Design Considerations for Roller
Compacted Concrete and
Rollcrete in Dams
by Robert W. Cannon
The traditional concepts of dam con-
struction during the past 10 years have
been altered by the uses of roller com
pacted eonerete and rollerete. How.
fever, there are diferences in the mate-
ial properties of rolled concrete and
Tollerete and theae are diseused. These
Gigferences should be omsidered in the
design of such structures.
Keywords: compacting: compressive
Asdngah concrete dam onereee dur
iy; elats mix properdoning roller-com-
acted concreter shear usengt tale
engi watercement rao
The concept of transporting,
spreading, and compacting con:
crete with lange equipment previ-
ously used for earthwork and com:
paction of rock fills has radically
changed traditional concepts of
dam construction during the past
10 years, The record setting: con-
struction of Willow Creek Dam by
the U.S. Army Corps of Engi
neers in Heppner, Ore., has at
tracted worldwide attention to the
possibilities for this type of place-
ment in dam construction.
‘The concept is equally applicable
for construction of slabs on grade
wherever there is sufficient
room for the equipment to oper
ate; however, such construction is
less dramatic and therefore at-
tracts less attention to its possibil
ities. The concept has become
known as roller compacted con-
50
erete (RCC) but has been referred
to by some as “rollerete.” In 1980,
ACI Committee 207 published a
state-of-the-art paper on the sub-
ject.!
‘The expression roller compacted
conerete was first used in 1975, in
the Tennessee Valley Authority's
‘General Construction Specifica-
tion No. G-48, for Roller-Com-
pacted Conerete.”” The specifiea-
tion was issued to “‘govern the
produetion and placing of no-
slump conerete by construetion
methods normally associated with
the placement of roller compacted
rock fills” The term rollerete was
first used in 1974 by John Lowe
III with reference to the emer-
gency placement of one-half mil-
lion eubie yards at Tarbela Dam.?
He deseribed rollerete as ‘‘con-
rete made from bank run gravel
and very little cement placed con-
tinuously by earthfill methods.”
‘The terms have been used inter-
changeably to date with no clear
distinction in terminology. The
basic difference in philosophy con-
cerns whether the end produet
should approach that of soil-ce-
ment or eonerete. This is particu-
larly true with reference to dam
construction. In the author's opin-
ion, there is a sufficient difference
in material properties between
rollerete and rolled concrete to re
quire a separation of the terms.
This paper will discuss the use of
both in dam construction,
Definitions
ACI Committee 116 defines con-
erete as ‘‘a composite material
which consists essentially of a
binding medium within which are
embedded particles or fragments
of aggregate; in portland cement
conerete, the binder is a mixture
of portland cement and water.”
Basically, the definition says that
the quantity of "binding medium”?
(paste) must be sufficient to embed
all the particles or fragments of
aggregate.
Tn the author's opinion, this im-
plies that: (1) the mix should have
sufficient paste to fill all aggre-
gate voids, to coat the entire sur-
face area of aggregates, and suffi-
cient excess to provide for proba-
ble variations in aggregate gra¢
tion, and (2) fine aggregates
should be separated from coarse
aggregates in aggregate handling
and batching procedures as a min-
imum measure of control.
Control of RCC consistency will
improve with the increase in num-
ber of separated coarse aggregate
sizes in the same fashion as with
conventional conerete.
Basically the difference between
rolled concrete and rollerete is in
CONCRETE INTERNATIONAL/DECEMBER 1985control of voids within the consol
dated material. Voids may be in
herent in the proportioning of the
no-slump mixture, may be the re-
sult of variations in aggregate
gradings, or may result from re-
ducing water contents to speed
placement. If the material con-
forms to the definition of con-
crete, paste will rise to the surface
with continuing passes of the vi-
rating roller. If the material has
insufficient paste to fill the voids,
continuing passes of the vibrating
roller will crush aggregates at the
surface.
Roller compacted concrete
(RCC) may be defined as concrete
which differs from conventional
no-slump conerete principally in
requiring a consistency stiff
enough to support the weight of
vibratory rollers, but having a suf-
ficient volume of paste to fully
consolidate under externally ap-
plied vibration,
Rollerete may be defined as a
conerete or soil-cement mixture
whose no-slump consistency may
‘equal or exceed that of roller com-
pacted concrete, but may fre-
quently contain an insufficent vol-
ume of paste to fill all voids and
coat all particles of aggregate.
Full consolidation may be as-
sumed when the compacted mate-
rial attains 98 percent of its maxi-
mum densification,
Compressive strength
When the material conforms to
the definition of concrete, investi-
gators have found that strengths
vary inversely with the water/ce-
mentitious ratios in the same fash-
ion as in conventional concrete.
This does not appear to be true
with rollerete of very high water!
cementitious material ratios as in-
dicated by Ernest K. Schrader in
his latest paper on Willow Creek.*
He states that conventional ideas
of water-cement ratio do not ap-
ply since strengths increased with
inereased water contents. The ex-
planation for this phenomenon
concerns the effect of entrapped
air voids on the strength of con-
crete.
5
Bea
Fig. 1 — Strength versus water-cement ratio.
In any batching operation, the
cementitious portion of a given
mix is held relatively constant.
Strength variations of solid eon-
crete are primarily a reflection of
variations in water demand. How-
ever, strength is also affected by
the air or void content of the eom-
pacted conerete. Most investiga
tors agree that a minimum
strength loss of five percent may
be expected for each one percent
inerease in voids.
For conventional concrete, this
effect is seen in the strength dif-
ferences in air entrained and non-
air entrained concrete for a given
twle ratio. The average water con-
tent of the interior conerete for
Willow Creek was approximately
215 pey (128 kgleu m) for 80 Ib (48
kg) of cement and 82 Ib (19 kg) of
fly ash, or an average water/ee-
mentitious material ratio of 1.92
by weight.
Concrete quality _ monitor
“CQM” test data from Exhibit 9.1
of Reference 6 indicates water
contents varied from a low of 189
1b to a high of 249 Ib. This is con-
siderably higher than the 182 Ib
water content in the mix investi-
gation program of reference." Wa-
ter-cement ratios of this magni-
tude are beyond published infor-
mation on conerete. Prior to Wil-
low Creek, Corps of Engineers
specifications generally limited
water-cement ratios to 0.8 in
dams.
In Fig. 1, the water-cement ra-
tio strength data from ACI 211°
(increased by approximately 20
CONCRETE INTERNATIONAL/DECEMBER 1985,
percent) has been extrapolated to
the average 28day strength data
for 6 x 12 tamped cylinders at
Willow Creek. It is quite apparent
from Fig. 1 that a large change in
w/c is required to have a signifi-
cant impact on the strength of
Willow Creek's interior concrete.
Approximately 17 Ib (7.7 kg) of
water is required to occupy 1 per-
cent of the entrapped air voids of
a cubie yard (0.76 cu m) of con-
crete.
The average 28-day strength of
the interior conerete at Willow
Creek was 1172 psi (8 MPa). Us-
ing the 5 to 1 ratio, a 1 percent in-
crease in entrapped air voids
would reduce strengths by about
60 psi (0.41 MPa). If, however, 17
1b (7.7 kg) of water were added to
offset the increase in void content,
the loss in strength would be only
40 psi (0.28 MPa) based on Fig. 1.
The result would be a net increase
in strength of 20 psi (0.14 MPa) by
adding 17 lb (7.7 kg) of water to
the mix.
From extension of the 211 curve
in Fig.1, it would appear that this
might hold true for any mix with a
water/eementitious material ratio
greater than 1.0 by weight. At
lower wie ratios, the increased
slope of the strength curve would
indicate inereasing strengths with
decreased water contents despite
a corresponding inerease in voids,
The optimum density studies by
John Lowe! shown in Fig. 1 and 2
do not confirm this. One possible
explanation is that the rate of
a55
te
5
IS
By
150
EJ
7
iE
IZ as
Ey COMPACTION] ENERGY:
EQUIVALENT] MOD AASHO
ze
IS 140!
3a 3 6 7 |
WATER CONTENT- % ORY WEIGHT
2500
7 mix a4 X54) PORTLAND CEMENT
e 53 “Sg rey as
= 2000 RI
5 Se
5 Sy
é %
2 a
e %
5 500 ey,
:
%
g
# 1000
3
3 j__WATER_ 0. 09
eevewrercvasty’ °° :
500 1
z 3 se 7 H
WATER CONTENT— % ORY WEIGHT
Fig. 2 — Compaction curve and compressive strength for rollerete mic.
strength loss versus voids is More rollerete has been placed
greater than 5 to 1 when the in- at Tarbela Dam than all other
crease in void content is accom- placements combined. In the 1975
panied by a decrease in paste vol- emergency placement of rollerete
tne for tunnel repairs, the eylinders
were wet sereened and compacted
Molding of test cylinders by tamping with a Kango bam
If the material has sufficient — mer.”* (An instrument commonly
paste to fill all the voids in the used in England for compacting
combined gradation of the aggre- test specimens of lean concrete.)
gates, it will fully consolidate un- The average 28-day strength of
der externally applied vibration. the 2 bag mix was 1800 psi (12.4
Full consolidation will be achieved MPa)
when paste surrounds all of the The 4 pef (0.64 kg/eu m) differ-
aggregate particles at the surface. ence in reported average wet and
If the mixture does not have dry densities indicates an average
sufficient paste to fill the aggre- water content of 108 pey (64 kg/ew
gate voids, it will not fully consoli- m) or an average w/e ratio of
date under vibration and some 0.55. In the 1978 placement for
‘other method of molding cylinders __ repair of the service spillway com-
must be employed for the tests to parative tests were run with eyl-
be representative of the average _ inders compacted with the Kango
conditions of the placement. hammer and with standard Amer
ican Association of State Highway
and Transportation Officials
(AASHTO) compaction proce-
dures,
Full consolidation was appar-
ently not achieved with the
AASHTO rammer as the average
strength was only 1835 psi (12.7
MPa) compared with 2700 psi
(18.6 MPa) with the Kango ham-
mer. A similar difference in
strengths was found between wet
sereened, tamped 6 x 12 and
tamped 9 x 18 cylinders at Willow
Creek. The optimum density tests
of Fig. 2 were compacted in a 14
in, (856 mm) compaction mold with
compaction equivalent to a modi-
fied AASHTO procedure.
The amount of compactive ef-
fort required to achieve full con-
solidation in molding eylinders is
dependent entirely on the rela-
tionship between the volume of
paste and the aggregate void sys-
tem. In John Lowe’s optimum
density study, the mixes with w/
(c+) ratios of 0.6 and higher meet
the definition of concrete and
could be consolidated by vibration.
The strengths of these mixes cor-
respond to normal. expectations
for conerete with their propor-
tions. The volume of paste as a ra-
tio of the air-free volume of mor-
tar for the 0.6 mix is 0.36. This
corresponds closely to the mini-
mum reeommendations of ACI 207
of 0.88 for interior mass mixes.
Tensile strength
The available test data on ten-
sile strength indicates the tensile
strength of horizontal cores from
RCC placements may exceed the
tensile strength of vertical cores
from the same placement by as
much as 20 percent. In comparing
test data, it is thus important to
consider the orientation of the test
specimen with respect to type of
test.
With vertical cores, a direct ten-
sile test would test tensile
strength of the horizontal plane
whereas a split eylinder tensile
test of a vertical core would test
strength in the vertical plane.
When comparisons are made of
tensile properties as a function of
52 CONCRETE INTERNATIONAL/DECEMBER 1985compressive strengths, it is impor-
tant to remember that the rela-
tionship also varies with eompres-
sive strength. The relationship is
also influenced by the method of
test.
Split cylinder tests generally
produce higher strengths than di-
rect tensile tests but are consid-
ered by many as being more rep-
resentative of actual tensile
strengths. Fig. 3 is a plot of ten-
sile /compressive strength test
data from cores taken from test
fills by the Tennessee Valley Au-
thority," the U.S. Corps of En-
gineers,"" the U.S, Bureau of
Reclamation (Upper Stillwater
Dam), the Construetion Industry
Research and Information Associ-
ation of London," and Willow
Creek Dam, The cores represent
tensile strength on the horizontal
plane. The 4.5vJé curve appears
to be the best fit for the RCC data.
Considering 6Vfé as best fit for
conventional concrete, the tensile
strength of RCC in the vertical di-
rection generally appears to be
about 75 percent of conventional
concrete with the same compres-
sive strength. In the horizontal di-
rection the strength of RCC might
be expected to approach that of
conventional concrete.
A lowering of the tensile/eom-
pressive strength relationship may
logically be attributed to reduced
aggregate bond due to the in-
crease in surface area of aggre-
gate and the decrease in paste vol-
ume. For rollerete, it naturally fol-
lows that even less paste volume
combined with increased voids will
further decrease the tensile prop-
erties of the material. This is indi
cated by the Willow Creek data in
Fig. 3.
Bond (tensile) strength on
horizontal joints
‘The principal factors which eon-
trol the bonding of one lift or layer
of conerete to another are: (1) the
condition of the surface to be
bonded, (2) the moldability of the
covering conerete, and (3) the
‘magnitude of work applied in con-
solidating the covering concrete. A
Fig. 3
pressive strength.
Vertical cores; tensile strength versus com-
Fig. 4 — Effect of exposure time on percent of bonded
joints,
number of misconceptions exist
concerning the achievement of
bond in RCC construction which
have led investigators to seek a
single sol
A summary of both published
and unpublished test data for bond
CONCRETE INTERNATIONAL/DECEMBER 1985
of vertieal cores is presented in
Table 1. The author was directly
involved in six of the seven refer-
enced placements. In addition to
the construction of Willow Creek
Dam, test fills have been con-
structed at Willow Creek and Up-
8TABLE 1 — Summary of vertical core test data on bond
Construction joints Lift joints
T Foints bonded
Joint
Maximum Number Bond strength
, Age size aggregate of Bonded
Prswem) | Gays) | in (im) cores (wervent) psi__ (Pa) Number Percent
TVA) % 30 (78) 12 75 188.16) 2100
TVA20) 7 15 Gay 2 % BS 2% 100
TVA2 1 15 G8) 3 25 =
TVA3@) 1 13 G8) 4 100 153.06) ”
TAs 1 13 G8) 8 13 °
VAS 1 1568) 2 0 290 2.0)
WESA(4) 2 30 (78) n 0 = 100
WES2) 45 aM 10100
LC) 1 18 (88) 20 % 180 (1.24) 20 65
16-265) 1 80 (78) EI “ 325 (038) 2 a
16-35) 1 80 (78) é 100 100 (0.88) 6 50
La) 1 Bo (8) 6 50 io (076) 6 8
| we76) 1 15 (8) 2 60 30214)
10-76) 30 (78) 8
Leas) 1 7 (is) n 9 25 (1.69)
LC) 1 30° (78) n 3 2m
WES.36) 1" 1% a) Tt °
WES r 13 (8) 2 °
WES-46) 1 15 G8) 2 58 6 (on)
WES 1 7% (9) 2 22 120 (082)
WES 1 15 (3 6 100 225 (1.55)
[ariam 6 15 G3 12 8
| cIRIAM, 6 13 Gay 2 o
| arian 1 15 (8) 92 8 235 (1.62)
CIRIAG) 2 1s GB 2 100 ao (097)
CIRIA® a 15 (8) 58 96
fronted copsraetn jane,
‘Goneete of umpable consistency
"Speak ibang wis with nerensed paste volume
Abbreviations: PrN), indents project, teat ll number, and reference number: TVA, Tennesae Valley Authority: WES, Waterways Expriment Station,
Ay Coe 2 egret “Gtlsbarg, Mien; UC, Cast Crank tat fils, U.S. Army Corps of Engineers, Pordand, Ore, distree CHRIA, Constraction
Induiry Resear and Informatlon Assocation, Lindo, England
per Stillwater dams. Only limited of laitance. The pereent of bonded TVA’s Tims Ford Dam" the time
test data is available to date on joints from Table 1 is plotted in lapse between lifts was appro:
core recovery for these place- Fig. 4 against time of exposure for mately 17 to 18 hr. Beeause of its
ments. covering mixes of maximum ag- location, the placement was
Condition of the surface gregate sizes of 3 in. (76 mm) plus shaded during exposure. No water
The first misconception arose and 1'% in. (88 mm) minus. If we curing was applied but the surface
from the lack of bleeding and lait- define a failure rate of 90 percent gave no evidence of drying prior
ance in RCC placements. Previous or more as a cold joint, cold joints to being covered.
experience, in conventional con- may be expected to oceur any time Ambient temperatures in Vicks-
struction clearly indicated that after final set of the conerete. burg, Miss., were definitely on the
achievement of bond was depen- Bond is clearly affected by the warm side for the third test fill
dent on the removal of laitance. It maximum size of aggregate. Pro- placement at WES."* Lack of bond.
also showed that the best methods _jecting the curves to zero, bond on some of the treated 1 day old
of joint treatment, such as wet would indieate an influencing ef- joints can be attributed to the
sandblasting, removed the least fect of aggregate size on critical stiffness of the covering mix.
amount of excess material from timing for cold joints but there However, at least one mix with a
the surface. It rather naturally does not appear to be any logical definite bondable consistency also
followed that if there was no lait- basis for this. failed to bond. On the fourth test
ance there was no need for joint ‘There is some indication that fill, the same mix bonded to an un-
treatment. cool, moist, ambient conditions treated 1 day old joint. The treat-
Testing of RCC placements may extend the timing of cold ment of cold joints appears to have
clearly indicates that cold joints joints somewhat beyond final set. little influence on the bonding of
are not dependent on the presence In the initial RCC placement at RCC placements.
st CONCRETE INTERNATIONAL/DECEMBER 1885,‘Test data indicates that the con-
crete covering a cold joint should
have a slumpable consistency to
achieve bond. This is indicated by
the successful bonding of @he in-
advertent, overwet coverimg mix
in TVA’s second test fill and the
significant inerease in bond of the
slumpable eonerete in the fourth
WES test fill. Testing also indi-
cates that saturating the joint with
water prior to covering may be
detrimental to bond development.
Moldability of covering conerete
The ability of conerete to mold
to any surface depends on the tex-
ture of the surface, the shape and
volume of coarse aguregate in the
mix, and the flowability of the
mortar, Once the lower lift has
completely hardened, bond is de-
pendent on the interlocking of the
paste of the covering mix into the
pore structure of the sound paste
of the lower lift. The flow of paste
at the surface improves with the
smoothness of the surface, de-
creased volume of coarse aggre-
gate, and the increase in volume
and consistency of the paste. If the
lower surface has not completely
hardened, bond will be improved
to the extent of remolding and
combining of the lower surface
paste with the paste of the cover-
ing mix.
For bond on cold joints, it ap-
pears imperative to use a bonding
mix of slumpable consistency.
Such eonerete ean still be consoli-
dated with large vibratory rollers
if spread in 3 in, thick layers and
covered with 8 to 10 in. of no-
slump consisteney conerete prior
to consolidation. A bedding mix is
not necessary if tensile bond
strength is not required sinee ade-
quate moldability for watertight-
ness and shear strength can be
achieved with proper mix design.
Minimum consolidation
In the author's opinion, one of
the problems in developing bond
strength on 1 day old joints is the
amount of emphasis which has
been given to establishing the
minimum number of vibrating
passes. In many instances, the au-
thor believes that bond would have
SHEAR-PRICTION FACTOR OF SAFETY OF 3
bor TT
WoL. Regte
Wels Rest
3.5
Var/vot
en ee eee
f¥e = 2000 pst (13.8 MPa)
prt = 200 psi (7.38 MPa)
TT T i
to Resist Sliding
to Resist Overturning |
‘VOLUME REQUIRED FOR SLIDING RESISTANCE AS A RATIO OF MINIMUM REQUIREMENTS
oo.
0 2 Wo €0 80 100 120 1h0 160 180 200
‘BOND STRENGTH IN PST
0.2 0.3 Ok 0.5 0.8 0.7 0.8 0.9 1.0
BOND SHRENGTH AS"A RATTO OF CoNGRBTE’ TENSILE StaENGTH
Fig. 5 — Effect of bond strength on volume requirements.
been substantially improved by
additional passes. The number of
passes of the vibratory roller is not
the controlling: economic element.
‘The bringing of paste to the sur-
face (for mixes with adequate
paste content to achieve bond) is
an indication of approaching max-
imum densification. However, it
may not be the right stopping
place for bond. Most investigators
stop at this point because paste
accumulates on the roller under
continued vibration after paste
reaches the surface. If a second
layer is placed over the first layer
before the first layer reaches ini-
tial set, the first layer will receive
the additional needed vibration
without the problem of paste ac-
CONCRETE INTERNATIONAL/DECEMBER 1985
cumulation. This was done in the
initial test fill at Tims Ford and
may explain the good bond with 8
in, aggregate,
Shear strength of horizontal
joints
‘The shear strength require-
ments for joints is normally based
on a factor of safety for a combi
nation of bond strength (cohesion)
plus friction. The coefficient of
frietion for eonerete sliding on a
relatively smooth conerete surface
is approximately 0.7. The plane of
failure increases with joint incli-
nation and bond. The coefficient of
friction is generally taken as 1.0
when intimate contact of the two
surfaces is assured. The coeffi-
51m
H = 200 ft. (61m)
He 150 fe. (yom)
. ev
i
t+
TT Chet
20 yo 60 80 100 120 140 160 180 200
TT
SHEAR-PRICTION FACTOR OP SAPETY AGAINST SLIDING
BOND STRENGTH OF HORIZONTAL JOINTS IN psi
Fig. 6 — Minimum section dam; effect of design head and joint strength
on shear-friction factor of safety.
cient of friction for the interior thor is not aware of any design
concrete at Willow Creek from agency” or organization which ac-
laboratory investigations was cepts a shear-friction factor of
0.88." A considerably higher value safety less than 3 for normal op-
was obtained from the relatively erating conditions, For net oper-
green lift joint of the test fill. This ig heads (H) less than 300 ft,
is not surprising, however, since Fig. 5 indicates that the size of the
the unconfined shear strength of dam will be controlled by mini-
concrete may range from 25 to 35 mum section requirements as long.
percent of the compressive as the minimum bond strength
strength. The design engineer (JS) exceeds the following:
should not be misled into thinking {18 Bond) =
such tests represent the strength i
Of dam eonstruction joints. henry
Bond strength has already been _Bond is also plotted as a ratio of
discussed. In the design of dams, concrete tensile strength in Fig. 5.
the importance of bond strength Under the best conditions, the
can be seen in a plot of bond _ strength of bond on hardened con-
strength requirements versus the struetion joints should not be as-
difference in volume requirements sumed greater than approxi-
for the size of dam to be con- mately 60 percent of the tensile
trolled by its stability against strength of the weaker material on
overturning or controlled by a _ either side of the joint. For roll-
minimum shear-friction factor of crete, the bond strength on hard-
safety of 8. (See Fig. 5.) The au- ened joints without special treat-
ment, is extremely questionable.
The shear-friction factor of safety
for rollerete construction should,
therefore, be based on frictional
resistance alone.
Durability
When the volume of paste in
RCC mixtures constitutes a suffi-
cient binding medium to embed all
the particles of aggregates, Fig.
2.2.1, 2.2.2, and 2.3.2 of Ref-
erence 1 clearly indicate com-
pacted densities of 98 to 99 per-
cent of theoretical air-free densi-
ties when compacted by vibration.
imilar results were obtained both
in the laboratory and from the
densities of test fill cores as re-
ported in Reference 17. It is thus
apparent that very little, if any,
air can be successfully entrained
into no-slump eonerete of the re-
quired stiffness necessary to sup-
port the weight of vibrating roll-
ers. The difference between RCC
and rollerete in resistance to
cycles of freezing and thawing
would be limited to their differ-
enees in permeability. Both mate-
rials will be subject to surface
spalling under severe exposure
unless protected by an insulating
layer of frost resistant concrete.
Without protection, the depth of
spallage would be deeper and oc-
cur sooner in undrained portions
of the more permeable rollerete.
No one knows at this stage of
‘experience with rollerete, what ef
fect percolating water will have on
material stability with time, We do
know that free lime is leached
from the cement jell as water pei
colates through cracks. The prit
cipal source of leakage with roll-
crete will be the horizontal con-
struction joints. If the leakage rate
is high ‘and the cementitious
binder is thin, there is a possibility
of eventual complete loss of binder
at the joint.
Joint friction is directly affected
by uplift conditions along the joint.
However, for bonded joints there
is a considerable time lag between
changes in head water and change
in uplift pressure depending on the
rate of percolating water. For a
56 CONCRETE INTERNATIONAL/DECEMBER 1985,quality joint, the time lag may be
a matter of weeks or months
whereas the uplift pressure on an
uncemented joint of high leakage
rate may adjust completely in a
matter of hours.
Fig. 6 shows the effect of design
head on the factor of safety
against sliding assuming compres-
sion over the entire base and uplift
pressure varying from a maxi
mum upstream to zero down-
stream. Minimum seetion require
ments are based on. stability
against overturning. From Fig. 6,
a 150 ft high dam designed for a
normal head of 130 ft would re-
quire a minimum bond strength of
approximately 55 psi for a design
shear-friction factor of safety of 3.
Under porous joint conditions the
safety factor would approach 14
under normal head water and ap-
proach 1.0 under a maximum flood
of 150 ft. Under these conditions,
the designer should be very con-
servative in his assumptions of
bond on horizontal joints.
Comparative costs
Cost comparisons may be mis-
leading unless two separate and
distinct designs are bid competi-
tively for the same job. U.S. Army
Corps of Engineer estimates for
alternate designs of rockfill and
rollerete for Willow Creek Dam
were competitive. Bid prices for
the rollerete dam were $11 million
less than Corps estimates. Costs
appear definitely to favor rollerete
over rockfill assuming Corps esti-
mates for rockfill were more accu-
rate than estimates for rollerete.
‘The basie difference in design
criteria for rollerete and rolled
conerete dams less than 300 ft
high is the difference in allowable
bond on construction joints. Mini-
mum section requirements for
dams less than 300 ft high, are
basically controlled by resistance
to overturning under maximum
design head. Sliding resistance is
generally based on maximum nor:
‘mal pool elevations because of the
extended time required to effect
changes in uplift on construction
Joints. For sliding resistance of a
minimum section, bond require-
ments will vary with design head
as indicated in Fig. 5. The differ-
ence in volume requirements be-
tween a minimum conerete sec:
tion and a rollerete section of un-
certain bond will decrease as the
difference between maximum de-
sign head and normal operating
head inereases. The low operating
head at Willow Creek was a major
factor in whatever economic ad-
vantage rollerete may have had
over rolled concrete.
‘The cost of rollerete for Willow
Greek was $20 per eubie yard ($26
per cubie meter) according to the
latest cost figures. Its volume was
1.4 times the minimum volume re-
quired for overturning at its max-
imum design head. The break-even
cost for a minimum section of
rolled concrete would be $28 per
‘cubic yard ($37/eu m). A minimum
section rolled concrete dam would
have required a minimum bond
strength of approximately 80 psi
(0.21 MPa) for equivalent stiding
resistance to Willow Creek. In the
author's opinion, the added cost of
‘a separation of fine and coarse ag-
gregate for control of gradation
and the increase in cementitious
material costs to provide the
needed minimum paste volume,
should not have exceeded $2 per
cube yard ($2.60/eu m)..
Upper Stillwater Dam in Utah is
a 275 ft (84 m) high rolled con-
erete dam. To insure the needed
bbond strength on horizontal joints,
specifications required that ‘80
percent of all test cylinders ex-
ceed 4000 pounds per square inch
(27.6 MPa) compressive strength
at one year.”” The higher than nor-
mal strength requirements re-
sulted from a lower than normal
relationship between tensile
strength and compressive
strength. (See Fig. 4.) The bid
prices for RCC ranged from $22 to
$30 per cubie yard ($29 to $39 per
cu m) for the 20 contractors bid-
ding. on the job. Approximately $2
pey ($2.60/eu m) should be added
to this price for sand which was
furnished under a separate con-
tract. A comparable design for
CONCRETE INTERNATIONAL/DECEMBER 1985
rollerete could require as much as
3 times the volume of Upper Still-
water to maintain the same factor
of safety against sliding depend-
ing on frictional resistance alone.
In the author's opinion, material
costs could have been substan-
tially reduced on both projects by
use of a more realistic specifica-
tion requirement for fly ash. In
particular, it makes no sense to
specify very low loss of ignition
limits for no-slump conerete since
entrainment of air is not consid-
ered. The primary function of fly
ash in RCC, beyond its long-range
pozzolanie strength contribution,
is to inerease the volume of the ce-
mentitious binder to the mini-
mums needed to coat all the parti-
cles of aggregate and fill all the
aggregate voids while lowering its
heat generating characteristics
This does not require long-term
storage or extensive prequalifica-
tion testing.
Conclusions
Designers need to be aware
that rollerete is not simply a lean
mixture of conerete which ean be
used in the design of concrete
dams without accounting for the
differences in its properties in
proportioning section require-
ments. The variability of the ma-
terial for both rollerete and rolled
conerete must be considered in as-
sumed values of bond for sliding
resistance. We have insufficient
experience, at this time, to judge
the stability of porous materials
subject to high pressure water
percolation when the water/ee-
mentitious materials ratios of
those materials exceed present ex-
perience levels.
‘The absence of laitance in RCC
and rollerete does not change
needed concern for bond on cold
joints, A covering mix of slump-
able concrete appears essential to
insure bond on eold joints. Supe-
rior bond is obtained when the
surface of the cold joint is rela-
tively free of moisture.
Exeept for differences in mate-
rial costs, there appears to be lit:
tle difference in the unit cost of
roller compacted eonerete in com-
stRecommended
Practice for
Evaluation of
Strength Test
Results of
Concrete,
(Reaffirmed
1983) (214-77)
Statistical methods
provide valuable tools for
assessing results of strength
tests, and such information
is also of value in refining
design criteria and
specifications. Discusses
briefly the numerous
variations that occur in the
strength of concrete and
presents statistical methods
useful in interpreting these
variations.
ACI Committee 214,
1977, 14 pp. $7.75; to ACI
members $5.50.
>
AMERICAN
CONCRETE
INSTITUTE
P.O. Box 19150
Detroit, Mi 48219
58
parison with rollerete. There is a
significant difference in volume
requirements, however, if the dif-
ferenees in properties are ac-
counted for in design.
References
1. ACI Commitee 207, ‘Roller Com
pacted Concrete,” (ACI 207.5R-80),
American Conerete Institute, Detroit,
1980, 22 pp,
2. "General Construction Specifica-
tion No. G-48 for Roller Compacted
Conerete,” Tennessee Valley Author:
ity, Knoxville, Aug, 1975, pp. 1+
's. “Repairs at Tarbela Dam Race
Spring Snowmelt,” Engineering
News-Record, V. 198, Oct. 31, 1974, p.
8
4, ACI Committee 116, “
and Concrete Terminology,” (ACT
116R-78), American Conerete Insti-
tute, Detroit, 1978, 50 pp.
'5, Schrader, Ernest, and MeKinnon,
Richard, “Construction of Willow
Greek Dam,” Concrete International:
Design & Construction, V. 6, No. 5,
May 1984, pp. 38-45
6. “Concrete Report; Willow Creek
Dam—World’s First All Roller Com.
pacted Concrete Dam,” U.S. Army
Corps of Engineers, Walla Walla Dis:
triet, Aug. 1984
7. “Willow Creek Lake, Heppner
Oregon—Supplement 1 to GDM 2
Phase Il—Maine Dam, Spillway, and
Outlet Works—V. 1," U.S. Army
Corps of Engineers, Walla Walla Dis
thiet, June 1981.
8. ACI Committee 211, “Standard
Practice for Selecting Proportions for
‘Normal, Heavyweight, and Mass Con-
crete (ACI 211.1-81)," American Con:
erete Institute, Detroit, 1981, 82 pp.
9, Lowe, John III, “Use of Rollerete
in Barth Dams,"” Proceedings, CIRIA
International Conference on Rolled
Conerete for Dams, Construction In-
dustry Research and Information As-
association, London, 1981, pp. Wl
Ws.
10. Chao, Paul C., “Tarbela Dam—
Problems’ Solved by Novel Con.
cretes,” Civil Engineering—ASCE, V.
50, No. 12, Dec. 1980, pp. 58-64.
i1. Gannon, Robert W., “Compac-
tion of Mass Concrete with Vibratory
Roller,” ACI JOURNAL, Proceedings V.
T1, No, 10, Oct. 1974, pp. 506-5138.
42. Cannon, R. W., "Bellefonte Nu-
clear Plant—Test for Compaction of
No-Slump Conerete Next to Form:
work,” Progress Report No. 2, Ten-
nessee Valley Authority, Knoxville,
‘Aug, 1974, 5 pp.
18. Cannon, R. W., “Bellefonte Nu
clear Plant—Roller Compacted Con-
crete—Evaluation of Core Test Re-
sults,” Report No. CEB-76-39, Ten:
nessee Valley Authority, Knoxville
1977, 6 pp.
14, Tynes, W. O., “Feasibility Study
of No-Slump Conerete for Mass Con:
crete Construetion,"” Miscellaneous
Paper No. €-78-10, US. Army Engi-
neer Waterways Experiment Station,
Vieksburg, Oct. 1973, 29 pp.
15. “Lost Creek No Slump Roller
Compacted Concrete Investigation,”
Final Report, U.S. Army Corps of
Engineers, Portland Distriet, North
Pavifie Division Materials Laboratory,
‘Troutdale, Mar. 1974, 8 pp.
16, Saucier, K. L., “Investigation of
No-Slump Roller Gompacted Conerete
(RCC) for Use in Mass Concrete Con-
struction,” Proceedings, CIRIA Inter-
national Conference on Rolled Con:
crete for Dams, Construction Indus-
try Research and Information Associ-
ation, London, 1981, pp. 9-1 thru 9-7.
17. Dunstan, M.R. H., “Rolled Con-
crete for Dams—Construetion Trials
Using High Flyash Content Con-
crete,” Technical Note No. 106, Con-
struction Industry Research and In-
Formation Association, London, May
1981, p. 34.
18, Design of Gravity Dams, U.S.
Bureau of Reelamation, Denver, 1976,
558 pp.
Recined and reviewed under Inet publi
Robert W. Cannon, a member of the
Inetitute for more than 20 years and a
Fellow, 18.0" consulting engineer in
Maryville, Penn. He ts a member of
ACT Committers 207, Mass Concrete:
540, Nucloar Structures; 359, Anchor
aie to Concrete: and teas recently ap
int the ne Concrete Materials
esearch Council. A graduate of Geor.
ra Intute of Techno, he retired
from the Tennessee Valley Authortt
‘in 1980 after 31 years as a research
nd design engineer specializing im
anerete dams and power plant sirua-
tes hte san edna om
such subjects us roller compacted con
Grete fy aah sea, and Slee! ombed-
‘ment
CONCRETE INTERNATIONAL/DECEMBER 1985