PB88-102892
Sampling Oil-Water Vixtures at OHM3ETT
(Oil and Hazardous Materials Simulated
Environmental Test Tank)
Mason and Hanger-Silas Mason Co., Inc.
Leonardo, NJ
Prepared Eor
Environmental Protection Agency, Cincinnati, OH
Sep 87
I
-------
P888-102892
Sampling Oil-Water Mixtures at OHM3ETT
(Oil and Hazardous Materials Simulated
Environmental Tost Tank)
Mason and Hanger-Silas Mason Co., Inc.
Leonardo, HJ
Prepared for
Environmental Protection Agency, Cincinnati, OH
Sep 87
™
-------
IJ'A/OOO/J-87/073
rA.vpi.7Nr; OIL-WATFH
AT OH.vrKrr
Vichael Porst
Mason 8r Hanger-Snas Maron Co., Inc.
P.O. Pox 31V
l.eor.ardo, !
-------
TECHNICAL REPORT DATA
£PA">OU/2-87yQ73
Sampling Oil -Water Mixtures at OHMSETT
"Michael Borst . >.
Mason & Hanger-Silas Mason Company, Inc.
Post Office Box 117
Leonardo, NJ 07737
Hazardous Haste Engineering Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45258
1 Hiw"pB8fil"i'S6'208 9 n/JS
» mt »rj»T a*T^_
St-pti-nb-t-r J987
e. M«»O«M,NC o«CAN.f*r,o», c&ot
» »fc«»0«M.N 0«CAN,J* ION «iro«T >u ^
10. mOCAAM (LlMtNT MO.
II CO-«T«ACT/GH»NT NO
68-03-3203
Final Report
14. SPONSORING AGfcNC* CO&t
68-03-3056
4
1
waterTnixturesrt deSCMbeS Procedures developed at OHKSETT.for sampling oil and '
.Two procedures for sampling in containers are discussed: grab and stratified
cMtifn»r hStl yl!eSe teehnl<»ues 'eqtiire stripping free-standing water from the i
thorn HI • ^\ I e grab Samp1e tf" hnic*ue ^quires that the renriaping fluids be i
s?on 9Tni T^^r"l^:^L^ !,'^le^^OU3h_t,he resu!ti"9 homogeneous «,«!- .
flu1ds Wre tested* The two "
tr.'Cttl1ltllHt/O?tH ENOCO TCMM3
UnclassifiArt
30. JICwOHY CLASS (Ttuiptiti
llnrla««*fi^
i
•
1
21. NO. OF PACCS 1
•*a
2J PHICI
-------
NCCT'J--
T>.ir> rejicrt has let-n reviewed Vy th» Hs-ardous Waste Kn*;ine«»rlr.«T
laboratory, U.S. Fr.viror^:«-ntfcl Protection A^f-r.cy, and ayproved for
publication. Approval dees r.ot firr.ify that the contents necessarily
reflect the views ar.i policies of the 'I.™. Fr.viroriner»t«l Protection A^er.cy,
nor dC'?-s mention of trade nar.es or co^zercial products constitute eridorse-
r.f-nt or recosr-.endaticn f&r use.
ii
-------
FC5-KVCW-
As hazardous vaste continut-s to bfr cne of the core proffin'-nt <>nvircr.-
concerns to the people of the United Stater and other countries
throughout the world, there are continuing needs for research to charae-
terirs problers and develop ar.d evaluate alternatives to addressing those
probler?. The programs of the Hazardous Waste ; r.gir.eerir.g Research
laboratory (HVKPL) are designed to contribute to satisfying these research
needs.
This report describes procedures developed at CHKSrTT for ser.pling
Kixtures of Circo Mediua test oil (a refined naphthenic oil) and fandy Kock
Pay water (salt content approximately 21 ppt). It includes a grab sample
technique, a stratified raspler and severej discrete sampling procedures.
The accuracy and applicability of each technique is addressed. This report
is submitted as partial completion of Contract Ho. 66-03-3056, Job Order
75, vhich was sponsored by the U.S. EPA. Further information nay be ob-
tained through the Releases Control Branch of the Hazardous Vaste Engineer-
ing Research Laboratory, Edison, Rev Jersey 08837.
Thoffias R. Eauser, Director
Hazardous Waste Engineering Research Laboratory
ill
-------
This report descries procedures developed at CHXI.'KTT for
oil ard water r.ixturer. The results of testing done to evaluate each s&n-
plir.g technique are presented.
Tvo procedures for -ssrpling In containers are discussed: grab and
stratified sampling. Foth of these techniques require stripping free-
star.dirir water fron the container bottcs. The grab sajr.ple technique
requires thut the reERir.irjr fluids be thoroughly r.ixed before iEr.erfir.g a
bottle through the resulting hocogenecus tjr.ulsion. The stratified sampling
procedure uses a sanple thief to capture e segmented cross-section of the
remaining fluids. The grab sample results proved to be vithin 3* of the
knovn relative oil content in all tests that were considered valid. The
grab sample results averaged less than ?* different fron the kr.ovn oil con-
tent over the rar.jre of 9 to 91? oil. The stratified sar.pler vas tested
using several procedures. In the most complete analysis, the stratified
sampler gave results vithin 3? of the kr.ovn oil content, averaging less
than ?1> different.
Two procedures for sailing flowing fluids were tested. The two sam-
pling tubes tested were installed immediately downstream of a series of
static mixers and a centrifugal pump. The sae-pllng ports were a simple
slotted tube and a pitot-shaped tube. The slotted tube results were always
vithin 8% of the grab sasple analysis and averaged less than U? different
from the grab sample analysis in all cases. The pltot-sheped tube gave
results vithin 1% in each case and averaged within UJ of the grab sample
analysis.
This work is submitted by Mason & Ranger-Silas Mason Co., Inc. as
partial coKpletion of Contract Ho. 68-03-3056, Job Order 75 which was spon-
sored by the U.S. Environnental Protection Agency. Further information nay
be obtained through the Releases Control Branch of the Hazardous Waste En-
gineering Research Laboratory, Edison, Kew Jersey 08837.
Iv
-------
CCXTXNTP
Forevord ....... . ...................... . .................. • .......... Hi
Abstract ................................... o ....................... - v
List of Figures [[[ vl
List of Tables [[[ »i
Section
1. Introduction ............................... . ............ 1
?. Conclusions .......... .................................. 3
3. Becomnendations ......................................... 5
1«. Sampling fron Containers ................................ 6
5. Sampling fron Flows ..................................... 16
Bltliography
Appendices
-------
So.,
1 Grajh of Grab Sarp.le Test Penults « 8
2 Construction Drawing of Stratified Sarpler 10
3 Graph of Stratified f-Birpler Test FesuKs Vsir.g Method 1 13
U Graph cf Stratified Sampler Test Fesults i'sir.g Method ? 13
5 Graph of Stratified Earpler Test Fesults L'sirg Method 3 I1*
6 Graph e" Stratified Sampler Test Results Usir.g Method U I1*
7 Cutaway Travirg of Static Mixer.... 18
8 Graph of Slotted Tube Test Results.....
9 Graph of Pi tot Tube Test Results.. 20
TABLES
1 Summary cf Grab Eacple Test Results •• 9
2 SuMtary cf Stratified Test Results Usiig All Methods 15
3 Oil Percentages for Grab and Both Dynamic Saeples 19
Yi
-------
INTRODUCTION
The U.S. Environmental Protection Agency cvr.s the Oil and Hazardous
Materials Simulated Environaental Test Tark (CKMSSTT) located in Leonardo,
Sew Jersey. ?ince its inception, OH.V3KTT has tested oil spill recovery and
cor.trol devices in an environmentally safe Banner. The daily dealings with
oil and water have led to the development of sairolirg techniques for saa-
plir/g mixtures of these two iraiiscible fluids. The need to know with
reasonable accuracy the relative proportions of oil and water ir. a con-
tainer required that a consistent means of saaplirig and quantifying the oil
content in containers and in flowir.g streaas be developed.
The procedures used at OKKSCTT to ceasure the relative oil-water con-
tent in a tank have developed since the facility was first cperaticnal in
197&. The earliest measure of the oil percentage in a container was
idealistic. A tape measure was used to sceasure the total height of col-
lected fluids and the height of the oil portion alone. The ratio of the
height of the oil portion to the total fluid height was considered the
relative oil percentage in the tank. Experience dictated that improvements
vere necessary. The first attempt to improve this measurement involved
stripping free-standing water before taking the second height measurement.
This procedural change reduced the error introduced by misreading the
oil/water Interface due to an oil coating on the sides of the container.
The next change accounted for water caught up in the oil after stripping.
A bottle was lowered through the remaining oil-rich liquid. The sample was
analyzed according to procedures outlined by the American Society for Test-
ing and Materials (ASTM) Procedure 1796.
The vork described in this report began vith another change that vas
Implemented to improve the quality of the sample of the oil-rich liquid.
This procedure involves mixing the oil and remaining vater to produce a
homogeneous emulsion of the oil-rich layer for sampling. Grab-saapling was
still used and is the first technique discussed here. The method of sam-
pling used most frequently during the testing of oil spill control and
removal equipment has been the grab sample technique. A second technique
was developed to avoid the intentional ewulslfication of the test oil by
using a stratified sampler known as the Johnson Sanpler. A third technique
used to sample mixed fluids flowing within a pipe or hose, is used at
OHMSETT vhen a real-time analysis is required. This technique uses a small
diameter pipe to divert a portion of the flowing mixture to a container for
analysis. The fourth technique employed is simply taking a cup and snatch-
Ing a portion of an open-air flow as the mixture free falls from the end of
the carrier. This last technique has been used Infrequently, and as a last
-------
renort vhen only rou^h estimates of oil content vere required and th^re vtis
no other seans of obtaining the desired sar.ple. The eost coi»on csffplir.g
r.eed during OH.VSJTT tests has been sanplin^ oil ard vater in a cont&ir.er.
TS'ierefore, grab sarples and stratified sarples have been cost conacrily
taken.
-------
KKCTIOS ?
CONCLUSIONS
Mixtures of oil and water can be sarpie?J to reflect accurately the
relative oil/v&ter content of either tanks or flcving streams. The sam-
pling technique that is chosen for use is dependent on the type of process
involved (i.e. "catch or dynaicic) and available resources.
SAMPLING FROM COSTA'SERB
Either the stratified sampler or grab sarpler can be used to obtain a
representative sample of the oil-water mixture in a container. It is in-
port ant to realize that as much of the water phase as is possible should te
removed before s&npllr.g because of the techniques used in the laboratory
analysis. The pear shape of the centrifuge tube allows for highly accurate
measurements of low water-content samples (less than 2* water) but intro-
duces a significantly greater measurement error ir. high water-content
(greater than 25* water) measurements. The grab sample technique will give
measurements within 3* oil of the actual oil content of the fluids in a
container. The stratified sampling technique gives measurements within 3?
of the actual relative oil content. Shortcuts will give measurements
within 15?, depending on which shortcut is used.
The stratified sampler, when constructed with a transparent sheath,
provides a visual cross-section of the materials and layers within the
tank. This would be of value when an immediate estimate of fluid quality
is needed. This is, of course, only feasible when the sheath is con-
ctructed of a material chemically compatible with the sampled material or
the general nature and reactivity of the material to be sampled is known.
These tests were conducted using only OHXSETT's Circo Medium test oil
and Sandy Hook Bay water. The results can presumably be replicated using
many other fluids. If more than one liquid is present, and the liquids are
not mutually completely soluble, the stratified sampler will not give a
representative sample in irregularly shaped tanks. In such a case, the
grab sample technique would be preferable.
The major drawback of the grab sample method Is the need to mix the
liquids thoroughly. This extends the time required for sampling, requires
that power be available to make mixing feasible, and binders later separa-
tion of the liquids.
Beither of these two techniques is readily applicable to inflatable
tank or bladder type storage, but stratified sampling would most likely
-------
give better results in this; type container provldc-d that depth-volume es-
timates could be established.
SAMPLING FROM FLOWS
The pitot-chnped-tube and the slotted-tube sampling port; pave tn
equally representative rer.ple of the flowing fluid when u"ed ir con:M nation
With the static mixers. The campling tubes
-------
SECTION 3
RECOKMESEATIONS
When sampling oil-water mixtures in containers, stratified sarplers
should be used vhenever the container is constructed vith vertical, paral-
lel sides. The time period allowed for gravity separation of the imis-
cible fluids should be as long as practical so that the container can be
re&dily stripped of free-standing vater.
In other saapling applications, grab samples should be taken follow-
ing the procedures outlined. A thorough mixing is necessary to create a
homogeneous ezulsion for ssapling.
When it is necessary to sample flowing mixtures of oil and vater,
either the pitot tube-shaped sampling port or the slotted-tube sampling
port can be used with equal accuracy. Whichever sampling port is chosen,
it should be located immediately downstream of the static mixer. The
choice of sampling port will be dependent on tbe specifics of the case in-
volved and the fabrication capabilities available. When used in the
field, several of both types should be readily available from a supply es-
tablished prior to the spill.
Each of the sampling techniques that was tested used only OHMSKTT's
Circo Medium test oil and salt water. If these techniques are to be used
vith different fluids froa these, a wide variety of mixtures should be in-
vestigated to ensure that the physical properties of the fluids involved do
not adversely affect the devices. The principal areas that need to be in-
vestigated further are the effects of viscosity, specific gravity, surface
tension, and mutual immiscibility. Although it is anticipated that these
results will be applicable for a wide range of mixtures, there may be cases
where these devices do not perforn equally well. For example, the
stratified sample thief may not be able to capture a sample in a highly
viscous fluid. If the fluid does not flow into the annular space during the
time that the sheath is removed, there vill not be a sample taken. It is
equally likely that the continuous phase of the emulsion will selectively
flov into tbe annular space giving a non-representative sample. Similarly,
if the flowing fluids have high content of stringy fibers, static mixers
•ay plug quickly, leading to excessively large pressure drops, making this
approach to sampling impractical.
Another extension of this work should be' a repeatability and
confidence-level investigation to confirm statistically the error of each
method. It may be necessary to use only one of the test fluids unless the
physical properties prove be of significant influence.
-------
CECTIOS U
SAMPLING FRCM CONTAINERS
When evaluating ar. oil skimmer, a mixture of oil and vater has been
typically purped into a collection container from the skinner. In order to
determine the recovery efficiency, throughput efficiency, and oil recovery
rate, the volumes of oil and water in the collection tank or tanks had to
be determined.
These containers have traditionally been free-standing vlth vertical
parallel sides so that the volume was directly proportional to the height
within the container. Vhea containers vere used that had variable cross-
sectional area, a height-volume calibration vas conducted so that volume
could be determined directly from fluid height measurements.
GRAB SAMPLES
Each tine a grab saeple was taken, the height of the total collected
fluids was neasured using a tape measure. This initial height was re-
corded. The collected fluids were then stripped of as much free-standing
water as possible using a valved spigot on the bottos of the tank. When
oil began to appear in the stripped water, the valve was partially closed.
The stripping process was continued at a reduced rate until the maximum
possible amount of water tad been removed. The remaining high-oil-content
fluids were emulsified for five minutes using a Lightning electric mixer.
A glass 100-ml bottle claeped to a rod was slowly lowred through the
homogeneous fluid from top to bottom and slowly raised out of the emulsion.
The bottle was capped for analysis later in the day. These saitples of oil
and water were analyzed in accordance with ASTM D-1796.
Test Procedures
Twenty-two tests were conducted by pumping oil into a 1.9-o flat-
bottom polypropylene barrel and adding bay water to produce mixtures of oil
and water ranging from 9? oil to 91? oil. The polypropylene barrel was
cylindrical, 1.2-m in diameter and 1.5-* high. The oil used was analyzed
and determined to contain less than 0.1? bottom solids and water, which was
assumed to be negligible. The height of the oil in the tank was measured
using a tape measure and sighting the height of the fluid through the
translucent tank walls and recorded. The test oil bad a viscosity of less
than 1,000 cSt at the ambient temperature. The height of the combined oil
and water in the tank was also measured and recorded. These two tank
soundings were used to determine the true oil percentage in the barrel.
The expected error in these two measurements yield an error in oil content
calculation of less than 1? oil.
-------
The fluids in the containers vere agitated slightly vith the mixer
and sampled as outlined. The percent oil in the container vas calculated
by nultlplying the vclurae after stripping by the percent oil In the grab
ftftsple as determined by laboratory analysis, dividing ty the initial
volume, and multiplying by 100.
Pesults
Of the ?;? tests, 20 vere considered valid. Two runs are discounted
because the height of the fluid renaining after stripping vas insufficient
for proper mixing, therefore the homogeneity of the fluid sampled is
suspect. A least-squares linear regression on both variables for all 22
points yields:
y = 1.00 x +1.61 r = 0.993
A second regression with the tvo data points discounted yields:
y = 0.97 x +P.03 r = 0.996
In both cases, y is the percent oil in the tank as analyzed, x is the knovn
percent oil by volume, and r is the correlation coefficient for the regres-
sion. The data are listed in Table 1 and graphed in Figure 1. The 20-
point results shoved that the oil content averaged less than 2% oil dif-
ferent from the knovn oil content. The 95* confidence licit shoved that
the difference should not be expected to exceed 3% oil (see Appendix B,
Table B-l).
The oil content determined should be expected to be higher than the
actual oil content. Both the regressive analysis and the student t-test
shov the bias. It is believed that the bias is introduced by tvo factors.
First, the electric mixer does not produce a truly homogeneous emulsion.
The container originally contained a higher oil-rich phase and the result-
ing emulsion also contained a higher oil content at the top than on the
bottom. The second cause is the insertion method of the 100-ml bottle.
The bottle first samples the higher layers of the container. If the bottle
is filled prior to reaching the bottom of the container, then the lover
level will not be sampled. The combination of these are systematic to
producing a higher-than-true oil determination. This is further supported
by the excluded tests. Observation shoved that the tank vas improperly
nixed. The fluids vere knovn to be nonhomogeneous. The resulting analysis
shoved much higher than knovn oil content.
The statistical analysis shovs that the points excluded are statisti-
cal outliers. This supports ignoring them in additional analysis. The
third data point is also a statistical outlier. It has not been discarded
because there vas no observed discrepancy in the test.
-------
K
*j
i
c
0
0
5
•
c
1
•
°
iUU -
90-
80-
70-
60^
50-
40-
30-
20-
10-
0,
(
'1
, ' i
I
X \
r I
0 a/ 1
Dx-"
V" !
x ;
/ i
X'o !
1.Q
' i
nx i
D' i
^ .
.*fl •
D''^ !
X i
y i
X 1
.* \
<
D /' i
0 S
a/
X }
/" i
x 1
r i \ \ \ '"i — ' i i i n r™ *1
) 20 «0 60 80 1C
KhomCI Content (X OS)
0 DtaofedDoto
Figure 1. Results of grab sample testing. Rote that the line drawn shows
perfect correlation and is not the results of curve fitting.
-------
TABI.E 1. GRAB SAMPLK QUALITY TKSTS (22 Tests)
Known
on
Content
U oil)
7U
6k
60
66
1.5
55
75
65
65
80
1.0
51
1.3
36
1.1
29
2U
9
16
9
11
90
Determined
on
Content
('• oil)
81-
73*
51*
67
1.6
56
77
87
66
82
1.0
50
37
39
31
23
13
19
11
15
90
Difference
(t oil)
-7
-9
6
_1
-1
-1
-2
-2
-1
-2
0
1
-1
-1
2
-2
1
-fa
-3
-2
-li
0
Absolute
Difference
(* oil)
7
9
6
j.
1
1
2
2
1
2
0
1
1
1
2
2
1
li
3
2
fa
0
Teets discounted due to improper mixing caused "by low height of
stripped fluids in the tank.
STRATIFIED SAMPLIJIG
The second means of sampling tanks containing both oil and vater at
OHKSETT is using the stratified sample thief known as the Johnson Sampler.
The device was developed by OHMSETT Test Director Michael G. Johnson to
avoid the intentional emulsiflcation of the test oil, as all of the test
oils are refurbished and reused at OEMSETT. This device (see Figure 2) was
constructed of three major segments: an inner core, an outer sheath, and a
handle. The inner core of the sampler was constructed so that each of the
annular segments captured 51 mm of sample height.
-------
Dclnils of a Stratified Sampler
(V) Sheath
£5) Supporting washers
(f) Wiper
Center Rod
[S) Spacer
© Handle
fj Bottom Stopper
Figure 2. Stratified sampler vith construction detail*
1C
-------
The samplers used at OKMSK7T for this study were each l.?-a long vith
a 3.2-iE handle. The size of the narpler was selected to r.etch the height
of the container to be caicpled. A Barpler that was too short would not
have provided a r^prc-nenthtive sas-ple of the entire tank. Ar. excessively
lorg parpler vc-ul oil content results tended to be high rather
than low as shown in Figure b.
Method 3, substituting the graduated cylinder analysis for the com-
plete ASTM analysis, yielded a consistently high oil content (see Figure 5)
because of the assumption that the oil phase Is pure oil. This method
11
-------
yielded results within lr?% of the actual oil content. The oil involved has
historically chcvn a tendency to retain soire water after ex;«osure, and 1C?
water voluae vould not be exceptional. Arbitrarily estlcating that the oil
phar.e contained 10$ vatpr will cut the statistical error in this analysis
in hair. The aaount of water retained by the oil should be expected to
vary frost oil to o53. Thic vould affect the above liirits, The addition of
the solvent to lower the apparent viscosity of the r.ixture combined vith
the enhanced gravity separation of the centrifuge ir. the A5TTM tethod mini-
mized this problem.
When Methods 2 and 3 were combined, the accuracy of the neasuren-ect
improved to within 6t of the actual value and gave a slightly lov average.
The scatter decreased and the correlation of the measured relative oil con-
centration with the known relative oil concentration in the container in-
proved (see Figure 6). It was felt, however, that this aay have been a
fortuitous coincidence rather than a predictable trend. These results vere
named as being the field data points and the technique as Method 1*. The
sane cautions applied to Method 3 should also apply to Method h.
The results of these tests are shown In Table 2 and Figures 3 through
6. The statistical analysis is given in Appendix B.
The complete analysis shows a bias. The oil content determined from
the sample taken with the stratified sample thief vas consistently lover
than the known oil content in the container. One possible explanation of
this bias Is that some portion of the oil originally in the sajsple does not
completely drain. This is particularly true of the sampler core. Although
long drain times were used, an oil costing undoubtedly remained. The con-
parative volume of the oil coating may account for the 1 to 3% bias in the
results.
12
-------
too
•0 -
r
j
70 -
•0 -
00-
40 -
30-
20 -
10-
40 M
Kno*n Ol OenCant (X OQ
100
Figure 3. Results of stratified sazpler testing using Method 1. Note
that the line drawn shows perfect correlation and Is not the
result of curve fitting.
100
00-
00-
70-
00
ao
4OH
1O-
o-
S
40
01
109
Figure l». Results of stratified sampler testing using Method 2. lote
that the line drawn shovs perfect correlation and is not the
result of curre fitting.
13
-------
100
•0
•0 -*
i
70-j
,J
r
40
l
90 -i
i
20 -
10 *•
O
a
s-"
20
• I T T—
40 00
^ KntHm Ot Ca>i»»iH (JJ Ol)
10O
Figure 5. Results of stratified saspler testing using Method 3. Rote
that the line drawn shovs perfect correlation and is rot the
result of curve fitting.
100-
70
3
I "
? 00-1
4O-
•0-
«0-
10-
toe
Figure 6. FesultB of stratified sampler testing using Method fc. Sote
that the line drawn shovs perfect correlation and is not the
result of curre fitting.
-------
ABLK
STPATIF1KD SAMPLES KKSULTS
Kr>cvn
Oil t
X
74
64
5U
67
1*6
56
77
87
66
82
1»0
50
44
37
39
31
23
13
19
11
15
90
1
Yl
73
60
60
66
46
54
71
63
65
79
40
48
45
35
39
29
21
9
18
10
10
91
Oil Percent
Method Number
'<•
y?
75
72
50
68
68
68
75
83
63
79
31.
55
50
40
40
34
20
12
20
12
8
91
3
Y3
87
82
77
82
54
68
99
99
83
97
51
74
56
50
52
40
33
20
25
17
19
100
4
Y4
74
70
n/a*
68
59
58
79
86
63,60+
81
41
45
48
32
45
30
19
6
11
4
6
98
Sample lost vben vind blew over graduated cylinder.
Two samples taken
Linear Regressions
Yl «
Y2 -
Y3 -
Y4 -
1.001
1.004
1.124
1.109
X -
x •»•
1.560,
1.273,
x + 6.594,
6.059,
X -
r
r
r
r
0.995
0.967
0.987
0.984
Method 1: Maximum stripping, complete ASTM analysis estimated error, 3% oil
Method 2: Ko stripping, complete ASTM analysis estimated error, 6% oil
Method 3: Maximum stripping, graduated cylinder analysis estimated error 15£ oil
Method 4: Ko stripping, graduated cylinder analysis estimated error 6% oil
15
-------
i-KTION 5
SAMPLI.VG FROM HvCVS
It has been occasionally desirable to saisple oil and vater flowing in
hoces or pipes when the nature of the flow is ur.Knovn. The fluids could be
flowing in turbulent or laainar regir.es, stratified laterally or radially,
or dispersed. It nay be at very high or low flow rates. This type of
sample has been of particular use when a reel-time analysis of the BkiEsser
performance was required or the skicaer evaluation had to be performed on
en operating skinnier in an actual field test at a spill of opportunity.
There have been three attempts to staple flowing oil and water at ORKSETT.
All of the attempts have addressed, to some degree, the nature of two phase
flow, a realistic sanple size, and ease of fabrication. All three of these
techniques have as their basis the interception and diversion of a po" tion
of the flow area to obtain a representative eaEple of the flowing fluids.
The first device tried was a holed sampler (perforated sampling
tube)* It was constructed using 6-ea (l/l»-lnch) pipe with holes of differ-
ing sizes drilled at regular intervals. These holes were chosen to te
standard twist drill sizes for ease of construction. The second and third
devices were a slotted tube and a pilot-shaped tube downstream from a
static mix?r on the high-pressure side of a pump.
UHMIXED FLOW
When OHMSETT conducted an offshore test of the Shell SOCK skinner
(see Lichte et al.. 198l), the discharge of a Tuthlll positive-displacement
pump was sampled for comparison with the samples taken from the collection
tanks. There was no official testing of the holed sampler in this program,
but the discrete values of recovery efficiency (percent oil) agreed within
about 15? of the values obtained using the tank soundings. Similar sam-
pling ports have been used at OHMSETT prior to the SOCK tests with similar
results.
The holed sampler was constructed using TO-OB (3-inch) scbedule-40
galvanized steel pipe as the bousing to natch the pump discharge size. The
sampling pipe was 6-om (lA-inch) pipe with holes drilled along its length.
The hole size and spacing was chosen based on readily available twist dri.il
bit sizes.
MIXED FLOW
A second method of dynamic sampling was tested in an effort to obtain
a more representative sample of oil and vater flowing together. The heart
-------
of the technique .ay In using a static mixt-r to produce a homogeneous dis-
persion and creating plug flew conditions vithln the pipe at the tianiplirg
port regardless of the initlhj flow conditions. It vac felt that the com-
pletely uniform dispersion in plug flow conditions would make the geoaetry
of the port Irrelevant to a erefit degree.
Two r.anpllr.p; ports vero constructed for testing in thece trials. The
first vhs shaped like a clecsic F'itot tube locked into the center of the
rain f lew-carry ing pipe. Thic vnr, generally designed around the specifica-
tions outlined by Underwriters' Labo atory specification 1501* but had
slight dimensional variations. The necond uaapling port was constructed
using a 6-nm (lA inch) pipe with a -ilot cut the width of a sav blade
(approximately 3 aan) along the long; '.udinal axis.
Static Mixer
The static mixer vas choser over dynsnic rcixers to eliminate extra
power requirements. The mixing energy was provided cy the puap. The unit
selected for this application was Model X020-Ol<0-l-003-22 manufactured by
Kocax Systems, Inc., Long Beach, California. The mixer was chosen to
provide a dispersion of 1000 nicron (l nun), or approximately ?OS of the
slot width. This size of the dispersion was chosen so that the oil would
coalesce readily after sampling to avoid any unnecessary eiaulsification of
the fluids. The static mixer used is shown in Figure 7.
Test Procedures
A Barnes Model 12CCG centrifugal pump vas fed an oil-water mixture
from two separate containers holding pure oil and water from Sandy Hook
Bay. The relative percentage of oil was set by the arbitrarily selected
positions on the spigot feed valves. The percent oil, by volume, varied
from 21? to 97? during the 12 comparison runs. The pump discharge vas
fitted with the three static mixers in series and the sampling ports. The
slotted tube vas followed immediately by the pitot tube. Samples were
taken at 60-second intervals, with the first 60 seconds of each test being
considered slop. The sampling ports vere constantly open. The fluid frm
the ports vas routed to a separate container when not filling the 200— ml
sample bottles.
Results
Four to ten samples vere taken for each test. The analysis of each
test vas averaged based on the assumption that there vas a constant pumping
rate during the test. The average oil content found in the discrete
samples vas then compared vith the percent oil determined by taking a grab
sample of the total puuup discharge , less the 60 seconds of slop.
The slotted-tube sampler and pltot-tube sampler each had an absolute
average difference from the grab sample analysis of approximately 3%. The
average algebraic difference from the grab sample analysis vas less than 1%
in each case. The absolute difference is indicative of the Inaccuracies of
the Ret hod, regardless of direction, i.e., a 1% high measurement or a It
17
-------
Figure 7. Longitudinal cutaway view of Komax static mixer.
-------
low measurement vculd give a 1% absolute difference. The algebraic dif-
ference incorporates the direction of the difference to reflect the fys-
tesdc bias of the nethcd. The 95? confidence interval shoved the sampling
and analysis procedures yielded oil contents within 9? of that found in the
grab sample. The data is plotted in Figures E and 9. The results are cor-
jared in Table 3.
TABU: 3. OIL PKR«.STAGLS FCR GRAB AND DYNAMIC SAMPLJS
Test
No.
A
B
1
2
3
U
5
6
7
8
9
10
Crab Sacple*!
* Oil
56
60
97
86
78
75
Ul
36
29
28
21
22
Pitot Tube
Average % Oil
52 (+M«2
58 (+2)
97 (+0)
8U (+2)
71 (+7)
68 (+7)
fco (+i)
1«2 (-6)
31 (-2)
32 (J.)
21 ( 0}
28 (-6)
Slot tube
Average ? Oil
52 (+M
59 (+1)
97 ( 0)
8U (+2)
73 (+5)
73 (+2)
k€ (-5)
1)1 (-5)
36 (-7)
31 (-3)
21 ( 0)
25 (-3)
Average absolute difference 3.W 3.2
Average algebraic difference 0.1»)S -0.8
•1 Post test analysis of the grab sample test reexilts shoved that these
measurements are within 3% oil of the actual oil content.
•2 Hvmbers shovn in parenthesis are the algebraic difference betveen grab
sample and dynamic samples
19
-------
8
I
™ ~™t H
100
OB Ow*ont
Figure 8. Results of slotted tube tests. Note that the line dravn shovs
perfect correlation and is no; the result of curve fitting.
100
100
<*«C
Figure 9. Results of pitot tube tests. Hote that the line dravn shovs per-
fect correlation and is not the result of curre fitting.
20
-------
BIBLIOGRAPHY
1. Perry, R.H. and Chilton, C.H. Chemical Engineer's Handbook, (5th ed).
McGraw-Hill Book Company, New York, New York, 1973.
2. Farlov, J.S. and Griffiths, R.A. "OKMSFTT Research Overview 1979-80."
In: Proceedings of the 1981 Oil Spill Conference, Atlanta, Georgia,
1981, pp. 655- 661.
3. Johnson, M.G. "The Stratified Sample Thief - A Device for Sampling
Unknown Fluids." In: Proceedings Management of Uncontrolled Waste
Sites, Washington, D.C., 1931.
k. Urban, R.W. and Graham, D.J. Performance Tests of Four Selected Oil
Spill Skimmers. E?A-600/2-T8-20i, U.S. Environmental Protection Agency,
Cincinnati, Ohio, 1978, 62 pp.
5. Lichte, H.W. et al. Tests of the Shell SOCK Skimmer Aboard USSS
Povhatan. U.S. Environmental Protection Agency, 1981.
6. "Standard Test Method for Water and Sediaent in Crude Oils and Fuel Oils
by Centrifuge". In: I960 Annual Book of ASTM Standards, Part 2l».
American Society of Testing and Materials, Philadelphia, PA, I960, pp 68-
71.
7. Mason, R. D., D. A. Lind. W. G. Marchal. Statistics: An Introduction.
Harcourt Brace Jovanovicb, Inc., Rev York, NY., 1933.
21
-------
APP:3: :X A
:T;~:-S KKVTPOV-'-MA?. PROTFCTIOS »/;:••?«• ry
-
The U.S. Environmental Protection Agency operates the Oil and Hazardous
Katerials Simulated Environmental Test Tank (OHKSETT) located In Leonardo, Kev
Jersey. This facility provides an environmentally safe place to conduct test-
ing and development of devices and techniques for the control and clean-up of
oil and hazardous material spills.
The primary feature of the facility is a pile-supported, concrete tank
vith a water surface 203 meters long by 20 meters vide and vith a vater depth
of 2.1» meters. The tank c*n be filled vith fresh or salt vater. The tank is
spanned by a bridge capable of exerting a horizontal force up to 151 kllonev-
tons vhile tovlng floating equipment at speeds to 3.3 meters/second (6.5
knots) for at least kO seconds. Slover speeds yield longer test runs. The
tovlng bridge is equipped to lay oil or hazardous Eaterials on the surface of
the vater several meters ahead of the device being tested, so that
reproducible thicknesses and vidths of the test slicks can be achieved vith
ninlmm interference by vind.
22
-------
The principal syotees of the tank Include a wave generator, a beach, and
ft filter system. The vove generator and absorber beach can produce regular
waves to 0.6 =«i-ter high and to ^5 aeters long, as veil as a series of 0.7
ceters high reflecting, complex vavec r.eant to clsulate the water surface of a
harbor. The tar* water is clarified by recirculeition through a UO cubic
ceter/hour diatOKaceous earth filter system to perslt full use of a sophisti-
cated undervater photography and video Imagery system ar.d to rer.ove the
hydrocarbons that enter the tank water as a result of testing. The tewing
bridge has a built-in oil barrier which is used to skim oil to the North end
of the tank for cleanup and recycling.
When the tank must be emptied for naintenance purposes, the entire water
volume of 9800 cubic meters is filtered and treated until it meets all ap-
plicable State and Federal water quality standards before being discharged.
Additional specialized treatment may be used whenever hazardous caterlals are
used for tests.
Testing at the facility is served from a 650 square seters building ad-
jacent to the tank. This building houses offices, a quality control
laboratory (vhich is very iarportant since test fluids and tank water are both
recycled), a small machine shop, and an equipment preparation area.
This government-owned, contractor-operated facility is available for
testing purposes on a cost-reimbursable basis. The operating contractor,
Mason fc Hanger-Silas Kason Co., Inc., provides a permanent staff of twenty
multi-disciplinary personnel. The U.S. Environmental Protection Agency
provides expertise in the area of spill control technology and overall project
direction.
For additional information, contact: Richard A. Griffiths, OHMSETT Project
Officer, U.S. Environmental Protection Agency, Hazardous Waste Engineering
Research Laboratory, Releases Control Branch, Edison, Kev Jersey 08837
Telephone: 201-321-6629.
23
-------
AWKKDIX B
STATISTICAL SUMMARY OK TKST RISULTS
Individual replicates of the tests were not perforned as part of this
program. There were sufficient Individual tests conducted to establish an
estimate of overall confidence l^iits across the oil content range
presuming that differences are .normally distributed.
2k
-------
TABLK B-l. STATISTICAL SUMMARY OK GRAB SAMP I JO T>ST RESULTS
OK
% Oil
7"*
6l»
60
66
"»5
55
75
65
65
80
to
51
k3
36
111
29
2k
9
16
9
11
90
OD
* Oil
81
73
5k
67
k6
56
77
87
66
82
kO
50
kk
37
39
31
23
13
19
11
15
90
Average
Standard Deviation
n
df
a
t
tails
nin
•ax
error range
OK-OD
% Oil
-7
-9
6
-1
-1
-1
-2
-2
-1
-2
0
1
-1
-1
2
-2
-1
-k
-3
-2
-It
0
-0.85
2.17
20
19
0.05
2.093
2
1.96
+0.06
±1.01
' CK-OD '
! ', on !
7"
9*
6
1
1
1
2
p
1
0
1
1
1
2
2
1
k
3
2
fa
0
1.85
I.k2
20
19
0.05
2.093
2
1.19
2.51
±0.66
OK Known oil content estimated at ±1? oil
OD Determined oil content
11 Data points excluded
25
-------
TABLK fc-?. STATISTICAL SUMMARY Of ETPATIFIKD SAXPI.K TEST PK
USIHG Mr.THOD 1
OK
% Oil
74
64
54
67
46
56
77
87
66
82
40
50
4U
37
37
31
23
13
19
11
15
90
05
% Oil
73
60
60
66
16
54
71
63
65
79
40
48
45
35
39
29
21
9
18
10
10
91
Average
Standard Deviation
n
df
a
t
nin
max
error range
OX-CD
% Oil
1
It
-6
1
0
2
6
it
1
3
0
2
-1
2
0
2
2
4
1
1
5
-1
1.50
2.45
22
21
0.05
2.080
0.42
2.58
±1.08
j rx-oD j
1 * Oil '
i
it
6
1
0
2
6
I
1
3
0
2
1
2
0
2
2
4
1
1
5
1
2.23
1.81
22
21
0.05
2.080
1.43
3.03
±0.80
OK known oil content estimated at ilf oil
OD determined oil content
-------
TABU: B-3. STATISTICAL SUMMARY OF STRATIFIED SAMPI.K TEST RESULTS
USING MKTKOD ?
OK
% Oil
7k
6k
5k
67
k6
56
77
87
66
82
ko
50
kk
37
39
31
23
13
19
11
15
90
OD
% Oil
75
72
50
68
68
68
75
83
63
79
3k
55
50
JiQ
•1 0
3k
20
12
20
12
8
91
Average
Standard Deviation
n
df
a
t
min
max
error range
OK-OD OK-OD |
% Oil % Oil
1
-1 1
-8 8
k k
-1 1
-22 2?
-12 1?
2 2
k k
3 3
3 3
6 6
-5 5
-6 6
-3 3
-1 1
-3 3
3 3
1 1
-1 1
-1 1
7 7
-1 1
-I.k5 kj»5
6.32 k.72
22 22
21 21
0.05 0.05
2.080 2.080
-k.26 2.36
1.35 6.55
+2.80 12.09
OK known oil content estinated at ±1% oil
OD determined oil content
27
-------
TABU: fe-i«. STATISTICAL SUMMARY OK STRATIFIES SAMPLE TT-ST RESULTS
USING KKTHGD 3
OK
% Oil
7l«
6k
5*
67
1<6
56
77
87
66
82
fco
50
U
37
39
31
23
13
19
11
15
90
CD OK -CD
* on %
87
82
77
62
5»»
68
99
99
83
97
51
7U
56
50
52
1»0
33
20
25
17
19
100
Average
Standard Deviation
n
df
a
t
•in
max
error range
Oil
-13
-18
-23
-15
- 8
-12
-22
-12
-17
-15
-11
-2U
-12
-13
-13
- 9
-10
- 7
- 6
- 6
- fa
-10
-12.73
5.35
22
21
0.05
2.080
-15.10
-10.35
±2.37
' OK-OD ;
| * Oil i
13
18
23
15
8
12
22
12
17
15
11
2U
12
13
13
11
10
7
6
6
b
10
12.73
5.35
22
21
0.05
2.080
10.35
15.10
±2.37
OK knovn oil content estimated at ±1$ oil
OD determined oil content
-------
T JLK B-5. STATISTICAL EfMMAKY OK STHATIK1KD SAMPLE TEST RESULTS
USING METHOD 1*
OK
? Oil
Tfc
6U
66
67
46
56
77
87
66
82
1*0
50
U
37
39
31
23
13
19
11
15
90
01)
% Oil
7*
70
60
68
59
58
79
86
63
81
1.1
1*5
1.8
32
1>5
30
19
6
11
It
6
98
Average
Standard Deviation
n
df
a
t
•in
max
error range
OK-OD
% Oil
0
-6
6
-1
-13
-2
-2
1
3
1
-1
5
-1*
5
-6
1
I*
7
8
7
9
-8
0.6*
5.56
22
21
0.05
2.080
-1.82
3.10
'2.1.6
i CK-CD ,
| * Oil i
0
6
6
1
13
2
2
1
3
1
1
5
It
5
6
1
1;
7
8
7
9
8
1..55
3.26
22
?1
0.05
2.080
3.10
5.99
OK known oil content estimated at +1? oil
OD determined oil content
29
-------
TABLK h-6. STATISTICAL NUMMARY OF PITOT-TUPE SAMPLING POPT
OK
I Oil
56
60
97
86
78
75
1»1
36
29
2.,
21
22
OD
J Cll
56
58
97
eu
71
68
kO
1.2
31
32
Pi
28
Average
Standard Deviation
D
a
t
Bin
nax
error range
OK-OD OK
% Oil %
4
2
0
2
7
7
1
-6
-2
-*
0
-6
O.J»2
J».21
12
0.05
2.201
-2.26
3.10
+2.1*6
-OD
Oil
h
2
0
2
7
7
1
6
2
U
0
6
3.^2
2.50
12
0.05
2.201
1.83
5.01
41.59
OK known oil content estimated at + lljt oil
OD determined oil content
30
-------
TABI.K B-7. STATISTICAL SUMMARY OK SLOTTED TWK SAMPLING POP7
OK
% Oil
56
60
97
86
78
75
10
36
29
28
21
22
OD
t Oil
52
59
97
81.
73
73
k6
hi
36
31
21
25
Average
Standard Deviation
a
t
Kin
max
error range
OK-OD
% Oil
It
1
0
2
5
2
-5
-5
-7
-3
0
-3
-0.75
3.63
0.05
2.201
-3.07
1.57
2.32
OK-OD 1
I Cilj
l»
1
0
2
5
2
5
5
7
3
0
3
3.08
2.10
0.05
2.201
1.75
lt.U2
1.33
OK known oil content estimated at + ±ljt oil
OD determined oil content
31
------- |