HERON is jointly edited by:

STEVIN-LAllORATORY of the
department of Civil Engineering,
Delft University of Technology,
HER N vol. 27
1982
no. 3

Delft, The Netherlands

and Contents
INSTITUTE TNO
for Building Materials and CONCRETE UNDER IMPACT LOADING
Building Structures. TENSILE STRENGTH AND BOND
Rijswijk (ZH), The Netherlands.
HERON contains contributions H. W. Reinhardt
based mainly on research work
performed in these laboratories Delft University of Technology

on strength of materials, structures Department of Civil Engineering

and materials science. Stevin Laboratory

Slevinweg 4, P.O. Box 5048, 2600 G A Delft, The Netherlands

Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2
Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 3
1 Introduction................................ 5
2 Study of the literature . . . . . . . . . . . . . . . . . . . . . .. 6
2.1 Loading rates associated with impact loads 6
2.2 Suitable methods of testing. . . . . . . . . . . .. 6
2.3 Effect of loading rate on tensile strength
and stress-strain diagram of concrete .... 7
2.4 Effect of loading rate on bond between
steel and concrete. . . . . . . . . . . . . . . . . . . .. 8
2.5 Effect ofrepeated impact loading on tensile
strength of concrete ... . . . . . . . . . . . . . . .. 9
2.6 Theories to explain the various influences 9
2.7 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . .. 9
3 Experimental research . . . . . . . . . . . . . . . . . . . . . .. 10
3.1 Method of investigation ................ 10
3.2 Research program ..................... 13
3.2.1 Once-only impact tensile loading ........ 13
3.2.2 Repeated impact tensile loading ......... 14
3.2.3 Bond tests ............................ 14
4 Results .................................... 15
EDITORIAL BOARD: 4.1 Once-only impact tensile loading ........ 15
J. Witteveen, editor in chief 4.2 Repeated impact tensile loading ......... 19
G. J. van Alphen 4.3 Bond tests ............................ 22
M. Dragosavii: 4.3.1 General .............................. 22
H. W. Reinhardt 4.3.2 Force-displacement relations ............ 23
A. C. W. M. Vrouwenvelder 4.3.3 Processing the results for ribbed steel . .. 28
L. van Zetten 4.3.4 Translating the results into the behaviour
for long bond lengths .................. 30

Secretary:
G. J. van Alphen
Stevinweg I
P.O. Box 5048
2600 G A Delft, The Netherlands This publication has been issued in close co-operation with
Tel. 0031-15-785919 the Netherlands Committee for Research, Codes and
Telex 38070 BITHD Specifications for Concrete (CUR-VB).
5 Comparison with codes and standards . . . . . . . . .. 31
5.1 Calculation of tensile strength from cube
(compressive) strength ........... i • . • •• 31
5.2 Repeated impact tensile loading. . . . . . . .. 35
5.3 Bond ................................ 36
6 Snmmary and conclusions. . . . . . . . . . . . . . . . . . .. 37
7 Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 40
8 References.............. . . . . . . . . . . . . . . . . . .. 41
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43

Publications in HERON since 1970

Abstract

Uniaxial impact tensile tests on plain concrete were carried out with the aid of Split
Hopkinson Bar equipment with stress rates of up to 60000 N/mm2. s. Various concrete
mixes were investigated under. dry and wet conditions. All the concretes showed an
increase in strength with increasing stress rate. At very high stress rates the strength
may attain twice the static tensile strength.
Repeated impact tensile loading reduces the strength considerably more than cyclic
loading does with conventional stress rates.
The bond between reinforcing steel and concrete was studied in pull-out tests with
short embedment length. The results showed the bond strength and stiffness of
deformed bars to increase with the loading rate, whereas plain bars and prestressing
strands were hardly affected by the loading rate.
It proved possible to formulate the tensile strength and the bond behaviour as a func-
tion of stress rate by means of a power function. Relations between compressive
strength and tensile strength are given for various stress rates.

Key words

Concrete, tensile strength, bond strength, impact loading, impact fatigue, testing
methods.

2
CONCRETE UNDER IMPACT LOADING. TENSILE STRENGTH AND BOND

Foreword

In recent years, exceptional loads on structures, such as impact loads, have claimed an
increasing amount of attention. This is bound up with the introduction of new types of
structure and with safety aspects, which received rather less attention in the past.
Despite the fact that offshore platforms, nuclear power stations and storage tanks
have been constructed of concrete, there are today still gaps in our knowledge of the
behaviour of this material at high rates of loading. The same applies to the reinforce-
ment of concrete foundation piles, the design of which can be placed on a really sound
basis only if the properties of the concrete and the bond between steel and concrete
when subjected to high loading rates are known.
The lack of adequate knowledge of these matters associated with material behaviour
prompted the CUR-VB to set up Committee C 35 "Concrete under impact loading" in
1976.

On completion of its research the Committee was constituted as follows:

Ir. 1. Schippers, Chairman
Ir. 1. 1. Eberwijn, Secretary
Ir. W. Haitsma
If. 1. 1. de Heer
If. 1. van Keulen
Prof. Dr.-Ing. H. W. Reinhardt
If. A. B. M. van der Plas, Mentor

If. W. H. M. van Lange participated in the Committee's work since its inception. He was
succeeded by Ir. 1. van Keulen in 1979.
The research was carried out in the Stevin Laboratory of the Delft University of Tech-
nology. If. H. A. Kormeling, Prof. Dr.-Ing. H. W. Reinhardt, Ir. E. Vos, Ir. A. W. de Vries
and Dipl.-Ing. A. 1. Zielinski were associated with these investigations.

Financial support has been provided by Stichting CUR-VB, Stichting Bouwresearch,

Betonson B.V., Haitsma Bouwindustrie (HB!) B.V., PIT Beton Heipalenfabriek Kam-
perland B.Y., Charcon Ringvaart B.Y., Schokindustrie B.V., Voorbij's Beton B.Y. and
IJsselmeer Beton B.V. This is gratefully acknowledged.

The present issue of "Heron" is based on CUR-VB Report No. 106 entitled "Concrete
under impact loading - tensile strength and bond".

3
Concrete under impact loading
Tensile strength and bond

1 Introduction

Impact load acting on concrete structures or parts thereof is of common occurrence.

Examples that come to mind are: collision of vehicles or vessels with bridge piers or
superstructures, collisions with offshore structures, aircraft crashes, explosions in or
near structures, and the effect of explosions of bombs or projectiles. Apart from these
exceptional loads, structures may also be subjected to functional impact loads, notably
exemplified by those acting on piles during the driving operation. These loads differ
from ordinary static loads in the very much shorter duration of loading, which in the
case of impact is measured in milliseconds or at most in seconds.
If the designer wishes to take account of impact loads, he will 'have to know whether
the properties of the materials - such as compressive strength, tensile strength, stress-
strain behaviour, bond between steel and concrete - are perhaps altered under the in-
fluence ofthe rate ofloading. If these properties were found to be impaired as a result of
high loading rates, it would mean that the safety based on the static structural prop-
erties would be reduced under impact loading. Conversely, if the properties were found
to become better under impact loading, it would be possible to design structures more
economically to meet those conditions and yet fulfil the safety requirements.
Questions such as these relating to material behaviour led to the setting up of CUR-
VB Committee C 35 "Concrete under impact loading". This Committee's task was
limited to the investigation of three important mechanical properties, namely, the
tensile strength of concrete, the stress-strain diagram of concrete in tension, and the
bond between steel and concrete. These three properties play an essential part in con-
nection with the cracking of reinforced concrete or prestressed concrete piles during
driving. But in all the other above-mentioned types of impact loading these properties
have a share in determining the degree and extent of cracking, crack width and crack
spacing. Furthermore, these properties significantly affect structural loadbearing
capacity under conditions of punching shear or bending shear. Tensile strength is also a
parameter associated with the formulation of the biaxial and triaxial strength of
concrete.
In most cases the behaviour of concrete under once-only impact loading will more
particularly be of importance.
In pile-driving, however, there is repeated impact loading. The number of impact
load applications will vary, depending on the length of the pile and on soil conditions,
but will usually not exceed 3000. Moreover, the wish to apply the results to piles was
what prompted the investigation of the effect of repeated loading on the tensile strength
of concrete.
Also, the present investigators' own results were compared with the information
given in foreign codes of practice.

5
 The development of a testing method was an important feature of the investigation
as a whole, and a full description of this has accordingly been included in the report.
Finally, on the basis of the research results obtained, recommendations for practical
use are offered.

2 Study of the literature

Before the actual research was started, a study ofthe available literature was undertaken
with regard to the following points:
- loading rates associated with impact loads;
- suitable methods of testing;
- effect of loading rate on the tensile strength and stress-strain diagram of concrete;
- effect of loading rate on the bond between steel and concrete;
- effect of repeated impact loading on the tensile strength of concrete;
- theories to explain the various influences.
These points will now be briefly examined.

2.1 Loading rates associated with impact loads

The loading rates (increase in stress per unit time) for various cases are given in Table 1.
Of course, these figures are very approximate and subject to considerable scatter,
because the mass and stiffness of the structure itself determine the rate of loading that
occurs at a section. The order of magnitude of the associated strain rate is also indicated
in the table.
Table I. Loading rates associated with various cases of loading
loading rate iT strain rate i;
loading case (N/mm 2 • ms) (1Js)
collision with vessel 10- 4 to 10- 3 10- 5
collision with vehicle 10- 3 to 10- 2 10- 4
gas explosion 10- 3 to 10- 2 10- 4
crashing aircraft 10- 1 to 10° 10- 2
earthquake 5.10- 1 to 10 2 10- 2 to 3 . 10°
pile-driving 10° to 3.10 1 3 . 10- 2 to 10°

The highest loading rates are in the region of 100 N/mm 2 • ms and occur in earthquakes.
In pile-driving the (tensile) loading rates are between 1 and 30 N/mm 2 • ms. The maxi-
mum strain rates are approximately 3/s (= 300%/s).

2.2 Suitable methods of testing

Three methods are commonly employed for the testing of tensile strength, namely, the
flexural test, the splitting test and the direct (or axial) tensile test. From the point of
view of execution, the flexural and the splitting test are much simpler than the direct

6
tensile test, but their disadvantages are that the stress distribution in the test specimen
cannot be measured and that the strains and strain rates in the specimen are not con-
stant. The known stress distributions are based on linear elastic theory, which is valid
only for low stresses in concrete. On the other hand, the direct tensile test with its axially
central load application gives a well-defined relation between tensile force and stress,
while the strains and strain rates or stress rates are constant over the cross-section of the
specimen. The literature gives no information on how a direct tensile test can be
performed at high loading rates.
The bond between steel and concrete can be determined in various ways, depending
on the purpose of the investigation. If a local relation between force and displacement
has to be determined, as is needed for calculating the bond stresses along a reinforcing
bar, a pull-out test with a short bond length is a suitable method. The effect of the load-
ing rate can be investigated by pulling an embedded bar out of the concrete at different
rates (speeds of pull-out) and measuring the force needed for this.

2.3 Effect of loading rate on tensile strength and stress-strain diagram of concrete
The effect that the rate of loading has upon the tensile strength of concrete has been
dealt with in a small number of publications [1 to 5], the results of which are shown in
Fig. 1. Evidently the direct tensile strenght increases as the loading rate is higher.

5 f /fo

I
I I

4 --'I'~~-----l!-~ Komlos

0----<'> H"ilmann
! ! *-._._* Tak"da
3---L---- ___ ~+ :;=: ~~~:~~dZ" +--------+---------f------+---~---
I b-----6 Hatano /
I ~BirkimN
I, *... ..
···x K~rm"ling et al.I

21-----+-- - i~------+-----+----~-+-- I --+---~

I *
! ,... x 'I I !
i i i * *-'-.. "...... :

I-------t--~_~~~f~~~=_I-I
~ air Plan:'pil" driving .. I
oL-~g----L~--~~~~~~~~~~,_--~~~~
8 410- 0 10'5 4 10. 10 ' 10 10' 10 0 (N/mm' ms)
l~·g 1~-7 1~-5 10 t 1'0' E: (1/5)

Fig. 1. Relative impact tensile strength as a function of loading rate.

7
It is to be noted that most of the tests were performed at rates ofless than 1 N/mm2. ms
and that only one investigation involving very high rates has been reported in the litera-
ture. The range between 1 N Imm 2 . ms < II < 100 N Imm 2 . ms, which is more particular-
ly of interest in connection with the behaviour of piles, was not investigated.
With regard to the effect of the loading rate on the stress-strain diagram there is lack
of agreement between the conclusions of the various investigators. They concur only on
the fact that the modulus of elasticity increases with higher loading rates. On the subject
of ultimate strain opinions are divided. Some investigators assert that it decreases with
higher loading rates - i.e., the material displays a more brittle type of behaviour -
whereas others conclude that the ultimate strain, like the strength, increases with
higher rates.

2.4 Effect of loading rate on bond between steel and concrete

Only two reports of researches concerned with the effect of loading rate on steel-to-
concrete bond were found [6, 7]. In pull-out tests on ribbed bars with a short bond length
of 112 mm the maximum bond stress was found to increase with the loading rate (see
Fig. 2). Although this trend appears reasonable, a correct interpretation of the results is

0.6 'max Ilc

0.5 -
"

0.4
""

0.3 -- - - - - - _.-0', ••• ----t-

d:::16mm
0.2 ly:::112mm

0.1 -.... ----.--.---t-------j

0 10' 5 10-3 10-2 10-1 10 0 10'

t (N/mm' ms)

Fig. 2. Effect of loading rate on pull-out resistance of ribbed reinforcing steel [6].
, Ilc
0.6

05

0.4

0.3

0.2

/'
0.1

Ob-~--~0~'O~1~--~~01--~~--~~--~~10·
O(mm)

Fig. 3. Relation between bond stress and pull-out at two pull-out rates [6].

8
rendered difficult by the considerable scatter they exhibit. As can be deduced from Fig.
3, for a particular bond stress the pull-out becomes less according as the rate increases.
The same trend emerges from the results of the other research [7], from which it was
inferred that at pull-out rates corresponding to a time of 10 ms the bond strength
approaches the uniaxial cylinder compressive strength.

2.5 Effect of repeated impact loading on tensile strength of concrete

No information on the effect of impact load repetition on concrete could be obtained

from the literature. For the sake of completeness, however, it is to be noted that a com-
prehensive report on compressive impact loading on concrete appeared fairly recently
in Sweden (8).

2.6 Theories to explain the various influences

From the study of the literature it emerged that some mechanical properties of concrete
at high loading rates have indeed been experimentally investigated, but that no sys-
tematic research on the effect of various parameters has been carried out. Hence it
would be of real value to have a theory or model to account for the phenomena in ques-
tion. Mihashi and Izumi [9] developed a theory based on fracture mechanics and pre-
supposing that there exists a particular failure probability per unit of time. Subject to
these conditions a generally-valid relation between loading rate and strength is derived:

(1)

wherefi is a material parameter that depends on the composition of the concrete, on the
temperature and on the climatological conditions. An extension of this theory to
repeated loading [10] gives a logarithmic relation between fatigue strength and the
number ofload repetitions. This theory is suitable for indicating the effect of the load-
ing, but it cannot predict the effect of the type of concrete, the temperature, the humid-
ity, etc. To determine these things requires experimental investigations.
For bond there exists as yet no physical model, apart from elastic analyses and nu-
merical treatments of the problem with the help of finite element programs. On the
basis of the theory ofMihashi and Izumi [9] and the assumption that the strength of the
concrete substantially governs the bond behaviour, it can be presumed that a relation as
expressed by (1) exists also for bond.

2.7 Conclusion
The conclusion drawn from the study of the literature was that as yet relatively little was
known about the behaviour of concrete under impact loading and that systematic
research would therefore be very useful.

9
3 Experimental research

3.1 Method of investigation

As Table 1 shows, the loading rates that may occur in pile-driving or in earthquakes are
of the order of 10 to 30 N Imm 2 • ms. In relation to a static tensile test, which is performed
at an average loading of 0.05 N/mm2. s, these impact rates are thus higher by a factor
ranging from 200000 to 600000. Such rates were not attainable with the hydraulic test-
ing equipment available in the Stevin Laboratory ofthe Delft University of Technology.
For this reason a new method of testing was sought.
In the end, a method based on the "Split Hopkins Bar" principle was chosen [11]. This
principle was developed as follows:
In a long bar of elastic material a tension wave is generated, which is transmitted into
a test specimen and which then continues into a second elastic bar, after which the
tension wave is damped out.

5b buffer
upper bar
guide
4 4 strain gauge
5a upper cooling jacket
5b lower coolingjacket
6 concrete test specimen
ro 74 xl00mm
7 working platform
8 counterweight
9 lower bar
10 frame
11 drop weight
12 coupling
13 uncoupling
14 lifting device
15 demping material
16 anvil
17 guide tube
18 pneumatic jack
19 frame base

Fig. 4. Schematic diagram and photograph of test set-up.

10
The loading arrangement is shown schematically in Fig. 4. The drop-weight 11 slides
along the lower bar (aluminium, 74 mm diameter) and strikes the thickened end (anvil)
of this bar, so that a tension wave is set up in the latter.
The maximum stress of the wave is determined by the height offall of the weight, the
magnitude of the weight, and the contact between the weight and the anvil. Greater
height of fall, greater weight and hard contact produce high stresses. The loading rates
and the wavelength are mainly governed by the contact between the drop-weight and
the anvil. In order to obtain a lower rate ofloading, a soft material (cardboard, rubber) is
used for the interposed layers. In this way rates ranging from 2 to 60 N/mm2. ms can be
obtained. As appears from Fig. 4, the tension wave is introduced into the lower bar and
then passes through the test specimen, which is glued with polyester resin between the
two bars. If the mechanical impedances of the specimen and the bars are equal, the
wave will pass undisturbed. But if the impedances differ, the wave will be partially
reflected. With normal concrete between aluminium bars the reflected proportion is 5
to 20%, while 80 to 95% passes. It is this passing proportion that loads the test specimen.
If the strain in the upper bar is measured, the average stress in the specimen can be
calculated from the condition that the same force must act at the top of the specimen
and at the underside of the upper bar. No.4 in Fig. 4 denotes a measuring position. The
strain of the specimen is measured either by proximity transducers (VRPT) or by glued-
on electric resistance strain gauges. When these two measurements have been syn-
chronized - the wave first passes through the specimen and only then reaches meas-
uring position 4 - the stress-strain (a-e) diagram of the concrete can be plotted.
The dimensions of the test rig have been so chosen that concrete with a maximum
aggregate particle diameter of16 mm can be properly tested (74 mm diameter bars) and
that even at low rates ofloading (2 N/mm2. ms) the reflections from the ends of the bars
do not reach the specimen before the wave on its initial journey has passed the speci-
men. This testing arrangement was employed both for the impact loading tests on con-
crete and for the bond tests.
The mechanism which, for repeated impact loading, raised and released the drop-
weight 16 times a minute is shown in Fig. 4 (Nos. 13, 14). The number of load applica-
tions up to failure was recorded.
The impact tensile tests on plain (unreinforced) concrete were performed on cylin-
ders, 74 mm in diameter and 100 mm in height, which had been drilled from a block of
concrete. The bond between steel and concrete was tested on pull-out specimens with a
short bond length. Because of the hoop tensile forces due to the bond it was necessary to
employ a concrete cylinder of 102 mm diameter instead of74 mm. The length of the cyl-
inder was 130 mm, and the bond length ofthe 10 mm diameter reinforcing bars, and of
the 9.6 mm diameter prestressing strands, was 30 mm.

Fig. 5 is a drawing which shows the test specimen glued to the aluminium bars - with
the aid of an adapter unit at the top and a steel plate underneath. The displacement of
the bar is measured in a recess in the adapter against the end of the bar with the aid of a
proximity transducer (VRPT). The force in the bar is determined with electric resis-

11
 --;-
a
<D

12
tance strain gauges. In this way the relation between the pull-out force and the displace-
ment of the steel relatively to the concrete at any particular time is established.
The results of the measurements were recorded with a transient recorder (Nicolet,
Explorer II), with a measuring frequency of2 MHz and a 4 k memory, and processed by
means of the HP 21 MX laboratory computer.
Static reference tests were performed on a hydraulic testing machine. These tests
were concerned with determining the static tensile strength of the concrete used.

3.2 Research program

3.2.1 Once-only impact tensile loading

On the basis of the preliminary study it was decided to apply loading rates varying be-
tween 3 and 30 N/mm2. ms. The effect of the following parameters on the impact
tensile strength was investigated:
- type of cement;
- cement content;
- water-cement ratio;
- maximum aggregate particle size;
- moisture content of the concrete;
- direction of loading in relation to the direction of casting of the specimen.

The type of aggregate was kept unchanged (rounded particles, grading curve as in Table
Al of Appendix A).
The age of the concrete at the time of testing was about 28 days, and the temperature
about 20°C.
In a preliminary program the parameters listed in Table 2 were investigated.

Table 2. Parameters investigated in the preliminary program

loading rate 3 and 30 N/mm 2 • ms
type of cement portland cement A, Band C
portland blastfurnace cement A and B
cement content 300, 325 en 375 kg/m 3
water-cement ratio 0,40 en 0,45
max. aggregate size 8, 16 and 24 mm

The results of these investigations showed that the cement type was of minimal influ-
ence, that the water-cement ratio should be studied within a wider range of values, and
that an aggregate particles size in excess of 16 mm resulted in too much scatter because
of the relation between such size and the test specimen diameter of 74 mm.
The parameters for the main program were chosen as listed in Table 3. The cement
employed was portland cement of class B, and the maximum aggregate particle size was
16 mm.

13
Table 3. Parameters investigated in the main program
loading rate 3 and 30 N/mm 2 . ms
cement content 325 and 375 kg/m 3
water-cement ratio 0.40 and 0.50
moisture condition of the concrete dry, wet
direction of loading parallel and perpendicular to the direction of concreting

The moisture condition designated as "dry" means that the test specimens were stored
wet for 14 days and then at 50% relative humidity; "wet" means that they were stored in
a humidity chamber up to the time of testing. All the specimens were, on the 14th day,
obtained by drilling them out of 200 mm cubes (preliminary program) or out of
300 mm x 600 mm x 250 mm blocks (main program). They were then finished by
sawing, to that their end faces were parallel and perpendicular to the longitudinal axis.
The results of the accompanying control tests on the concrete mixes are given in Tables
A2 and A3 of Appendix A.

3.2.2 Repeated impact tensile loading

The parameters for repeated loading were identical with those ofthe main program with
once-only loading, with one exception.
The loading rate was lower, being 2 to 6 N/mm2. ms, with an average of 5
N Imm 2 • ms. The number of load applications up to failure was chosen between 1 and
1000, this latter number being regarded as a target, as is usual in fatigue testing. The
results of the control tests are given in Table A3 of Appendix A.

3.2.3 Bond tests

In these tests, too, the principal parameter is the loading rate expressed as an increase in
bond stress with time. For this purpose four ranges ofloading rate were chosen, namely,
100 to 160 N/mm2. ms, 20 to 40 N/mm2. ms, approx. 0.08 N/mm2. ms and 0.3 x 10 - 3
N/mm 2 .ms.
The bond stress is defined as the pull-out force divided by the circumference times
the bond length (average bond stress).
The lowest rate can be regarded as constituting static load. As the effect of the loading
rate was expected to differ for different types of steel, three types of steel were investi-

Table 4. Steels used in the pull-out tests

yield point or 0.1 % proof stress modulus of elasticity
type of steel (N/mm2) (kN/mm 2)
plain, 010 mm 285 207
ribbed (Hi-bond), 0 10 mm,
relative rib area 0.076 445 212
prestressing strand 09,6 mm 1730 200

14
gated, namely, plain and deformed (ribbed) reinforcing bars and strand prestressing
tendons.
The mechanical properties of the steels employed are listed in Table 4.
The bond behaviour was investigated for three grades of concrete with average cube
strengths of 22,45 and 55 N/mm2 respectively. The cement employed was portland
cement class B, and the maximum aggregate particle size was 16 mm.
The overall review of the pull-out testing program is presented in Table 5. It is to be
noted that the tests indicated in the first two columns were performed with the testing
technique described (see 3.1), whereas the slow tests were performed on an electro-
hydraulic testing machine.

Table 5. Pull-out testing program

concrete compressive bond stress rate (N/mm 2 . ms)
strength ,t;m
(N/mm2) 100-160 20-40 0.08 0.3 10- 3
22 ribbed ribbed ribbed ribbed
45 ribbed ribbed ribbed ribbed
plain plain
55 ribbed ribbed ribbed ribbed
plain plain
strand strand

4 Results

4.1 Once-only impact tensile loading

All values of the tensile strength at high rates ofloading are g~eater than those obtained
under static loading. This is the general result of the investigations and is in agreement
with the theory (see formula (1)). How much greater the impact tensile strength is than
the static tensile strength will depend on the composition of the concrete and on the
loading rate. The results will be discussed in this chapter, with reference to a double
logarithmic relation of the following form:

InJ=A+B In a (2)

a
whereJis the tensile strength at the loading rate and A and B are coefficients. Formula
(2) was chosen because of the ease of computer processing it offers and is identical with
formula (1), putting B = (1 + )3) - 1 and A = In to - (1 +)3) - \ while to and ao are the
tensile strength and the loading rate in the static test.
All the results were statistically analysed with a view to verifying the reliability of
formula (2). The correlation coefficient and the 95% confidence interval were calcu-
lated. The results ofthe statistical analysis are given separately in Table A4 of Appendix
A. Some of the static tests were splitting tests and some were direct (axial) tensile tests.
From comparison tests it emerged that the splitting tensile strength is a few per cent
higher than the direct tensile strength. This difference is neglected here, however.

15
 9 f(N/mm 2 )

3 F-----t-=---r'... -
<j0/0

10' 10'
(; (N/mm'ms)

Fig. 6. Relation between tensile strength and loading rate.

Before the various influence parameters are discussed, the results of all the static and
impact loading tests will be treated as though they belonged to one statistical popula-
tion. The regression analysis of the 323 results yields the relation between tensile
strength and loading rate, as has been plotted in Fig. 6. The mean tensile strength for
static testing (0-= 10- 4 N/mml. ms) is 3.05 N/mm 1 , and for impact testing (0-= 10 2
N/mml) it is 5.50 N/mml.
From the same calculation are also obtained the bounds between which 90% of all the
anticipated results should be situated. These bounds are indicated by dash lines. From
Fig. 6 it is also apparent that, in view of the scatter of the results and the slope of the
lines, a distinct increase in tensile strength will be achieved only as a result of a sub-
stantial increase in loading rate.
A somewhat more general interpretation of the results can be obtained by plotting
the relative strength - i.e., the ratio between the strength at a particular loading rate and
the static strength - against the ratio of the associated loading rates.
In Fig. 7 the values on the horizontal axis begin with 1 (unity), corresponding to the stat-
ic test. With increasing rate the strength also increases and attains at 10 6 a mean value of
1.80 times the static strength. The value which 5% of the results will fail to reach is 1.35,
and the value which will be exceeded by 5% of the results is 2.37. From Fig. 7 it also
emerges that for a loading rate ratio of 10 3 the 5% lower limit just coincides with unity.
Furthermore, it can be shown statistically that in only 0.05% of the cases there will be no
increase in strength at all ifthe loading rate ratio is increased to 10 6 . The general conclu-
sion is, accordingly, that even at high loading rates there will almost certainly be an
increase in tensile strength.
To follow this approximate general interpretation, the various test parameters will
now be separately examined. It should be noted from the outset, however, that all of
these have much less effect on the result than the loading rate has.

16
 4 fI fa

2,5

1,5 :'0% 1-
1_--1---
0,75
---,,-,

~ ,--:-.
--'- f
! --~

_~',,__-:-IOl~'
.-
%--I
--------- ----,,--

- ' " ,-'

i
-----------

~ - -- ------::;.....-----:---

-=-,' __ .... "....

'

_,-" i
• I
n = 323
0,5 1.-_ _- ' -_ _ _' - -_ _- ' -_ _ _' - -_ _- ' -_ _---'

1 10 10 2 10 3 10' 10 5 10 6
0/00
Fig. 7. Relative strength as a function of relative loading rate.

In this research the maximum aggregate particle size ranged from 8 mm to 24 mm.
From the results it emerged, too, that a larger particle size resulted in a lower impact
tensile strength, but bearing in mind that the scatter of the results increased. As regards
the water-cement ratio, a decrease in this ratio tends to be associated with an increase in
impact tensile strength, but then on the other hand the ratio between the impact tensile
strength and the static tensile strength decreases. The cement content is found to have
relatively little effect on the results, though these do indicate that the impact tensile
strength increases if the cement content is higher. There was no discernible effect ofthe
type of cement, nor of the moisture condition of the concrete, upon the impact tensile
strength. On the other hand, the effect of the direction of loading is very considerable. In
tests performed in the direction perpendicular to the direction of casting the impact
tensile strength is 20% higher than in tests parallel to the direction of casting the
concrete. It was attempted to establish a clear relation between the cube (compressive)
strength and the impact tensile strength, but without success. Evidently the parameters
determining the static compressive strength do not similarly affect the impact tensile
strength.
The stress-strain diagram (a-[ diagram) of the concrete was determined under static
tensile loading and also in the impact tensile test. The average diagram based on four
static and twelve impact tests is shown in Fig. 8. It is notable that the modulus of elastici-
ty and the ultimate strain in impact loading tests are larger than in static loading tests.
While the static modulus of elasticity (secant modulus determined at the origin of the
diagram) was 25500 N/mml at a= 2 N/mm 2, in the impact test it was 39 500 N/mm2 at
the same level of stress. At a= 5 N/mml this latter value had decreased to 30000
N/mm 2. The strain associated with the highest stress is 0.014% in the static test and
0.024% in the impact test. These results therefore show that concrete behaves in a more
rigid, but not in a more brittle manner when subjected to impact loading.
If the underlying factors governing the results of the tests with once-only impact
loading are to be elucidated, it will be necessary to consider the composition and struc-

17
 - - i; = 30 N/mm 2 .ms
- - - - i; = 10- 4 N/mm 2 .ms

-«%.)

Fig. 8. Stress-strain diagrams for static loading and impact loading.

ture of the concrete in general. Concrete is a composite material consisting of a matrix

(hardened cement paste + fine particles) with coarser particles embedded in it. The
strength of the matrix and particles, the bond between matrix and particles, and the mix
proportions are what determine the behaviour of this composite material. Besides,
concrete always contains microcracks, chiefly at the interfaces (boundary surfaces) of
the matrix and the particles embedded in it [12].
Having regard to this structure it can reasonably be supposed that these cracks grow
during loading and that failure occurs as soon as they exceed a certain length. The ques-
tion is how the loading rate can affect crack growth. For cracks to grow in size a certain
time is required, which is available in static loading tests and even more so in creep
tests. Under such circumstances the cracks will extend to zones where the tensile
strength or the bond strength is lowest. For this reason the long-term (sustained load)
strength is lower than the short-term static strength.
In the case of impact loading, where failure takes place within a few milliseconds, the
cracks do not always have an opportunity to seek the zones of least resistance and may
therefore instead make their way into higher-strength zones. Also, with cracks exten-
ding very rapidly branching of the crack tip may occur [12]. Both aspects demand more
energy: hence the measured tensile must be higher as the rate of loading is increased.
The effect of the loading rate diminishes with decreasing difference in mechanical
behaviour between the matrix and the particles, e.g., if the water-cement ratio is low.
The ratio between impact tensile strength and static tensile strength accordingly de-
creases from 1.60 for mix 25 (water-cement ratio 0.50) to 1.50 for mix 23 (water-cement
ratio 0.40). A similar effect should occur in concrete with a low cement content, because
the influence exercised by the hardened cement paste itself becomes less. In concrete
containing large aggregate particles there often occur cavities due to water segregation

18
under these particles, so that their bond to the matrix is poor. The probability of a crack
forming under the particles and extending from particle to particle is greater according
as the particles are larger, so that the effect of the loading rate diminishes. Conversely,
with small aggregate particles the cracks are often compelled to intersect the particles,
which requires more energy. The same aspect manifests itself in comparing the results
of tests in which the direction ofloading is parallel to, or perpendicular to, the direction
of casting. In the former case, i.e., where loading direction and casting direction coin-
cide, the cracks tend easily to pass round the particles, whereas with loading perpendic-
ular to the casting direction the cracks are often compelled to intersect the particles.
Fracture surfaces of test specimens consistently reveal this difference in behaviour.
The stress-strain diagram calls for some further comment. Differences in stiffness in
the initial stages of loading, where crack propagation still plays only a subordinate part,
must be sought in the response of the visco-elastic hardened cement paste. As soon as
cracking dominates, the arguments presented with regard to the strength now equally
apply with regard to the stiffness behaviour: there is greater re.,sistance to deformation
and therefore greater stiffness. All the same, the ultimate strain for impact loading is
greater than for static loading. This is due to the fact that with impact a crack does not
get an opportunity to find the easiest path so as to cause the specimen to fail at its
weakest section. Instead, cracks will begin to develop in various sections, with the result
that the average ultimate strain over the length of the specimen becomes greater. An
indication ofthe correctness of this argument was obtained at very high loading rates, at
which some specimens underwent simultaneous failure at two sections.

4.2 Repeated impact tensile loading

Repeated impact tensile loading can be regarded as a fatigue test, with constant ampli-
tude, in which the load cycles to failure correspond to a number of impact load applica-
tions to failure. The designation "impact fatigue" would appear most appropriate. From
fatigue tests it is known that the relation between the maximum stress and the associ-
ated number of load applications to failure is a simple function of the following form:

(3)

where Al and BI are constants which depend on the material and on the test conditions.
Now if the maximum stress is referred to the static tensile strength, formula (3) can be
written as follows:

(4)

where A2 and B2 likewise are constants.

All the test results have been subjected to a regression analysis with the formulae (3)
and (4), the results of which are contained in Table AS and A6 of Appendix A.

An example of a particular concrete mix composition (mix 23) is presented in Fig. 9,

19
where the mean regression line and the 2.5% and 97.5% bounds have been plotted. The
decrease in the number of load applications to failure with increasing maximum stress
is clearly manifest.

The results of 89 repeated impact tests (impact fatigue tests) are embodied in Figs. 10
and 11, not taking account of the different mix compositions. From Fig. 10 it emerges
that the mean absolute value of 4.3 N/mm2 found in once-oply loading decreases to 2.0
N/mm2 for 1000 load applications. According to this relation the absolute value will
decrease to zero for 750 000 load applications. It is very unlikely, however, that this will
indeed occur, and therefore the validity of thus extrapolating outside the range actually
covered by testing must be called in question. Further tests with a low loading level
would have to be performed in order to obtain more certainty in the matter.

cern. PB
6r------- ---- ------ cem.cont. 375
wI c 0.40
5 _+-___-+-______ Dmax. 16
hum. dry
load/cast.dir. II
4~-- ~~~----4-

2r-----+----"--'= --
O~1------~~------~r-------1~O~----~104
N
Fig. 9. Relation between maximum stress and the number of impact loads to failure for a par-
ticular concrete composition.
8 °mox (N/mm')

tested ---!- extrapolatec n ,,89

61---+---+--------+---f- ._-- ~-- 62 : ~·~~mm2 ms

__.
!
----1----_ _ __ --_
4
--94 1
2r- __~~~-~5~O~%~--._+--~~--+_---~~

N
Fig. 10. Relation between maximum stress and the number of impact loads for all the results.

2 °mox /to

1.5t---'-4~........
:---+----i--+--+--+--

O.5t----r---........
-'!'"---=;::-----t--""-....;t-........ ;;c--+---+---j
----t---.........

10'
N
Fig. 11. Relation between relative maximum stress and the number of impact loads for all the
results.

20
 In Fig. 11 the relative maximum stress is presented as a function of the number of
impact loads applied. Since the maximum stress for impact fatigue loading is referred to
the static tensile strength (omax/fo), the relative value for one impact load is larger than
unity, this being, after all, the effect of the loading rate upon the strength in the case of
once-only impact loading. After 40 impact load applications this loading rate effect has
been nullified, for then the relative strength has been reduced to unity. After about
5000 load applications the mean impact tensile strength has been reduced to half the
static tensile strength.
The results show considerable scatter, as is indeed normally found in fatigue testing.
Fig. 11, for example, shows that with once-only loading the impact tensile strength in
5% of the cases is not necessarily higher than the static tensile strength. With repeated
loading the impact tensile strength in 5% of the cases has already decreased to half the
static tensile strength after about 100 load applications. On the other hand, in 5% of the
cases the impact tensile strength is still as high as the static tens.ile strength even after
1000 load applications.
In view of the range of scatter it would appear meaningful to seek an interpretation of
the results for the respective mixes separately. For then the effect of the concrete com-
position and the direction of loading in relation to the direction of casting can be
studied.
The effect of a particular parameter can be ascertained with the aid of Tables A5 and
A6 of Appendix A. For this purpose the coefficients Al and BJ offormula (3) or A2 and B2
offormula (4) are governing quantities. The test program was so planned that there were
pairs of series in which only one parameter was changed, thus enabling the results to be
directly compared.
The water-cement ratio was varied between 0.40 and 0.50 for a cement content of375
kg per m 3 of concrete. From the results it emerges that for the higher water-cement ratio
the relative impact tensile strength under once-only loading and also under repeated
loading increases. Although the decrease in strength with the number of impact load
applications is the same, the ratio between the maximum stresses after 1000 load appli-
cations and after one load application (see Table A5 of Appendix A) is 0.55 for a water-
cement ratio of 0.40 (mix 23), whereas the corresponding value is 0.62 for a water-
cement ratio of 0.50 (mix 25). This difference becomes even more clearly manifest if the
relative strengths (see Table A6 of Appendix A) are compared: 0.63 as against 0.92. It
can be inferred that the increase in tensile strength as determined by the static test and
brought about by a low water-cement ratio is absent in the case of repeated impact
loading.
The cement content was 325 kg/m 3 in mix 21 and 375 kg/m 3 in mix 25, the water-
cement ratio being 0.50. The results show that the concrete with the higher cement
content has a higher impact tensile strength and also undergoes less tensile strength
reduction with the number of load applications. Thus the ratio between Omax 1000 and
Omax I is 0.40 for 325 kg/m 3 and is 0.62 for 375 kg/m 3 cement content. Referred to the
static value these ratios correspond to 0.56 and 0.92 respectively. What emerges there-
fore is that a leaner concrete is more sensitive to repeated impact loading than a con-

21
 crete containing more cement, although both have very nearly the same cube strength
(see Table A3 of Appendix A).
The moisture content ofthe concrete was varied through the storage conditions of the
specimens. The "wet" specimens were immersed in water up to the time of testing,
while the "dry" ones were allowed to dry for two weeks. These respective treatments
were found to have virtually no effect on the tensile strength obtained with once-only
impact loading.
On the other hand, the effect of the direction of loading in relation to the direction of
casting was very pronounced. Perpendicularly to the casting direction of the concrete
the impact tensile strength is higher than parallel to that direction, and in that case the
decrease in impact tensile strength with increasing number ofload applications is great-
er. The ratio between Omax 1000 and omax 1 is the same in both cases. It can be inferred that
when specimens are tested perpendicularly to the casting direction the strength is
always higher than when they are tested parallel to that direction.
This description of the results will now be followed by a look at the underlYingfactors
governing the behaviour observed. Fatigue is the formation of cracks which are at first
stable and then grow in an unstable manner until failure occurs. For the comparison of
various types of concrete the range of stable crack growth is more particularly of
interest. If it is presupposed that a brittle material can undergo only very little plastic
deformation, this implies that a crack - which may be a microcrack that has developed
from a pore or from the contact zone between an aggregate particle and the hardened
cement paste - will grow more rapidly than in a material which can deform plastically.
In proportion as the concrete is more brittle, its sensitivity to repeated loading will be
greater.
From the results reported above it emerges that concrete with a low water-cement
ratio and a low cement content is most sensitive to decline in strength when subjected
to repeated loading (impact fatigue loading). This is also the combination of influences
that produce the most brittle behaviour. Having regard to the moisture content it might
be supposed that wet concrete will behave in a less brittle manner and therefore attain a
higher fatigue strength. Actually, this influence was evidently so slight as not to be
measurable.
It can be concluded that if concrete is required to have a high impact tensile fatigue
strength, it should be as tough as possible. This may mean that it is preferable to specify
a lower cube (compressive) strength for the sake of obtaining greater toughness and
thus higher fatigue strength.

4.3 Bond tests

4.3.1 General

As has been described in 3.1, the bond between the steel and the concrete was deter-
mined by tests in which a 10 mm diameter bar (or a 9.5 mm seven-wire strand) was
pulled out of a concrete cylinder, the bond length being 30 mm. By keeping the bond

22
 t
deformed bar

L..~------""i strand
~ _ _ _ _ _ _-; plain bar

J
I
t
5

Fig. 12. r-o relation for three steels (schematic).

length as short as possible a stress distribution that is as nearly constant as possible is

obtained along the bar and is not disturbed by a high peak stress at the pull-out end of
the bar and a marked drop in stress at the other end. Admitte~ly, this is not the stress
distribution that will occur in actual practice; it must instead be regarded as an ideal
situation, the results of which are suitable for the calculation of the bond stresses in
practical cases. For that purpose it is necessary to know the relation between the pull-
out force and the displacement between the steel and concrete.
Distinctly different force-displacement relations occur with, respectively, plain bars,
deformed (ribbed) bars and strand tendons (local r-o relations); these are shown in
Fig. 12.

The three well-known bond mechanisms - namely: adhesion, friction and shear - are all
present, though in varying degrees. With plain bars the adhesion is first manifest and
subsequently, when this has been overcome, friction. In contrast with this, the bond
between ribbed bars and concrete is due mainly to the shear resistance of the concrete
between the ribs, breakdown ofthe bond being associated with shearing ofthis concrete
and also with internal cracking of the concrete around the bars. The bond developed by
a strand is due mainly to friction and partly to shear resistance.
The essential question with which this research was concerned was whether the local
r-o relation could be affected by the rate ofloading and whether this effect, if any, would
be the same for every type of steel and grade of concrete. Should there indeed be found
to exist an influence exercised by the loading rate, this could be of importance with
regard to the bond length of a bar embedded in concrete, the crack spacing and the crack
width in a reinforced concrete structure.
The results of the tests will now first be discussed; next a formula expressing the
effect of the loading rate will be established; and finally the background to the behav-
iour revealed by the tests will be examined.

4.3.2 Force-displacement relations

A pull-out test yields two primary diagrams: a force-time (or stress-time) relation and a
displacement-time (or slip-time) relation.

23
 1: 6
(N/mm') (mm)
20. 0.20. ~--'------'-~-'-~--'-'-~-'-~----r---'----'

150.15

10. 0.10.

50.0.5 -

o 20 30 40 50 60. 70
t (5)
Fig. 13. Stress-time and displacement-time relations of ribbed steel at low rate, for three
concretes.
1: 6
(N/mm') (mm)
35 0.35 ,----,-------r-----,--,---,---,

250.25 r-----i---t-----t----+-I

20 0..20

15 D.15 +-----I'-t+-

10 0.10.

o 200 300 400 500 600

t (>L 5)
Fig. 14. Stress-time and displacement-time relations of ribbed steel at very high rate, for three
concretes.

The desired force-displacement relation is obtained by synchronization of these two

primary relations. Two examples of primary diagrams are given in Figs. l3 and 14: the
first for a low, the second for a very high rate, in each case for three grades of concrete.
The results presented in Fig. l3 were obtained on a servo-hydraulic testing machine
with force control. The bond stress-time relation is therefore a straight line, and the
pull-out depends on the resistance of the bond zone. A different situation exists with
regard to the results presented in Fig. 14, which were obtained with the test equipment
described in 3.1. In this case, after an adjustment period has been passed, the stress-
time and the displacement-time lines in a particular range are approximately straight.

24
 Whereas the displacement-time relation continues to rise uniformly, the stress-time
relations deviate at some particular time and remain almost constant. The testing
method can be regarded as approximately strain-controlled.In order to arrive at com-
parable results, the rate of stress is defined as the slopeofthe T-t line for a displacement of
0.01 mm.
These primary diagrams have been converted into T-O relations. All the results
obtained are presented in Table A7 of Appendix A. They will now be further discussed
with reference to diagrams.

Ribbed steel
To reveal the effect of the most important variable - the loading rate - the averages of
three bond stress-displacement lines at four loading rates, for a low-strength concrete,
have been plotted in Fig. 15. The effect of this variable - as a parameter in the form of
the relative loading rate referred to the static test - manifests itselfin an increase in pull-

!
25

20

i dyn/1:s tat
5r----- I--~
3.53)(10 5

1--- 1.32)(105

10 /" 2,64)( 10 2

/; t::::1~-
r-
---
~
jfc ,,22.7 N/mml

o 0.02 0.04 006 0.08 0.10 0.12

J
0.14
I
0.16 .018 .020
6(mm)
Fig. IS. r-o relation at four loading rates for concrete with a cube strength of 22.7 N/mm 2 •

25

0~~0~.o~2~.o~04~~00~6~0~.o~8~O~.10~~0.1~2--0~.1~4~O~.16~~ill~8~Q20
6(mm)

Fig. 16. T-O relation at four loading rates for concrete with a cube strength of 54.8 N/mm 2 •

25
out resistance, although the characteristic shape of the lines is preserved.
A similar effect emerges from Fig. 16, relating to a high-strength concrete. Here, too,
the stiffness is greater according as the loading rate is higher, but in the absolute sense
the effect of this rate is less pronounced. Because of the high cube strength the bond
stresses are of course higher than for concrete of lower cube strength.
The effect of the compressive strength of the concrete is additionally illustrated in
Figs. 17 and 18. The higher the compressive strength, the better the bond. This is a well-
known phenomenon. But what also emerges is that the quality of the concrete (its
strength class) has less effect on the bond strength according as the loading rate
increases.

30'( N/mm')
f

25
I
I-- - 54.8

------
f--
20
I
. V
15
/, 4Lo.9

/~ ~
10

5
f __l- 22.7

o
V
0.02 0.04 006 0.08 0.1.0 0.12 01 4 016 0.18 Q20
o(mm)

Fig. 17. r-o relation for r= 0.3· 10~3 N/mm2. ms for three grades of concrete and ribbed steel.

-----
54.S

25
V

20
/ / l----- r- ._.
44.9

,/

/ /
--
15

1.0 II /"
22.7

5
'V
.0 0.02 .004 006 Q.o8 010 0.12 0.14 016 0.18 Q20
o(mm)

Fig. 18. r-o relation for r= 140 N/mm2. ms for three grades of concrete and ribbed steel.

26
 6' ( N/mm')

2~-------+--------+---------+
¢ 10. plain

ob-------~------~~----~~------~0,20
5(mm)

Fig. 19. r-o relation for plain steel at two rates and for two grades of concrete,

6' (N/mm2)

4
Iy r---

3/8 " strand

fc'" 55N/mm 2

r
2
t
- 40N/mm ms
0.3 ·10-'N/mm'

o 0,05
T
0,10 0,15 0,20
5(mm)

Fig. 20. r-o relation for prestressing strand tendons at two loading rates.

Plain steel
Average values of r-o relations (see Fig. 19) indicate that the loading rate has no signific-
ant effect on bond strength and bond stiffness. From the results of this research, which
were in agreement with data published in the literature, it emerged that no further
experimental investigations on plain steel reinforcing bars were necessary. It was con-
cluded that such reinforcement is not sensitive to the rate of loading.

Strand tendons
The overall character of the results obtained with strand prestressing tendons is similar
to that obtained with plain bars, i.e., the effect of the loading rate is negligible (see Fig.
20). For this reason no further investigations on strand were carried out.

27
4.3.3 Processing the results for ribbed steel
The results obtained for ribbed reinforcing bars will be processed in the same way as is
done with the results of tensile tests on plain (unreinforced) concrete. A formula of the
general type:

T ( i )~ (5)
~= To
is analogous to formula (1) or (2). Here Tis the bond stress associated with a certain dis-
placement and with a certain loading rate i; TO and TO are the corresponding values for the
static test, while 17 is the parameter which represents the effect of the compressive
strength of the concrete and is a function of the displacement.
Statistical analysis of all the test results for ribbed steel yielded the following formula
for
0,7(1- 2,56)
17= (6)

where 6must be substituted in mm and!: in N/mm 2• The correlation coefficient for this
relation was 0.96 for the limits 0 < 6 < 0.2 mm.
The relation between the formulae (5) and (6) expresses the fact that the bond stress
for a certain displacement and a certain concrete compressive stress increases with the
rate of loading. This increase is less for small displacements and low compressive
strength. The bond strength is significantly dependent on the compressive strength.
The effect of the loading rate upon the bond strength diminishes according as the com-
pressive strength of the concrete is higher. As it is known that concrete is more brittle
with higher compressive strength, it can alternatively be stated that the effect of the
loading rate becomes less according as the concrete is more brittle in its behaviour.
In Fig. 21 the formulae (5) and (6) have been plotted in graph form for three displace-
ments and five grades of concrete (mean cube strengths). It clearly emerges that the
effect ofthe loading rate increases according as the concrete is oflower strength and also
in cases where the displacement ofthe steel in relation to the concrete is small. For prac-
tical purposes this means that the effect of the loading rate is greatest immediately after
the formation of a crack, for then the displacement is still small. When the crack is
pulled wider open, it matters much less whether the load is applied at a faster or slower
rate. Lower strength makes the concrete more sensitive to these phenomena, as is
indeed expressed by formula (5).
The results of the bond tests can be explained quite simply by considering the
mechanism of the bond between ribbed reinforcing bars and concrete. After an initial
very small displacement the adhesion is destroyed and then the ribs on the bar begin to
bear against the concrete. In consequence, high concentrated stresses develop under
the ribs, and these stresses may exceed the cube (compressive) strength of the concrete.
This in turn causes cracking within the concrete, and the ribs undergo displacement in
relation to the concrete.

28
 2,6!=1:=/~1:0,,===!,=~~==c=c=c=c==-============
2,4

1,4f-----+-----+--/"----+___L:.--~_¥' sa
50

1.2

10 ' 10' 10 3 10'

0= 0.1 mm
1,8 deformed bar ----+-----f-----+---c-.
1,6;1-----!------I----------I--

10' 10' 10 3 10' 10 5

-----+
2 1: 11:0

8 ~;f~;~~rl J
I
1.61-----
~
P
1.4 j.----' --
~ t---::::::: l----R:::
l-
t==-:i:: I-
2
~~ ---- ~
60
I-

-
10 ' 10' 10 3 10'

Fig. 21. Relation between bond stress, loading rate, concrete quality and displacements for
ribbed steel (double logarithmic scale).

29
 Both from the present research and from the literature it emerges that the compres-
sive strength of the concrete and its tensile strength (cracking) are affected by the load-
ing rate. So it is not surprising that the bond exhibits the same behaviour.
It is not possible to state with certainty to what extent this result is valid for all types of
deformed reinforcement. Two aspects are, however, important in connection with
assessing this: first, the fact that the valuefR (relative rib surface area)* is within narrow
limits for all deformed bars [15]; second, that the mechanism underlying shear bond
behaviour is discernibly manifest.
The valuefR is the principal parameter for bond [14] which determines, among other
matters, what mechanism will govern the bond behaviour. For ribbed steel, for whichfR
is between 0.065 and 0.1, the mechanism as described above will develop. Hence it can
be inferred that the research results are valid also for other types of deformed bars than
those used in the tests, but having the above value for fR. The second aspect relates to
the effect of the loading rate upon the mechanical properties of concrete. As has been
shown, these are sensitive to variations in the loading rate. With different reinforcing
bars the loading rate will have a similar effect on bond, provided that the mechanism is
the same in all cases. In view of what has been said above, it would appear justified to
conclude that the research results for bond behaviour can permissibly be applied more
widely than the experiments strictly authorize.

4.3.4 Translating the results into the behaviour for long bond lengths
With the aid of pull-out tests on specimens with short bond Rehm derived a differential
equation with which the distribution of the steel stresses and bond stresses along an
embedded bar can be calculated [14]. This equation, based on the linear elastic theory,
is as follows:

(7)

where:
o = displacement of steel in relation to concrete
x = distance from free end of bar to section considered
ok = characteristic diameter of bar
n = ratio of the moduli of elasticity of steel and concrete (modular ratio)
12 = percentage of reinforcement
Es = modulus of elasticity of steel
r( 0) = bond stress as a function of displacement

* jjR= AA,
where;' is the projected area of the rib in the longitudinal direction of the bar and Am is the
product of the bar circumference and the rib spacing.

30
 O~-------5~O--------1~OO--------~15-0-------2~OO
X(mm)

Fig. 22. Steel stresses along an embedded bar under static loading and under impact loading.

If the function r( 6) is also dependent on the loading rate, i.e., r( 6, i), the differential
equation (7) wi111ikewise be dependent on the loading rate.
The result of such a calculation for two types of concrete, for a given steel stress of
400 N/mm2 at the free end of the bar, is shown in Fig. 22. The continuous line, which
relates to impact loading, is steeper for both concretes than the dash line for static load-
ing. This means that in the case of impact loading the steel stress decreases more rapidly
on account of the higher bond stiffness and that the bond length is therefore shorter. It
also means that for a high rate of loading the crack spacing in a tension member would
have to be smaller and the crack width would have to be less than in the case of static
load of the same magnitude. This interpretation does not take inertia effects into con-
sideration.

5 Comparison with codes and standards

5.1 Calculation of tensile strength from cube (compressive) strength
If no additional tensile tests are performed on concrete, the tensile strength can be cal-
culated from the cube strength. For this purpose there exist simple formulae derived
from a large number oftest results. The Netherlands code of practice for concrete VB
1974 [17] employs the following relation:
(8)

where:
ftk = characteristic tensile strength of the concrete
!ck = characteristic cube strength
The coefficient 0.87 has been introduced in order to take account of the greater scatter
in the tensile strength as compared with that in the compressive strength.
The CEB-FIP Model Code [18] adopts a different relation for the purpose, namely:

ftm = 0.26!ck2j3 (9)

31
where:
ftm = mean tensile strength of the concrete
fck = characteristic cube strength

In a general way the formulae (8) and (9) can be written respectively as follows:

ftk = a + bick (10)

(11)

The coefficients a, b, c and d comprise all the factors that may affect the ratio between
tensile and compressive strength, e.g., the composition of the concrete, the tempera-
ture, the moisture content of the test specimen, the method of testing and the rate of
loading.
If the true effect of the loading rate were introduced, then either the above-
mentioned coefficients could be made rate-dependent or an extra term be added to take
account of the loading rate. The former alternative appears attractive because the
known relation between cube strength and tensile strength of concrete then remains
unchanged, only the numerical values being somewhat changed. In that case, however,
it is presupposed that there exists a close relation between tensile strength and com-
pressive strength at high loading rates. In the course of the research this was found not
to be entirely so; other parameters, such as the water-cement ratio and the cement con-
tent, resulted in better correlation than the cube strength did.
Retaining the simple relation between tensile strength and compressive strength for
practical convenience implies that the scatter of the ratios must increase with increasing
rate ofloading. This phenomenon can be allowed for by so determining the coefficients
that the lower results are given more weight than the higher ones. This is a safe approxi-
mation that underrates some results.
There is another aspect that must be given attention in connection with the deter-
mination of the coefficients in the formulae (10) and (11). In the research, particular
concrete mix compositions were employed, which resulted in particular mean values
and standard deviations. Because of the limited scope of the investigations and the fact
that the scatter of laboratory tests differs from that of tests conducted under practical
conditions, it is not possible to calculate the characteristic values. The coefficients are
accordingly determined with the aid of the mean values.
Taking account of the two above-mentioned aspects, the following formulae were
established, for which purpose it was attempted to round off the coefficients to values
that could be conveniently memorized. Thus the VB formula becomes:

static, 00 fbm = 0.87 (1 + io fern) (12a)

a 3
-:-= 10 fbm = 0.87(3.15 + fofem) (l2b)
ao
a 6
-:- = 10 fbm = 0.87 (4.60 + -to fern) (l2c)
00

32
ibm andfc·m are the mean values of the static tensile strength and of the static compressive
strength (cube strength) of concrete respectively (in N Imm 2).
The CEB-FIP formula is modified to the following:

static, ao ibm = 0.26(fcm - 10)2/3 (13a)

a 1
~=1O' ibm = 1.00(fcm - 10)2/5 (13b)
ao

a 6
~=1O ibm = 2.40(fcm- 10)1/5 (13c)
ao
According to the CEP-FIP Model Code the factor (fem -10) is equal to the characteristic
value. It is possible to interpolate for other loading rates than those indicated.
The formulae (12) and (13) are presented graphically in Fig. 23. Both express the facts
that emerged from the investigations, namely, that the tensile strength increases with
higher loading rate, but also that this increase becomes less pronounced with higher
strength of the concrete (greater brittleness). For a loading rate which is 10 6 times as
high as in the static test (i.e., 100 N/mm2. ms) the impact tensile strength is doubled for
a mean cube strength of30 N/mm2 and is increased 1.5-fold for a mean cube strength of
50 N/mm 2. Fig. 23 also shows that the CEB-FIP formula always predicts higher values
than the VB formula.
A comparison between the proposed conversion value and the results obtained from
the regression analysis with the aid of which the test results were processed is made pos-
sible in Fig. 24. It emerges that the VB line gives a good approximation of the static
results and that the tensile strength at higher loading rates is underestimated. This latter
aspect is intentional, however, in view of the not very close relation between compres-
sive strength and tensile strength. The CEB-FIP line overestimates the static strength,
at least in this calculation with the constant difference of 10 N/mm2 between the mean
and the characteristic value for the compressive strength. For a= 10 - 1 N Imm 2 . ms the

ftmlN/mm2)

6 --l--L----+---+--f----+--I
5

o~--~----~----~--~----~--~-J
o 10 20 30 1.0 50 60
fem IN/mm2)

Fig. 23. Relation between concrete tensile strength and cube strength according to VB formulae
(12) en CEB-FIP formulae (13) at three loading rates.

33
 ftm ( N/mm')

---

Fig. 24. Comparison of experimental results with the relation between concrete tensile strength
and cube strength according to VB formulae (12) and CEB-FIP formulae (13) at three
loading rates.

CEB-FIP line likewise overestimates the results. Only for a= 10 2 N/mm2. ms does the
relation appear to be closer to reality, but in view of what has been noted above this, too,
is to be regarded as something of an overestimation.
A better representation of the test results is obtained with the CEB-FIP formula if the
difference between the mean and the characteristic compressive strength is incorpora-
ted in the coefficient c and not taken into account merely by applying a shift of 10 N/mm2
to the compressive strength axis. From Table 2.2 in the CEB-FIP Model Code's expla-
natory notes it can be deduced that:

frO.05 = 0.75frm (14)

Let fr 0.05 be designated as the characteristic strength; then:

frk = 0.75· O.26/ck2j3 (15)

or:
(16)

Associated with formula (16) are two statistically similar values. If this same expression
is used also for mean values - as in the VB formula - then the formulae (13) are con-
verted into the following relations:
static, ao frm = O.20/cm2j3 (17a)

a 3
--;- = 10 (17b)
ao

a 6
--;- = 10 (17c)
ao

34
 Itm (N/mm2)

°O~---1~O----2~O----~30-----4~O----5~O----~
lem (N/mm2)

Fig. 25. Relation between concrete tensile strength and cube strength according to VB formulae
(12) and modified CEB-FIP formulae (17) at three loading- rates.

From Fig. 25 it emerges that this modified CEB-FIP line yields almost the same results
as the VB line does and thus provides a better representation of the test results, i.e.,
adopts a more conservative approach.
It is proposed that either the formulae (12) based on the VB 1974 Netherlands code
for concrete or the formulae (17) based on the CEB-FIP Model Code be adopted, in
which case the same functions can be used for calculating the mean values or the char-
acteristic values, depending in whether the mean or the characteristic value is intro-
duced for the cube (compressive) strength.

5.2 Repeated impact tensile loading

The tensile strength under repeated impact loading is especially of importance for
structures which have to be uncracked under service conditions. An example of such a
structure is a concrete road. Fig. 26 shows a Smith diagram which is used in concrete
road pavement design and in which the upper stress limit and the mean stress are
presented in relation to the short-term strength. The continuous lines relate to short-
term tests, while the dash lines indicate the effect of long-term (sustained) loading.
The factor 0.8 is the ratio between sustained load strength and short-term strength.
For a zero lower stress limit, according to these diagrams, a fatigue strength ofO.5fo can
be expected (for N = 2 X 10 6 cycles). For N = 100 this would be 0.8fo. All these values are
based on test results obtained at normal rates of loading.
The mean values of the results obtained with repeated impact tensile loading are in-
cluded in Fig. 26. It emerges, first, that for a low number ofload applications the results
are higher than in the case of normal loading rate; second, that the effect of the number
of load applications is much greater for impact loading. For 10 3 load applications the
results are already lower, and for 10 4 load applications the strength attains only 0.42fo.

35
 1,2i-'om""o"-'x'--I-CI°T--_---,_ _--,-_ _r-_--,-_ _71
.10'

o,81----+--

0,61----+-

O,L.I-----I'~---c¥--,++

0,2---- of

o 0~---0~2~--~--~-----L----~--~
Om 11 o

Fig. 26, Smith diagram for repeated tensile loading of concrete.

In general it can be inferred from these results that the diagrams which can be used for
normal loading rates give too favourable a picture of the behaviour under repeated
impact loading. At high loading rates concrete is evidently more sensitive to repeated
loading.

5.3 Bond

In the VB 1974 and in the CEB-FIP Model Code the bond strength is given as a linear
function of the tensile strength, bond strength being defined for the present purpose as
the average bond stress along the anchorage length of a bar loaded to a stress equal to
the yield point of the steel. This cannot be directly compared with the values which, in
the present research, were obtained for short bond lengths. For this reason, too, it is not
possible to give a similar quantitative relation.
It can, however, be proposed to use the formulae for static values, for which purpose
the static tensile strength is replaced by the tensile strength associated with a particular
loading rate. This last-mentioned strength can be obtained either from Fig. 25 or from
the formulae (12) and (17).
From a comparison of Figs. 25 and 21 it can be inferred that this procedure will lead to
acceptable results. In both diagrams the increase in tensile strength with higher-
strength concrete and higher loading rate shows the same trend. The order of magnitu-
de of the increase is also in agreement. Only the fact that the ratio rjro is dependent on
the absolute displacement (small displacement, large effect) rules out an exact quanti-
tative comparison.
This proposal is valid only for deformed (ribbed) reinforcing bars. In the investiga-

36
tions for plain bars and for strand prestressing tendons no effect of the loading rate was
found, and it is accordingly recommended that for these types of steel the static values
should always be adopted in the calculations, irrespective of the rate of loading.

6 Summary and conclusions

In connection with the increasing need for information on the effect of the loading rate
upon the mechanical properties of concrete, experimental research on the axial tensile
strength of concrete and on the bond between reinforcement and concrete was carried
out.
In order to attain very high loading rates - up to 600 000 times as high as those normal-
ly applied in "static" loading tests - a special testing rig, based on the Split Hopkinson
Bar principle, was developed and built. The same rig can, after some simple adaptation,
also be used for pull-out tests for measuring the bond between steel and concrete within
200 lIS.
The investigation comprised three parts:
- determination of the uniaxial tensile strength and the stress-strain diagram of con-
crete under single impact tensile loading;
- determination of the uniaxial tensile strength of concrete under repeated impact ten-
sile loading;
- determination of the bond behaviour between steel and concrete under single impact
tensile loading.

In the first part of this research program the following parameters were investigated:
- loading rate;
- type of cement;
- cement content;
- water-cement ratio;
- maximum size of aggregate particles;
- moisture condition of concrete;
- loading/casting direction.

From all the test results it emerges that the rate ofloading has a significant effect on the
tensile strength of concrete. The tensile strength increases with the loading rate and the
relationship between them is double-logarithmic. For example, when concrete is load-
ed 1l million times as faster than in the static test, the mean tensile strength can be
expected to be 80% higher. Actually this value ranges between 35 and 110% and
depends on the introduced test parameters as follows:
- type of cement: no distinct effect;
- water-cement ratio: a higher water-cement ratio results in higher values of the ratio
impact/static tensile strength;
- cement content: a combination of a higher cement content and a higher water-
cement ratio seems to result in a higher relative impact tensile strength;

37
- maximum size of aggregate particles: larger particles result in a lower relative impact
tensile strength;
- moisture condition of concrete: no distinct effect;
- loading/casting direction: loading perpendicularly to the casting direction results in
higher values of the relative impact tensile strength.

The stress-strain diagram of concrete under impact tensile loading shows an initially
steeper slope and a larger ultimate strain than the diagram obtained under static load-
ing. This means that the behaviour of concrete does not become more brittle with in-
creasing rate of loading.
Repeated impact tensile loading appeared to have a great influence on impact tensile
strength. After 100 loading cycles the gain in strength due to the high loading rate
vanishes, and after ca. 1000 impacts the tensile strength is reduced to only 70% of the
static strength. These mean values, too, are dependent on the composition of the con-
crete. Thus, a higher water-cement ratio and a high cement content are favourable to
the impact tensile strength under repeated loading, while the moisture content of the
concrete has no effect. Load application at right angles to the direction of concreting
results in higher fatigue strength. In this part of the research the loading rate was kept
constant.
In investigating the background to these phenomena it is necessary to consider the
process of crack growth in the concrete. According as the hardened cement paste is
more brittle, failure will occur more easily under repetitive loading, and according as
the loading rate increases, influences such as crack branching and forced fracturing of
particles will become more important.
The bond resistance between reinforcement and concrete was investigated by means
of pull-out tests with short embedment length. The variables in this investigation were
the type of steel, the cube compressive strength of concrete and the rate of pull-out. It
was found that the influence ofloading rate on the bond resistance of deformed bars was
similar to that on the tensile strength of concrete. For plain bars and prestressing
strands this effect is negligible. The loading rate has the greatest effect in the case oflow
strength concretes, especially at small displacements between reinforcing steel and
concrete. This result can be interpreted such that the effective bond length of a
deformed bar decreases with increasing loading rate.
After the analysis of experimental results it was attempted to extend the relation be-
tween tensile strength and cube compressive strength of concrete - as conceived in the
Netherlands code VB 1974 and in the CEB-FIP Recommendations - to comprise the
rate ofloading. Having due regard to the scatter displayed by the results, a proposal for a
linear relationship and for a non-linear relationship is made, for which the coefficients
are introduced as dependent on the rate of loading.
The results of repeated impact loading tests were compared with those of normal fati-
gue tests. It emerged that the commonly employed Smith diagrams would give too
favourable picture for impact fatigue.
Considering bond resistance it is proposed that the relationships given in VB 1974

38
and in the CEB-FIP Recommendations should be used. The effect of the rate ofloading
can be neglected for plain bars and for prestressing strands. For deformed bars subject-
ed to monotonically increasing load, however, the tensile strength associated with a par-
ticular loading rate should be used instead of the static tensile strength.
For applying the results of this research in actual practice the relevant cases would
include the shear strength of slabs and beams under impact loading, the punching shear
strength of slabs, the crack width, the crack spacing and the deflection of reinforced con-
crete structural members. By means oflarge-scale investigations [21] it has been shown
that impact loads cause an increase in the magnitude of the failure load, in accordance
with the results of this research. Another relevant possibility is to reinforce foundation
piles in order to prevent brittle fracture.
Taking account of the complexity of this problem it does not appear appropriate to
introduce the impact tensile strength values into the commonly used formulae before
investigations will show how that should be done. This fact leads to a suggestion for
further research: to investigate the behaviour of piles during pile-driving, both theore-
tically and experimentally, with particular reference to the behaviour during and after
cracking. Another direction in which research may be continued is that of the multi-
dimensional behaviour of concrete under impact loading. This in order to establish a
failure criterion for concrete at high rates ofloading. When that problem will be solved
for ordinary concrete, attention must be turned to fibre-reinforced concrete and polym-
er concrete, since these concretes may have a significant part to play in structures which
can be subjected to impact and explosion loads.

39
7 Notations

a, b, c, d coefficients
f tensile strength
fa static tensile strength
ftk calculated characteristic tensile strength of concrete
ftm calculated mean tensile strength of concrete
h cube strength (compressive strength)
hm mean cube strength
hk characteristic cube strength
T(O) bond stress as function of displacement
n number of test specimens
n modular ratio of steel and concrete
r correlation coefficient
x coordinate
A,B coefficients
E modulus of elasticity
N number of load applications to failure
jJ coefficient
o displacement (pull-out)
c strain
i: strain rate
coefficient
reinforcement percentage
stress
loading rate (stress rate)
loading rate in static test
Gmax upper stress limit for repeated loading
T bond stress for a certain displacement
TO bond stress in static test
i- loading rate for bond
TO loading rate in static bond test
characteristic diameter of a plain or a deformed bar

40
8 References
1. KOMLOS, K., Investigation of rheological properties of concrete in uniaxial tension. Material-
priifung 1970, nr. 9, pp. 300-304.
2. HEILMANN, H. G., H. HILFSDORF and K. FINSTERWALDER, Festigkeit und Verformung von
Beton unter Zugspannungen. Deutscher Ausschuss fUr Stahlbeton, Heft 203, Berlin 1977.
3. TAKEDA, J. and H. TACHIKAWA, Deformation and fracture of concrete subjected to dynamic
load. Proceedings International Conference "Mechanical behaviour of materials", Kyoto
1971, Vol. IV.
4. KVIRIKADZE, O. P., Determination of the ultimate strength and modulus of deformation of
concrete at different rates of loading. International Symposium "Testing in situ of concrete
structures", Budapest 1977, pp. 109-117.
5. BIRKIMER, D. L. and R. LINDEMANN, Dynamic tensile strength of concrete materials. Journal
of the American Concrete Institute, title no. 68-8, January 1971.
6. HJORTH, 0., Ein Beitrag zur Frage der Festigkeiten und des Verbundverhaltens von Stahl und
Beton bei hohen Beanspruchungsgeschwindigkeiten. Dissertation Technische Universitat
Braunschweig, 1976.
7. HANSEN, R. J. and A. A. LIEPINS, Behaviour of bond under dynamic loading. Journal of the
ACI (1962), no. 4, pp. 563-583.
8. FAGERLUND, G. and B. LARSSON, Betongs slaghallfasthet. Cement- och betonginstitutet,
Forskning Fo 4:79, Stockholm 1979.
9. MlHASHI, H. and M. A. IZUMI, Stochastic theory for concrete fracture. Cement and Concrete
Research 7 (1977), pp. 411-422.
10. MIHASHI, M. and F. H. WITTMANN, Stochastic approach to study the influence of rate ofload-
ing on strength of a concrete. HERON 25 (1980) no. 3.
11. KOLSKY, H., An investigation of the mechanical properties of materials at very high rates of
loading. Proceedings Phys. Soc. Sec. B62, (1949), pp. 676-700.
12. STROEVEN, P., Some aspects of the micromechanics of concrete. Dissertation Delft University
of Technology, 1973.
13. MCCLINTOCK, F. A. and A. S. ARGON, Mechanical behaviour of materials. Addison-Wesley,
Reading, U.S.A., 1966.
14. MARTIN, H. and P. NOAKOWSKI, Verbundverhalten von Betonstahlen. Aus unseren For-
schungsarbeiten IV, Technische Universitat Miinchen, December 1978, pp. 75-77.
15. REHM, G., Uber die Grundlagen des Verbundes zwischen Stahl und Beton, Deutscher Aus-
schuss fUr Stahlbeton, Heft 138, Berlin 1961.
16. EISENMANN, J., Betonfahrbahnen. Wilhelm Ernst & Sohn, Berlin/Miinchen 1979.
17. Voorschriften Beton VB 1974, NEN 3861, Nederlandse Normalisatie-instituut (Netherlands
Standard Institute), Delft.
18. CEB-FIP Model Code for Concrete Structures. Comite Euro-International du Beton, Paris
1978.
19. KORMELING, H. A., A. J. ZIELINSKI and H. W. REINHARDT, Experiments on concrete under
single and repeated uniaxial impact tensile loading. Stevin report 5-80-3, Delft University of
Technology, May 1980.
20. Vos, E. and H. W. REINHARDT, Bond resistance of deformed bars, plain bars and strands un-
der impact loading. Stevin report 5-80-6, Delft University of Technology, September 1980.
21. Symposium "Stossbeanspruchung von Betonkonstruktionen", Universitat Dortmund, Sep-
tember 1980.

41
Appendix A

Table AI. Grading of aggregates with different maximum grain size

cumulative percentage
retained (m/m) at maximum grain size
sieves acc. to NEN 2570
sieve size (mm) 8 mm 16 mm 24 mm
24
16 31
8 29 60
4 30 50 78
2 50 64 80
1 65 75 90
0,5 80 82 94
0,25 93 95 97
0,125 100 100 100

Table A2. Composition and results of the control tests on the concrete of the preliminary
program
splitting tensile
cube strength strength
maximim coefficient coefficient
cement particle type water- of of
content size of cement fern variation hpi variation
mix (kg/m3) (mm) cement ratio (N/mm 2 ) (%) (N/mm 2) (%)
1 325 16 PA 0,45 51,04 4,1 3,52 11,1
2 325 24 PB 0,45 52,15 1,6 2,95 5,4
3 375 24 PB 0,40 57,05 2,3 3,20 15,6
4 375 24 PB 0,45 48,28 3,9 3,24 9,9
5 325 16 PB 0,45 47,05 3,2 2,93 8,5
6 375 16 PB 0,40 53,53 3,4 2,97 6,4
7 375 16 PB 0,45 50,09 3,4 3,18 17,9
8 325 16 PC 0,45 60,84 1,2 3,66 7,4
9 375 16 PC 0,40 62,53 1,2 3,38 6,8
10 316 16 PB1A 0,48 43,82 3,1 2,65 6,4
11 316 16 PB1B 0,45 49,14 2,2 3,03 11,5
12 364 16 PB1B 0,41 52,24 1,9 3,03 4,3
13 358 16 PB1B 0,45 51,10 3,6 3,37 4,7
14 290 16 PA 0,52 37,23 4,2 2,58 10,1
15 325 8 PB 0,48 41,51 0,8 2,79 6,4
16 375 8 PB 0,43 47,96 3,8 2,63 5,7
17 352 8 PB 0,48 41,60 4,7 2,69 4,5
18 375 16 PC 0,45 59,78 2,8 3,25 7,4

43
Table A3. Composition and results of the control tests on the concrete of the main program
splitting
cube strength tensile strength
maximum coefficient coefficient
cement particle water- of of
content size moisture cement fem variation /'Pl variation
mix (kg/m)) (mm) condition ratio (N/mm2) (%) (N/mm2) (%)
19 325 16 dry 0,40 61,14 1,4 3,33 9,6
20 325 16 wet 0,40 60,10 0,9 3,44 8,7
21* 325 16 dry 0,50 47,80 6,2 2,95 10,0
22* 325 16 wet 0,50 44,79 5,5 3,00 11,3
23* 375 16 dry 0,40 58,25 6,3 3,31 10,1
24 375 16 wet 0,40 54,66 7,2 3,50 16,0
25* 375 16 dry 0,50 45,73 3,4 3,06 9,7
26 350 8 dry 0,40 54,20 4,2 2,88 5,7
* used also for repeated impact loading

44
 Table A4. Calculated values of the impact tensile strength, obtained from the regression analysis based on: In! = A + BIn o.
impact tensile strength
for a loading rate 0 in
regression N/mm 2 ·ms
number
coefficient 95% confidence belt for
of tests 0=0,00010=3 0=30 !(o= 30)
mix results A B r2 In! B (N/mm 2) (N/mm 2) (N/mm 2) !(o= 0,0001)
1 9 1,552 0,0324 0,87 ± 0,182 ± 0,0106 3,50 4,89 5,27 1,50
2 9 1,375 0,0320 0,81 ± 0,223 ± 0,0135 2,94 4,10 4,41 1,50
3 10 1,407 0,0276 0,70 ± 0,254 ± 0,0145 3,17 4,21 4,49 1,42
4 9 1,367 0,0226 0,53 ±0,315 ± 0,0185 3,19 4,02 4,24 1,33
5 9 1,546 0,0520 0,95 ± 0,172 ± 0,0101 2,91 4,97 5,60 1,92
6 9 1,538 0,0495 0,96 ± 0,157 ± 0,0092 2,95 4,92 5,51 1,87
7 9 1,623 0,0521 0,91 ± 0,247 ± 0,0146 3,14 5,37 6,05 1,93
8 9 1,604 0,0329 0,82 ± 0,233 ± 0,0136 3,67 5,15 5,56 1,52
9 9 1,652 0,0471 0,92 ± 0,201 ± 0,0118 3,38 5,49 6,12 1,81
10 8 1,418 0,0486 0,95 ± 0,165 ± 0,0103 2,64 4,35 4,87 1,84
11 9 1,538 0,0487 0,83 ± 0,332 ± 0,0194 2,97 4,91 5,49 1,85
12 9 1,606 0,0539 0,98 ± 0,121 ± 0,0070 3,03 5,29 5,98 1,97
13 9 1,638 0,0471 0,87 ± 0,252 ± 0,0158 3,33 5,42 6,04 1,81
14 9 1,392 0,0483 0,80 ± 0,336 ± 0,0210 2,58 4,24 4,74 1,84
15 9 1,543 0,0566 0,96 ± 0,163 ± 0,0102 2,78 4,98 5,67 2,04
16 8 1,576 0,0675 0,96 ± 0,205 ± 0,0132 2,60 5,21 6,08 2,34
17 9 1,535 0,0590 0,99 ± 0,083 ± 0,0052 2,69 4,95 5,67 2,11
18 9 1,563 0,0424 0,87 ± 0,225 ± 0,0142 3,23 5,00 5,51 1,71
19 II 1,492 0,0370 0,93 ± 0,140 ± 0,0073 3,16 4,63 5,04 1,59
20 14 1,504 0,0311 0,85 ± 0,179 ± 0,0083 3,38 4,65 5,00 1,48
21 15 1,526 0,0519 0,91 ± 0,205 ± 0,0100 2,84 4,87 5,49 1,93
22 15 1,493 0,0431 0,90 ±0,173 ± 0,0084 3,00 4,66 5,15 1,72
23 26 1,482 0,0322 0,80 ± 0,196 ± 0,0068 3,28 4,56 4,91 1,50
23* 22 1,599 0,0447 0,90 ± 0,186 ± 0,0069 3,27 5,20 5,77 1,76
24 19 1,507 0,0350 0,77 ± 0,213 ± 0,0098 3,27 4,69 5,08 1,55
25 38 1,457 0,0374 0,84 ± 0,183 ± 0,0055 3,05 4,47 4,87 1,60
26 7 1,401 0,0373 0,92 ± 0,208 ± 0,0130 2,89 4,23 4,60 1,59
26* 13 1,533 0,0513 0,90 ± 0,209 ± 0,0115 2,89 4,90 5,51 1,91
.j::>.
V1
* loading direction perpendicular to casting direction
oj:>.
0\

Table A5. Calculated values of the strength after repeated loading, obtained from the regression analysis based on: amax = A + BIn N.
regression 95% impact tensile strength after
number coefficients confidence belt for N load applications (N/mm2)
of test amaxW == roool
mix results A B r2 °max B N=1 N= 10 N= 100 N= 1000 amax(N = 1)

21 13 4,142 -0,360 0,72 ± 1,580 ± 0,150 4,14 3,31 2,48 1,65 0,40
22 14 4,512 -4,080 0,80 ± 1,343 ±0,131 4,51 3,57 2,63 1,69 0,37
23 35 3,757 -0,242 0,65 ± 0,808 ± 0,063 3,76 3,20 2,64 2,08 0,55
23* 7 5,649 -0,356 0,93 ± 0,596 ± 0,116 5,65 4,83 4,01 3,19 0,56
25 16 4,513 -0,245 0,78 ± 0,674 ± 0,075 4,51 3,95 3,38 2,82 0,62
* loading direction perpendicular to casting direction

Table A6. Calculated values of the strength after repeated loading, obtained from the regression analysis based on: amax/fa = A + B In N
95% confidence belt for relative impact tensile strength after
number regression coefficients N load applications (N/mm2)
of test °max
mix results A B r2 -y;;; B N=1 N= 10 N= 100 N= 1000
21 13 1,404 -0,122 0,72 ± 0,535 ± 0,051 1,40 1,12 0,84 0,56
22 14 1,504 -0,136 0,80 ± 0,447 ± 0,041 1,50 1,19 0,88 0,56
23 35 1,135 -0,073 0,65 ±0,144 ± 0,018 1,13 0,97 0,80 0,63
23* 7 1,706 -O,108 0,93 ± 0,180 ± 0,033 1,71 1,46 1,21 0,96
25 16 1,475 -0,080 0,78 ± 0,220 ± 0,023 1,47 1,29 1,11 0,92
* loading direction perpendicular to casting direction
 Table A7. Results of the pull-out tests
splitting :rOdUIUS bond stress T for displacement 0 = .. mm
type cube tensile behav- loading
of strength strength elasticity 0,01 0,03 0,05 0,10 0,15 0,20
iour at rate
steel code* mix (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm 2 . ms) failure**

ribbed 0806 23,4 2,00 30,9 8,32 11,34 12,60 13,27 13,27 12,85 108
steel, 0812 23,4 2,00 30,9 6,50 9,62 10,96 12,32 12,90 13,19 104
010
0807 23,4 2,00 30,9 6,17 9,07 10,38 11,42 11,80 11,70 50,1
0808 23,4 2,00 30,9 4,84 7,44 8,41 9,95 10,68 11,20 40,1
0811 23,4 2,00 30,9 6,69 8,73 9,74 10,89 11,66 12,25 27,2
0905 20,9 1,73 26,2 5,48 7,43 8,41 9,17 9,62 9,90 0,777.10- 1
0906 20,9 1,73 26,2 5,57 7,70 8,50 9,53 9,98 10,34 0,757.10- 1
0804 23,4 2,00 30,9 4,13 6,00 7,18 8,86 9,88 10,61 0,806.10- 1
0805 23,4 2,00 30,9 4,73 6,70 7,78 9,44 10,52 11,38 0,931.10- 1
0901 20,9 1,73 26,2 3,72 5,41 6,31 7,04 7,58 7,89 0,289.10- 3
0801 23,4 2,00 30,9 2,63 5,00 6,00 7,47 8,37 9,04 0,300.10- 3
0802 23,4 2,00 30,9 3,32 4,58 5,25 6,35 7,03 7,52 0,296.10- 3
0707 2 44,5 2,88 34,9 12,35 17,08 19,66 22,61 23,43 23,35 107
0708 2 44,5 2,88 34,9 10,31 16,90 19,56 21,93 22,27 22,10 117
0709 2 44,5 2,88 34,9 12,54 17,59 19,84 21,75 21,63 21,07 124
0710 2 44,5 2,88 34,9 9,62 14,52 16,38 18,29 19,22 19,73 23,4
0711 2 44,5 2,88 34,9 12,35 16,82 18,68 25
0712 2 44,5 2,88 34,9 10,32 15,39 17,28 19,10 19,61 20,03 23,9
0602 2 46,4 3,03 34,4 10,55 15,20 17,10 19,19 20,03 20,46 30,3
0701 2 44,5 2,88 34,9 9,21 14,69 17,49 20,65 22,24 23,01 0,823.10- 1
0702 2 44,5 2,88 34,9 8,74 13,92 16,22 18,50 19,39 19,82 0,868.10- 1
0703 2 44,5 2,88 34,9 10,05 14,73 16,46 18,14 18,89 19,39 0,860.10- 1
0705 2 44,5 2,88 34,9 8,80 12,82 14,52 16,51 17,63 18,33 0,296.10- 3
0706 2 44,5 2,88 34,9 9,09 12,90 14,53 16,46 17,63 18,40 0,294.10- 3
0613 2 46,4 3,03 34,4 7,06 10,71 12,27 14,10 15,04 15,74 0,328.10- 3
""'--..l" 0614 2 46,4 3,03 34,4 7,77 10,99 12,47 14,28 15,37 16,01 0,297.10- 3
 Table A7 . continued
.j::.
00 splitting ~OdUIUS bond stress r for displacement 0 = .. mm
type cube tensile loading behav-
of strength strength elasticity 0,01 0,03 0,05 0,10 0,15 0,20 rate iour at
steel code* mix (N/mm2) (N/mm2) (N/mm 2) (N/mm 2) (N/mm 2) (N/mm 2) (N/mm 2) (N/mm2) (N/mm 2) (N/mm 2 • ms) failure**
1201 3 53,9 3,34 34,8 16,22 20,50 22,65 25,28 26,60 25,85 158 1
1202 3 53,9 3,34 34,8 13,67 20,12 23,52 27,88 30,07 31,22 172 1
1105 3 55,2 3,88 34,7 12,73 19,02 22,35 25,56 26,27 25,95 159 1
1106* 3 55,2 3,88 34,7 20,17 23,53 25,18 27,50 28,65 29,60 2ll 2+3
1101 3 55,2 3,88 34,7 11,36 17,71 20,35 23,63 25,40 26,40 19,9
1102 3 55,2 3,88 34,7 11,40 17,00 19,55 23,77 26,40 22,7
1103 3 55,2 3,88 34,7 12,37 18,01 21,01 26,40 18,2
1210 3 53,9 3,34 34,8 9,08 13,86 16,56 20,61 23,08 24,89 0,877 . 10-1 3
1211 3 53,9 3,88 34,7 10,23 15,52 18,72 24,06 27,59 30,54 0,872.10- 1 3
1111 3 55,2 3,88 34,7 9,31 14,61 17,78 22,21 25,06 26,88 0,841.10- 1 3
1112 3 55,2 3,88 34,7 10,66 16,24 19,83 25,07 28,13 29,73 0,852.10- 1 3
1203 3 53,9 3,34 34,8 8,38 12,81 15,73 20,64 24,02 0,287.10- 3 3
1205 3 53,9 3,34 34,8 8,05 12,95 15,82 20,27 23,09 24,88 0,289.10- 3 3
1108 3 55,2 3,88 34,7 9,90 14,69 17,42 21,73 24,73 26,70 0,294.10- 3 3
1110 3 55,2 3,88 34,7 8,86 13,59 16,19 20,24 22,79 24,65 0,297.10- 3 3
plain 0714 2 44,5 2,88 34,9 4,50 5,10 40
steel, 0607 2 46,4 3,03 34,4 5,20 6,05 0,3.10- 3
010 0608 2 46,4 3,03 34,4 4,35 4,90 0,3.10- 3
1003 3 59,1 3,53 34,9 3,65 3,65 40
1005 3 59,1 3,53 34,9 3,20 3,05 40
1007 3 59,1 3,53 34,9 3,75 4,05 0,3.10- 3
1008 3 59,1 3,53 34,9 3,50 3,65 0,3.10- 3
pre- 1303 2 55,5 3,38 5,50 5,50 5,65 5,90 6,25 6,45 40
stress- 1304 2 55,5 3,38 3,60 4,55 5,00 5,70 5,75 5,80 40
ing 1301 2 55,5 3,38 5,65 5,65 5,60 5,55 5,50 5,50 0,3.10- 3
strand, 1302 2 55,5 3,38 5,50 6,00 5,95 6,00 6,15 6,20 0,3.10- 3
09,6
* key to code: first two digits indicate casting number, last two digits indicate specimen number
** failure behaviour: 1. slipping
2. splitting
i t"n<i1" "r""k