UNITED STATES GENERAL ACCOUNTING OFFICE

WASHINOTON, D.C. 20648

RESOURCES, COMMUNITY.
AN0 ECONOMIC DEVELOPMENT
DIVISION

B-214229 FEBRUARY 17, 1984

123436

The Honorable John 0. Marsh, Jr.

The Secretary of the Army
Dear Mr. Secretary:
Subject: Thermal Analysis of Mass Concrete Structures:
Lock 6 Dam No. 26 (Replacement) Project
(GAO/RCED-84-87)
During
our survey of the U.S. Army Corps of Engineers Lock
& Dam No. 26 (Replacement), Project, we noted that Corps design
engineers were using outdated thermal analysis techniques in
attempting to predict temperature distribution and resulting
thermal stresses in the mass concrete to be used in this proj-
ect. The thermal analysis techniques being used are based on
methodology and programs developed about 15 years ago. State-
of-the-art finite element computer programs which aid in pre-
dicting the potential location of cracks from thermal stresses
in mass concrete structures are currently used by large consult-
ing engineering firms and federal agencies, such as the Nuclear
Regulatory Commission and the Bureau of Reclamation, U.S.
Department of the Interior, in designing concrete structures.
Corps officials acknowledged that the Corps lacks experience in
the use of state-of-the-art thermal analysis programs.
Manohar Singh of our office brought these observations
(deta?:; are included in enclosure I) to the attention of Corps
officials in the Office of the Chief of Engineers, the Director-
ate of Research and Development, the Waterways Experimentation
Station, and various division and district offices. The offi-
cials contacted acknowledged their lack of experience in the use
of state-of-the-art thermal analysis programs. As a result the
Corps, in September 1983, began consultations with Bureau of
Reclamation officials on their experiences with finite element
programs for thermal analysis of concrete structures. Also,
Corps officials advised us that based on our observations, a
4-year (fiscal years 1984-87) research and development program
has been initiated for the study of concrete cracking. The
estimated cost for this program is about $510,000.

(085678)
B-214229

Corps officials stated that it would not be practical to use

state-of-the-art programs on the first stage of the Lock E; Dam
No. 26 (Replacement) Project because it is about 85 percent com-
plete. The Corps will, however, examine the feasibility of using
state-of-the-art programs in predicting temperature and strain-
stress distributions in the projects second stage. Further, we
were advised that the Corps acknowledges that an expanded manual
is needed to provide guidance on thermal analysis and that, over-
all, the Corps needs to reassess its capability to provide direc-
tion to its district offices on thermal analysis of mass concrete.
Because of actions taken and planned by the Corps as a result of
our work, we are not making any recommendations.
Our purpose is to bring to your attention the actions taken
and promised by the Corps. We would appreciate being advised
of any Corps actions and the Corps progress in applying state-
of-the-art thermal analysis programs to future mass concrete
structures.
Copies of this report are being sent to the Director, Office
of Management and Budget, and to appropriate congressional commit-
tees having oversight responsibilities for Corps activities.
Sincerely yours*r,

Enclosure

2
ENCLOSURE I ENCLOSURE I

THERMAL ANALYSIS OF MASS CONCRETEfSTRUCTURES:

LOCK & DAM NO. 26 (REPLACEMENT) PROJECT
BACKGROUND
Lock and Dam No. 26 (Replacement) Project was authorized by
Public Law 95-502. The project is located on the Mississippi
River, 200.78 miles.upstream from the confluence of the
Mississippi and Ohio Rivers and about 2 miles downstream from the
existing Locks and Dam No. 26. The Corps has stated that the
existing structures needed to be replaced because they (1) were
not large enough to handle anticipated traffic on the Mississippi
River and (2) had a history of structural problems.
The dam, including a navigation lock, will have a total
length of about 1,350 feet. The concrete structure will be
founded on steel H-piles driven into bedrock. The navigation lock
will be 1,200 feet long and 110 feet wide and have a design depth
of 9 feet. The lock will be located on the Illinois side of the
river and have a maximum lift of 24 feet. The lock will have a
lift gate on the upstream side and a miter gate downstream. Con-
struction of the Lock and Dam No. 26 (Replacement) Project will
extend until 1989. The project is in various stages of design and
construction. As of October 1, 1983, the Corps had estimated the
project's construction cost to be $833 million.
A primary concern relating to mass concrete construction is
the control of thermal cracking. During construction, heat is
produced by concrete during the critical days after the concrete
is placed. Subsequent thermal gradients occur due to cooling at
rates depending especially on external temperature. Concrete
temperature change causes proportional volume change that, if
restrained, either externally or internally by the mass concrete
itself, will result in cracking. Such cracking is of concern
because it can occur at any time during or after construction.
Cracking in mass concrete structures is undesirable as it
affects the water tightness, durability, appearance, and internal
stresses of the structures. Cracks vary from extremely small, or
hairline, surface cracks which penetrate only a few inches into
the mass, to irregular structural cracks of varying width which
completely cross construction blocks, to the regular contraction
joint with a relatively uniform opening which separates the con-
struction blocks in a concrete monolith. These cracks have
economic impacts because of the cost of rehabilitating navigation
and flood control structures. According to the Corps' Walla Walla
District officials, the Corps has spent about $500,000 thus far
in attempting to repair a single extensive crack resulting from
thermal stresses in the Dworshak Dam in Idaho.

3
ENCLOSURE I ENCLOSURF J

Officials of the Portland Cement Association advised that

various techniques to reduce the potential for thermal crackina
have been developed. These include reducinq the potential tempcr-
ature rise by limitinq the heat of hydration1 of cements used,
minimizinq cement content, or replacing part of the cement with
fly ash. Other measures include precoolinq the aggregate to
reduce placement temperature of insulating surfaces to control
thermal absorption or loss, or postcooling the concrete mass with
cooling pipes to reduce peak temperatures or thermal qradients.
According to a professor of engineering at Ohio State University,
since the 1930s, numerical methods have been developed to predict
temperature distribution and resulting thermal stresses in mass
concrete structures. These approaches permit the evaluation of
various ways of controlling temperature during dam construction.
The professor advised that the finite element method (FEM) of
analysis has superseded other numerical methods of investigating
thermal analysis.
The finite element method was initially developed to analyze
problems in structural mechanics. FEM involves replacinq the
actual continuum by a finite number of discrete subregions called
elements. For the purposes of structural analvsis, a structure
is defined as a system of nodal points (joints) interconnected
by discrete elements. Using FEM of analysis, joint displacements
and the internal stresses in the structural elements can be deter-
mined given (1) the joint loading, (2) the geometry of the struc-
ture (location of joints), and (3) the stiffness properties of the
structural elements.
The development of FEM as an analysis tool began with the
advent of the electronic digital computer. For the numerical
solution of a structural or continuum problem, it is necessary to
establish and solve algebraic ecuations that govern the response
of the system. During the past 2 decades, FEM of analysis has
become a popular and effective technique for the computer solution
of the governing equations for complex problems in enqineerinq.
OBJECTIVES, SCOPE, AND METPODOLOGY
Our review objectives were to (1) examine the effectiveness
of the U.S. Army Corps of Fnqineers' programs to predict tempera-
ture distributions and resulting thermal strain-stress distribu-
tions in mass concrete, (2) examine the state-of-the-art finite
element computer programs for thermal analysis that are beina used

lNewly placed concrete underqoes a rise in temperature due to the

exothermic reaction of the cementing materials. The amount of
heat of hydration aenerated will depend on the amount and type of
cementitious materials in the concrete.

4
ENCLOSURE I ENCLOSURE I

by major consulting engineering firms and other federal agencies

such as the Bureau of Reclamation, U.S. Department of the
Interior, (3) determine whether such state-of-the-art programs
can be modified and applied to perform thermal analysis of the
Lock & Dam No. 26 (Replacement) Project, and (4) examine the
Corps' capability to use state-of-the-art programs to perform
thermal analysis of mass concrete structures. Manohar Singh,
a Doctor of Science, was the project manager of this review.
To fulfill our objectives, we examined the latest engineering
guidance on thermal analysis. In addition, we discussed the
state-of-the-art finite element computer programs on thermal
analysis with (1) large consulting engineering firms that apply
these programs in their engineering projects, (2) a nonprofit
organization, the Portland Cement Association, Chicago, Illinois,
(3) academicians at the University of California in Berkeley,
California; Massachusetts Institute of Technology in Cambridge,
Massachusetts; and the Ohio State University in Columbus, Ohio,
who developed the first finite element method of analysis for
temperature distributions and thermal stress analysis of mass
concrete, (4) engineers at the Nuclear Regulatory Commission and
the Bureau of Reclamation, and (5) Corps officials at Washington,
D.C., the St. Louis and Walla Walla Districts, the North Pacific
and Lower Mississippi Valley Divisions, and the Waterways
Experimentation Station (WES), Vicksburg, Mississippi.
The review was performed in accordance with generally
accepted government auditing standards.
THE CORPS' APPROACH TO CONDUCTING
THERMAL ANALYSIS OF MASS CONCRETE
Our review showed that the Corps' Engineering Manual EM
1110-2-2000 (dated Sept. 1982), Engineering and Design, Standard
Practice for Concrete, g ives minimal treatment to thermal analysis
ot mass concrete structures when compared with the Engineering
Monograph No. 34 (dated 1981), Control of Cracking in Mass
Concrete Structures, of the Bureau of Reclamation.
Elements discussed in the Corps' manual include design
studies and reports , preparation of contract plans and specifica-
tions, construction preparation, and concrete construction inspec-
tion. The manual discusses the thermal analysis of mass concrete
in a very general way. The Bureau's monograph is a comprehensive
treatment of temperature control studies; design consideration;
construction requirements; thermal properties of concrete for
various dams; temperature variations of flat concrete slabs ex-
posed to yearly, bimonthly, and daily temperature variations on
both surfaces; the range of actual reservoir temperatures under
operating conditions; the increase of temperature due to solar

5
ENCLOSURE I EMCLOSUREI

radiation; and the effect of initial temperature on heat of

hydration.
In 1967, a graduate student2 at the University of
California, Berkeley, California, developed a finite element
computer proqram called 2-Dimensional (2-D) Stress Analysis with.
Incremental Construction & Creep3 to predict thermal strain-
stress distribution for the Dworshak Dam. The Corps' Walla Walla
District funded the development of the proqram as a part of its
desiqn and construction of Dworshak Dam. The temperature distri-
bution histories for input to this proqram were generated by the
finite difference meth0d.l In 1968, Dr. Edward L. Wilson of the
same university developed another finite element computer proqram
called Determination of Temperatures Within Mass Concrete Struc-
tures (DTMCS) for the Corps' Walla Walla District to predict the
temperature distribution during and after construction of Dworshak
Dam. The output from the DTMCS program was used as input to the
Stress Analysis with Incremental Construction & Creep program.
Program inputs included concrete placement schedules, ambient air
condition, material properties of concrete, geometry of the struc-
ture, and insulation considered necessary for thermal strain con-
trol. While these programs were deemed to be a major advance in
state-of-the-art thermal analysis, according to a Corps official
at the Waterway Experimental Station, they were not effective in
helping the Corps to avoid thermal cracking at the Dworshak Dam.
The Corps still uses the programs developed at the University of
California in 1967 and 1968.
STATE-OF-THE-ART THFRMAL
ANALYSIS FINITE ELEMENT PROGRAMS
In discussions with major consulting enqineering firms and
academicians, we found that these firms use some modifications of

2Dr. Ranbir Sinqh Sandhu, currently a professor of engineerinq at

the Ohio State University, Columbus, Ohio.
3The Portland Cement Association defines creep as the slow
increase of deformation due to sustained load and shrinkaqe in a
stress-independent, lonq-time deformation caused by drvinq. In
concrete, creep deformation exceeds several times the 'instantan-
eous elastic deformation. Shrinkage is of similar maanitude as
the elastic deformation. Creep properties change with the aqe of
the concrete. Accordinqly, the effects of creep in structures
can be very sianificant.
4This is another alternative to replacinq differential equations
by a set of algebraic equations. This method was popular before
the finite element method of analysis came into being.

6
ENCLOSURE I ENCLOSURE I

the Structural Analysis Program (SAP)5 of the University of

California. Executives of these firms confirmed that the state-
of-the-art finite element programs are the ADINAT (temperature),
ADINA (strain-stress), and the SAP-83 (strain-stress) programs.
The SAP-83 program was developed by Dr. Wilson of the University
of California. The SAP-83 program has nonlinear analysis capabil-
ity in 2 dimensions and 3 dimensions. Thermal analysis experts at
one of the major consulting engineering firms consider SAP-83 to
be a very efficient program.
The ADINAT and ADINA programs are proprietary. They were
developed by Dr. Klaus-Jurgen Bathe of the Massachusetts Institute
of Technology. ADINAT is a heat transfer program compatible with
ADINA. Both ADINAT and ADINA have 2-D and 3-D capabilities. The
programs can be used in linear and nonlinear, steady-state, and
transient heat transfer analysis. Nonlinear conduction and bound-
ary convection and radiation conditions can be taken into ac-
count. According to Dr. Bathe, ADINAT with 9 nodes/element has
been used on an experimental basis, and ADINAT can be modified in
approximately 2 days to introduce 9 nodes/element instead of the
current 8 nodes/element for predicting temperature distributions
in mass concrete. Both the SAP-83 and ADINA programs can perform
analysis for incremental construction, creep, thermal expansion,
and cracking.
In 1976, another computer program called Determination of
Temperatures in Construction (DETECT) was developed by Dr. Wilson
and a graduate student at the University of California, Berkeley,
to predict temperature distributions in mass concrete with 8
nodes/element. It is a linear finite element program. According
to experts in a major consulting firm, this program with the 8
nodes/element capability did not provide accurate results of tem-
perature distributions in mass concrete of a large dam. This pro-
gram is being enhanced with a 9 nodes/element capability to pro-
vide accurate temperature distributions. These experts believe
that when the DETECT program of predicting temperature distribu-
tion is modified with 9 nodes/element, SAP-83 will provide highly
accurate results of strain-stress distribution in mass concrete.
An official of the Nuclear Regulatory Commission advises that
the Commission uses the University of California-developed SAP
IV6 and SAP V finite element programs to determine the thermal

51n 1969, the development of the series of programs SAP-I to

SAP-IV was initiated.
6This program was developed by Drs. Wilson, Bathe, and Peterson
at the University of California in June 1973 and revised in April
1974.

7
ENCLOSURE I EP'CLOSUREI

strain-stress distributions in mass concrete of nuclear-related

structures. The Bureau of Reclamation has been using the Heat
Flow finite element proqram to determine temperature distribution
in mass concrete structures for the last 10 years. This program
was also developed at the University of California. It is a 2-D,
nonlinear program. Bureau officials stated that they used this
program on the Upper Stillwater Dam, which is in the construction
bid stage. Temperatures in the test sections (in-situ conditions)
correlated with the temperatures predicted by the Heat Flow pro-
gram. Also, the Bureau developed a finite element program called
the Arch Dam Stress Analysis System (AIXAS) to determine strain-
stress distribution in arch dams only. It is a 3-D, linear pro-
gram in which creep is factored indirectly. For stress distribu-
tion in arch and gravity dams, the Bureau uses the 3-D SAP-IV
program. Recently, the Bureau used the ADINA program to determine
strain-stress distributions in the Upper Stillwater Dam. We noted
that the Bureau is satisfied with the performance of the Heat Flow
program and the ADINAT program for temperature distributions and
the ADSAS, SAP-IV, and ADINA finite element proqrams for strain-
stress distribution in mass concrete structures.
Corps officials acknowledged that they lack experience in
using the state-of-the-art programs on thermal analvsis of mass
concrete that are being used by major enqineering firms and by
the Bureau of Reclamation. Furthermore, the capability of Corps
headquarters and WES to provide direction on thermal analysis to
the numerous district offices is severly limited. Only two staff
engineers and one research physicist are available for this
effort.
EFFECT OF NOT USING STATE-CF-THF-ART
THERMAL ANALYSIS PPOGRAMSON THE
PROJECTS'S FIRST STAGE
In 1981, for the first staqe of the concrete dam construc-
tion at the Lock and Dam No. 26 (Replacement) Project, WES used
the 1968 DTMCS program after some modification to predict terrpera-
ture distributions, and the Stress Analysis with Incremental Con-
struction & Creep program to determine strain-stress distribution
as functions of time, incremental construction, materials, and
construction parameters. Not using state-of-the-art finite ele-
ment programs for thermal analysis of mass concrete in the desiqn
of the project could lead to the development of deep cracks durina
and after construction; this could lead to safety, engineering,
and economic problems. According to an academician, state-of-
the-art finite element proqrams can predict whether crackinq in
mass concrete will occur and the location of cracking.
Due to several reasons cited by thermal analysis experts, we
have concerns regardins the adequacy of the proqram the Corps used

8
ENCLOSURE I PNCLOSURE I

in conducting thermal strain-stress analysis of the Lock & Dam

No. 26 (Replacement) Project's first stage. The Corps' current
thermal analysis program is inadequate because the finite element
methodology used limits the program to linear analysis only. The
proqram also
--allows changes in temperature but assumes that the proper-
ties of concrete will not chanqe, i.e., the specific heat7
and conductivity 8 of concrete will remain constant;
--does not allow for radiation from concrete surfaces;
--predicts temperature distribution in mass concrete with
only 4 nodes/element? and
--predicts temperature distribution in 2 dimensions only,
therefore lacking the 3-D capability of predicting all-
encompassing temperature distribution in mass concrete.
THERMAL ANALYSIS OF THE
PROJECT'S SECOND STAGE
In March 1983, WES submitted a proposal to the St. Louis
District Office (STLD) to perform thermal strain-stress analysis
of the second stage of the Lock 6 Dam No. 26 (Replacement) Proj-
ect. The work was scheduled to be done in fiscal year 1984 at a
cost of $100,000. A minimal 3-D analysis was also part of the
proposal so that WES could compare both the results with the 2-D
analvsis.
Bxperts from major design and engineerinq consulting firms
having direct field experience in designing and constructinq larqe
concrete dams and academicians accredited with developinq the
state-of-the-art thermal analysis proqrams have told us that the
computer program in the VES proposal will be inefficient. The
program will be inefficient because it (1) does not factor in
realistic properties of materials, such as temperature dependence
of suecific heat and conductivity of concrete and incremental
construction, and (2) factors in creep in concrete in onlv a
rudimentary manner.

'Specific heat is the ratio of the amount of heat required to

raise a unit weight of the material 1 degree to the amount of
heat reauired to raise the same weiqht of water 1 degree.
*Thermal conductivity is a measure of the ability of the material
to conduct heat and is defined as the ratio of the rate of heat
flow to the temperature aradient.

9
ENCLOSURE I ENCLOSURE I

AS a result of our advising WES officials of the

inefficiencies in the program proposed, in June 1983 WES sent
another proposal to the STLD to perform thermal analysis of the
project using a state-of-the-art finite element program. In
October 1983, STLD sent a letter to the Lower Mississippi Valley
Division with a recommendation that this latest WES proposal be
accepted. Further, STLD recommends instrumentation of the lock
during construction so that temperature distribution in the lock
structure can be determined.
VIEWS OF RESPONSIBLE OFFICIALS
Office of the Chief of Engineers
Officials of the Corps' Office of the Chief of Engineers
stated that they recognized the need for the Corps to become more
familiar with the improvements that have taken place in the area
of thermal analysis of concrete structures. The Corps has begun
consultations with the Bureau of Reclamation on its experience
with the state-of-the-art finite element programs on thermal
analysis of mass concrete. Based on our observations, the Corps
has initiated a 4-year research and development program for the
study of concrete cracking (estimated to cost $510,000). For
fiscal year 1984, WES will spend $110,000 to examine the state-of-
the-art finite element programs on thermal analysis and to examine
analytical models of concrete cracking.
Corps officials stated that it would not be practical to use
the programs on the first stage of the Lock & Dam No. 26 (Replace-
ment) Project since it is about 85 percent complete. However, in
fiscal year 1984, the Corps will examine the state-of-the-art
finite element programs such as Heat Flow, DETECT, and ADINAT for
temperature distribution and SAP-83 and ADINA programs for strain-
stress distribution in the second stage of the project.
We were further advised that the Corps acknowledges that an
expanded manual is needed to provide guidance on thermal analysis
and that the Corps needs to reassess its capability to provide
direction to its district offices on thermal analysis of mass con-
crete structures.
Directorate of Research & Development
Officials of the Corps' Directorate of Research and Develop-
ment believe that research and development aspects of thermal
analysis need to be strengthened.

10