UDC 666.97.017;620.

173

Concrete subjected to high rates of loading in

.
compreSSIon
B. P. Hughes, PhD, BSc, CEng, MICE, MIStructE and R. Gregory, PhD, BSc, CEng, MICE

UNIVERSITY OF BIRMINGHAM: DEPARTMENT OF CIVIL ENGINEERING*

SYNOPSIS effect of rate of stressing upon the compressive

Compression tests on concrete specimens have been strength of concrete are compared in Figure 1. Both
carried out at very high rates of loading with a specially Watstein(4) and Evans(5) found similar increases in
constructed drop-harnmer apparatus which enabled load- strength for average rates of stressing exceeding 830
time and strain-time curves to be determined. The effects N/mm 2 s compared with strengths at 830 N/mm2 s.
of varying the length and platen restraint of the speci- Evans attributed the differences in his results at lower
men, mix proportions, maximum aggregate size, age at rates of loading from those of other investigators to
test and type of cement have all been investigated and the type of testing machine used, and suggested that
compared with results for both static tests and earlier the crushing strength was essentially constant for
tests at high rates of loading. hydraulic testing machines at all loading rates up to
0·05 s to failure. However, other factors which varied
Introduction in the investigations shown in Figure 1, such as size
and shape of test specimen and constituents of the
Two types of test to determine the behaviour of
concrete, could also be significant. Although the
materials subjected to shock loading, 'impact tests' and
aggregate/cement ratios for Watstein's tests varied
'tests at high rates of loading', have been carried out
between 3·5 and 7·5, only Evans attempted to draw
for many years. The forces applied to the test speci-
separate curves for different aggregate/cement ratios
mens in impact tests were, in general, not well defined
of 3, lO, 14 and 18. No investigations have been car-
and the 'impact strength' was measured in arbitrary
ried out at very high rates of loading, e.g. greater than
units, whereas in tests at 'high rates of loading' the
about 200 kN/mm2 s t (or faster than about 250 (.LS to
applied forces and, in many cases, the deformation of
failure), and it is evident that little is known of the
the specimens also, were measured.
effect of factors such as type of aggregate or cement,
Tests at high rates of loading not only assist predic-
mix proportions, etc.
tion of failure stresses and strains in components sub-
jected to high rates of loading but also provide
valuable information in the development of failure
Stress waves
criteria of materials, especially those which exhibit Various types of stress wave can theoreticalIy be
time-dependent strains. propagated through a body(6). The velocity of propa-
Although the earlier studies concerning impact gation of a stress wave through a bounded solid has
strength of concrete were concerned with the frequent been shown to be dependent upon the dimensions of
failures during driving of precast piles, there are the solid and the wavelength(6,7). Figure 2 shows the
numerous other examples, such as structures exposed theoretical relationship between the actual velocity of
to earthquakes or explosions, foundations of drop a stress wave through an elastic medium, c, the elemen-
forges and carelessly handled precast units. tary velocity, co, the diameter of the rod, d, and the
The results of earlier investigations(1-5) into the wavelength A. It can be shown that Co = (<./p)!, where
<. = Young's modulus and p is the density of the
medium. The elementary velocity(6,8) is the velocity of
*Dr Hughes is Reader in Concrete Technology and a Senior
Lecturer in Civil Engineering. Dr Gregory is now with the
Upper Tame Main Drainage Authority. tThe alternative units GNjm' S are numerically identical.

25

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research: Vol. 24, No. 78 : March 1972

E 200 E
E E
.l Z
Z • Jones and Richart 0
'"
6
>-
-<
lS0
o Thaulow
o Moore
.l Authors (see later)
1".-., " '"', ~

'"
>-
-<
I
I
>- l'> Watstan >-
"
Z
w
160
'V Evans
} right-hand scale
"Z
w

'"t;; '"t;;
~
140 ~

0 0
w w

"-<>- 120 "-<>-

Z
Z w
w
U U

'"
w
Q. 100
'"
W
Q.

-< -<
v>
v>
-< -<
I SO I
>- >-
"Z
w
"
w
Z

'">-
v>
60
0-1 loe 1010
'"
I-
v>

AVERAGE RATE OF STRESSING-kNjm's

Figure 1: The effect of rate of stressing upon the compressive strength of concrete.

1-0

0-9

O-S

.f-
Co
0-7 v
o
~01
I~+-+-~~~~~~a~0-2
0-6
~ -0-)
- ~04
0-5 ~------0~--4------~0--S-------1~-2--------1~-6--------1~-S--------2L-0----~~ 0-5

dj).

Figure 2: Relationship between clco and dl). for various values of Poisson's ratio. (After Bancroft (7 ) .

a longitudinal stress wave of infinite wavelength along (suffixes 1 and 2 refer to the first and second media
a rod (i.e. when dJA is small). respectively). An experimental investigation by Ripper-
Ifa stress wave meets a change in cross-section or an ger and Abramson(lO), however, showed that the actual
interface between two media, part of the wave is re- magnitudes of the reflected and transmitted waves
flected at the interface and part continues into the were higher than those predicted by theory.
second medium. A simple theoretical approach (9,10)
has shown that the forces in the incident, reflected and Apparatus and experimental technique
transmitted waves, FI> F/ and F z' respectively, are
Compression tests at high rates of loading can be
related as follows
carried out on metals and plastics with quite small
specimens, long slender input and output bars being
used to provide and measure the 10ads(11). This tech-
nique facilitates accurate control and measurement of
loading and strain in the specimens.
Concrete is not as homogeneous as metals; hence
larger test specimens and therefore larger test appar-
where p, A and c are the density, cross-sectional area atus become necessary. Specimens 102 mm "quare
and elementary velocity for a particular medium were selected and the drop-hammer apparatus shown

26

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 COli rete ubj ('led to high rates oJ loading ;n compre ion

in Figur 3 \ a d ign d < nd on (ructed. arly t I

u ing < hamm r n tru ted fr m parate teel block
bolted t gether \ r ~ lind t be un ti fa try, nd
o a new hammer \ cighing appr im tely 106 kg wa
on tru ted from < iogle block of mild teel .
hamm r c uld fall through a height f 0 er I· m
aligned b two I mm diameter mild- teel guide rail.
he hammer hOll ed three bearing tw running along
one rail aod one long the oth r.
I tri al-r i tan e train gauge mounted on the
load oillmn, a h wn in Hgure 4, mea ur d th I ad.
The train on om f the pc i men aI
mea ured byele trica l-r i tan train gauge. II the
gaug w r onn ted into uitabl Wh al t ne bridge
circuit, and th out- f-balance oltage during the
te t were di played n an ill pe r nand
recorded by an 0 cillo cope 35 mm till camera. he
imple and highly !fecti e tri.gg ring dice ho\ n in
figure 4 enabled a uitabl oltage to be fed into the
eternal trigger ~ r a ingl \ ep f the 0 illo cope
when the no e of th hammer truck th platen a e
the pccimen. Th c.amera hutter wa operated man-
ually before and ft r til imp<et.

p rim ntal programm

Figur~ J : Gt'IItra/ view 0/ drop-hamm r apparalu .
The eITe t of arying cement/wat r ratio • g , type
and ize of coar aggr gat cement c ntent and
oar /fL ne aggregate ratio \ er all in e ligated. E -
cept when the coar e/fine aggregate rati \ a pe i-
fically aried, all the conerele had the ptimum
coar e aggregate content: that i , the volume fraction
f coar e aggregate, a, which pro ide rna imum
workability for given olume fraction of cement, c,
and water, 1\1(1 2 ).
The coar aggregate were hame Valley gra el
fr m k nd n, e; cru hed lim tone from
hipping odbury, louce t hire; rll hed granite
from Mount orrel Leice ter hire· and rll hed ba alt
from I e Hill , hrop hire. hame alley and from
o
kendon wa II ed throughout. Both ordinary
P rtlaod and high- lumina cement re II d.
The ab orption of the aggregate w determined,
in gen ral, by u ing an a curate turated-air tech-
nique lU ) but that for th 0 kendon coar aggregate
wa obtained a et out in B J2 414 ). he required
==bot"""
plat.,.
--4-11-.-1
-- + +--;lL=:1:::0-
'i eight of aggr gat and water for each mi wer
10>4 colli ... - - t--t---i
placed in an airtight drum and left overnight to en ure
th.t the air-<iry aggr gat \: ere fully aturated. The
con. nt f th drum were placed in tbe mi er the
~ 1I0~ ing day and the cement wa add d imm diat I
before mi ing.
Elcv n lO2 111m ubes er ea t : three for unia ial
compre i e tr ngth t t, u ing MGA pad 415); and <On<rete but
eight for rapid te t (i.e. t t t high rat of loading
four with MGA pad and fOllr \ ithout. ur u
\ ere al 0 t ted by mean of reen' balli tic pendu- Figure 4 : Diagrammatic repre enlalion of Iht drop-hammu
lum Cl6 ) • . our 244 x 102 x 102 mm pri m were al 0 apparalus and triggering cirCllil.

27

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of COllcreJe Re earch : Vol. 24, o. 7 : Mar II 1972

ca t for rapid te ting from a few mi e . he pecim n The energy ab orbed b 244 mm long pri m at
were cured in moi tairatl7! ± 1° and tie t95 % f: ilure wa con id rably greater than that for cube .
eelati e humidity until due for te ting. Hence th nergy remaining in the hammer aft r fail-
ure of the pri m , and ther fore the damag cau ed by
Mod of failur the follow-through of the ha mm r, wa Ie. wo
The failure pattern for ube and pri m were con of concr te generally r main d one adjacent to
imilar to tho e found in tatic te t . each pI ten , a h wn in igur 6 and 7 for pri m
.1 w-friction' MGA pads, developed by Hughe from both the original 'built up' hamm r and the
and Bahramian cu1 to reduce the effect of platen fric- tatic te t . he failure were imilar and occurred in
tion in tatic te t , were in e tigat d for u e in rapid th central portion of the pri m with a 'cone' of
te t u ing both cube and pri m . The pri m te t al 0 concr te at ea h end . igure ho that failure for
gave an indication of the po ition of the maximum a few t t on pri m with MAp d , u ing the
compr i e tre in the pecimen . he initial te t , original 'built up' hammer, con i ted of c mplete
u ing 'a made MGA pad with a Ii ral I yer f cru hing of the lower part onl , indicating that lover
Moly lip grea betw en the Meline and aluminium tr e e i ted higher up th pri m.
heet howed that ery little, if any gr a e wa The abo e te t with M A ad indicated that
forced out. Only a ery thin film of grease remain tr riation along th length of the pecimen
after failure in tatic te t, ince any urplu i clod d could cau e failure in one part only of a long (pri m)
a oon a a mall load i applied. The highly i ou pe im n. Tn te t ithout MGA pad, the mod of
beha iour in rapid t t could cau lateral c pan ion failure of both cube and pri m in th dr p hammer
in contra t to lat ral r traint, and w eliminated by w r imilar to tho e in tatic t t, probably becau
loading the MGA pad (to about 50 ) betv e n of the effect of platen friction .
tccl platen prior to u e in th t.e t , to force out the It can th refore be concluded that tr ariation
exce grea and leave oly a • olid' low fri tion layer along the length of cube at fai lu re ere Je than for
on each urface of the adjacent m mbrane. omplete pri m . In addition , the u in rapid te t of G
d (ruction of the pecimen did not occur for ome pad from which the e c grea e had been removed
of the tronger concretes and it wa po ible to ob r e by pre-loading wa a ati factory a in tatic t t,
the typical 'con 'of unbroken concr t that are ob- that i ,an appro imation to uniaxial compre ion wa
tained when cube are te t d without G pad. obtained. Hence the Mol lip grea cud in the MGA
Only one or two in tance of partial de truction pad did not 'br ak do n' at high rate of loading.
occurred for cubes with MGA pad ,wher ertical hiddi ter and al ero ll 7) found that hort alum-
cracking imilar to that obtained in tatic te t could inium bar te ted bet ccn teel ar at high r te of
be seen, a in Figure S, which how both a cube from loading did not 'barrel' when th end had been
the rapid t t with the original 'built up' hammer and lu ricat d ith a fine lay r of geea e containing
a cube feom the tatic te t, u ing MGA pad . m Iybdenum di ulphid, hich agr e with th above.

Figllr~ 5: V~rricol crocking Inol occurrtd in on~ or IWO cQS~ 0/ CuMS wilh MGA pod .

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Concrete ubjecled 10 high ro/(' oJloadillg ill compression

Figure 6 : Pri m afttr fui/url! ill ropid It I ( Iro ...n in direclion Figure 7: Prism afrtr faJiU/(' in srO(l( I I (~"Q""" ill direclioll
(tsred). /l·sted).

Figurt 8: Pri illS afttr failure ill rapid /e /. howi"g compint shatttring of lite /OM;tr por/i()It
onl>, ( hown in dirulion IfSI('d).

29

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research: Vol. 24, No. 78 : March 1972

Typical traces Comparisons of load-time traces obtained by re-

Figure 9 shows typical load and strain traces for leasing the hammer from a low height onto the load
cubes with MGA pads obtained in tests on 'strong' column and corresponding traces obtained on a 102
concretes. The strain traces increased steadily to a mm square steel prism of approximately the same
peak with small oscillations for all concretes and indi- height indicated that errors introduced by treating the
cated the average longitudinal strain over the central load column and platen above it as a single unit were
5,[ mm of the cubes. The load traces for 'weak' con- small. This is possibly due to the fact that the value of
cretes had two distinct peaks, the second peak having -0'76 for the wave reflected at the interface between
a greater amplitude than the first. For the 'strong' the load column and the steel base is the most signi-
concretes, the first peak tended to become indistinct. ficant. Successive transmissions and reflections of a
A residual oscillation of the load-time traces after wave at and above the top change in section of the
the specimen had failed and the load had returned to load column would be expected to reduce the effect of
zero can be seen in Figure 9 and more clearly in such a wave very considerably by the time it returned
Figure 10 which shows typical load-time curves. This to the gauged position.
residual effect is due to successive reflections of the If the stress waves reflected down through the speci-
stress wave at the base and top of the load column men from the top of the specimen and parts of the
after failure. hammer are considered as part of the stress wave
caused by the impact, Figure 11 explains diagram-
Evaluation of the true failure stresses matically the passage of the stress waves through a
Two stress waves were generated when the hammer specimen and load column. In cases a, b, c and d, the
struck the specimen, one passing up the hammer and load column measures directly the load at the base of
the other down through the specimen and load column the specimen after allowance is made for the time lag
into the base. Since a stress wave is split into a reflected for the stress wave to travel from the base of the
wave and a transmitted wave at each interface or specimen to the gauges on the load column. In case e,
change in section, the initial wave passing down the the indicated load is increased, owing to the wave
specimen rapidly became a complicated series of str'ess reflected from the base of the load column. Case f
waves travelling up and down the system. Table 1 becomes even more complex, since a tensile wave has
shows the proportions of a unit incident wave (F = 1) been reflected from the top of the load column. Point
reflected and transmitted at each discontinuity, for B in Figure 10 is given by a rapid increase of load,
waves travelling either up or down through the appar- after either a low rate of increase, or decrease, of load.
atus. These were calculated by using equations 1 and 2. Figure 11 implies that point B would coincide with the

TAB LEI: Reflection and transmission coefficients for a unit wave incident on each interface.

Wave direction
Interface Upwards Downwards
Reflected wave Transmitted wave Reflected wave Transmitted wave

Top of hammer 1·00 0·00 - -

Hammer body: nose plate 0·57 1'57 -0'57 0·43
Nose plate: top platen 0·00 1·00 0·00 1·00
Top platen: concrete 0·66 1·66 -0,66 0·34
Concrete: bottom platen -0·66 0'34 0'66 1·66
Bottom platen: load column -0·25 0·75 0·25 1·25
Load column -0,31
0·31 1'31 0·69
(top change in section)
Load column 0·69
-0·31 0·31 1·31
(bottom change in section)
Load column: steel base -0·76 0·24 0·76 1·76
Steel base: concrete base -0·16 0·84 0·16 1·16

ASSUMED PROPER TIES

Steel Concrete
Density (kg/m 3 ) 7800 2300
Velocity of wave (m/s) 5180 3600

30

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 CQI/erele ubjecled 10 high raIC' of loading in compressioll

ngur~ 9. T)plcal load-I;ml' alld

Slra;II·lilll/' IraCI',f.

o
o<
""

(a) (h)

o
<
~

Figllrl' 10 : /../'lCh~s o/Iypical

TIME 'oad·I;III/, curves.
(c) (d)

lOAD
«Of"pr.",on)

rm----~...-
(0)
E
(b)
~
(c)

Figure J I : The problelll

simplijil'd: IIII' compr~ ion
dur 10 IrOllSmi((l'd olld
re/frNed ,,·a.'e ill "arioll
(d) co rs.

3\

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research: Vol. 24, No. 78 : March 1972

arrival of the reflected stress wave from the base of the

load column. However, the times between the begin-
ning of the curves at 0 and the points B were not
consistent. They varied between 65 and 125 [LS, al-
though the majority were between 75 and 95 ;1.S. A
common feature was that, although the time interval
for 0 to B varied considerably, the period of oscillation
of the residual wave was remarkably consistent at
about 139 I1.S. This compared with a calculated value
of 100 [LS, based on an assumed stress-wave velocity in
steel given by the elementary velocity of 5180 m/s. The
calibration of the oscilloscope time base was verified
as being accurate. Hence the reduction to about 72 %
of the elementary velocity could be explained by the
fact that d/A (see Figure 2) was effectively about 0·9 for
the steel column.
Corrections for the wave reflected from the base of
the load column were therefore applied by using both
a constant time from 0 (called "modification 1")
and the point B itself (called "modification 2")
as the starting times for the correction~ The best
continuity between the initial part of the recorded
curve and the corrected curves was obtained for the
latter, but even this starting time provided poor con-
tinuity in some cases.
Figure 12: Typical stress-time curve for 'weak' concretes.
All the load-time curves displayed four stages. A
rapid increase in load from 0 to P, a slight increase or
even decrease in load from P to B, a further rapid
increase in load from B to Q and then a decrease in
load after the peak at Q. Usually the coincidence be-
tween modifications 1 and 2 was good. However, Hence the failure stresses for prisms were less reliable.
Figure 12 shows a load-time curve with two distinct The prism tests were therefore discontinued since the
peaks, which was a common feature for weak con- modes of failure of prisms with and without MGA
cretes. The first peak occurred before the arrival of the pads confirmed that the low-friction pads were effec-
reflected wave and therefore gives the actual failure tive at high rates of loading for all specimens.
load; the second peak is caused by the reflection of the The relationship between the static uniaxial com-
first peak from the base of the load column. pressive strength, feu, and the failure criterion for
The continuity between the initial portion of the modification 2 for cubes with MGA pads,/cz, is shown
curve (OP) and the corrected parts (after B) was good in Figure 14.
in about 75 % of the cases, but a distinct trough was Figure 15 shows the results of the tests in which the
found in the modified portions of the curves in some failure stresses in drop-hammer tests on cubes with
cases, as can be seen in Figure 13. The first peaks on MGA pads were given by the first peak. Each result
the load-time traces clearly defined the failure stresses is the average of at least three cubes. The times of
for the weaker specimens only. Two criteria for failure loading in these tests were from 33 to 62 I1.S and the
were therefore adopted for the stronger specimens: average rates of stressing were between 8 IO and 1830
one given by the point P (corresponding to the first kN/mm 2 s, with an average of 1300 kN/mm 2 s. The
peak in the weaker specimens), and the other as given ratios of rapid strength to static strength were between
by modification 2. Point P therefore indicates a min- 1·52 and 2·31 with an average of 1·92. This average
imum failure stress (which equals the actual stress for value, for a static rate of stressing of 0·230 N/mm 2 s
the weaker specimens). Modification 2 indicates the which is close enough to be plotted to the left-hand
probable failure stresses (if greater than that for point scale in Figure 1, shows close agreement with the trend
P). obtained by Watstein.

CHANGES IN CEMENT jWATER RATIO

Results Figure 16 shows how the rapid strength varied with
The traces for prisms were more variable than those c/w ratio for mixes containing four different coarse
for cubes or cubes with MGA pads owing, presumably, aggregates. Notice that the uniaxial. compressive static
to the stress variations along their greater length. strengths followed the expected trends for changes ·in

32

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Concrete subjected to high rates of loading in compression

260
3500

240

section 1 section -4
220 3000

200

180 2500

160

NE
2000
E
Z HO ~
X
~
I Z
~
w
120
«
'">- '"t;;
~

1500
100

80
1000

60

40
500

100 150 200 250 400

TIME-[.LS

Figure: 13 Typical stress-time and strain-time curves (cube with MGA pads).

120 90

0 0
110
80
o
o
100
70

90
60

80
E 50
E
E Z
E
Z 70 I~
~ 40
l
'""
60
30

50
20
o gravel
'V granite

40 D basalt
~ limestone
o basalt
30 6. limestone

o granite with high 70

alumina cement
20
10 80
fcu-Njmm 1
Figure 15: The relationship between the static uniaxial
Figure 14: The relationship between the static uniaxial compressive strength (feu) and the failure stresses in cubes with
compressive strength (fcu) and the strength of cubes with MGA MGA pads for the concrete in which the failure stresses were
pads according to modification 2 (fcz). given by the first peak (fcp).

33

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Vol. 24, No. 78 : March 1972

120
clw in all cases, the limestone aggregates giving the
higher strengths followed by basalt, granite and gravel
110
respectively.
100
",-;//~/
/'..4
",'" , CHANGES IN CEMENT CONTENT
90
/Jl" / , Figure 17 shows the effect of changes in cement
/ ,/ / content. The uniaxial compressive strength, at a given
80 I
/ 1/, / clw ratio, would be expected to decrease with increas-
E
E
,, ,0 ing cement content, owing to the phenomenon of
Z
J,
70 ,/ ~ bleeding. However, some mixes with low cement con-
I
tents had poor workabilities and excessive air voids
,
V>
W I
'"
!;;
60
I were present; this explains the increases in static
w

'":::J....
~
SO I' strength with cement content shown for gravel and
granite. It can be seen that the effect of bleeding was
reduced in the mixes with alb constant, because of the
~O
larger fine aggregate content, which confirms earlier
f"
work (18).
30 gravel 0

granite 'I
20
basalt 0
limestone t::.
10
granite with 0
high alumina
cement
0
O-~ 0-5 0-6 0-7 0-8 0-9 Figure 16: The relationship between strength and c/w ratio for
c/w concretes made with various coarse aggregates.

110

0- _ _ _ c/w= 0-9
--.;.;-
---0
100

90

80

E 70
E

I
V>
w
alb constant
'0
~ 60 ~
~_c/.. ",O
w

'":::J.... --::::: -~-~

~ so

~o

cfW 0·7, alb constant

~C/W O'7,o_tJ
o
30 f"
gravel 0
granite tJ

20 basalt 0
limestone t::. -tr-
Figure 17: The effect of cement
content upon the strength of
concrete made with various
VOLUME FRACTI'ON OF CEMENT-% coarse aggregates.

34

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Concrete subjected to high rates of loading in compression

CHANGES IN MAXIMUM AGGREGATE

SIZE
Comparisons were made for maximum sizes of 25, 120 fa feu
granite 'iJ ~
19 and 10 mm basalt; 25 and 19 mm gravel and
limestone l:1 -fr-
limestone; and 19 and 10 mm granite. No large static 110
granite with high 0
strength differences were found for the different alumina cement

aggregate sizes used, and no definite relationship was I

apparent between strength and aggregate size for either I

,
I

the rapid or static tests. I ",'

CHANGES IN COARSE AGGREGATE

CONTENT
E
Z
J
E
90

80
/:;- ------... -
/:;,'"

,/
/
I
"
"".I"

An increase in fine to coarse aggregate ratio can '"w ... 'V

'"
increase the concrete strength slightly because there is !;;
w 70 "V----
a slight decrease in bleeding(18). In the present tests, '"
::;)
....
this expected increase was often offset, and in some ~
60
cases even reversed, by the increase in voids due to the
poorer workability of the mixes with low coarse aggre-
gate content. The results showed no definite relation- 50
ship between strength and aggregate content in either
the rapid or the static tests.
40

CHANGES IN AGE
30
Figure 18 shows the results for feu and.fc2 at 7, 28
and 91 days for concretes made with two aggregates
20
and ordinary Portland cement. In all cases, the static 1 28 91
strength increased steadily with age, whilst the rapid AGE-days

strengths were inconsistent.

Figure 18: The effect of age upon the strength of concrete
(c/w = 0·7).
HIGH-ALUMINA CEMENT CONCRETES
The effect of changes in c/w ratio upon the strength
of concrete is shown in Figure 16. It can be seen that,
in rapid and static tests for cubes and cubes with MGA
pads, the strength increased considerably with increase
in c/w. Figure 18 shows that the static strength in- can be overcome by using a longer load column.
creased steadily with age from 1 to 91 days. The Measurement of the load as close to the specimen as
strengths for rapid tests on cubes also increased with possible will ensure that the load measured by the
age, whereas rapid tests on cubes with MGA pads load column is a good approximation of the load at the
showed strengths at 3 and 7 days slightly less than at base of a specimen, with only a small time lag between
1 and 91 days for high-alumina cement concretes. the specimen being subjected to a load and that load
being recorded by the load column. However, it was
STRAIN MEASUREMENTS recommended by Krafft(ll) that the length of a load
column should be twice the diameter of the specimen
Strain measurements for rapid tests were made on
if reasonably uniform surface strains are to be ob-
two specimens from each mix. The strain-time curves
tained. rt would therefore not be possible in tests on
were almost linear. The average rate of straining was
102 mm specimens to move the gauges on the load
found to be between 13·0 m/m sand 44·5 m/m s. No
column closer to the specimens than those used in the
correlation was apparent between average rate of
present tests, without adversely affecting the accuracy
straining and either uniaxial compressive strength or
of the load measurements.
mix proportions in the mixes used.
Conclusions
Discussion The results of the present tests at very high rates of
The experimental technique and apparatus used in loading appear to agree with the trend obtained by
the drop-hammer test, with the exception of the load- Watstein at slower rates of loading. Although the
column, were found to be satisfactory. The problem interpretation of the results for the stronger concretes
of the reflected waves from the foot of the load column was less certain than for the weaker concretes, all tests

35

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research: Vol. 24, No. 78 : March 1972

appeared to indicate a ratio of rapid strength to static 8. TIMOSHENKO, s. Theory of elasticity. New York, McGraw-
strength of about 1·9. This ratio was largely indepen- Hill Book Co. Inc., 1934. pp. 416.
9. HAUSER, F. E. Techniques for measuring stress-strain rela-
dent of cement/water ratio, age and cement content, tions at high strain rates. Experimental Mechanics. Vol. 6,
but less independent of the type of coarse aggregate. No.8. August 1966. pp. 395-402.
10. RIPPERGER, E. A. and ABRAMSON, A. N. Reflection and trans-
mission of elastic pulses in a bar at a discontinuity in
ACKNO WLEDGEMENTS cross-section. Third Midwestern Conference on Solid Mech-
anics, University of Michigan, April 1957.
The authors wish to thank Professor S. C. Redshaw 11. KRAFFT, J. M. Instrumentation for high-speed strain mea-
and all other members of the Civil Engineering Depart- surement. SHEWMAN, P. G. and ZACKAY, v. F. (Editors).
ment, University of Birmingham who assisted in this Response of metals to high velocity deformation. New York,
Interscience Publishers, 1961. pp. 9-48. Metallurgical
investigation, and the Science Research Council for
Society Conferences Vpl. 9.
providing a Research Studentship. 12. HUGHES, B. P. The rational design of high-quality concrete
mixes. Concrete. Vol. 2, No.5. May 1968. pp. 212-222.
(Also Concrete Reprint No.1.)
REFERENCES 13. HUGHES, B. P. and BAHRAMIAN, B. An accurate laboratory
test for determining the absorption of aggregates. Materials
1. JONES, P. A.and RICHART, F. E. The effect of testing speed on
Research and Standards. Vol. 7, No. 1. January 1967.
strength and elastic properties of concrete. Proceedings of
pp. 18-23.
the American Society for Testing Materials. Vol. 36, Part 2.
14. BRITISH STANDARDS INSTITUTION. BS 812:1967. Sampling
1936. pp. 380-392.
and testing of mineral aggregates, sands alldfillers. London.
2. THAULOW, s. Rate of loading for compressive strength tests.
pp.39-46.
Betong. Vol. 38, No. I. 1953. pp. 11-15.
IS. HUGHES, B. P. and BAHRAMIAN, B. Cube tests and the uni-
3. MOORE, O. L. Report of working committee C-I on plastic
axial compressive strength of concrete. Magazine of Con-
mortar tests for portland cement. Proceedings of the
crete Research. Vol. 17, No. 53. December 1965. pp.
American Society for Testing Muterials. Vol. 34, Part 1.
177-182.
1934. pp. 322-360.
16. HUGHES, B. P. and GREGORY, R. The impact strength of
4. W ATSTEIN, D. Properties of concrete at high rates ofloading.
concrete using Green's ballistic pendulum. Proceedings of
Symposium on impact testing. Philadelphia, American
the Institution of Civil Engineers. Vol. 41. December 1968.
Society for Testing Materials, 1955. pp. 156-169. Special
pp.731-750.
Technical Publication No. 176.
17. CHIDDISTER, J. L. and MALVERN, L. E. Compression-impact
5. EVANS, R. H. Effect of rate of loading on some mechanical
testing of aluminium at elevated temperatures. Experi-
properties of concrete. WALTON, W. H. (Editor) Proceedings
mental Mechanics. Vol. 3, No.4. April 1963. pp. 81-90.
of a Conference on the Mechanical Properties of Non-
18. HUGHES, B. P. and ASH, J. E. The effect of mix proportions
Metallic Brittle Materials, London, 1958. London, Butter-
and aggregate dust upon the compressive strength of con-
worths Scientific Publications, 1958. pp. 175-192.
crete. Magazine of Concrete Research. Vol. 20, No. 63.
6. KOLSKY, H. Stress waves in solids. Oxford, Clarendon Press,
June 1968. pp. 77-84.
1953. pp. 211.
7. BANCROFT, D. The velocity of longitudinal waves in cylin-
drical bars. Physical Review. Vol. 59, No.7. April 1941. Contributions discussing the above paper should be in the hands
pp. 588-593. of the Editor not later than 30 June 1972.

36

Downloaded by [ NEW YORK UNIVERSITY] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.