Document 6534345

SHRP-P-636
Portland Cement Concrete
Core Proficiency Sample Program
G.W. Steele
Steele Engineering, Inc.
Tornado, West Virginia
Strategic
Highway Research Program
National Research Council
Washington, DC 1993
SHRP-P-636
Contract P-001
Program Manager: Neil Hawks
Project Manager: Cheryl Richter
Production Editor: Elizabeth Lucks
Program Area Secretary:
Cindy Baker
May 1993
key words:
compressive strength
core sample
laboratory accreditation
poisson s ratio
portland cement concrete
static modulus
tensile strength
Strategic Highway Research Program
National Academy of Sciences
2101 Constitution Avenue N.W.
Washington, DC 20418
(202) 334-3774
The publication of this report does not necessarily indicate approval or endorsement of the fmdings, opinions,
conclusions, or recommendations
either inferred or specifically expressed herein by the National Academy of
Sciences, the United States Government, or the American Association of State Highway and Transportation
Officials or its member states.
© 1993 National
350/NAP/593
Academy
of Sciences
•
Acknowledgments
The research described herein was supported by the Strategic Highway Research Program
(SHRP). SHRP is a unit of the National Research Council that was authorized by section
128 of the Surface Transportation and Uniform Relocation Assistance Act of 1987.
iii
Contents
Summary of Research ...............................................
1
Appendix I, Correspondence dated 11/2/89 from Iowa
describing concrete mixes ..........................................
3
Appendix II, Correspondence dated 11/8/89 from Iowa
to all participating laboratories .......................................
6
Appendix III, Preliminary report dated 2/90 from Iowa
concerning results of PCC core testin .................................
14
Appendix IV, Correspondence dated 2/27/90 to
statistician concerning analysis of test data .............................
23
Appendix V, Correspondence dated 10/5/90 to
statistician concerning analysis of test data .............................
25
Appendix VI, Memorandum data 12/21/90 from
statistician on variance component analysis
(contains matrix of all test data) .....................................
28
Appendix VII, Memorandum dated 3/19/91 from
statistician on precision statements ...................................
52
Appendix VIII, Documents relating to the authorization
of SHRP laboratory to proceed with testing of LTPP cores ..................
57
Appendix IX, AASHTO/ASTM format precision statements ....................
60
v
-
Abstract
This document provides a description
Proficiency
Sample Program.
and process
of the Portland
Cement
Concrete
Core
vii
Executive
Summary
One element of Quality Assurance (QA) for laboratory testing that was deemed to be of
key importance by SHRP, as a result of Expert Task Group (ETG) recommendations, is the
American Association of State Highway and Transportation Officials (AASHTO)
accreditation program (AAP) for laboratories. All laboratories providing long-term
pavement performance (LTPP) testing services were required to be accredited by AAP.
Most of the laboratory tests on LTPP field samples were addressed by the AAP, which
includes on-site inspections of equipment and procedures, and participation in applicable
proficiency sample series. However, a few critical tests in the SHRP LTPP studies were
not fully addressed. After extensive consultation and careful study, it was determined that
supplemental programs should be designed to provide assurance that quality test data would
be obtained by using approaches similar to those provided by AAP for other tests.
One supplemental program approved for implementation was the Portland Cement Concrete
(PCC) Core Proficiency Sample Program. The program was designed to provide precision
data concerning the static modulus of elasticity, poisson's ratio, splitting tensile strength,
and compressive strength.
The PCC core program was modeled after the familiar Cement and Concrete Reference
laboratory (CCRL) proficiency sample programs at the National Institute of Standards and
Technology (NIST). The core samples were prepared and distributed to participants, the
raw test data was collected and entered into a matrix for analysis, and an interim report to
participants was distributed for SHRP by the Iowa Department of Transportation's Office of
Materials.
Two different PCC mixes were prepared by the Iowa Materials Laboratory and cast into
forms that would allow 4 in. diameter by approximately 9 in. length cores to be obtained
for testing. All cores were taken, cured and shipped in accordance with standard practice.
Twelve cores were sent to each participating laboratory for testing at age 56 days, six from
each mix.
Instructions to the laboratories directed that two cores from each mix be tested in
compression, two from each mix be tested for splitting tensile strength, and two be tested
for static modulus of elasticity and poisson's ratio. A single operator was to perform the
same test on both samples. Different operators could be used for different tests. Explicit
directions were included concerning procedures to be followed for each test.
Raw test data were returned to Iowa for matrix entry and preliminary reports to participants.
Subsequently, preliminary scatter dia:zrams an5 individualized tables of results were
distributed to the cooperating laborat_:_ries.
A 3 1_ in. floppy disk containing the raw test data along with the core sample identification
key was prepared by the Iowa Materials Office and forwarded to the SHRP Quality
Assurance Engineer when all data had been received. The floppy disk was then transmitted
to the SHRP Statistician for final analysis and determination of test precision.
The statistician's initial report indicated that potential outliers existed in the data which
should be investigated. An investigation was conducted by the Quality Assurance Engineer.
It was determined that the outliers should be set aside in the final analysis.
The SHRP authorization to proceed with tests of LTPP field samples was issued based on
results of the proficiency sample tests.
Precision statements were derived and drafted in the standard AASHTO/ASTM
format for use by standards writing committees as they deem appropriate.
Appendices to this report contain the complete set of supporting documents for this
program as listed in the table of contents.
Thirteen (13) laboratories participated in this program. Each participant has made a
substantial contribution to the successful completion of SHRP research in the LTPP
program. Participants were:
Florida Department of Transportation, Gainesville, Florida
Iowa Department of Transportation, Ames, Iowa
Federal Highway Administration, Denver, Colorado
California Department of Transportation, Sacramento, California
West Virginia Department of Transportation, Charleston, West Virginia
Law Engineering, Atlanta, Georgia
National Aggregates Association/National Ready Mix Concrete Association,
Silver Spring, Maryland
Bureau of Reclamation, Denver, Colorado
Waterways Experiment Station, Vicksburg, Mississippi
Concrete Materials and Technical Services, Skokie, Illinois
CANMET, Ottawa, Ontario, Canada
Wiss, Janey and Eisner, Northbrook, Illinois
New York Department of Transportation, Albany, New York
APPENDIXI
,,9
l
,
,"
_.
Form 000032
3-86 H-8687
IOWADEPARTMENTOFTRANSPORTATION
800 Lincoln Way, Am,
es, Iowa 50010
515/239-1649
Ref. No. 435.24
November 2, 1989
Garland W. Steele
President,Steie Engineering,Inc.
Box 173
Tornado, West Virginia 25Z02
Dear Garland:
Attached is informationon the "Precisionand Accuracy
Determinationfor P.C. Concrete Core Testing'. The concrete
Iowa's 6-4 mix is c_monly used for low traffic
county roads.
The
mix on
is October
used on19
primary
and on
interstate
paving.
was C-4
poured
and cored
October 26
and 27.
forwill
yourbe
review
in you
the next
reek
two. Please lettesting"
me know
We
sending
a draft
ofor
"instructionsfor
if you have any questions about our plans.
Sincerely,
Cement & Concrete
Kevin Jones
Engineer
C_
KJ:sh
Attach.
co: B. Brown
4
L_
'_
v _-_-v-I"
MLR-89-11
Concrete Pour
October 19, 1989
8:00 a.m.
Special InvestigationsLab
Materials
Cement - NorthwesternType I
Coarse Aggregate- Martin Marietta Ferguson
Fine Aggregate- Halletts Materials,Ames
Air EntrainingAgent - Protex AES
Mix
No.
B-4
C-4
Cement
Ibs.
Coarse
Agg.
Ibs.
Fine
Agg.
Ibs.
492
1558
1558
5.33
624
1499
1495
5.75
Test Results
Mix
No.
W/C
Slump
Air
B-4
.60
2.0"
4.8%
C-4
.49
1.5"
4.9%
?
Air Entraining
Agent
Oz.
APPENDIX II
•
o
J
°
•
Materials
Office
lowa Department
of Transportation
Cement and Concrete
Laboratory
800 Lincoln Way
Ames, Iowa 50010
(515) 239-1649
Sponsored by
Strategic Highway
Research Program
(SHRP)
November
TO:
8, 1989
Participating
SUBJECT:
P.C.C.
Laboratories
Core Samples for Testing
The samples for precision and accuracy determination for P.C.C.
core testing are being shipped to you the week of November 27th.
Shipment will consist of two boxes each containing six p.c.c.
cores.
The samples are packaged in plastic bags to retain the
moisture and are identified as No. 1 and No. 2. The core samples
upon arrival should be cured in lime water as per ASTM C-192
until the test date of December 14, 1989.
If you do not receive the samples or if the samples you receive
are seriously damaged, notify us immediately and the necessary
replacement will be sent.
The tests should be conducted on December 14 if possible and the
test results sent to me. After receiving the test results from
each participating laboratory, the results will be sent to AMRL
for analysis.
The results of the AMRL analysis will be sent to
each participating laboratory.
Instructions for testing and the necessary sheets for reporting
the test results are enclosed.
Please read these instructions
carefully before proceeding with the tests.
7
Sincerely,
Kevin B. Jones
Cement & Concrete Engineer
Materials Office
Iowa Department of Transportation
Enclosures
I
IOWA DEPARTMENT OF TRANSPORTATION
P.C.C. Core Samples for Testing
A total of twelve P.C.C. Core samples will be sent to each
participating laboratory.
Six cores will be from each mix.
Two
concrete cores from each mix will be tested by each participating
laboratory for (a} compressive strength, (b} splitting tensile
strength and (c} static modulus of elasticity.
It is recommended
that one operator make the same test on both samples
APPLICABLE
DOCUMENTS
AASHTO
T22-88I
Compressive
Specimens
AASHTO
T24-86
Obtaining and Testing
Beams of Concrete
AASHTO
T198-88I
Splitting Tensile Strength
Concrete Specimens
AASHTO
T67-85
Load Verification
AASHTO
T231-87I
Capping
ASTM C469-87
Strength
of Cylindrical
Drilled
and Sawed
of Cylindrical
of Testing
Cylindrical
Cores
Concrete
Concrete
Machine
Specimens
Test Method for Static Modulus of Elasticity
and Poisson's Ratio of Concrete in Compression
INSTRUCTIONS
-
COMPRESSIVE
STRENGTH
TESTING
This test shall be conducted
following modifications:
-
as per AASHTO
T22-881
except
for the
-
The diameter (D} of the test specimen shall be determined to
the nearest 0.01 inch by averaging two diameters measured by
a caliper at right angles to each other at about the midheight of the specimen.
-
Measure the length of specimen before capping (LO) and after
capping (L) to the nearest 0.1 inch prior to testing.
The
length shall be determined by averaging four measurements
equally spaced around the specimen.
The length of the
specimen when capped, shall be as nearly as practicable twice
its diameter.
Section 6.2 of AASHTO T24-86 for specimen end
preparation shall be followed.
-
Test specimens shall be sawed on both the top and bottom ends
of the core to achieve the desired L/D ratio of approximately
2.00.
(Use the length of the capped specimen to compute the
L/D ratio}.
-
-
AASHTO T231-87I
procedure
for capping hardened concrete
specimens shall
be followed
for capping both ends of
specimen.
Neither
end of test specimens when tested
shall
depart
from perpendicularity
to the axis by more than 0,5 o
(equivalent
to 1/8 inch in 12 inches).
-
Type of fracture
AASHTO T22-88I).
-
SPLITTING
should be reported
TENSILE
STRENGTH
(Refer to Fig. 2 of
TESTING
-
This test shall be carried out in accordance
except for the following modifications:
with AASHTO
T198-88I
-
Measure the diameter (D) and the length (L) of the test
specimens to the nearest 0.01 inch following section 5.2 of
AASHTO T198-88I.
-
The test specimen shall be sawed or ground to achieve a
uniform length, and the end surfaces shall conform to section
6.2 of AASHTO T24-86.
The L/D ratio shall be nearly as
possible to 2. Test specimens shall be trimmed as not to
exceed 1-1/4 inch at the bottom of the specimen and up to 1
inch at the top of the specimen.
(The finished ends are not
to be capped_.
-
Type of fracture
AASHTO T22-88I).
-
STATIC MODULUS
should be reported
OF ELASTICITY
(Refer to Fig. 2 of
TESTING
This test shall be performed in accordance
except for the following modifications:
with ASTM C469-87
-
The diameter (D) and the length (L) of the test specimen
shall be determined in the same manner as described for
compressive strength testing.
-
L/D ratio of the specimen shall be determined in the same
manner as described for compressive strength testing.
-
Ends of the test specimen
87I.
-
The test specimen shall be weighed prior to testing and the
weight recorded to the nearest gram.
The unit weight (CW)
shall be calculated to the nearest 1 pcf by dividing the
weight of the specimen by its volume using the dimensions
determined above.
-
Deformation should be measured by a linear variable
differential transformer (LVDT).
•
shall be capped
as per AASHTO
T231-
-
The
value of
elasticity
testing
shall
correspond
to S _0._orof the
the modulus
ultimateof load
determined
in the
compressive strength testing.
-
Calculate the modulus of elasticity to the nearest 50,000
and poisson's ratio to the nearest 0.01.
-
REPORTING
-
The test results shall be reported on the attached forms and
returned to the Iowa Department of Transportation
as soon as
possible.
/0
psi
_J
E
E
o
¢.J
QJ
4-- S..i.a ,,,.v"
_,J U I.i..
I°c-
l I/I ,IJ
i ClJ O_.---_._-.4
O
0C
QJ;
0,I_
_
N
0
C,J
=
•
.e-"OIL
4-_
e-I---
(.0_"
I-- LI-
_
l'-- OC
__
ev"
O_ I':-I
¢.j
e-
e-
_
U') 4.-_ _b,.,-.,,- •
4-J
b0
O
r._
¢.Jt_
•"
o5
I--.
ca_
¢/)
O-
O "
r'_ .,.-
O..
4-z_J -,---.._--
(M
C_.
c
_-
_
!-,-
el)cM
o
_
•,-- :,,3 e--- e_O
O
4-_
e-" S.
IM -_'--'
e- ---',. ¢"
_
e-
E
_¢4-
4.a
_
,._
._Ie,_(.J
O
6_._
_
Q.;
G;
_
"u'_
",--
P:£_
_J
S.-
E
r'_
rO
-e-- _
_
_
_
O
O
_
%0
o
_.
s--
ro
_
4.J
_
_
I--
I_
_::
OJ 4-_-,-e'_.r-v" c" %.
.--I
f--
,,--,
._
o _J _J 4.,_>,_CD
r'3 f__-,.- e" _ O
c_ oo4.a
(-)O')
I--.
_
II
z
z
X3
4.a
QJ
E
E
o
u
o 0
c'- S.. ;_
Gu¢._
__..
o
C_
E
uO
m
0
(-_
O
_
"_
l--
he"
e_
e-
i.-b-
_
bJ_.
e- I/'l
.,_ _.__ ....,
4-:
3
g
d
o
_
N
E
•,-.
-_ 0,
O.-J
,_
0
---
_"
_,n
o°"_"
$.
_
0
• 4_ ---,
tJ O'_
..J e"
QJ c" ._...
U
e'_Q;
_"
_
o__o
"35--,'- e--
OJ
,-.,--
I--
_.,---_
QJ .---_"
.,-
"O
•,--_,_,-- e_ _.;O
_ _j c-_
_0.....e-__
_
.,-
E
S.-
(I;
4-_
_.
_-
0
(1;
g._
_" _" _-
.J
--
OOE
_" ¢,J _" a,-,.,¢:;0¢=E
(1;
0-
E
_J
¢n
__¢
•
L.
=.-
o
O
0,1 e-
_) _4.J (1,t
S- S- 0"_
.l-m
4..m_-tO _
I
>-
u..
_
P-U(23
I.=JI--,.
I'-- C_
O _ "--_"
-,--
_
I'--
_ 0 v_ ._,.
O
_-" I.=J
_"
Lt,J"---,"
_
°
o_.,.
g
•,-
_0
S.
_nk_
_¢_-
_ ;Z= ¢z_
E ¢"
•r- .l_
0J
e"
_
¢-
e- .j
0
_
_-
"_
_
.,
¢'_
E_ C3 ¢"_-_"
_
¢,J
""
_
E
"o
(M 4-_ .,-- C3 -r_ ¢" £,..t
_
_
I_
=0
4._
O
_
r,.0
¢_
0-
o_
(13
II
qJ
I--
e..
_O_:
",-- _,--o.(J
0
s-w_oo_
,,-
_" (..3 _" _--_O0 _
0
_J
I--
I--
_
_
_-_
_
_
z
_,
¢_
(I)
o_
RESULTSOF
PCC CORETESTING
MLR-89-11
FEBRUARY1990
.
IOWA DEPARTMENT OF TRANSPORTATION
HIGHWAY DIVISION
OFFICE OF MATERIALS
AMES, IOWA 50010
(515) 2,59-1649
/5
Summary
of Results
- General
In most cases, averages,
standard
deviations,
and coefficients
of
variation
are given with all results
reported,
and then with one
or more outlying
results
omitted.
In some cases two or more recalculations,
with laboratories
omitted,
have been done for the
same test; and in these cases, all of the laboratories
omitted
in
previous
recalculations
are also omitted
in subsequent
ones.
Results
omitted
are values which are more than three standard
deviations
from the mean of one or both samples.
In most cases,
elimination
of these outlying
results
has little effect
on the
average,
but may have a more pronounced
effect on the standard
deviation
and coefficient
of variation.
Scatter
A set
of
scatter
diagrams
Diagrams
is supplied
with
the
report.
The manner
of preparing
scatter diagrams,
and their interpretation,
is described
in the Crandall
and Blaine
paper, published
in the 1959 ASTM Proceedings.
Each of the laboratories
will
receive
a complete
set of diagrams.
In those instances
where the
laboratory
was unable
to report
results,
diagrams
will still be
furnished.
A scatter
diagram
is plotted
for each test method by taking
the
results
received
from each laboratory
and plotting
the value for
the odd numbered
samples
on the X, or horizontal
axis, against
the value for the even numbered
samples
on the Y, or vertical
axis.
To locate your point, just plot as you would when plotting
any scatter
diagram.
The vertical
and horizontal
lines of
dashes,
which divide
the diagrams
into samples
respectively.
The
first
line of print
under the diagram
includes
the test number,
as given on the data sheet, the test title,
and the number
of
data points
on the diagrams.
The number
of plotted
points
may
not agree with the total number of data pairs
included
in the
analysis
because
a few points may be off the diagram,
and some
points
may represent
several
data pairs,
which
are identical.
Laboratories
whose points
are off the diagram
will have a rating
of + 1 for that particular
test.
As described
in Crandall
and Blaine,
a tight circular
pattern
of
points
around the intersection
of the median
lines is the ideal
situation.
A stretching
out of the pattern
into the first
(upper
right}
and third (lower
left) quadrants
indicates
some kind of
bias or tendency
for laboratories
to get high or low results
on
both samples.
Examination
of the scatter
diagrams
indicates
strong evidence
of bias on almost all tests.
/6
,
Each laboratory receives an individualized Table of Results.
The
Table of Results shows the test number, test title and the
reporting unit in the first three columns.
Thereafter, it lists
in order, the laboratory's results for the odd and even numbered
samples, overall averages for the odd and even numbered samples
and the laboratory's ratings for the odd and even samples.
(See
reverse for an explanation of the scatter diagrams.}
The laboratory ratings, shown in the Table of Results for the
individual laboratory, were determined in the manner described by
Crandall and Blaine, using a rating scale of 1 to 5 instead of 0
to 4. The ratings nave no valid standing beyond indicating the
difference between the individual laboratory result and the
average for a particular test.
The table which follows, details
ratings and the averages.
Ratings
5
4
3
2
1
the relationship
Range (Number of
Standard Deviations}
Less than i
i to 1.5
1.5 to 2
2 to 2.5
Greater than 2.5
between the
Number (Per 1000
of Laboratories)
that might have
greater variation
317
134
45
12
--
The sign of the rating merely indicates whether the result
reported was greated or less than the average obtained.
In cases where some of the laboratories' results are eliminated,
averages, standard deviations, coefficients of variation, and the
ratings of the other laboratories' results, are recalculated,
using the data remaining after the elimination.
• t
_
q
_
Lo
•
,--i
oo
•
•
N
/g
COMPRESSIVE
STRENGTH
RESULTS
CONCRETE
SAMPLE
NOS.1 & 2
i
1000
_
;
6,800"
O iO
:
="6,600•
O
o i
o o
q,D
in,
0
• ............................
0
..O
..........................................o..-._
U:
_ 6,400.
a2
:
X
"
°6,200.
i
_5,
i
•_I' '8001
'
I
'
4600
TESTNO. 1
'
I "
i
,
I
'
'
' I' '
'
_
"
'
I
'
'
'
I
i
4800 5000 5200 5400 5600 5800
SAMPLE
NO.I : COMP.
STR-56D
PSI
COMPSTR-56D
SAMPLENO. 1 AVO 5390.8
,
6000
13 POINTS
S.D. 265.2 C.V. 4..92
SAMPLENO. 2 AVG 64.66.2 S.D. 270.3
,
C.V. 4..18
SPLITTING
TENSILE
STRENGTH
RESULTS
CONCRETE
SAMPLE
NOS.1 &:2
00
'
:
0
7_'
_7_,
i
!
:
6_-
!o
:0
_6_,
o
o
:o
0
!
o
•
eeeeoeeeeeee_eooeeeooeeeeeeeoeee_egeele_oeeae_ooeoeJeoeooee
0
5_.
.
ooe_eoeeoe
•
o
•
eee
•
.et
eee
•
::
z
i
4_.
o
i.
o
,,,,
300
i,,,,
350
i
i,,,,
400
i,,,,
4_
i,
,i,
500
I,,,',
5_
I,,,'
6_
I''
6_
''
7_
SkMPLE
NO.1: _NSILE
Sm-_D _1
TEST NO. 2
TENSILESTR-56D
12 POINTS
SAMPLE NO. 1 AVG 521.2
S.D. 63.3
C.V.
12.15
SAMPLE NO. 2 AVG 589.5
S.D. 91.7
C.V.
15.55
2.0
STATIC
MODULUS
RESULTS
CONCRETE
SAMPLE
NOS.1 & 2
.
5.50
0
O
O
gSO0.
_
::
•
'_
X
:
:.
,_ 4.50"
...........
o..
U"a
•
Go
eo
o
0
0
i
•
•oQ
Qao•.
8 _i .
•...
•
..
egeeloeeoeQuJ
•
•
QeeeeOOOlO•OOQe
•QOneQQD..ee**
_J
4.00..
Oql,
•
c,_
0
_
0
:
•
_. 3.50'
a.
300
....
3.00
'
I
"
'
"
'
I
"
"'
'
'
I
'
"
'
"
I
'
'
"
3.50
4.00
4.50
5.00
SAMPLE
NO.I : MODULUS-58D
151X 1,000,000
TESTNO.3
STATICMOD-56D
SAMPLE NO. 1 AVG 4.145
S.D. 0.459
5.50
10 POINTS
C.V. 11.08
SAMPLE
NO.2 AVG 4.400 S.D. 0.402 C.V. 9.15
2./
"
POISSON'S
RATIO
RESULTS
•CONCRETE
SAMPLE
NOS.I& 2
0.260-
o
O
0.240-
o
O
='=0.220-
o
Z
O
¢/'}
i.,.eee..ooeo,oo,e,.,,e,.,..eo...o.ool,e,
oo.t.
o,q-.eo.o..,eoot'o.
'''''''eoo=eo''°''"
5 0.200Oo
,5 0.180.
o
_J
..,I
Q.
_ 0.160"
o
0.140.
C,
0.120 . , • , . , ' , ' , ' , ' ,
0.120 0.1400.1600.180 0.200 0.220 0.2400.260
SAMPLE
NO.I:POISSON'S
RATIO
TESTNO. 4.
POISSON'SRATIO-56D
7 POINTS
SAMPLENO. 1 AVG 0.207
S.D. 0.039
C.V. 18.74
SAMPLENO. 2 AVG 0.206
S.D. 0.04.6
C.V. 22.33
,ZZ.
APPENDIX IV
February
27,
1990
Robin
High
TRDF
2602
Dellana
Lane
Austin,
TX 78746
Dear
Robin:
Subject:
SHRP
Portland
Program.
Cement
Concrete
Core
Proficiency
Sample
Enclosed
is a floppy
disk
with
the raw
data gathered
from
the
subject
program.
Three
pairs
of cores
for each
of two mixes
were
distributed
to
participating
laboratories.
One
pair
from
each
mix was
to be
tested
in
compression,
one in split
tension,
and
one pair
in static
modulus
including
poisson's
ratio.
Replicate
sets of values
for each
of the
four properties
for each
mix were
obtained.
However,
all
values
were
not determined
by
all
laboratories.
Therefore,
the degrees
of freedom
available
differs
for each
of the properties
to be evaluated.
Please
review
the
data
and
I
will
be
in
contact
with
concerning
the analyses
you
feel would
be most
appropriate.
sure that
Virgil
could
provide
some
valuable
guidance
also.
Yours
very
truly
Garland
W.
President,
Steele,
Steele
P.E.
Engineering,
enclosure:
floppy
disk
Box
173 • Tornado,
West
Virginia
Inc.
25202
• Tele
(304)
727-8719
you
I am
APPENDIXV
Z,E'
October
Robin
TRDF
2602
1990
High
Dellana
Austin,
Dear
5,
TX
Lane
78746
Robin:
Subject:
SHRP
Portland
Program.
Cement
Concrete
Core
Proficiency
Sample
The
report,
dated
5/21/90,
on components
of variance
analysis
of
data
gathered
in the subject
program
has
been
carefully
reviewed.
As suggested
on
page
4,
the
two
possible
outlier
values
in
compressive
strength
data
have
been
investigated.
Both
the
laboratories
involved
in the
performance
of the tests
and the
laboratory
responsible
for
preparation
of
the
samples
were
queried
concerning
possible
assignable
causes
that
may
have
affected
the two values.
The final
conclusions
are based
on the
recollections
and
comments
of
the participating
parties.
Each
felt that
the values
were
the
result
of
an
assignable
cause,
however
there
was
not
agreement
(and
there
was no objective
information)
that
would
lead
to the
positive
identification
of
same.
The
first
possible
outlier
(page
2, figure
2) in the
compressive
strength
data
was
likely
the
result
of
misidentification
of
a
specimen
or
cross-identification
of two
specimens.
The
second
possible
outlier
in
the compressive
strength
data
was
likely
the result
of an error
in recording
the
dial
reading
or mis-identification
of specimens.
Based
upon
the above,
it is recommended
that
the data
be retained
and
placed
in the
SHRP
data
bank.
However,
the two values
in
question
should
be
identified
in
the
data
bank
as probable
outliers,
and
the components
of variance
analysis
placed
in the
data
bank and
used
by SHRP
should
exclude
these
two values.
The
suggestions
on
page
I0
of
the
report
concerning
tensile
strength
values
were
of
greater
concern,
since
it was
felt
that
their
implementation
would
be quite
time
consuming
and relatively
costly.
Careful
review
of the presentation
of data
in the report
then
revealed
that
replicate
columns
2
and
3
under
tensile
strength
in
figure
2
on
page
2
have
been
transposed.
Investigation
of the original
worksheets
verifies
that
column
2
data
should
be moved
to column
3 and
column
3
data
should
be
moved
to
results.
a problem
strength).
specimens
column
2.
This
will
substantially
change
the
end
It should
be noted
that
lab I will
likely
still
present
(ie
reporting
mix
1
strength
greater
than
mix
2
However,
the
possible
cross-identification
of
noted
previously
would
involve
one
or
more
of these
cores.
Box 173.Tomado,
Z_
WcstVirgmia
25202-Tclc.(304)727-8719
The
aforementioned
review
transposition
of values
and
Poisson's
ratio
should
be interchanged
It is
glitch
also
probable
that
these
when
the data
was
transpositions
transferred
Please
call
after
completion
of
strength,
modulus,
and
Poisson's
determine
the appropriate
course
of
obtained
therefrom.
Let
Yours
me
know
very
Garland
W.
President,
cc:
Adrian
if
you
have
any
questions
truly
Steele,
Steele
Pelzner
revealed
that
the
same
has
ocurred
for the modulus
of elasticity
in
figure
2
on page
2.
Columns
2 and
in each
case.
P.E.
Engineering,
Inc.
resulted
from
a
from one system
to
the
ration
action
3
computer
another.
analysis
for
tensile
at which
time
we can
based
upon
the
results
concerning
the
above
items.
..,. ..
TRD
.
TECH MEMO:
AU-127
AUTHORS:
Robin High'l_
DISTRIBUTION:
Garland
SUBJECT:
Variance
Component
Concrete
Core Proficiency
A proficiency
variance
cement
engineering
modulus
Steele,
cores.
material
determined
by the indirect
compression
data
and
test.
Poisson's
analysis
proficiency
observed
between
phase
program
various
were
laboratory
an
strength,
testing
tensile
for two mixes
The
tensile
estimated
strength,
(medium
compressive
and high
strength
strength,
from
an
test program
was
modulus
unconfined
as well
as the
in Figure i.
cement
analysis
It.will
properties
the
of four
Portland
performs
to establish
involved the determination
are presented
the
Cement
Portland
ratio)
ratio
of
first
21, 1990
of
laboratories.
in the four materials.
and within
testing
of the proficiency
tests
P-O01
in SHRP-LTPP
tension test, while
The scope
FILE:
of SHRP Portland
(compressive
different
from the material
The
with
and Poisson's
thirteen
December
Sample Program
was undertaken
properties
at
elasticity,
Analysis
The test program
of elasticity,
DATE:
Bill Hadley
assclated
strength)
of
_
test program
components
concrete
_/
Lf
concrete
of
the
then assess
variations
core
sample
variance
(ANOVA)
the magnitudes
of the
(o2L_ and 0 2 respectively)
for each property.
Two
replicate
laboratories
did
not
estimate
cells
another,
are
the
modulus
the
summarizing
of
capability
design
were
the various
available
2602 Della_
to
cores
conduct
were
the
and Polsson's
not
material
estimate
and Polsson's
Lane
to
of elasticity
this is not a major
elasticity
concrete
for the two mixes and four material
have
in
sets
filled
properties
limitation.
the
(see
sources
provided
tests.
test
Several
each
required
therefore,
Figure
i).
will not be
fewer
of variability
• Telepho.4
512 1 327-4211 • Pax
the
compared
degrees
the
ratio than for the compressive and tensile
Austin, Texu
the
to
some of the
Since
for
of
laboratories
procedures
ratio;
However,
to
with
data
one
of freedom
modulus
of
strengths.
512 I 328-7246
o
E
d
d
_•
d
d
_
d
d
d
d
d
d
_
d
d
d
d
d
d
T"
Figure
I.
Concrete
core
proficiency
sample
test
data.
i
ESTIMATION
The
impact
OF VARIANCE
experimental
on
how
statistical
the
the
property.
From these
computation
variances
Four
mean
laboratories
memorandum
which
provided
Analyses
are
determined
of
for
variance
each
components
material
for
Details
to check
The
between
about
their
the similarity
for each material
property
of
with
B.
strength
The factorial
have
unknown;
been
tests
shown
revealed
when compared
all
*
in
strength
balanced.
outllers
FIEure
these
A
in the data.
and
occur
potential
values
the Other compressive
1
thirteen
appear
outliers
to
strengths
in
deviate
of the same
laboratories.
alternatives
outliers.
First,
variability
inherent
they
in
using different
are likely to be possible
may
be
the
test
an
extreme
operators
within
the
which
in different
test
causes
manifestation
procedure
true, these values should be retained
observations
an
from
1 is completely
causes behind
compressive
with
obtained
two possible
with
The possible
two
were
in Figure
designated
these
The following
justification
this
procedure
variance
were performed
proceed.
An analysis
are
the
in
a direct
Strength
mix from different
other
used
the
squares
has
properly
sections.
in Appendix A.
in Appendix
I and 9.
substantially
s--_-rize
collected
(02) can be estimated.
review of the data
values
were
should
statistics
laboratories
compressive
prellmlnary
data
approach
following
s-mmary
reported
laboratories.
performed
A and
across laboratories
Compressive
analysis
the
are explained
the results
remain
the
expected
(_LAS) and within
These
in
the
analysis
in Appendix
with
which
statistical
presented
along
under
behind
briefly
results
table
destgn
concepts
explained
COMPONENTS
program.
of
occurs
the
when
laboratories.
and processed
of the two
random
tests
If this were
in the same manner
Removing
these
would result in lower estimates of the process
are
data
as the
without
variability
than
should be reported.
On the
from
the
other
prescribed
hand,
the
outlying
experimental
values
procedures
3/
could
or
be
may
the
result
represent
of
deviations
an
error
in
,.
calculating
the
or
actual
are
reasons
found.
Only
identified
as
testing
recording
outlying
in
this
manner
data
necessary
the
1.
expected
mean square and solving
between
This
component
lower portion
values,
implies
the
significant
estimated
correct
Contact
A,
standard deviation
within
for compressive
The estimated within
laboratory
it
variation,
implies
laboratories.
similar
From
These
between labs is F-3.15.
it
indicate
some
values
Although
between
average
for
effect
error
to
its
are reported
of
the
difference
at the _ - 0.01 level.
compressive
and high
strength
strength
levels,
to be
for ERROR.
outllers,
considers
two
of O2LaB is
found
testing
this value
at each
the
The
strength
measure
the results
across
exists.
the
analyses
large,
Table
7
The
from laboratory
i.
the final
of variance
omitting
13
of
laboratories.
outlier
when
of
significance
at this time concerning
additional
desirable
is not extremely
laboratories
values
there has been no determination
first
for
is larger than the between
are
will be shown to be due to the possible
The
method
and
numbers
value
difference
investigated.
are
to be an excellent
_,
the
of the potential
values
for all compressive
indlca_es
Since
in
used
the estimate
A small
F-ratlo
variation
difference
with
variation,
test
the
variation
does
responsible
LAB and MIX was not found
is considered
" 42748.9.
1
the
""
strength.
average
Table
confidently
of variation
average
laboratories
laboratory
_LAB
be
laboratory
source
the
between
between
that
action.
a commonly
from the medium
The interaction
error(s)
i.
indicates
to vary
investigate
strength
for
each
in Table
and therefore has been combined
of repeatability
status
with
52 compressive
from
difference
any
an appropriate
for o2tas and G2.
I
to
observations
components
is u - (79630) 112- 281 psi which
since
to
Appendix
square
Table
which are expected
been verified.
data
and
the
all
in
estimates
of
essential
the two levels of the factor MlX is significant
result
have
for
variance
the mean
is
establish
As explained
respective
It
eliminated.
table
of equating
The
should
to
consists
as the variance
values
and
of variance
in Table
calculating
these
is
The analysis
values.
for
erroneous
laboratory
presented
numerical
were
the presumed
'-
Table I.
Analysis
of variance
for compressive
strength,
all clara.
°.
DEGREES
SOURCE
OF FREEDOM
SUM OF
MEAN
SQUARES
SQUARE
LAB
12
3007508.81
250625.73
MIX
1
15025275.08
15025275.08
38
3025940.42
79630.01
ERROR
EXPECTED
MEAN
SQUARE
Oa + 4OaUW
Oa +
_(MIX)
Oa
Variance
Component
u2_s
02
Estimate
42748.9
79630.0
DEGREES
..
SUM OF
MEAN
SQUARES
SQUARE
13
18032783.88
1387137.22
17.42 =
lAB
12
3007508.81
250625.73
3.15 "
MIX
1
15025275.08
15025275.08
188.69 "
38
3025940.42
79630.01
51
21058724.31
SOURCE
OF FREEDOM
MODEL
ERROR
CORRECTED
TOTAL
Q
Significant
at the 0.01
level.
F-RATIO
outlier
when
from laboratory
omitting
I, while
the presumed
The analysis
outliers
of variance
MIX 2 from the laboratory
the within
laboratory
gives an estimated
of
the between
impact
of
estimates
varlance
testing
from MIX 2 (laboratory
Table
3.
between
laboratories,
previous
case.
reduced
since
than
data
The within
the estimated
the expected
Based
I
certain.
standard
number
is
an
These
established
has
the
accepting
of
66984.55.
This result
psl.
The estimate
to O2LAs -- 16616.8.
a large
The
reduction
in both
of 5090
the values
from Mix
1 (laboratory
9) is presented
component
nearly
may be
and significance
the same
variance,
02-
considered
as
the
49754.7,
normal
core
Tables
for
results
for
the
has been
further
testing
variation
outlying
the analysis
is less
tests.
2 and
the
in
level
of a - (49754.7) 1/2 - 223 psi
whereas
of
made
The estimate
when omitting
deviation
from
reduced
of
derived
for concrete
outlier
status
before
table
what
too 2-
2.
5090
of variance.
laboratory
on the results
laboratory
value
remain
and now lles below
in Table
the results
the value
of O - 258.8
has been
of the variance
_LAB'
from omitting
deviation
I) and 6335
The estimates
1 and 9.
is presented
variance
of variance
from laboratories
has been reduced
of the two components
The analysis
s-mmarizes
derived
standard
one omitted
analysis
cable
1 results
laboratory
this
the second
3 it is likely
value
from
data
values
of variance
the value
laboratory
needs
results
9
to
be
from
is
less
further
in either
Table
2 or Table 3.
Tensile
Strength
The analysis
strength
of variance
from twelve laboratories
of the factor MIX on tensile
ratio
(F-26.08).
effect
and variance
of the difference
The between
across
laboratories
_L_"
in Table
were estimated
strength
varlatlon,
estimates
in Table 4.
is indicated
components
in tensile
laboratory
in the test results
are presented
strength
These variance
component
for the tensile
The significance
4 by the large
after
due to the factor
5167.96,
is greater
indicates
than
removing
Fthe
MIX.
the variation
the variation
within
Table
2.
Analysis
with
of
one
DEGREES
SOURCE
variance
possible
OF FREEDOM
for
outlier
cospressive
SUM OF
MEAN
SQUARES
SQUARE
LAB
12
1585469.40
132122.45
MIX
1
15560753.72
15560753.72
37
2478428.45
66984.55
ERROR
strength
removed.
EXPECTED
MEAN
SQUARE
02 + 3.92 _tU
O_ +
_(MIX)
Variance
Component
_LAB
16616.8
02
66984.55
DEGREES
SOURCE
OF FREEDOM
MODEL
13
Estimate
SUM OF
MEAN
SQUARES
SQUARE
17863230.37
1374094.64
LAB
12
1585469.40
132122.45
MIX
1
15560753.72
15560753.72
37
2478428.45
66984.55
50
20341658.82
ERROR
CORRECTED
* Significant
TOTAL
at the 0.01
level.
F-RATIO
20.51 "
1.97
232.30.
Table
3.
Analysis
of
variance
with two possible
DEGREES
SOURCE
OF FREEDOM
for
compressive
outllers
removed.
SUM OF
MEAN
SQUARES
SQUARE
LAB
12
1346096.76
112174.73
MIX
I
16216287.41
16216287.41
36
1791170.68
49754.74
ERROR
strength
EXPECTED
MEAN
SQUARE
02 + 3.84 02LAS
02 +
Q(MIX)
02
• °
Variance
Component
Estimate
02
16255.2
LAB
02
49754.7
DEGREES
SOURCE
OF FREEDOM
MODEL
13
SUM OF
MEAN
SQUARES
SQUARE
18395392.94
1415030.23
LAB
12
1346096.76
112174.73
MIX
1
16216287.41
16216287.41
36
1791170.68
49754.74
49
20186563.62
ERROR
CORRECTED
* Significant
TOTAL
at the 0.01 level.
F-RATIO
28.44 *
2.25
325.92 *
laboratories
(o _ -
standard
2046.59)
deviation
is
by
estimated
indicated by the siEnlficant
test scores when
for subsets
of
differences
in average
ModuIus
relatively
narrow
techniques
scale, roundoff
Results
modulus
difference
Including
was
found
this factor
for
between
laboratory
larger
than the within
result
indicates
This
difference
(F-18.74).
modulus
is
Results
Polsson's
are glve
ratio.
the
small
wlth
variance
component
larger than the within
result indicates
is also indicated
ratio
digits
given
and lie in a
and mixes.
Since
are reported
of
In
analysis
data
variance
the effect
on
a
test results
(F-13.82).
two types
are estimated.
is nearly
across
the large
of
The
five times
This
laboratorles.
F-ratlo
across
5 for
a significant
(o2 - 0.04176).
of measurements
5 by
that
of the
or testing
component
in Table
in Table
elasticity
for LAB
laboratories
for
7.
analyses
F-ratlo
of variance
for MIX
reported
indicates
difference was found for this factor on Polsson's ratio
laboratory
7 s---,=rlzes the
response
(G2LAB - 0.19077)
in average
in Table
be different
and Folsson's
indicates
modulus
removes
component
indicated
from
The
factor
due to laboratories
The differences
are summarized
for MIX
of uniformity
also
Therefore,
estimates.
of variance
F-ratio
laboratory
a lack
also
and a very small range of data may not give
the
variance
is
Ratio
efficiently
in the model
mixes before the variation
will
across all laboratories
found from the analysis
The
testing
feature
Table
of elasticity
most
errors
the
In Table 4.
together.
of these variance
of elasticity.
This
only a small number of values.
work
estimates
45.24.
data
for tensile strength.
range when compared
of
very accurate
this
to only one or two significant
the data assume
continuous
From
across laboratories
grouped
for modulus
this analysis
variance
O -
and Poisson's
in Figure 1 were reported
2 1/2.
for LAB (F-10.88)
test results
The values reported
be
compared
laboratories
of Elasticity
of
to
F-ratlo
the average
the
a factor
(_L_
laboratory
variance
a lack of uniformity
in the analysis
" 0.0011805)
of measurement
of variance
table
that
no
(F-O.02).
IS more
component
In Table
slgnlflcant
The between
three
times
(02 - 0.0003537).
This
across
than
6 for
laboratories
and
for the factor LAB ( F -
Table
4.
Analysis
of
DEGREES
SOURCE
variance
for
SUM OF
OF FREEDOM
tensile
strength.
MEAN
SQUARES
EXPECTED
SQUAFI
MEAN SQUARE
LAB
11
244959.23
22269.02
o2 +
MIX
1
50778.59
50778.59
oa +
34
69583.93
2046.59
ERROR
""
3.913
O2u_
_(MIX)
o2
Variance
Component
o 2LAB
5167.96
0.2
2046.59
DEGREES
SCIURCE
IAB
MIX
MEAN
SQUARES
SQUARE
12
295737.82
24644.82
12.04
II
244959.23
22269.02
10.88 "
53381.49
53381.49
26.08
34
69583.93
2046.59
46
365321.74
1
ERROR
CORRECTED
" Significant
SUM OF
OF FREEDOM
MODEL
TOTAL
Estimate
at the 0.01 level.
F-RATIO
"
..
Table
5.
Analysis
DEGREES
SOURCE
OF FREEDOM
of variance
SUM OF
MEAN
SQUARES
SQUARE
for modulus.
EXPECTED
MEAN
SQUARE
LAB
8
6.2591
0.78239
02 + 3.8824 02L_
MIX
1
0.5114
0.51144
02 +
25
1.0439
0.04176
02
ERROR
_(MIX)
Variance
Component
SOURCE
Estimate
02u_
0.19077
oa
0.04176
DEGREES
SUM OF
MEAN
OF FREEDOM
SQUARES
SQUARE
F-RATIO
MODEL
9
6.7705
0.7523
18.02 "
LAB
8
6.2591
0.7824
18.74
°
MIX
1
0.5772
0.5772
13.82
*
ERROR
25
1.0439
0.04176
CORRECTED TOTAL
34
7.8145
• Significant at the 0.01 level.
Table
DEGREES
SOURCE
OF FREEDOM
6.
Analysis
of variance
for Polsson's
SUM OF
MEAN
SQUARES
SQUARE
ratio.
EXPECTED
MEAN SQUARE
LAB
5
0.024306
0.004861
02 + 3.8182 G2LAB
MIX
i
0.00000952
0.00000952
o2 +
0.005659
0.0003537
O2
ERROR
16
_(MIX)
..
Variance
Component
02 _B
0.0011805
0_
0.0003537
DEGREES
SUM OF
MEAN
SQUARES
SQUARE
6
0.024315
0.0040525
11.46 *
IAB
5
0.243056
0.048611
13.74 *
MIX
1
0.00000784
0.00000784
16
0.005659
0.0003537
22
0.029974
SOURCE
OF FREEDOM
MODEL
ERROR
CORRECTED
* Significant
Estimate
TOTAL
at the 0.01
level.
F-RATIO
0.02
Table
¢o_presslve
7.
Analysis
of
laboratol_"
means.
Strength
LAB:
9
10
2
6"
11
5
12
8
3
6
7
13
1
MF..a_: 6181 610S 6082 6080 6035 6032 5995 5955 5901 5882 5860 5778 5185
GROUP :
Tensile
Strength
LAB:
13
3
12
MEAN:
683
627
600
8
2
589
584
II
582
9
574
I
530
4
524
5
6
522
462
i0
404
GROUP:
Modulus
LAB:
MEAN:
13
5.00
6
5
4.72
4.48
I0
9
II
12
4.21
4.17
4.16
4.12
2
3.75
7
3.59
GROUP:
Poisson's
LAB:
MEAN:
GROUP:
Ratio
13
0.2525
"-
5
0.2433
6
0.2425
I0
0.2075
9
0.1875
12
0.1650
ratio
for
equals
13.74).
Po_._son's
ratio
ILLUSTRATING
test
Differences
are
If two sample
means
results
two
population
means
the two-sample
available
from more
including
its
true means?
If each
all tests have
a
probability.
therefore
two
the
requlre
7
to
be
Groups
of
concerned
continuous
suggest
are
llne does
the
true
deviation
to
difference.
which
from
lle
means
from
Nontransltlve
confidence
means
an
are
is
of
the
respective
is
significance
to
control
the different.
comparison
results.
one another.
on either
these
sample
also.
this
of two means and
which
The average
from
exceed
test
largest
to indicate
which
The averages
to be
end of the row.
averages,
are
Probability
concerning
are underlined
intervals,
occur
their
in a row and ranked
are necessarily
results
each
that either the means
laboratory
insufficient
then
all of them having
limits and do not conform
these
that
that
for
the probability
contain
different
means
different
interval
Interpretatlons
not appear underneath
do not imply the population
it indicates
make
are presented
those
control
When interpreting
rather
used
with
If data are
What
comparisons
of tests.
the true
level _, then the probability
than for a single
laboratory
most
mean.
from the
is tested
interval.
each giving
the conclusion
statistically
are not statistically
with
than
they
whether
a confidence
simultaneously
types
from each laboratory
about
to one another
population
that
are computed
a confidence
multiple
alternative
ones
standard
of
reaches
are producing
to smallest.
inferences
are equal
u is less
goal
special
is
then
test has the significance
needs to be stronger
laboratories
to
will
laboratories
COMPARISONS
deviations
be constructed,
intervals
across
7.
laboratories,
true
The procedure
Table
results
than
results
MULTIPLE
or by constructing
The
same versus
statements
laboratory
significance
alone.
Table
MEANS:
corresponding
that all
test
and their standard
as easily
probability
in
laboratories,
t-test
can Just
level
AMONG
of each
laboratory
the
s-mm_rized
DIFFERENCES
from
in average
there
the
If one
is evidence
two
or
three
to the rest of the data.
means which
equal
size
For
are underlined
to one another
was used
example,
equal;
to detect
if
the
a
row
"-
contains
may
be
middle
seven
values,
significantly
the
three
different
from
may not be siEnlflcantly
these means belong
largest
means
one
another,
different
to both groups
and
which
the
yet
three
the
from each other
smallest
four
means
means
even
in
though
the
some of
are different.
"-
SUMMARY
Test
across
equal
results
all
compressive
laboratories
to
the
laboratories
Polsson's
value
One possible
when using
the
consequence
were
within
Large
found
to
laboratory
variations
cause
for these
specialized
in estimating
from the analysis
be
in
results
test
same
nearly
across
of elasticity,
are
differences
evaluate
properties
of variance presented
and
in
these material
to further
these material
the
was
results
tests to estimate
laboratories
nearly
variation
strength, the modulus
A query of the cooperating
of the differences
a direct
that
expected.
procedures
properties.
and
strength
were found for tensile
ratio.
laboratory
extent
for
here.
the
should be
REFERENCES
I.
Dixon, W.J.,
and Massey,
Ed., McGraw-Hill,
2.
"Conducing
F.J., Int_oducV$o_
New York,
Analysis,
3rd
N.Y., 1969.
an InterlaboratoryTest
Test Methods
V@ Statistical
for Construction
Program
Materials',
to Determine
ASTM
the Precision
Standard
Practice
of
C 802-
80.
3.
"SAS User's
Guide:
Statistics',
Version
5, SAS
Institute,
Cary,
NC,
1985.
4.
Milliken,
Learning
G. A.,
and Johnson,
Publications,
Belmont
D. E.,
Analysis
Californla,
of Messy
Data,
Lifetime
1984.
• t
APPENDIX A
EXPERIMF__AL
The
design of the
were controlled
The
layout
and are assumed
of the factors
primary variables,
space
experiment
relative
statistical
their
laboratories
analysis.
includes
to account
and
to this study
two classification
for variations
levels
is shown
variables
and
serve
in Figure
as the independent
results
which
in the test results.
1.
(LAB) and mixes (MIX), represent
Inconclusive
to occur if other variables,
DESIGN
The
two
the inference
variables
from this experiment
in the
are
which were not measured or controlled,
likely
influenced
the results.
FACTOR
CLASSIFICATIONS
The experimental
since
the
requires
data
design used to relate these factors requires
collection
the application
is classified
plan
implies
of specific
as either fixed
and also by the definition
A variable
is defined
in the test plan.
were
chosen
levels.
as
entire
to be fixed
medium
difference between
A factor
is designated
population
included
of
to be
designated
as having
computed.
made
Each primary
The decision to designate
and
variable
an effect
of the study
or
interest
a random
random
if all levels of interest
only two levels of MIX were
high
strength
which
are included
defined.
implies
two
to be fixed and an effect
They
specific
summarizing
the two levels can be computed.
as random
are chosen at random.
considered
be
of the factor.
The factor MIX is designated
the average
can
on the inferential objectives
In this analysis
either
inferences
analysis methods.
or random.
in one of these two ways is based
what
explanation
are
whenever
included
only a few elements
in
The laboratories
sample
effects
of
..
the
design,
and
chosen for this
all possible
from
the
component
ones
design
are
They
are
laboratories.
and the variance
the
o_m
will be
Factors
defined
the
whose
levels
as crossed
levels
of
possible
to
appear
with
one
MIX and
find
any
for
LAB are
within
each
variance
ability
used
for
an
this
the
of
objective
variances
which
separated
into
effects
of
with
one
of
these
levels
mix.
of
the
factors
proposed
another
two
Two
the pure error,
_,
it
is
..
how the error
in
were
made
the analysis
of
s-mmarlzes
the
which
for the same
design
factors.
measurements
used
are
since
determines
The method
themselves
variance
cannot
parts
laboratories).
The
(one random
component
be measured
assignable
model
and
of
_
directly.
interest
laboratory
one fixed).
_
and mix.
total
sources
deduce
variation
(between
two-factor
model
values
of
is to be
and
with
within
mixed
is:
+ ai + _j + _ijk
- mean
of the ith laboratory
Ej
- effect
of the jth mix
_ijk " residual
of
components
square
from
then
solving
commonly
(fixed)
used
for between
each
source
for OZLAB and GZ.
component
(random)
(random)
the most
mean
variance
to
The form of the model
- effect
variance
is
The
is the
ai
One
analysis
to different
Yijk "
estimates
methods
for
calculating
and within laboratory
of variation
These
provided
to its
two numbers
in Tables
(I-1, .... 13) and T be their
the expected
components
average.
of the mean
have the following
squares
values:
the respective
consists
of equatlngthe
expected
mean
are reported
I through
Let T|.. -_s_-Yij k, the sum of all measurements
data
in
in this test plan
each
remaining
reason,
is estimated.
estimate
the
COMPONENTS
The
where
level
crossed
of
of the tests to repeat
VARIANCE
all
laboratory
laboratory
provides
For
combination
the within
every
another.
The type of replication
term
with
made
Then it can be
and
in the set of
5.
in the ith laboratory
shown
for laboratories
square
and
that
for balanced
for the residual
..
E [MSLABS
] -- E [ _
(Z|..
E [MS.s:] - E [ _
To
calculate
variance
their
as
the
variance
given
expected
in
mean squares,
For unbalanced
slightly
Tables
data
different.
2 through
properties;
This
6.
still be reasonable
1 through
and solve
-
0 _ + bn 02LAB
02 and
02LAs,
compute
6,
these
_o
set
the
situation
occurred
components
analysis
equal
of the two equations
in the analyses
data points
of the true values.
of
to
simultaneously.
s-mmarlzed
will loose some of their
if the number of missing
good estimates
the
expressions
two equations
the factor bn in the first
The variance
however,
]
(Y_Jk" y)Z ] . oZ
components
Tables
T)2
is small
is
in
optimal
they will
APPENDIX
An important
the similarity
The
part
determine
a thorough
of measurement
investigation
variations
of
within
analysis
variances
initially
assumes
different
of
this
data
(homogeneity)
the
laboratories
if there is sufficient
B
hypothesis
to reject
to
within
that
are the same.
evidence
is
investigate
laboratories.
all
measurement
The objective
this hypothesis
is to
based
on
the observed data.
Estimated variances
the following
formula
within
each of the laboratories
for all four material
-
from the ith laboratory
of test results
The data used for these calculations
refers
to the ith laboratory
Tables
estimated
these values
Table
variances
are based
should
not be
B-I the variances
of 200 in laboratory
within
is the consequence
174,050
considered
5 to a high
of
in MIX 2 is considerable
Although not nearly
determination
wide
accurate
strength
1.
The subscript
subscript
and
ranges
outlier.
J refers
most
three
are
listed
of values.
estimates.
(Nij
i
to
9.
in
Since
i) - I),
For example,
from MIX 1 range
in
from a low
An even wider
This
large
variance
from MIX
The
second
largest
variance
larger than the remaining
of these numbers
mix
(i.e.,
in laboratory
is observed.
alone
variances.
some major
Based
differences
in
likely exist.
as severe,
of the other
laboratory
of 415,872
a possible
laboratory variation
in Figure
on one degree of freedom
of compressive
on the visual observations
within
each
are observed
range for MIX 2 of 800 to 793,800
2
are given
and the jth mix
2 (J - 2).
B-I through B-4 from which
these computations
and the j_h mix
from the ith laboratory
(i - 1 ..... 13) and the
either MiX 1 (J - I) or MIX
The
- i)
[ E Yijk2 - NijYij2 ] / (Nij - i)
where Yijk " the k th observation
Nij - the number
using
tests:
(Y|]k " Yi])2 / (Nil
O_j2 -Z
were calculated
wide ranges in variances
material
properties
also exist in the
as well.
Whenever
only
""
one observation
(NIj -
i)
was available
it
deficiency
was
not
three
observations,
sample
sizes,
maximum
already
normality
calculate
B-2, B-3,
have
not work
observed
with
leads
to
the
makes
more
variation
efficient
of
the
of variances
power
when
in each laboratory,
require
working
at least
with
small
In this
study,
from each cell.
The wide
range
conclusion
that
the
assumption
tests for homogeneity
a
of
of
of variances
of
sent to each laboratory,
the within
data.
This
laboratory
method
will be the same for both MIX 1 and MIX 2.
the two mixes
the formula
variance
assumes
Combining
to compute
it is
the
which
testing
the data
this pooled
from
estimate
is:
oi/where
- 1) * oi_z+
NS - (nil + ni2 - 2).
previous
estimates
each mix.
pooled
estimate
rather
where
is
divided
the
by 2.
This
of
compressive
the
two
They are based
homogeneous
laboratories.
and
material
are more
Values
identified,
strength,
are
the
reasonable
quite
number
average
of tests
of
the two
performed
on
previous
estimates
on two degrees
within
each
of freedom
(NS - 2)
appear
be
efficiency.
pooled
than
for laboratories
still
is a weighted
sizes are the same for each mix so the
than one and have much greater
For
- 1) * OizZ] / NS
are the
the sample
sum
(nil
estimate
the weights
In this analysis
laboratory
were
[(nil
..
and B-4 by a blank.
mixes were
estimate
use
This
are not feasible.
Since two cores from different
a pooled
variance.
data.
Therefore,
methods
estimated
nonnormal
were available
on these statistical
compute
an
sufficient
well
may not be Justified.
possible
for any of these three properties
tests for the homogeneity
do not
or do
to
in Tables
of two replicates
values
based
possible
is indicated
Most statistical
from any cell
the
individual
i and 9, where
large.
tests are much more comparable
estimates
Estimates
to
estimates
the potential
from
to one another.
the
more
for all
outllers
other
three
°
Table B-I.
Within
laboratory
variances
COMPRESSIVE
MIX 1
LAB!
1
2
3
4
5
6
7
8
9
10
11
12
13
Nil-1
1
1
1
1
1
1
1
1
1
1
I
1
1
Table B-2.
Oil 12
Within
1
1
I
i
I
I
I
1
i
1
1
i
i
laboratory
MIX i
I
2
3
4
5
6
8
9
10
11
12
13
Nil-1
I
I
1
0
1
1
1
I
1
1
1
1
0i112
968.0
4512.5
40.5
450.0
312.5
0.0
420.5
162.5
200.0
50.0
3528.0
POOLED
ai12 2
Ni2-1
145800.0
22050.0
92450.0
1800.0
200.0
22050.0
9800.0
96800.0
415872.0
105800.0
14450.0
168200.0
6050.0
NS
793800.0
33800.0
49612.5
1250.0
11250.0
48050.0
7200.0
12800.0
49612.5
20000.0
174050.0
3200.0
800.0
variances
2
2
2
2
2
2
2
2
2
2
2
2
2
- Tensile
1
I
1
1
1
1
1
1
1
1
1
1
a1_
469800,00
27925,00
71031,25
1525 00
5725 O0
35050,00
8500,00
54800,00
232742.25
62900.00
94250.00
85700.00
3425.00
Strength.
STRENGTH
MIX 2
Ni2-1
Strength.
STRENGTH
MIX 2
TENSILE
LAB i
- Compressive
POOLED
Oi122
NS
364.5
50.0
3698.0
264.5
50.0
112.5
112.5
924.5
2.0
1250.0
1250.0
1200.5
2
2
2
1
2
2
2
2
2
2
2
2
Oi_
666.25
2281.25
1869.25
264.50
250.00
212.50
56.25
672.50
82.00
725.00
650.00
2364.25
Table B-3.
Within
laboratory
variances
MODULUS
Nil-1
2
5
6
7
9
I0
II
12
13
1
1
1
1
1
1
1
1
1
0tll 2
0,005
0,005
0,045
0 0025
0 0041
0 005
0 045
0 005
0 0113
Table B-4.
Within
MiX 2
Ni2-1
Oi12 2
1
0
1
1
1
1
1
1
1
0.0
0.0
0 0
0 005
0 0313
0 0013
0 02
0 0113
laboratory
variances
POISSON'S
MIX I
LAB i
Nii-I
5
6
9
I0
12
13
1
1
1
1
1
1
Oill 2
00005
00125
0002
0002
00005
0002
of Elasticity.
OF ELASTICITY
MIX 1
LAB i
- Modulus
POOLED
NS
2
1
2
2
2
2
2
2
2
0
1
1
1
1
1
Ratio.
RATIO
Oi122
.0
.00005
.00005
.00045
.00005
S/
0.0025
0.005
0.0225
0.01225
0.0045
0.0181
0.0231
0.0125
0.0113
Poisson's
MlX 2
Ni2:l
Oip2
POOLED
NS
Oip2
1
2
2
2
2
2
.00005
.00063
.000125
.000125
.000400
.00025
.
APPENDIX VII
•
TECHMEMO:
AU-127
(Addendum
Precision
Robin
High,
Virgil
Precision
Statements
AUTHORS:
SUBJECT:
Proficiency
The
core
within
_onn._
_.
results
presented
modulus
of
The
SHRP
that
and
value
random
from
type
impl_es
each
of
Compressive
the
the
laboratory
core
by
- The
Construction
2602
DeUana
SHRP
the
concrete
AU-127
dated
5as_d
_,^n the
_tate=_nts,
stength,
CONCRETE
are
The
standard
from
the
same
between
differ
from
tensile
than
2
2 G
CORE
based
strength,
SAMPLES
on
the
results
standard
deviation
deviations
limits
laboratory
one
another
wlthin.-laboratoryslngle
results
in
should
not
represent,
C670,
strength
operator
sample
Practice
preclslcn
FOR
two
more
for
only
are
of
for
given.
selected
measurement
made
_,e
an
the
This
measurement
5%
for
on
at
the
time.
Strength
same
ASTM
Core
Memorandum
compressive
difference
_I11
Therefore:
'
Technical
cores.
observations
compressive
numbers
1991
Concrete
components
statements
sample
that
concrete
Precision
These
19,
ratio.
precision
and
two
for
STATEMENTS
proficiency
between
latter
in
Polsson's
PRECISION
measurement
difference
March
Cement
variance
given
report,
wlthin-laboratory
individual
same
laboratory
were
elasticity,
concrete
Portland
,_,-..,._
a_-_.._-_-_"pro--Ide=
in
WITHIN-LABORTORY
Statements)_(_
Anderson]_Y"
for
SHRP
between
samples
21,
DATE:
Samples
and
proficiency
D_.cember
-
for
has
of
the
been
two
Precision
more
IS
and
to
deviation
be
a
conducteG
laboratory
by
the
stan_rd
found
properly
same
differ
respectively,
Preparing
operator
on
the
same
2 _G
-
D2S
limits
as
for
258.8.
tests
than
Statements
-
Test
by
Aura.in, Temm
•
Te/ephoee
512
I 32"/-42_!
•
Fix
512
the
concrete
732.0.
described
Methods
Materials.
lane
for
I _2942,.i6
in
for
Tensi_
Strength
Precision
- The wlthln-laboratory
single operator
tensile
strength has been found to be a - 45.24.
results
of two properly
in the same laboratory
differ
These numbers
represent,
AST_
C670,
Practice
Construction
Modulus
by more
Therefore,
tests by the same operator
on the same concrete
sample should
not
than 2 v_ o - 128.0.
respectively,
for Preparing
the lS and D2S limits as described
Precision
Statements
in
for Test Methods
for
standard deviation
for
of Elasticity
- The within-laboratory
modulus
of
Therefore,
same
numbers
Practice
Construction
Poisson's
C670,
results
should
represent,
single operator
elasticity
operator
sample
ASTM
conducted
for
Materials.
Precision
These
standard deviation
of
has
in the same
for Preparing
found
two properly
not differ
respectively,
been
be
o
conducted
laboratory
by more
to
than
on
-
tests
the same
2 _O
0.204.
by
concrete
- 0.578.
the IS and D2S limits as described
Precision
Statements
the
in
for Test Methods
for
standard deviation
for
Materials.
Ratio
Precision
- The within-laboratorysingle
Polsson's
results
Ratio
has been
of two properly
in the same laboratory
differ
These numbers
represent,
ASTM
C670,
Practice
Construction
by more
found
respectively,
to be a - 0.0188.
conducted
Therefore,
tests by the same operator
on the same concrete
than 2_o
for Preparing
Materials.
operator
sample should not
- 0.0532.
the IS and D2S limits as described
Precision
Statements
for Test
Methods
in
for
BETWEEN-LABORTORY
PRECISION
STATEMENTS
The between-laboratoryvarlance
given
in Technical
standard
deviations
different
between
differ
Memorandum
limits
laboratories
one measurement
are
FOR SHRP CONCRETE
components
AU-127,
are derived
selected
These
between
imply
two
two observations
that
from
the difference
at random from each of two laboratories
_T,ebetween-laborato_ys!ngle
compressive
289.14.
strength
Therefore,
from one concrete
will
+ o2) only 5% the time.
has
operator standard deviation
been
the results
sample
These numbers
represent,
respectively,
ASTM Practice C670, for Preparing
Construction Materials.
found
to be
of properly
conducted
(O2L_ + 02 ) - 817.81
tests
should
from each
the iS and D2S limits as described
Precision
Statements
for
_OaLA B + O2 .
in each of two laboratories
not differ by more than 2 V2
other.
for Test
in
Methods
for
operator standard deviation
for
Strength
Precision
- The between-laboratoryslngle
tensile
strength
Therefore,
has been
the results
found
to be _O_LA s + 02 -- 84.94.
of properly
conducted
concrete sample in each of two laboratories
by more
These numbers
ASTM
The
Strength
Precisl_n
Tensil
in this section.
values
from each other by more than 2J2(o2L_
Compressive
SAMPLES
fort he concrete core samples,
for the difference
given.
CORE
Practice
Construction
represent,
C670,
than 2
respectively,
for Preparing
Materials.
2 (O2
tests
from one
should not differ
+ 02 ) - 240.24 from each other.
the IS and D2S limits as described
Precision
Statements
for Test
Methods
in
for
Modulus
of Elastl¢i_y
Precision
- The between-laboratoryslngle
modulus
of
elasticity
Therefore,
concrete
by more
These numbers
represent,
the
has
results
sample
than
in each
2 J2
been
of
standard
found
properly
the
conducted
" 1.364
Statements
for
- 0.482.
tests
should
from
each
IS and D2S limits
Precision
deviation
to be
of two laboratories
(02LAS +02)
respectively,
ASTM Practice C670, for Preparing
Construction Materials.
Polsson's
operator
from
not
one
differ
other.
as described
in
for Test Methods
for
single operator standard deviation
for
Ratio
Precision
- The between-laboratory
Poisson's
ratio
Therefore,
concrete
by more
These numbers
represent,
ASTM
C670,
Practice
Construction
the results
sample
found
to be
of properly
_02LAS + 02 - 0.0392.
conducted
in each of two laboratories
than 2 _2
(O2LA s +02)
respectively,
for Preparing
Materials.
has been
--
0.1108
Statements
should not differ
from each other.
the IS and D2S limits
Precision
tests from one
as described
for Test Methods
in
for
APPENDIX Vlll
March
W.
Charles
Greer,
Law Engineering
396 Plasters
Avenue
Atlanta,
GA 30324
Dear
26,
1990
Jr.
NE
Charles:
Subject:
SHRP
PCC
Core
Proficiency
Sample
Program.
I am
pleased
to
advise
that
SHRP,
based
upon
test
results
from
the subject
program,
has
authorized
your
laboratory
to proceed
with
the
testing
of
portland
cement
concrete
(PCC)
cores
from
field
sections
of the
LTPP
project
in accordance
with
required
protocols.
It
was
noted
that
your
laboratory
achieved
a rating
of five,
using
the
Cement
and
Concrete
Reference
Laboratory
(CCRL)
approach
to
analysis,
on
all
tests
included
in the
subject
program.
This
was
indeed
an
excellent
performance.
As you
proceed
with
these
tests
on SHRP
field
samples,
the same
internal
quality
control
practices
should
be followed
that
were
used
when
testing
the PCC proficiency
samples,
thus
providing
confidence
in
the data
generated.
Yours
very
truly
Garland
W.
President,
cc:
Steele,
Steele
Adrian
Box
P.E.
Engineering,
Inc.
Pelzner
173 • Tornado,
West
Virginia
25202
• Tele.
(304)
727-8719
$
13
LABS
PARTICIPATE
IN
SHRP
PCC
CORE
PROFICIENCY
SAMPLE
PROGRAM
Final
results
of the
Portland
Cement
Concrete
Core Proficiency
Sample
Program
were
recently
forwarded
to
the 13 participating
laboratories
by
the
Iowa
Department
of Transportation
Office
of
Materials.
Over
150
4" by 8" cores
were
shipped
for determining
the precision
of tests
to be performed
on concrete
pavement
cores
from the LTPP
study.
The program
was designed
to obtain
data
on the static
modulus
of
elasticity,
poissons
ratio,
splitting
tensile
strength,
and
compressive
strength.
Detailed
data
analysis
is now under
way by
statiscal
consultants
at
TRDF.
However,
preliminary
data
analysis
and
laboratory
ratings
were
determined
using
the
widely
recognized
Cement
and
Concrete
Reference
Laboratory
(CCRL)
approach
i.
The best
laboratory
rating
under
this proceedure
is a
5, indicating
that
a
laboratory's
test
result
is less than
one
standard
deviation
from
the mean
of all results.
The preliminary
analysis
indicated
concrete
testing
(Law Engineering,
in each category
of
the program.
SHRP has
directed
Law
to proceed
LTPP
field
samples.
Laboratories
participating
Florida
Transportation,
FL
Department
of
Gainesville,
Iowa
Department
Transportation,
Ames
Federal
Administration,
,
Highway
CO
of
CA
Department
Charleston,
of
WV
Engineering,
Atlanta,
GA
National
Aggregates
Association/National
Ready
Mix
Concrete
Association,
Silver
Springs,
MD
i 1959
ASTM
this
program
Bureau
Co
were:
of
Waterways
Vicksburg,
Reclamation,
Experiment
MI
Denver,
Station,
IA
Department
Sacramento,
West
Virginia
Transportation,
Law
of
Denver,
California
Transportation,
in
that the
SHRP
laboratory
for
Atlanta)
achieved
a 5 rating
Based
upon
this performance,
with
the concrete
core tests
on
Proceedings,
Concrete
Technical
CANMET,
Canada
Materials
Services,
Skokie,
Ottawa,
Wiss,
Janey
Northbrook,
IL
Ontario,
and
New
York
Department
TranspoM_t_o_,AiB_y,
W. Charles
Greer,
Jr.
Law Engineering
S96 Plasters
Avenue
NE
Atlanta,
GA 30324
Crandall
and
Blaine
and
IL
paper.
Elsner,
of
NY
Compressive
Strength
Precision
The
within-laboratory
single
operator
standard
deviation
for
compressive
strength
of PCC
cores
has
been
found
to
be
a =
A258.8.
Therefore,
results
of two
properly
conducted
tests
by
the same
operator
in
the same
laboratory
on
the same
concrete
sample
should
not differ
by more
than
292 a = _732.0.
The
between-laboratory
compressive
strength
{(a_lah+a:)
=
A289.14.
tests
from one concrete
not
differ
by
more
other.
single
operator
standard
deviation
for
PCC
cores
has
been
found
to
be
Therefore,
results
of properly
conducted
sample
in each of two
laboratories
should
than
2{(2(a:iab+a2))
= B817.81
from each
of
These
numbers
represent,
respectively,
the
as
described
in
ASTM
Practice
C670,
Statements
for Test Methods
for Construction
Splittinq
Tensile
ADIS
and
Preparing
Materials.
_D2S limits
Precision
Strenqth
Precision
The
within-laboratory
single
operator
standard
deviation
for
splitting
tensile
strength
of PCC cores
has been
found
to be a =
A45.24.
Therefore,
results
of two
properly
conducted
tests
by
the same
operator
in
%he same
laboratory
on
the
same concrete
sample
should
not differ
by more
than
292 a = _128.0.
The between-laboratory
single
operator
standard
deviation
for
splitting
tensile
strength
of
PCC
cores
has been
found
to be
{(O21ab+O2
) =
A84.94.
Therefore,
results
of properly
conducted
tests
from one concrete
sample
in each
of two
laboratories
should
not differ
by
more
than
2{(2(a21ab+a2))
=
B240.24
from each
other.
These
numbers
as
described
Statements
for
&
represent,
respectively,
the
in
ASTM
Practice
C670,
Test
Methods
for Construction
ADIS and
Preparing
Materials.
5D2S limits
Precision
4
Modulus
of
Elasticity
Precision
"
The
within-laboratory
single
operator
standard
deviation
for
modulus
of elasticity
of PCC cores
has
been
found
to
be
a =
A0.204.
Therefore,
results
of two
properly
conducted
tests
by
the same
operator
in
the same
laboratory
on
the same
concrete
sample
should
not differ
by more
than
2q2 a = 50.578.
The
between-laboratory
modulus
of
elasticity
q(a_[_b+a_)
tests
from
not differ
=
_0.482.
one concrete
by more
than
single
of
PCC
operator
cores
standard
has
been
for
be
Therefore,
results
of properly
conducted
sample
in each
of two
laboratories
should
2_(2(a21_+a_))
= 51.364
from each
other.
These
numbers
represent,
respectively,
the
as
described
in
ASTM
Practice
C670,
Statements
for Test
Methods
for Construction
Poisson's
deviation
found
to
_DIS
and
Preparing
Materials.
_D2S
limits
Precision
Ratio
Precision
The
within-laboratory
single
operator
standard
deviation
for
Poisson's
ratio
of PCC
cores
has
been
found
to be a = _0.0188.
Therefore,
results
of two
properly
conducted
tests
by
the same
operator
in
the
same
laboratory
on
the same
concrete
sample
should
not differ
by more
than
2_2 a = _0.0532.
The between-laboratory
Poisson's
ratio
of
A0.0392.
Therefore,
single
operator
PCC
cores
has been
results
of properly
concrete
sample
in
more
than
2_(2(_21ab+_2))
each
of
=
standard
deviation
found
to be _(a21ab+a2)
conducted
tests
from
two
laboratories
should
80.1108
from
each other.
These
numbers
represent,
respectively,
the
as
described
in
ASTM
Practice
C670,
Statements
for Test
Methods
for Construction
not
ADIS
and
Preparing
Materials.
differ
for
=
one
by
_D2S
limits
Precision
Long-Term
Pavement
Performance
Advisory
Committee
Chairman
Marshall 1_ Thompson
William J. MacCreery
W.J. MacCreery, lnc.
University of Illinois
David Albright
Exxon Chemical
Kenneth P_ Wardlaw
t
Alliance for Transportation
Corporation
Research
Marcus Williams
Richard Barksdale
Georgia Institute
H.B. Zachry Company
of Technology
"
Liaisons
James L. Brown
Pavement Consultant
Albert J. Bush, III
USAE Waterways Experiment
Robert L. Clevenger
Colorado Department
of Highways
Station
Louis M. Papet
Federal Highway Administration
Ronald Collins
Georgia Department
of Transportation
John P. Hallin
Federal Highway Administration
Guy Dore
Ministere des Transports
de Quebec
Ted Ferragut
Federal Highway Administration
Charles E. Dougan
Connecticut Department
of Transportation
Frank R. M¢Cullagh
Transportation Research Board
MeRaney Fulmer
South Carolina Department
of Highways and Public Transportation
Expert Task Group
Paul E. Bcnson
Marlin J. Knutson
American Concrete Pavement
Association
Hans Jorgen Erlman Larsen
Danish Road Institute, Road Directorate
California Department
of Transportation
James L. Brown
Pavement Consultant
Lyle Calvin
Oregon State University
Kenneth H. McGhee
Consultant Civil Engineer
John P. Hallin
Raymond K. Moore
University of Kansas
Federal Highway Administration
Newton Jackson
Richard D. Morgan
National Asphalt Pavement
Washington State Department
of Transportation
Association
Alex Kazakov
William IL Moyer
Pennsylvania Department
of Transportation
Ontario Ministry
of Transportation
Walter P. Kilareski
David E. Newcomb
University of Minnesota
Pennsylvania
Transportation
Institute
Richard A. Lill
Charles A. Pryor
National
Stone Association
Robert L. Mason
Southwest Research Institute
Cesar A.V. Queiroz
The World Bank
William D.O. Paterson
World Bank
Roland L. Rizenbergs
_
Kentucky
Transportation
Cabinet
Gary K. Robinson
Arizona
Department
Federal Highway Administration
of Transportation
Department
of Transportation
Ted M. Scott
American
Richard M. Weed
New Jersey DOT
Frederic R. Ross
Wisconsin
James A. Sherwood
Trucking Association
Amir N. Hanna
SHRP Staff