in the References.
The most useful paper obtained for this study was a report of the Commission
of European Communities on "The Certification of Five Hydrocarbon Materials
for the Determination of Flashpoint (Temperature Range 15 to 65°C)" by D. Lewis,
L. Haemers and W. Karcher (see Appendix C). In that certification, a set of
pure hydrocarbons, including £-xylene and covering the temperature range of
approximately 15 to 65°C was selected with the aim of certifying their flash
17
-------
point temperatures. In this way certification was achieved over the temperature
range (15-65°C) of required temperature since the certified materials have
flash points at intervals of 10 to 15°C. These reference materials were
intended mainly for the calibration of the various closed-cup-flash point
apparatuses which are currently in use. The certification was carried out in
the framework of the "Reference Materials and Methods" program of the Community
Bureau of Reference (BCR) and Metre program of the Joint Research Centre of
the Commission of the European Communities.
Previous experience obtained by the participants of the BCR program,
indicated that only by the use of equilibrium methods for closed-cup flash
point determination could reference materials certified for flash point be
produced. The results obtained with nonequilibrium methods were known to give
significant variations between types of apparatuses and different procedures.
The amount of work involved to certify each material for a wide range of
different apparatus and methods was such that the simpler more universal
equilibrium methods were preferable.
The equilibrium methods for flash point selected for the certification
procedure were as follows:
a. "Rapid Tests for Flash Point", Institute of Petroleum, Method IP303/74.
The section of the method applying to "Flashpoint Determination
Section 2.2" was followed. This method is also available as ASTM
D3278-73.
b. "Flash Testing Using the Cup of Any Standard Closed Cup Apparatus,"
Institute of Petroleum Method IP304/74. The section of the method
applying to "Flashpoint Determination Section 2.2" was followed.
This method is also available as ISO Standard 1523, BS3900 Part A9,
1973, etc.
The flash points- were determined by twelve different laboratories throughout
Europe. A correction for atmospheric pressure variations was applied. A
statistical evaluation of the results was performed. The results obtained for
p_-xylene was 26.0 + 0.6°C with a tolerance interval of +1.5°C. This value was
used in this study to evaluate the accuracy of the methods under consideration.
RESULTS FOR PHASE I AND II
The six standards and simple samples selected for use during Phase I of
this task were analyzed in triplicate and the results are shown in Table 5,
after correction for barometric pressure.
Preliminary Experiments
Preliminary experiments with waste samples acquired during Phase I were
performed (see Table 2). During these tests, it was determined that all of
the .wastes have a flash point below room temperature as measured with the
Setaflash method. These results are shown in Table 6. A modification of
Methods 1010 (Pensky-Martens) and 1020 (Setaflash) was implemented to allow
for flash point measurements below 21°C (70°F). The modification consists of
cooling the testing cup with dry ice, introducing the sample once the testing
cup is cold, then allowing the temperature of the testing cup to raise slowly
as the experiment begins. This modification has proven successful for a very
wide range of temperatures using the Setaflash method. For the Pensky-Martens
18
-------
TABLE 5. CLOSED-CUP TRIPLICATE ANALYSES OF STANDARD
COMPOUNDS, PHASE I
Observed Flash point corrected for
Barometric Pressure, °C (°F)
Sample
n-Butanol
p_-Xylene
Cyclohexanone
Ethanol (15%)/
Water (85%)
n-Nonane (20%) /
n-Tetradecane (80%)
n-Nonane (5%)/
n-Decane (95%)
Pensky-Martens
36
37
34
28
27
27
43
44
42
41
39
41
62
62
61
47
49
48
( 97)
( 98)
( 94)
( 82)
( 80)
( 80)
(110)
(111)
(107)
(106)
(102)
(106)
(143)
(144)
(141)
(117)
(120)
(119)
Setaflash
37
37
36
27
27
27
45
45
45
39
39
38
63
63
63
48
47
47
( 98)
( 98)
( 97)
( 80)
( 80)
( 80)
(113)
(113)
(113)
(103)
(102)
(100)
(146)
(145)
(146)
(118)
(117)
(117)
19
-------
TABLE 6. WASTE SAMPLES ACQUIRED DURING PHASE I FOR WHICH THE
PENSKY-MARTENS METHOD WAS DETERMINED IMPRACTICAL
RTI Code Number Description Approximate Measured
Flash Point, °C (°I)
2258-14-01-1 Clear, pink/orange liquid containing gasoline, £9 (£48)
mineral spirits, paint, paint lacquer, alcohol,
acetone, and xylene. Expected flash point:
~70°F. Source: Triangle Resources, Inc.
2258-14-01-2 Brownish liquid with suspended solids containing £8 (£46)
xylene, methyl ethyl ketone, and butyl
e. Expected flash point: ~70-80°F.
Triangle Resources, Inc.
2258-14-01-3 Clear, pink/yellow liquid containing hexane, £7 (£45)
isooctane, petroleum ether, ethyl ether, acetone,
methanol, chloroform, water, dioxane, methylene
chloride, and tetrahydrofuran. Expected flash
point: ~70°F. Source: Triangle Resources, Inc.
2258-14-01-4 Clear pinkish liquid containing methanol, ethyl £8 (<46)
ether, chloroform, water tetrahydrofuran,
dioxane, and methylene chloride. Expected flash
point: <70°F. Source: Triangle Resources, Inc.
2258-14-02-1 Dark liquid, oil based containing cellulose £9 (£48)
nitrate lacquer, water, other unknowns. Expected
flash point: ->-700F. Source: Pope Air Force
Base.
2258-14-02-2
2258-14-03-1
2258-14-03-2
2258-14-05-1
2258-14-05-2
2258-14-05-3
2258-14-05-4
2258-14-04-1
2258-14-04-2
Dark liquid containing denatured alcohol and other
unknowns. Expected flash point: >70°F. Source:
Pope Air Force Base.
Yellow liquid containing paint thinner. Expected
flash point: unknown. Source: Triangle
Resources, Inc.
Dark, cloudy liquid containing 1,1 ,1-trichloroethane
and other unknowns. Expected flash point: unknown.
Source: Triangle Resources, Inc.
Two layer liquid: Top, cloudy brownish yellow;
Bottom, oily dark clear. Expected flash point:
Unknown. Source: local industry.
Clear black homogeneous liquid. Expected flash
point: unknown. Source: local industry.
Pale greenish cloudy liquid; two layers. Expected
flash point: unknown. Source: local industry.
Pale brown clear liquid. Expected flash point:
Unknown. Source: local industry.
Waste flammable liquid solvents. Expected
flash point: Unknown. Source: RTI.
Waste flammable chlorinated solvents.
Expected flash point: Unknown. Source:
RTI.
16 (60)
0-4 (32-39)
7-9 (44-48)
-11- -5 (13-23)
£-11 (£13)
£-10 (£14)
£-13 (£9)
£-7 (£20)
£-2 (£29)
Measured using Method 1020, Setaflash, closed-cup.
20
-------
method, however, it was determined that this procedure is practical only down
to about 10 to 13°C (50° to 55°F) due to the fact that Pensky-Martens equipment
has a very massive cup which cools and reheats very slowly making the operation
impractical.
The decision was made that waste samples with measured flash points below
55°F would be spiked with a waste oil and/or CC14 in order to raise the tempera-
ture of measurement above the minimum permitted by the Pensky-Martens method.
This decision was based upon the fact that the purpose of this task was to
determine accuracy (measured for standard samples) and precision of the methods
under consideration and not to test the wastes. The fact that some of the
collected wastes had low flash points was purely circumstantial and spiking
them with a waste oil would not alter the goal of this task.
Waste Mixtures
A 5-gallon waste motor oil sample was collected from a local service
station. Varying amounts of this waste oil were added to those waste samples
which had low flash points. Mixtures of the wastes and carbon tetrachloride
were also prepared.
The mixtures were then tested using the Setaflash method which required
only a 2 mL sample per determination. The results of these tests are shown in
Table 7. In order to raise the flash point above room temperature, a consider-
able amount of spiking material was required, in most cases more than 70% of
either used motor oil or CC14. These results demonstrated that the spiking of
the waste samples was'impractical and this strategy was, therefore, abandoned.
The new set of waste samples acquired are shown in Table 3 and are discussed
below.
New Set of Waste Samples
The waste samples listed in Table 3 were used for testing of both
Method 1010 (Pensky-Martens) and Method 1020 (Setaflash). The results
of these tests are shown in Table 8.
All the data in Table 8 has been corrected for atmospheric variations.
The data obtained by the Setaflash method for waste 2258-14-08-3 were corrected
by adding 1°F in order to compensate for the difference obtained for the p_-
xylene control which was 1 °F below that value allowed by the Method 1020
(Setaflash) description.
During the performance of Method 1010 (Pensky-Martens), two of the control
experiments using g-xylene were not performed: waste 2258-14-08-3 was tested
immediately after the control for waste 2258-14-06-3, and waste 2258-14-
07-11 was tested after the control for 2258-14-07-4.
Several wastes tested by the Setaflash method showed no measurable flash
point below 110°C (230°F). This phenomenon was observed primarily on those
wastes that contained substantial amounts of an aqueous phase and had a large
surface tension as determined by empirical observation; the high surface
tension prevented the waste from dispersing evenly throughout the cup and
since only 2 mL of sample was required, part of the cup remained empty. If
agitation was provided, then a flash occurred (note that the Pensky-Martens
method includes stirring). If a larger flame than the one specified by the
method was applied, a flash was observed, however, this flash was usually at a
21
-------
TABLE 7- MIXTURE OF WASTE SAMPLES AND USED MOTOR OIL OR CC14
Mixture
2258-14-01-1 and Motor Oil
2258-14-01-2 and Motor Oil
2258-14-01-3 and Motor Oil
2258-14-01-4 and Motor Oil
2258-14-02-1 and Motor Oil
2259-14-03-1 and CC14
2258-14-03-1 and Motor Oil
2258-14-03-2 and CC14
2258-14-03-2 and Motor Oil
2258-14-04-1 and Motor Oil
2258-14-04-2 and Motor Oil
Motor Oil
Amount (%) Approximate
% Waste % Spiking Agent Flash Point,3
°C (°F)
80
50
20
76
50
20
25
68
76
90
70
50
70
50
20
90
70
50
70
50
20
92
88
--
20
50
80
24
50
80
75
32
24
10
30
50
30
50
80
10
30
50
30
50
80
8
12
100
-8 - [-3](18-27)
3-4 (34-39)
9-13 (49-55)
7-11 (45-51)
11-13 (51-56)
19-21 (67-69)
-6 - [-2] (21-28)
0-4 (32-40)
6-11 (43-51)
<8 (<46)
N/A
<9 (<49)
6-8 (43-46)
13-14 (55-58)
13-18 (56-64)
<8 (<46)
<13 (<55)
(<42)
11-13 (52-55)
11-14 (52-58)
12-18 (54-56)
< -7 (<20)
-8 - [-2] (18-29)
>110 (>230)
Determined by the Setaflash method.
22
-------
TABLE 8. TRIPLICATE RESULTS FOR FLASH POINT DETERMINATION
OF FLAMMABLE WASTE, PHASE II
3
Waste Number
2258-14-06-1
£-Xylene
2258-14-06-2
jD-Xylene
2258-14-06-3
£-Xylene
2258-14-08-3
£-Xylene
2258-14-06-5
£-Xylene
2258-14-08-2
£-Xylene
2258-14-08-1
£-Xylene
Pensky-Martens
Method 1010
°C (°F)
61.1 (142)
62.2 (144)
62.8 (145)
27.2 ( 81)
27.2 ( 82)
65.0 (149)
60.0 (140)
62.8 (145)
26.7 ( 80)
26.7 ( 80)
72.8 (163)
71.1 (160)
73.9 (165)
27.8 ( 82)
26.7 ( 80)
34.4 ( 94)
33.3 ( 92)
35.0 ( 95)
N/Db
N/DD
92.2 (198)
94.4 (202)
90.0 (194)
27-8 ( 82)
28.3 ( 83)
46.7 (116)
48.9 (120)
50.0 (122)
27.2 ( 81)
28.3 ( 83)
15.0 ( 59)
12.9 ( 57)
12.2 ( 54)
27.8 ( 82)
26.7 ( 80)
~ Setaflash
Method 1020
°C (°F)
60.6 (141)d
58.3 (137)
60.0 (140)
26.7 ( 80)
26.7 ( 80)
54.4 (130)d
53.9 (129)
54.4 (130)
26.7 ( 80)
26.1 ( 79)
NFe
NF
NF
26.7 ( 80)
26.1 ( 79)
36.1 ( 97)f
36.1 ( 97)
37.2 ( 99)
25.6 ( 78)
26.1 ( 79)
85.0 (185)
85.0 (185)
85.6 (186)
26.7 ( 80)
26.1 ( 79)
48.9 (120)
50.6 (123)
51.7 (125)
26.1 ( 79)
26.7 ( 80)
30.0 ( 86)
NF8
NF
26.7 ( 80)
26.1 ( 79)
(continued)
23
-------
TABLE 8 (concluded)
Waste Number
2258-14-07-9
p_-Xylene
2258-14-07-10
£-Xylene
2258-14-07-12
p_-Xylene
2258-14-07-4
p-Xylene
2258-14-07-11
p_-Xylene
Pensky-Martens
Method 1010
°C (°F)
57.8 (136)
60.0 (140)
57.8 (136)
26.7 ( 80)
27.2 ( 81)
62.2 (144)
63.3 (146)
63.3 (146)
28.3 ( 83)
27.2 ( 81)
55.6 (132)
57.8 (136)
54.4 (130)
26.7 ( 80)
27.8 ( 82)
51.7 (125)
48.9 (120)
48.9 (120)
27.8 ( 82)
26.7 ( 80)
57.8 (136)
56.7 (134)
56.7 (134)
N/DC
N/D
Setaflash
Method 1020
°C (°F)
NF6
NF
NT
26.1 ( 79)
26.7 ( 80)
NFG
NI-
NE
26.7 ( 80)
26.1 ( 79)
NFe
NT
NF
26.1 ( 79)
26.1 ( 79)
NF6
NF
NF
26.1 ( 79)
25.6 ( 78)
NF6
NF
NF
26.7 ( 80)
26.1 ( 79)
Data corrected for barometric pressure changes.
Data for 2258-14-08-3 determined immediately after data obtained for
2258-14-06-3.
CData for 2258-14-07-11 determined immediately after data obtained for
2258-14-07-4.
iTlash points found by (a) shaking cup after sample injection, (b)
applying large flame (^3/4 in height); (c) increasing flame appli-
cation time to no greater than 10 seconds.
eNo flash observed, at any temperature below 110°C (230°F), following
the procedure described in Method 1020. If sample is shaken and/or
a larger flame is applied for approximately 10 seconds, a flash will
occur at temperatures below 110°C (230°F). Flash point observed may
be a function of flame size.
Data for 2258-14-08-3 was corrected for the £-xylene data difference
(1°F).
^Volatile components of this sample evaporated before testing.
Residual does not flash.
24
-------
higher temperature than expected. This observation may constitute a severe
interference to the Setaflash method because it may determine that some
ignitable wastes do not flash.
The Setaflash method did not determine a flash point for two of the
three replicates performed on waste 2258-14-08-1. This was attributed
to the evaporation of some volatile component before the measurement was
performed and the residue did not flash. This phenomenon may be enhanced
by the small amount of sample required (2 mL).
-STATISTICAL ANALYSIS
Phase I Results
A statistical analysis (10) on results shown in Table 5 was performed
to determine several factors. Included were the mean flash point values
(see Table 9), the precision of the values (see Table 10) and the equi-
valency of the values measured using the two methods (see Table 11).
The standard deviations for g-xylene and cyclohexanone as measured by
the Setaflash method were not calculated because the values measured
were equal.
The precision of the methods were then compared using the "F" test.
This involves calculating the ratio of the standard deviations squared,
i.e., variances (V), and comparing the value to a tabular value. This
test could not be applied to two of the measurements because the SD was
zero for the Setaflash experiment. The precision of the measurements
(excluding the g-xylene and the cyclohexanone results) made with the two
methods were not statistically different at the 95% confidence level
based on these calculations.
The mean values were then compared using Student's t-test. This
involves calculating a pooled standard deviation using the formula:
Sp =
(Nb-l)s*
- 2
1/2
Where N and N, are the number of analysis results in the data sets for
3 n
Pensky-nartens and Setaflash measurements, respectively, and S and S,
are the standard deviations of these data sets. Next a maximum Difference
in X and X, (mean values for the sets of data) which could result at
the 9%% confidence level purely by chance is calculated, i.e.:
C ts (1/N + 1/N. )1/2
- p a b
Where t is the a tabular value based on N , N, , and confidence level.
If the measured difference is less than this calculated value, then the
means do not differ at the 95% confidence level. Upon application of
this test, cyclohexanone and n-nonane/n-tetradecane yielded statistically
different values using the two methods.
25
-------
TABLE, 9. FLASH POINT MEAN VALUES, 95% CONFIDENCE INTERVALS, AND STANDARD DEVIATION, PHASE I4
Pensky-Martens , °C (°F)
Sample
n-Butanol
|>-Xylene
Cyclohexanone
Ethanol (15%) /Water (85%)
n-Nonane (20%)/n-Tetradecane (80%)
n-Nonane (5%)/n-Decane (95%)
Mean
35.7
(96.3)
27.1
(80.7)
42.9
(109.3)
40.4
(104.7)
61.5
(142.7)
48.2
(118.7)
Conf. Int.
2.9
(5.2)
1.6
(2.9)
2.9
(5.2)
3.2
(5.7)
2.1
(3.8)
2.1
(3.8)
S.D.
1.2
(2.1)
0.6
(1.2)
1.2
(2.1)
1.3
(2.3)
0.8
(1.5)
0.8
(1.5)
Setaflash, °C (°F)
Mean
36.5
(97.7)
26.7
(80.0)
45.0
(113.0)
38.7
(101.7)
63.2
(145.7)
47.4
(117.3)
Conf. Int.
0.8
(1.4)
0.0
(0.0)
0.0
(0.0)
2.1
(3.8)
0.8
(1.4)
0.8
(1.4)
S.D.
0.3
(0.6)
0.0
(0.0)
0.0
(0.0)
0.8
(1.5)
0.3
(0.6)
0.3
(0.6)
dFor statistical calculations °F units were used.
-------
TABLE 10. TEST FOR DIFFERENCE IN PRECISION, PHASE I
Compound
n-Butanol
p_-Xylene
Cyclohexanone
Ethanol/Water
n-Nonane/
n-Tetradecane
n-Nonane/
n-Decane
V2
12.9
NCb
NCb
2.28
6.96
6.96
Difference in precision
at 95% confidence level
No
NCb
NCb
No
No
No
Ratio of variances, where Vx ^ V2.
No precision was calculated due to 0.0 standard deviation for
Setaflash data.
27
-------
TABLE 11. TEST FOR DIFFERENCE IN MEANS, PHASE
00
Compound
n-Butanol
£-Xylene
Cyclohexanone
Ethanol/Water
n-Nonane/
n-Tetradecane
n-Nonane/
n-Decane
AXb
1.40
0.67
3.7
3.0
3.0
1.40
J1 +^
tSp \Na Nb
3.47
1.84
3.34
4.45
2.63
2.63
pooled
1.53
0.81
1.47
1.96
1.16
1.16
Statistical Difference
Means at 95% Confidence
No
No
Yes
No
Yes
No
in
Level
All statistical calculations were performed using the data expressed in °F.
Difference in means.
See text for details.
Pooled standard deviation.
-------
Phase II Results
For the results of the waste samples tested during Phase II, the
same statistical analysis was used as in Phase I. The mean values and
standard deviation for both methods are presented in Table 12 where it
can be seen that in all but two cases, Pensky-Martens values are higher
than those of Setaflash. All the statistical calculations were performed
using the data expressed in °F because these are the units in which the
experimental apparatus are calibrated.
The precision of the data was determined as before using the F
test. The results (Table 13) show that at the 95% confidence level, two
of the measurements (2258-14-06-2 and 2258-14-06-5) are significantly
different and, in both cases, the precision of the Setaflash experiment
is better than that of the Pensky-Martens experiment.
The mean values of the data obtained using both methods were compared
using the Student's "t" test. The results (see Table 14) show that in
all cases but one (2258-14-08-2) the mean values are significantly
different at the 95% confidence level. In addition, the data obtained
for p_- xylene using both methods was compared with the data obtained from
a reliable literature source (see Appendix C). The published flash
point, obtained in a round-robin study with 107 mean values, is 26 +
0.6°C (78.8 + 1.2°F) and a tolerance interval of +1.5°C (2.7°F). The
result of this comparison indicates that the Pensky-Martens method is
significantly different from the literature value while the Setaflash is
not significantly different from the literature value at the 95% con-
fidence level.
29
-------
TABLE 12. FLASH POINT MEAN VALUES 95% CONFIDENCE INTERVALS AND STANDARD DEVIATION, PHASE II'
Pensky-Martens, °C (°F)
Sample
2258-14-06-1
2258-14-06-2
2258-14-08-3
2258-14-06-5
2258-14-08-2
£-Xylene
Mean
62.1
(143.7)
62.6
(144.7)
34.3
(93.7)
92.2
(198.0)
48.5
(119.3)
27.4
(81.3)
Conf. Int.
2.1
(3.7)
6.2
(11.2)
0.1
(3.7)
5.5
(9.9)
4.3
(9.7)
0.3
(0.5)
S.D.
0.8
(1.5)
2.2
(4.5)
0.8
(1.5)
2.0
(4.0)
1.5
(3.1)
0.6
(1.1)
Setaflash, °C (°F)
Mean
59.6
(139.3)
54.3
(129.7)
36.5
(97.7)
85.2
(185.3)
50.4
(122.7)
26.3
(79.3)
Conf. Int.
2.9
(5.2)
0.8
(1.5)
1.8
(3.0)
0.8
(1.5)
3.4
(6.2)
0.1
(0.2)
S.D.
1.2
(2.1)
0.3
(0.6)
0.6
(1.2)
0.3
(0.6)
1.4
(2.5)
0.3
(0.6)
All statistical calculations were performed using the data expressed in °F.
-------
TABLE 13. TEST FOR DIFFERENCE IN PRECISION, PHASE II
Sample
2258-14-06-1
2258-14-06-2
2258-14-08-3
2258-14-06-5
2258-14-08-2
|>-Xylene
(Vj \ Difference in precision
TT- I at 95% confidence level
1.96
56
1.56
44
1.54
3.36
No
Yes
No
Yes
No
No
Ratio of variances, where Vj :> V2
31
-------
TABLE 14. TEST FOR DIFFERENCE IN MEANS, PHASE II
(JO
Sample
2258-14-06-1
2258-14-06-2
2258-14-08-3
2258-14-06-5
2258-14-08-2
£-Xylene
AXa ts
P
4.4
15
4.0
12.7
3.4
2.0
WN + N.
I a b
4.13
7.29
3.09
6.49
6.40
0.53
pooled
1.82
3.21
1.36
2.86
2.82
0.86
Statistical Difference
Means at 95% Confidence
Yes
Yes
Yes
Yes
No
Yes
in
Level
All the statistical calculations were done with the data expressed in °F.
-------
REFERENCES
1. Solid Waste Manual, Test Methods for Evaluating Solid Waste, Phy-
sical/ Chemical Methods, SW846, 2nd Edition (1982).
2. "Method Evaluation, Improvement, and Development for Hazardous
Waste Analysis and Characterization," EPA Contract No. 68-03-3099,
W.A. #4, 7, and 8 (1983).
3. "Solvents Guide," C. Martens (ed.), Intersicence, NY (1963).
4. Fisher Scientific Data Sheet.
5. Johnston, James C., "Estimating Flash Points for Organic Aqueous
Solutions," Chemical Engineering, November 25 (1972), p. 122.
6. NFPA Publication 325 M, "Fire Hazard Properties of Flammable Liquids,
Gases and Volatile Solids."
7. Lenoir, J.M., "Predict Flash Points Accurately," Hydrocarbon Pro-
cessing, January 1975, pp. 95-99.
8. Affens, Wilbur A. and McLaren, George W. , "Flammability Properties
of Hydrocarbon Solutions in Air," Journal of Chemical and Engineering
Data, Vol. 17, No. 4 (1972), pp. 482-488.
9. EPA Contract No. 68-03-3099, W.A. #8, "Evaluation of Test Methods
for Wastes with Flash Points Below 60°C."
10. Blaedel, W.J., Meloche, V.W., "Elementary Quantitative Analysis,
Theory and Practice," Row, Peterson Co., Evanson, IL (1957).
11. Wu, D.T. and Finkelman, R. , "A Mathematical Model for the Prediction
of Closed-Cup Flash Points." Org. Coat. Plast. Chem. 38, 61-67
(1978).
12. Smith, R. , Wood, J. , Fredrick, R.H., Hfling, J. , Briscoe, R.C.,
Saake, E.J., Lynch, D.T., Jr., and Carlson, D. , "Effect Obtained on
Flash Points with the Addition of Polar Solvents to Xylene." J.
Paint Technol. 44, 38-42 (1972).
13. Walsham, J.G., "Prediction of Flash Points for Solvent Mixtures."
.Advan. Chem. Ser. 124, 56-69 (1973).
14. Ellis, W.H., "Solvent Flash Points - Expected and Unexpected." J.
Coat. Technol. 48, 44-57 (1976).
33
-------
15. Barnard, A.J., Jr. and Lance, R.C., Jr., "Flash Point Determination
Calibration Standards." Patent: United States US 4033897.
16. Lewis, D., Haemers, L., and Karcher, W. , "The Certification of Five
Hydrocarbon Materials for the Determination of Flash Point (Tempera-
ture Range 15 to 65°C)." Comm. Eur. Communities EUR 6102 (1979).
34
-------
APPENDIX A
METHOD 1010
35
-------
METHOD 10101
PENSKY-MARTENS CLOSED-CUP METHOD
1.0 Scope and Application
1.1 Method 1010 uses the Pensky-Martens closed-cup tester to determine
the flash point of fuel oils, lube oils, suspensions of solids, liquids that
tend to form a surface film under test conditions, and other liquids.
2.0 Summary of Method
2.1 The sample is heated at a slow, constant rate with continual
stirring. A small flame is directed into the cup at regular intervals with
simultaneous interruption of stirring. The flash point is the lowest temper-
ature at which application of the test flame ignites the vapor above the
sample.
3.0 Interferences
3.1 Ambient pressure, sample homogeneity, drafts, and operator bias can
affect flash point values.
4.0 Apparatus
4.1 Pensky-Martens Closed Flash Tester, as described in Annex Al of ASTM
Method D93-77. (Automatic flash point testers are available and may be
advantageous since they save testing time, permit the use of smaller samples,
and exhibit other advantages. If automatic testers are used, the user must be
sure to follow all the manufacturer's instructions for calibrating, adjusting,
and operating the instrument. In any cases of dispute, the flash point as
determined manually shall be considered the referee test.)
4.2 Thermometers: Two standard thermometers shall be used with the
ASTM Pensky-Martens tester.
4.2.1 For tests in which the indicated reading falls within -7* to
+110* C (20* to 230' F), inclusive: either (1) an ASTM Pensky-Martens
Low Range or Tag Closed Tester Thermometer having a range from -7* to
+110* C (20* to 230* F) and conforming to the requirements for Thermometers
9C (9F) and as prescribed in ASTM Specification El, or (2) an IP Thermo-
meter 15C (15F) conforming to specifications given in Annex A3 of ASTM
.093-77.
*This method is based on ASTM Method D93-77. Refer to D93-77 or D93-80
for more information.
36
-------
2 / CHARACTERISTICS - Ignitability
4.2.2 For tests in which the indicated reading falls within 110"
to 370* C (230' to 700* F): either (1) an ASTM Pensky-Martens High
Range Thermometer having a range from 90" to 370* C (200* to 700* F) and
conforming to the requirements for Thermometers IOC (10F) as prescribed
in Specification El, or (2) IP Thermometer 16C (16F) conforming to
specifications given in Annex A3 of ASTM D93-77.
5.0 Reagents
5.1 Calcium chloride.
5.2 p-Xylene reference standard.
6.0 Sample Collection, Preservation, and Handling
6.1 All samples must be collected using a sampling plan that addresses
the considerations discussed in Section One of this manual.
6.2 Samples shall not be stored in plastic bottles since volatile
materials may diffuse through the walls of the bottle.
7.0 Procedure
7.1 Preparation of samples: Samples that do not contain volatile
contaminants shall be prepared in the following manner. NOTE: If the sample
is suspected of containing volatile contaminants, the treatment described in
7.1.1 and 7.1.2 should be omitted.
7.1.1 Samples of very viscous materials may be warmed until they
are reasonably fluid before they are tested. However, no sample should
be heated more than is absolutely necessary, and no sample should ever
be heated to a temperature that exceeds 17* C (30* F) below the sample's
expected flash point.
7.1.2 Samples containing dissolved or free water may be dehydrated
with calcium chloride or by filtering through a qualitative filter paper
or a loose plug or dry absorbent cotton. Warming the sample is permitted,
but it shall not be heated for prolonged periods or above a temperature
of 17* C (30* F) below the sample's expected flash point.
7.2 Routine procedure
7.2.1 Thoroughly clean and dry all parts of the cup and its
accessories before starting the test. Be sure to remove any solvent
that was used to clean the apparatus. Fill the cup with the sample to
37
-------
1010 / 3
be tested to the level Indicated by the filling mark. Place_the lid on
the cup and set the latter in the stove. Be sure to properly engage the
locating or locking device. Insert the thermometer. Light the test
flame and adjust it to a diameter of 5/32 in. (4 mm). Supply the heat
at such a rate that the temperature as indicated by the thermometer
increases 5" to 6" C (9* to 11* F)/min. Turn the stirrer 90 to 120 rpm,
stirring in a downward direction.
7.2.2 If the sample is expected to have a flash point of 110" C
(230* F) or below, apply the test flame when the temperature of the
sample is from 17* C (30* F) to 28* C (50* F) below the expected
flash point and thereafter at a temperature reading that is a multiple
of 1* C (2* F). Apply the test flame by operating the mechanism on the
cover which controls the shutter and test flame burner so that the flame
is lowered into the vapor space of the cup in 0.5 sec, left in its
lowered position for 1 sec, and quickly raised to its high position.
Do not stir the sample while applying the test flame.
7.2.3 If the sample is expected to have a flash point above 110* C
(230* F), apply the test flame in the manner just described at each
temperature that is a multiple of 2* C (5* F), beginning at a temperature
of 17* C (30* F) to 28* C (50* F) below the expected flash point.
NOTE: When testing materials to determine if volatile contaminants are
present, it is not necessary to adhere to the temperature limits for
initial flame application as stated in 7.2.2 and 7.2.3.
7.2.4 Record as the flash point the temperature read on the
thermometer at the time the test flame application causes a distinct
flash in the interior of the cup. Do not confuse the true flash point
with the bluish halo that sometimes surrounds the test flame at applica-
tions preceding the one that causes the actual flash. The actual flash
will have occurred when a large flame propagates itself over the surface
of the sample.
7.3 Determination of flash point of suspensions of solids and highly
viscous materials
7.3.1 Bring the material to be tested and the tester to a tempera-
ture of 15* _+ 5* C (60* _+ 10* F) or 11* C (20* F) lower than the estimated
flash point, whichever is lower. Turn the stirrer 250 +_ 10 rpm, stirring
in a downward direction. Raise the temperature throughout the duration
of the test at a rate of not less than 1* nor more than 1.5* F (2 to 3" F)/
min. With the exception of these requirements for rates of stirring
and heating, proceed as prescribed in Section 7.2.
7.4 Calculation and report
7.4.1 Observe and record the ambient barometric pressure at the
time of the test. When the pressure differs from 760 mm Hg (101.3 kPa),
correct the flash point as follows:
38
-------
APPENDIX B
METHOD 1020
39
-------
METHOD 10201
SETAFLASH CLOSED-CUP METHOD
1.0 Scope and Application
1.1 Method 1020 make use of the Setaflash^Closed Tester to determine
the flash point of paints, enamels, lacquers, varnishes, and related products
and their components that have flash points between 0* and 110' C (32* and
230* F) and a viscosity lower than 150 stokes at 25* C (77* F). Tests
at higher or lower temperatures are possible.
1.2 The procedures may be used to determine whether a material will or
will not flash at a specified temperature or to determine the finite tempera-
ture at which a material will flash.
2.0 Summary of Method
2.1 By means of a syringe, 2 ml of sample is introduced through a
leakproof entry port into the tightly closed Setaflash Tester or directly
into the cup that has been brought to within 3* C (5* F) below the expected
flash point.
2.2 As a flash/no flash test, the expected flash point temperature may
be a specification (e.g., 60" C). For specification testing, the temperature
of the apparatus is raised to the precise temperature of the expected flash
point by slight adjustment of the temperature dial. After 1 min, a test
flame is applied inside the cup and note is taken as to whether the test
sample flashes or not. If a repeat test is necessary, a fresh sample should
be used.
2.3 For a finite flash measurement, the temperature is sequentially
increased through the anticipated range, the test flame being applied at
5" C (9* F) intervals until a flash is observed. A repeat determination is
then made using a fresh sample, starting the test at the temperature of the
last interval before the flash point of the material and making tests at
increasing 0.5* C (1* F) intervals.
3.0 Interferences
3.1 Ambient pressure, sample homogeneity, drafts, and operator bias
can affect flash point values.
iThis method is based on ASTM Method D327-78.
40
-------
2 / CHARACTERISTICS - Ignitabillty
4.0 Apparatus and Materials
4.1 Setaflash Tester as described in Appendix XI of ASTM Method 3278-78.
4.2 Thermometers conforming to specifications given in ASTM Method
3278-78. Test to determine that the scale error does not exceed 0.25* C
(0.5* F). A magnifying lens significantly assists in making temperature
observations.
4.3 Glass syringe: 2 +_ 0.1-ml capacity at 25" C (77* F), to provide a
means of taking a uniform sample. Check the capacity by discharging water
into a weighing bottle and weighing. Adjust plunger if necessary. A dispos-
able syringe of equal precision nay be used.
4.4 Cooling block: Aluminum (described in Appendix X2 of ASTM D3278-78)
which fits snugly within the test cup for rapid cooling of the sample cup.
4.5 Barometer.
5.0 Reagents
5.1 p-Xylene: Reference standard for checking the Setaflash Tester.
5.2 Cooling mixture of ice water or dry ice (solid C02) and acetone.
5.3 Liquefied petroleum gas.
5.4 Heat transfer paste.
6.0 Sample Collection, Preservation, and Handling
6.1 All samples must be collected employing a sampling plan that
addresses the considerations discussed in Section One of this manual.
6.2 The sample size for each test is 2 ml. Obtain at least a 25-ml
sample from the bulk source and store in a nearly full, tightly closed clean
glass container or in another container suitable for the type of liquid being
sampled.
6.3 Erroneously high flash points may be obtained if precautions
are not taken to avoid loss of volatile materials. Do not open sample
containers unnecessarily and do not transfer the sample to the cup unless its
temperature is at least 10* C (20* F) below the expected flash point.
Discard samples in leaky containers.
6.4 Do not use plastic bottles since certain volatile compounds can
diffuse through the walls of the bottle.
41
-------
1020 / 3
7.0 Procedure
7.1 Prior to Initial use and after removal of the thermometer, insert
the thermometer into its pocket with a good heat transfer paste.
7.2 To help in making the necessary settings during a test, determine
the relationship between the temperature control dial and thermometer readings
at intervals not over 5" C (10* F) throughout the scale range of the heater
before the initial use.
7.3 Place the tester in a subdued light and in a position where it is
not exposed to disturbing drafts. Provide a black-coated shield, if necessary.
7.4 Read the manufacturer's operating and maintenance instructions on
the care and servicing of the tester. Observe the specific suggestions
regarding the operation of its various controls.
7.5 Check the accuracy of the tester by determining the flash point of
the p-xylene reference standard in duplicate (Appendix X3). The average of
the results should be 27.2* +_ 0.8* C (81* _+ 1.5* F). If not, remove the ther-
mometer and observe whether sufficient heat transfer paste surrounds the
thermometer to provide good heat transfer from the cup to the thermometer.
7.6 Ambient to 110* C (230* F).
7.6.1 Inspect the inside of the test cup, lid, and shutter mechanism
for cleanliness and freedom from contamination. Use an absorbent tissue
to wipe clean, if necessary. Lock the cover lid tightly in place.
7.6.2 Switch the tester on, if not already at stand-by. To
rapidly approach the specification flash temperature of the charged
sample, turn the heater dial fully clockwise causing the heater signal
(red) light to glow. When the thermometer indicates a temperature of
about 3* C (5* F) below the specification or target flash point tempera-
ture, reduce the heat input to the test cup by slowly turning the heater
control dial counter-clockwise until the signal light goes out.
NOTE: When the correct temperature is dialed on the temperature controller,
the elapsed time to reach it may be greater than when turned full on,
but less attention will be required in the intervening period.
NOTE: The test cup temperature is stable when the signal light slowly
cycles on and off.
7.6.3 Determine the barometric pressure to determine the corrected
specification temperature at that barometric pressure.
7.6.4 After the test cup temperature has stabilized at the specifi-
cation or target flash point, charge the syringe with the sample to be
tested and transfer the syringe to the filling orifice, taking care not
to lose any sample. Discharge the sample into the test cup by depressing
42
-------
4 / CHARACTERISTICS - Ignitabillty
the syringe plunger to Its lowest position, then remove the syringe.
the sample has a viscosity gVeater than 45 SUS at 37.8* C (100* F) 01
If
equivalent of 9.5 cSt at 25* C (77* F), discharge the contents of the
syringe directly into the cup. Immediately close tightly the lid and
shutter assembly.
7.6.5 Set the 1-min timing device by rotating its knob clockwise
to the required setting. In the meantime, open the gas control valve
and light the pilot and the test flames. Adjust the test flame size
with the pinch valve so as to match the size of the 5/32-in. (4-mm)
diameter flame gauge.
7.6.6 After 1 min has elapsed, observe the temperature. If at the
specification temperature (accounting for the differences of the barometer
reading from 760 mm), apply the test flame by slowly and uniformly
opening the slide fully and closing completely over a period of approxi-
mately 2-1/2 sec. Watch for a flash. (NOTE: The sample is considered
to have flashed only if a comparatively large blue flame appears and
propagates itself over the surface of the liquid. Occasionally, particu-
larly near the actual flash point temperature, application of the test
flame may give rise to a halo; this should be ignored.)
7.6.7 Turn off the test and the pilot flame. Clean the apparatus
in preparation for the next test.
7.7 0* C (32* F) to ambient
7.7.1 If the specification or target flash point is at or below ambient
temperature, cool the sample to 5* to 10* C (10* to 20* F) below that point by
some convenient means.
7.7.2 Cool the tester to the approximate temperature of the sample
by inserting the cooling block filled with a cooling mixture into the
sample well. Dry the cup with a paper tissue to remove any collected
moisture prior to adding the sample. (CAUTION: Be careful in handling
the cooling mixture and cooling block; wear gloves and goggles. Mixtures
such as dry ice and acetone can produce severe frost bite.) (CAUTION:
Be careful in inserting the cooling block into the tester cup to prevent
damage to the cup.)
7.7.3 Introduce the sample as in 7.6.4. Allow the temperature to
rise under ambient conditions or increase the temperature of the cup by
rotating the heater controller clockwise slowly until the specification
temperature adjusted for barometric pressure is reached. Determine
whether the sample flashes as 1n 7.6.5 and 7.6.6.
7.7.4 Turn off the test and pilot flames. Clean up the apparatus.
-------
1020 / 5
7.8 Ambient to 110* C (230* F)
7.8.1 Preliminary or trial test: Follow steps 7.6.2 to 7.6.5
omitting the barometric reading and using an estimated finite flash
point instead of a specification flash point temperature.
7.8.2 After 1 min has elapsed, observe the temperature, apply the
test flame by slowly and uniformly opening the slide fully and closing
completely over a period of 2-1/2 sec. Watch for a flash. (NOTE: The
sample is considered to have flashed only if a comparatively large blue
flame appears and propagates itself over the surface of the liquid.
Occasionally, particularly near the actual flash point temperature,
application of the test flame may give rise to a halo; this should be
ignored.)
7.8.3 Finite flash point: If a flash is observed, proceed as
below.
7.8.3.1 Using a temperature of 5* C (9' F) lower than the
temperature observed in 7.8.2, repeat 7.8.1 and 7.8.2. (CAUTION:
Be careful in inserting the cooling block into the tester cup to
prevent damage to the cup.) If a flash is still observed, repeat
at 5* C (9* F) lower intervals until no flash is observed. (NOTE:
Never make a repeat test on the same sample. Always take a fresh
portion for each test.)
7.8.3.2 Repeat 7.8.1 and 7.8.2 with a new sample, stabilizing
the test cup temperature at the temperature at which no flash
occurred previously. Observe whether a flash occurs at this
temperature. If no flash occurs, increase the temperature at
0.5* C (1* F) intervals by making small incremental adjustment to the
temperature controller and allowing 1-min intervals between
each increment and the flash point test. Record the temperature at
which the flash actually occurs. Record the barometric pressure.
Turn off pilot and test flames and clean up tester.
7.8.4 Finite flash point: If no flash point is observed in 7.8.2,
proceed as follows.
7.8.4.1 Using a test temperature of 5* C (9* F) higher than
the temperature observed in 7.8.2, repeat steps 7.8.1 and 7.8.2.
(NOTE: Never make a repeat test on the same sample. Always take a
fresh portion for each test.) If no flash is observed, repeat at
5' C (9' F) higher intervals until a flash is observed.
7.8.4.2 Repeat step 7.8.3.2 with a new sample.
44
-------
6 / CHARACTERISTICS - Ignitability
7.9 0" C (32* F) to ambient temperature
7.9.1 Preliminary or trial test: Cool the sample to 3" to 5" C
(5* to 10* F) below the expected flash point.
7.9.2 Cool the tester to approximately the temperature of the
sample by inserting the cooling block filled with a cooling medium into
the sample well.
7.9.3 Insert the sample as in 7.6.4. Set the 1-min timing device.
After 1 min, apply the test flame by slowly and uniformly opening the
side fully and closing completely over a period of approximately
2-1/2 sec. Observe for a flash. Record the temperature.
7.9.4 Finite flash point: If a flash is observed, proceed as
follows.
7.9.4.1 Cool a new sample and the sample cup to 5* C (9* F)
below the previous temperature (7.9.3). After 1 min, check for a
flash as in 7.9.3. If the sample flashes, repeat test at 5" C
(9* F) lower intervals until no flash is observed.
7.9.4.2 Repeat with a new sample, cooling both sample and
tester to the temperature at which the sample did not flash. After
1 min, observe whether a flash occurs at this temperature. If not,
increase the temperature at 0.5* C (1* F) intervals by making small
incremental adjustments to the temperature controller, allowing
1 min between each increment and the test for the flash point.
Record the temperature at which the flash actually occurs. Record
the barometric pressure.
7.9.5 Finite flash point: If no flash point is observed proceed
as follows.
7.9.5.1 Using a test temperature of 5* C (9* F) higher than
the temperature observed in 7.9.3, repeat step 7.9.3. (CAUTION:
Be careful in inserting the cooling block into the tester cup to
prevent damage to the cup.) If no flash is observed, repeat at
5* C (9* F) higher intervals until flash is observed.
7.9.5.2 Using a new sample, repeat 7.9.4.2 until a flash
occurs. Record the temperature at which the flash occurs and the
barometric pressure.
7.10 Cleanup of apparatus and preparation for next test
7.10.1 To prepare for the next test, unlock the lid assembly of
the tester and raise to the hinge stop. Soak up liquid samples with an
45
-------
1020 / 7
absorbent paper tissue and wipe dry. Clean the underside of the lid and
filling orifice. A pipe cleaner may be of assistance in cleaning the
orifice.
7.10.2 If the sample is a viscous liquid or contains dispersed
solids, after soaking up most of the sample add a small amount of a
suitable solvent for the sample to the cup and then soak up the solvent
and wipe clean the interior surfaces of the cup with an absorbent tissue
paper. (NOTE: If necessary to remove residual high boiling solvent
residues, moisten tissue with acetone and wipe clean.) (NOTE: If any
further cleaning is necessary, remove the lid and shutter assembly.
Disconnect the silicone rubber hose and slide the lid assembly to the
right to remove. If warm, handle carefully.)
7.10.3 After the cup has been cleaned, its temperature may be
rapidly increased to some stand-by value by turning the temperature
control dial to an appropriate point. (NOTE: It is convenient to hold
the test cup at some stand-by temperature (depending on planned usage)
to conserve time in bringing the cup within the test temperature range.
The cup temperature may be quickly lowered by inserting the aluminum
cooling block filled with an appropriate cooling mixture into the cup.)
7.10.4 The syringe is easily cleaned by filling it several times
with acetone or any compatible solvent, discharging the solvent each
time, and allowing the syringe to air dry with the plunger removed.
Replace the plunger, and pump several times to replace any solvent vapor
with air.
7.11 Correction for barometric pressure
7.11.1 When the barometric pressure differs from 760 mm Hg
(101.3 kPa), calculate the flash point temperature by means of the
following equations:
Calculated flash point = C + 0.03 (760 - P)
= F + 0.06 (760 - P)
where:
C, F = observed flash point (*C or *F)
P = barometric pressure (mm Hg).
7.11.2 Likewise determine the corrected specification flash point by
the following equation:
C = S - 0.03 (760 - P)
F = S - 0.06 (760 - P)
-------
8 / CHARACTERISTICS - IgnitabiHty; Corrosivity
where:
C, F = flash point to be observed to obtain the specification flash
point at standard pressure (S)
S * specification flash point.
7.12 Report
7.12.1 When using the flash/no flash method, report whether the
sample flashed at the required flash point and that the flash/no flash
method was used.
7.12.2 If an actual flash point was determined, report the average
of duplicate runs to nearest 0.5* C (1* F) provided the difference
between the two values does not exceed 1* C (2* F).
8.0 Quality Control
8.1 All quality control data should be available for review.
8.2 Duplicates and standard reference materials should be routinely
analyzed.
8.3 The flash point of the p-xylene reference standard must be determined
in duplicate at least once per sample batch. The average of the two analyses
should be 27* +_ 0.8* C (81* +_ 1.5* F).
-------
APPENDIX C
BCR Information
48
-------
COMMISSION OF THE EUROPEAN COMMUNITIES
BCR information
The certification of five hydrocarbon
materials for the determination
of flashpoint (temperature range 15 to 65 C)
(BCR reference materials Nos 41, 42, 43, 44 and 45)
D. LEWIS. K3r Mond Division, Northwich
L. HAEMERS, JBG tepra, Cetis, Italy
W. KARCHER. URG Petten,
Joint Research Centre
Petten Establishment - Netherlands
-------
Published by the
COMMISSION OF THE EUROPEAN COMMUNITIES
Directorate-General
Scientific and Technical Information and Information Management'
Bailment Jean Monnet
LUXEMBOURG
LEGAL NOTICE
Neither the Commission ol the European Communities nor any person acting on
behalf of the Commission is responsible for the use which might be made ol the
following information
A bibliographical slip can be found at the end of this volume
(C) ECSC-EEC-EAEC, Brussels-Luxembourg, 1979
Printed in Belgium
50
ISBN 92-825-0684-3 Catalogue number: CD-NV-78-010-EN-C
-------
TABLE OF CONTENTS.
1. Introduction 5
2. Certification Procedure 5
3. Participants y
4. Materials g
5. Flashpoint Determination Procedures 9
6. Statistical Evaluation of the Results -j-j
7. Correction for Atmospheric Pressure Variations 14
8. Conclusion 15
9. References
51
-------
1. INTRODUCTION.
Flammable liquids have for a considerable period been
classified for transport by reference to their flash-
point temperatures as determined by specified apparatus
and defined temperature boundaries. Efforts to achieve
international standardisation has led to a number of
national and international investigations of the
behaviour of liquids of high purity in different apparatus (1-4)..
The results of these investigations have shown that, by
the use of a specified equilibrium method and any closed
cup flashpoint apparatus cup and cover, flashpoint values
with a given material are the same for a considerable
number of apparatus types.
Flammable .liquids for flashpoint certification
purposes should have long term stability, resistance to
contamination by water and similar fluids, safety in use
and ease of flashpoint determination. These requirements
are met by pure hydrocarbons, but the flashpoint
temperatures obtained do not coincide with the defined
flashpoint classification boundaries. Mixtures could
be made to give the required flashpoint values at the
classification boundaries, but these would not satisfy
other requirements. Accordingly, a set of pure hydro-
carbons covering the temperature range of approximately 15 to 65 C
has been selected with the aim to certify their flashpoint
temperatures. In this way cprtification can be achieved either
side of a required temperature since the certified materials have
flashpoints at intervals of 10 to 15°C.
These reference materials are intended mainly for the
calibration of the various closed cup flashpoint apparatus
which are currently in use.
2. CERTIFICATION PROCEDURE.
The certification has been carried out in the framework
of the "Reference Materials and Methods" programme of
52
-------
the Community Bureau of Reference (BCR) and METRE
programme of the Joint Research Centre of the Commission
of the European Communities. The technical project
has been prepared and supervised within the BCR group
of specialists "RM for flashpoint determination" and
the. BCR-working subgroup 'Petroleum and Related Products'
(chairmen Messrs, van Straten and Rosado respectively;
technical secretary W. Karcher).
The certification programme was discussed at a number
of meetings with attention being paid to previous
experience with flashpoint evaluation programmes made
available to the specialist group. This indicated that
only by the use of equilibrium methods for closed cup
«
flashpoint determination could reference materials
certified for flashpoint be produced. The results
obtained with non-equilibrium methods were known to give
significant variations between types of apparatus and
different procedures and the work involved in attempting
to certify each material for a wide range of different
apparatus and methods was such that the simpler more
universal equilibrium methods were preferred.
The equilibrium methods for flashpoint selected for the
certification procedure were as follows:
2.1. "Rapid Tests for Flashpoint", Institute of
Petroleum, Method IP303/74. The section
of the method applying to "Flashpoint
Determination Section 2.2" was followed.
This method is also available as ASTM D3278-73.
2.2. "Flash Testing Using the Cup of Any Standard
Closed Cup Apparatus", Institute of Petroleum
Method IP3O4/74. The section of the method
applying to "Flashpoint Determination Section
2.2" was followed. This method is also
available as ISO Standard 1523, BS3900 Part A9,
1973, etc.
53
-------
Five hydrocarbons were selected for the certification
procedure covering a temperature range of approximately
15 to 65°C which satisfy the most important temperature
criteria applied to flashpoint. Further certification
work to higher and lower temperatures was envisaged for
later flashpoint certification programmes.
The results of the certification determinations were
called for without the application of any correction for
barometric pressure together with values of the barometric
pressure in order that the variation of flashpoint values
with barometric pressure could be examined.
The results of the certification procedure and the
statistical analysis of them have been discussed at a
meeting attended by representatives of the collaborating
laboratories.
3. PARTICIPANTS.
3.1. Provision and distribution of the materials:
Commission of the E^opean Communities, „ ,
Joint Research Centre, Petten Establishment, Netherlands
3.2. Flashpoint Determinations:
-Elf, Centre de Recherche, Solaize (F)
-Commission of European Communities, JRC Petten
Establishment, Petten (Netherlands)
-Esso Research Centre, Abingdon (United Kingdom)
-Health and Safety Executive, Buxton (United Kingdom)
-ICI Ltd., Mond Division, Northwich (United Kingdom)
-Institute for Industrial Research and Standards,
Dublin (Ireland)
-Labofina SA, Brussels (Belgium)
-Laboratory of the Government Chemist, London
(United Kingdom)
-Laboratoire National d'Essais, Paris (France)
-Physikalisch-Technische Bundesanstalt,
Braunschweig ( Federal Republic of Germany)
-Shell, "Nederl. Raff. Rotterdam (Netherlands)
-Staatspr^venanstalten, Copenhagen (Denmark)
3.3. Statistical evaluation of the results:
Cetis, JRC Ispra, Commission of the European
Communities, Ispra (Italy)
54
-------
4. MATERIALS.
4.1. Materials selection.
Saturated hydrocarbons were selected as being readily
available at adequate purity levels and having been
shown by flashpoint evaluation programmes carried out by
API, BSI and others as having a satisfactory stability
in storage, minimal contaminant pickup risk and not
presenting unusual hazards in carrying out flashpoint
determinations (corrosion, toxicity, etc.). Some
consideration was given to the use of hydrocarbon mixtures
to enable specified flashpoint temperature boundaries to
be obtained. This possibility was discarded because of
the problems associated with differential evaporation of
one component and the production of batches having a
satisfactory homogeneity.
The selected hydrocarbons were:
:8H18
:8H10
:9H20
n-Octane CRHI n boiling point 125.7°C
p-Xylene C8H1O boiling point 138.4°C
n-Nonane coHon boiling point 150.8°C
n-Decane C1OH22 boiling point 174.1°C
n-Undecane C11H24 boiling point 195.9 C
The availability of these materials from established
laboratory chemical suppliers was such that various
purities ranging from 95% to 99.5% were offered. Enquiries
were made from three suppliers and materials ordered to a
minimum purity of 99.O%, as more expensive materials of
ultrapurity were considered inappropriate for flashpoint
reference materials.
4.2. Material homogeneity and bottling.
Each material was provided in alternative bottle sizes
of 1OO ml and 5OO ml. The former size was considered
appropriate for work involving a rapid tester such as the
Setaflash where about 20 to 3O m^s required for a
flashpoint determination. The 5OO ml size was adopted
-------
for the convenience of users of standard flashpoint
apparatus cups according to IP3O4/74 where between 5O ml
and 85 ml is required for each filling of the apparatus.
The total batch of each hydrocarbon ordered was supervised
by a BCR representative during the manufacturers' purity
checking stage and the filling, sealing and identification
marking of the individual 1OO ml and 500 ml bottles
ordered by BCR. The batch size of the five materials
was chosen in line with expected demand:
-2OO 1 of p-xylene (400 bottles of O.I and
320 bottles of O.5 1)
-1OO 1 of n-octane and n-nonane (20O bottles of
O.I and 160 bottles of O.5 1)
-5O 1 of n-decane and n-undecane (100 bottles of
O.I and 80 bottles of O.5 1)
4.3. Impurities in the reference materials.
It was established from the suppliers that none of the
reference materials contained significant quantities of
impurities of widely different flashpoints (less than 0.2%)
and that the impurities present mainly consisted of
adjacent members of the hydrocarbon series. Each bottle
of the reference materials was given an individual code
reference so that any suspected impurity variations could
be examined in detail if required.
5. FLASHPOINT.DETERMINATION PROCEDURES.
5.1. Requirements for certifying laboratories.
Each laboratory was supplied with 5 bottles of 100 ml and
1 bottle of 500 ml of each material and requested to provide
a minimum of six determinations on each of the five
materials with the determinations covering two or more
different operators using the Setaflash rapid tester and
one or more other standard closed cup. Any nationally
accepted true closed cup apparatus could be used as the
alternative apparatus to the Setaflash (e.g. Tag. Abel.
5$6
Abel-Pensky and Pensky-Martens).
-------
5.2. Flashpoint determinations.
Flashpoint determination procedures have been nationally
and internationally standardised for a considerable time
and are regularly updated. The equilibrium methods
IP3O3/74 for the Setaflash and IP304/74 for the use of
other standard flashpoint apparatus cups were recognised
as adequately defining the methods to be used for the
certification procedure with the following additional
requirements:
1) All results were to be reported to BCR regardless
of the requirements for rejection of results on
the basis of repeatability limits in
IP303/74 and IP304/74.
2) The results were to be reported without the
application of a barometric pressure correction
or any thermometer correction.
3) The barometric pressure at the laboratory at the
time of carrying out each determination was to be
recorded and reported.
4) Any known thermometer correction for the
temperature concerned was to be reported.
Reference to IP303/74 and IP304/74 should be made for
full details of established procedures for flashpoint.
Adequately skilled operators were requested for the
programme and also the' use of apparatus in good condition.
All determinations were to be reported to the nearest O.5 C.
5.3. Collection of results.
All the results from the 12 laboratories were collected
by JRC Petten who applied the barometric correction of
^0.1°C for each-*- 4 mbar (3mm Hg) barometric pressure change
(given in IP303/74 and IP304/74) before listing the
results for statistical analysis. In addition, the
uncorrected flashpoint temperatures and the observed
barometric pressures were listed in order that the influence
of barometric pressure correctior^^ould be verified.
-------
5.4. Clarification of results.
Where inadequate data was reported, clarification of
the results was obtained by JRC Petten from the
laboratories concerned.
Due to an oversight, some results were reported by
methods which were non-equilibrium (using the standard
methods for the Abel and Pensky-Martens apparatus).
These showed a significantly different picture and as
a result a total of 3 laboratories were asked to carry
out a number of non-equilibrium methods for comparison
with the certification equilibrium results.
6. STATISTICAL EVALUATION OF THE RESULTS.
6.1. Equilibrium methods - General comments.
Approximately 1OO-120 determinations were received for
each of the 5 hydrocarbons of which 43% were obtained
with the Setaflash apparatus, 17% to 19% with each of
the Abel, Abel-Pensky and Pensky-Martens apparatus and
2% with the Tag apparatus.
Analysis of the results from individual laboratories
both as "units" and in terms of the individual operator's
results showed that all the variations were generally
acceptable and not rejectable on account of too high a
difference between different operators, different
apparatus or different laboratories.
6.2. Results using different flashpoint cups and the
Setaflash apparatus.
Except for the determinations using the Tag cup (only 2
results for each material), an adequate number of results
were obtained to compare the Abel, Abel-Pensky and
Pensky-Martens cups with each other and with the Setaflash.
In the case of all five hydrocarbons, the results of
using these four apparatus were comparable with each other.
The conclusion was reached that, using equilibrium methods
and for pure hydrocarbons giving flashpoints in the range
15°C to 65 C, the Setaflas^agAbel cup, Abel-Pensky cup
and Pensky-Martens cup give results within the statistical
-------
limits. The limited number of Tag cup results also
show excellent agreement.
6.3. Representation of statistical results.
For each of the five hydrocarbons tested, a mean value
ofrthe equilibrium results has been determined, m.
The allowance for variation in results to be anticipated
in using the reference materials for calibration purposes
can be expressed in two ways. The results obtained are
corrected for barometric pressure (see section 7) and then
are compared with the mean value m. If only one result
is obtained, their mean x should lie in the band m +2s_.
~~ x\
However, if a specified number n of repeat determinations
• *"^
in a laboratory are used to obtain an experimental mean
value x , this can be tested for adequate calibration of
the apparatus, method and operator by the formula:
"u
Values of s_ and s^ are assigned to each material from
Xx i
the statistical analysis.
6.4. n-Octane results. (see Table 1)
From the 125 equilibrium results for n-octane, a corrected
mean value of 14.O°C was derived. The variation parameters
were determined to be:
SR=0.7°C
s_ = 0.6°C
6.5. p-Xylene results. (see Table 1)
A tot.al of 107 equilibrium results were received for
p-xylene which resulted in a corrected mean value of 25.9 C.
The corresponding variation parameters were derived as:
SR = 0.6°C
s = 0.5°C
59
-------
6.6. n-Nonane results. (see Table 1)
A total of 112 equilibrium results produced a corrected
mean value of 32,O°C with variation parameters as follows:
SR = 0.8°C
s = 0.6°C
6.7. n-Decane results. (see Table 1)
For this material, 11O equilibrium results were analysed
to give a mean value of 49.1 C and variation parameters
as follows:
,R-i.oPc
s_ = 0.8°C
6.8. n-Undecane results. (see Table 1)
Analysis of the 115 equilibrium results gave a mean value
of 62.9 C with variation parameters as follows:
SR=
= 1.0°C
6.9. Non-Equilibrium results, (see Table 2}
A number of results using standard non-equilibrium methods
for the Abel apparatus (standard IP170/70) and for the
Pensky-Martens apparatus (standard IP34/67) were analysed
(reasons for these results are given in section 5.4. above).
Approximately 10 determinations were received for each of
the five materials which were analysed separately after
applying the standard barometric correction given in
section 5.3. above. The non-equilibrium results were
found to be far more variable than the equilibrium results
"and to show significant differences between the two
apparatus tested. Typical values for !=R (as defined above)
for the non-equilibrium methods were from 1.3°C to 2.8°C
compared to the O. 6°c to 1.1°C values for the equilibrium
methods. The Rapid Tester apparatus is based on
an equilibrium method and hence non-equilibrium results
/-Art
for this apparatus by an established method are not available,
-------
6.10. Comparison of certification procedure results with
accepted standards for flashpoint equilibrium methods.
From IP303/74 for the Setaflash apparatus, the defined
repeatability criteria for the range 20°C to 7O°C is + 0.5°C
between operators and the reproducibility between two
laboratories is + 1.5°C to 2.7°C (dependent on the mean
temperature value of two results being compared).
IP304/74 for equilibrium methods using standard cups
specifies repeatability as 1°C and reproducibility as 3.5°C.
These are based on 95% confidence limits. The variation
parameters obtained for the certification procedure results
are seen to be in close agreement to the repeatability and
reproducibility criteria published for IP3O3 and 304 methods.
7. CORRECTION FOR ATMOSPHERIC PRESSURE VARIATIONS.
The uncorrected results for each of the five hydrocarbons
were statistically examined for the effect of variations
in the barometric pressure prevailing at the time of
obtaining each result. As the flashpoint of a pure liquid
is equivalent to the lower flammability limit in air of the
same liquid under horizontally expanding flame propagation
conditions (which is constant as a volume % value), the
correction in temperature value is related to the vapour
pressure characteristics of the liquid. The correction
values given in published flashpoint methods are mean
values for a range of flammable liquids and do not
necessarily represent the optimum corrections for specified
reference materials. As the five materials concerned have
standard boiling points ranging from 125°C to 195°C, it
was considered possible that the optimum barometric
pressure could be either a function of the flashpoint or
the standard boiling point temperature values.
The conclusion reached from this part of the statistical
examination of the results was that the data available did
not justify a variation in the correction value of + 0,1°C change
for each + 4 mbar (3mm Hg) barometric pressure change for
61
-------
any of the hydrocarbons tested.
The range of atmospheric pressure variations covered
by this calibration procedure is from 0.951 to.1.031 bar
(715 mm Hg to 775 mm hg)
8. CONCLUSIONS.
8.1. Flashpoint determination methods.
The results of this certification procedure have
confirmed other published conclusions that the use of
equilibrium flashpoint methods enables equivalent
results to be obtained using a range of standard flash-
point apparatus cups. The older standard non-equilibrium
methods for the same apparatus do not give adequate
reproducibility to enable reference flashpoint materials
to be prepared for these methods.
8.2. Atmospheric pressure correction.
The published atmospheric pressure correction (IP303/74
and IP304/74) have been found to be satisfactory with
these BCR reference materials.
8.3. Flashpoint results.
From the statistical analysis, the following results have
been evaluated and are certified accordingly by rounding
off the mean value into the nearest 0.5 or l.O degree.
sample n-Octane p-Xylene n-Nonane n-Decane n-Undecane
Mean Flash- 14. O . 26. O 32. O 49. O 63. O
point ( C)
Tolerance +1.5 +1.5 +J. . 5 ±2.O +2.O
Interval (°C)
(2)
Estimation O.7 ' O.6 O.8 l.O 1.1
of SD (°C)
R (3)
Estimation O.6 O.5 O.6 O.8 l.O
of s (°C)
r (4)
n (number 125 io? H2 11O 115
of measure-
ments)
p (number 12 H 12 12 12
of laborat-
ories) 62
Remark; Tolerance interval= t.S (t95* = 2)
-------
(1) m is the unweighted mean of n individual measurements
obtained by p accepted laboratories participating in
the certification.
(2) For any given measurement, there is a 95% probability
of the results falling into the range covered by the
tolerance interval.
(3) s_ is the uncertainty, viz. estimated standard deviation
of reproducibility which affects each individual
measurement. It accounts for the precision of the
participating laboratories as well as for any inhomogeneity
of the material.
(4) s is the uncertainty, viz. estimated standard deviation
r of repeatability for individual laboratories.
63
-------
REFERENCES.
J.R. Hughes "The Need for an International
Standard on Flammability", Journal of the
Institute of Petroleum, 55, pp 380-7 (1969).
L.H. Bell "New Flash Test Methods", Journal
of the Institute of Petroleum, 57, pp 219-
230 (1971).
H.A. Wray "New Flash Point Tester for the
Paint Industry - Setaflash", Journal of
Paint Technology, 45, pp 44-54 (1973).
W. Weber "Erfahrungen mit dem Flammpunkt-
gerat Setaflash", Erdol und Kohle Erdgas
Petrochemie vereinigt mit Brennstoffchemie,
2_7, pp 17-22 (1974) .
H. Fangmeyer, L. Haemers, J. Larisse
Statistical approach for collaborative tests
and reference material certification procedures
EUR 5621, 3 (1977)
64
-------
European Communities - Commission
EUR 6102 — The certification of five hydrocarbon materials for the determina-
tion of flashpoint (temperature range 15 to 65°C)
Joint Research Centre - Petten Establishment
Luxembourg : Office for Official Publications of the European Communities
1979 _ 18 pp. - 21.0 x 29.7 cm
BCR information series
EN
ISBN 92-825-0684-3
Catalogue number: CD-NV-78-010-EN-C
BFR DKR DM FF LIT HFL UKL USD
100 17,50 6,40 14 2700 7 1.80 3.40
The flashpoints ol five pure hydrocarbon materials (n-octane, p-xylene,
n-nonane, n-decane and n-undecane) have been determined by equilibrium
methods in an interlaboratory exercise, involving eleven laboratories of the EEC
member countries and two laboratories of the Commission of the EEC. The
methods used were IP 303/74 (or ASTM D 3278 - 73; Rapid Tester) and IP
304/74 (or ISO 1523; Abel, Abel-Pensky, Pensky-Martens, Tag Cups). For each
hydrocarbon material, more than 100 individual measurements have been ob-
tained. As a result, the flashpoints of these materials are certified at the follow-
ing temperatures :
- n-octane 14.0°C (Sr = 0.6°C)
- p-xylene 26.0°C (Sr = 0.5°C)
— n-nonane 32.0°C (Sr = 0.6°C)
- n-decane 49.0°C (Sr = 0.8°C)
- n-undecane 63.0°C (Sr=1.0°C)
(Sr is the uncertainty, viz. estimated standard deviation of repeatability for in-
dividual laboratories).
This report describes the experimental details of the interlaboratory measure-
ments and the certification procedure.
65
-------
APPENDIX D
QUALITY ASSURANCE REPORT
66
-------
QUALITY ASSURANCE REPORT
Title: Evaluation of Ignitability Methods (Liquids)
Sponsor: Environmental Protection Agency
Cincinnati, Ohio 45219
Contract No.: 68-03-3099
RTI No.: 321U-2258-14
Task Leader: Dr. Mirtha Umana
Study Dates: March 1984-July 1984
To the best of my knowledge, the QA/QC requirements described in
the study QA Project Plan were implemented and satisfied.
Robert W. Handy, Ph.IVr Date
Task Quality Assuran^ Officer
67
------- |