Beruflich Dokumente
Kultur Dokumente
LICENTIATE THESIS
Erik Nordstrm
Licentiate thesis
Institutionen fr Vg- och vattenbyggnad
Avdelningen fr Konstruktionsteknik
ERIK NORDSTRM
PREFACE
Funding of the literature study and field exposure tests were made by SveBeFo,
ELFORSK and the Swedish Road authorities together. All other work was
generously financed by ELFORSK.
This work is also part of the research consortium Vg/Bro/Tunnel financed by
NUTEK, SBUF, Cementa, Elforsk, LKAB, NCC, PEAB and Skanska. Participation
in the consortium made the close cooperation with the department of Civil
engineering at Lule University of Technology possible. It also prepared the ground
for getting interesting contacts with other representatives from the industry as well as
professors and postgraduate students from other universities.
The type of investigations performed in this study with different type of tests and
long-time experiments also make many persons being connected to the work. Many
thanks to:
G
MSc Pr Hansson, Ericsson Radio Access for frutiful discussions and great
colleague-ship during his time at Vattenfall Utveckling AB.
The always interested reference groups connected to the parts funded by SveBeFo.
Peter Mjrnell, Bekaert for supplying me with cold drawn wire for making the
fibres.
Georg Danielsson, Lule University of Technology for help with the uni-axial
tests and cracking of samples.
My family and especially my beloved wife Maria who is as much part of this work
as I am, since we together managed to get through both work, studies and extreme
family situations during this work.
Erik Nordstrm
VI
ABSTRACT
A combination of the sprayed concrete technique and steel fibre technology gives
obvious advantages when saving the work needed to place conventional
reinforcement. In rock strengthening applications this is most accentuated.
Sprayed concrete in general, made by skilled worksmen, will recieve a high quality
and good durability. Durability requirements can also be found in todays regulations
with demands on service-life of more than 100 years. Since the steel fibre
reinforcement in wet-mix sprayed concrete has been common practice only since the
late 80:s questions could be raised regarding the resistance to corrosion. It has
previously been proved that steel fibres show an excellent durability against corrosion
in homogenous concrete. At conditions where conventional reinforcement show high
rates of corrosion the steel fibres can still be unaffected. Fibres have a smaller size
than conventional reinforcement and they seem therefore to be better protected by the
alkaline environment the concrete give. Smaller cathode area compared to the anode
area is another argument to the better resistance against corrosion.
The high quality combined with relatively thin layers applied in sprayed concrete
structures give rise to imposed deformations by shrinkage, which is a common reason
for cracks. In the design of steel fibre reinforced sprayed concrete (SFRSC) for e.g.
rock strengthening purposes the fibres are used both to minimize crack widths from
shrinkage and to obtain a sufficient post-crack behaviour. A system with bolts and
SFRSC is depending on a long-term residual strength capacity.
The purpose with the thesis is to investigate the mechanisms ruling initiation and
propagation, or possibly to explain the higher resistance against corrosion of steel
fibres in cracked concrete.
Field inspections of old SFRSC show that the amount of corrosion is limited after 515 years of exposure. Even with presence of high chloride concentrations the attack
seemed small. In all the inspected structures the amount of fibres crossing the cracks
was very small.
Two different approaches to study the corrosion of steel fibres in cracks have been
tested. Cracked beams of SFRSC have been exposed in field at three different sites.
Crack width, fibre length, mix-composition, accelerators and spraying technique
(wet-/dry-mix) are parameters beeing tested. After 2.5 years of exposure mainly
samples exposed along a motorway with direct splashing of water containing de-icing
salts show corrosion on fibres crossing the crack. A loss of 15-20% of the fibre
diameter in the outer 25 mm is common.
Laboratory studies with accelerated exposure tests have also been performed. The
purpose is to develop a technique to isolate parameters in a better way than in field
and to perform exposure tests in a more controllable environment. In addition a
VII
useful technique combined with a correlation to the field exposures could make it
possible to imitate longer real exposures in a shorter period of time and by this
estimate the long-time behaviour. Mainly the same behaviour as in field, with
increased corrosive attack with increased crack width and fibre length, could be seen
in the laboratory exposures. The influence of fibre length accentuate the importance
of the anode- /cathode ratio for the rate of corrosion which also have been stated for
conventional reinforcement. A very rough estimation is that the laboratory exposures
accelerate the exposure with about 10 times compared to the motorway environment.
As mentioned the steel fibres are supposed to be able to carry load during the entire
service-life. A discussion about how knowledge about the rate of corrosion could be
used in a design situation is also presented in the thesis. To counteract loss of loadbearing capacity due to fibre corrosion e.g. extra amount of fibres or an increase of
the layer thickness could be prescribed in the mix-design.
VIII
accelererade frsken mste genomfras p ett stt som mjliggr en korrelation med
fltfrsken. Drmed kan lngre tids verklig exponering efterliknas p kortare tid och
p s stt mjliggrs en bedmning av lngtidseffekter av korrosion. Huvudsakligen
uppvisas samma beteende som i fltexponeringarna med kat angrepp med kande
sprickvidd och kad fiberlngd. Inverkan av fiberlngd betonar vikten av anod/katod-frhllandet fr korrosionshastigheten vilket ocks pvisats fr konventionell
armering. En mycket grov uppskattning r att laboratorieexponeringarna ger ca. 10
gnger acceleration jmfrt med normal exponering i motorvgsmilj.
Som nmnt tidigare frvntas stlfibrerna kunna bra last under hela konstruktionens
livslngd. En diskussion kring hur kunskap om korrosionshastigheter skulle kunna
anvndas vid dimensionering presenteras ocks. Fr att motverka frlust av
lastbrande frmga p.g.a. fiberkorrosion skulle t.ex. extra mngd fibrer eller kad
skikttjocklek kunna freskrivas vid proportionering.
TABLE OF CONTENTS
PAGE
ABSTRACT
VII
IX
NOTATIONS
1
INTRODUCTION
1.1
1.2
1.3
1.4
1.5
1
1
2
2
2
Service-life requirements
Fibres and the sprayed concrete technique
Steel fibre corrosion
Research significance
Disposition of the thesis
XV
Introduction
Methodology for status control
Collection of object information
General inspection
Detailed inspection
Laboratory investigations
Selection of structures
Results of inspections
Structures exposed in mild conditions
Structures exposed in medium conditions
Structures exposed in severe conditions
Conclusions
3
3
3
3
3
4
5
5
8
8
11
12
13
15
3.1
3.2
3.2.1
3.2.2
3.3
3.3.1
3.3.2
3.3.3
3.4
3.5
15
17
17
18
20
20
21
22
23
23
Corrosion in general
Reinforcement corrosion in homogeneous concrete
Initiation by carbonation
Initiation by chloride ingress
Reinforcement corrosion in cracked concrete
Crack width
Anode- / cathode- area ratio
Self-healing of cracks
Influence of steel quality on reinforcement corrosion
Previous investigations on corrosion in cracked SFRC
XI
23
24
25
26
27
29
4.1
4.2
4.3
4.3.1
4.3.2
4.3.3
4.3.4
4.3.5
4.4
4.4.1
4.4.2
4.4.3
4.4.4
4.4.5
4.5
4.5.1
4.5.2
4.5.3
4.6
29
29
29
30
30
32
33
35
39
39
39
41
43
43
45
45
45
46
47
Background
Scope
Methodology
Spraying method
Concrete composition
Manufacturing of samples
Exposure environment
Evaluation after exposure
Results
Climatic conditions
Residual strength
Chloride content
Carbonation
Corrosion on fibres
Discussion
Residual strength
Chloride content
Extent of corrosion
Conclusions
49
5.1
5.2
5.2.1
5.2.2
5.2.3
5.2.4
5.3
5.3.1
5.3.2
5.4
5.5
5.6
49
49
49
50
52
54
54
54
56
57
58
59
Scope
Methodology
Concrete composition
Manufacturing of samples
Exposure environment
Evaluation after exposure
Results
Fibre corrosion
Exposure environment
Correlation to the field exposure tests
Discussion
Conclusions
XII
LOAD-BEARING CAPACITY
61
6.1
6.2
6.3
6.3.1
6.3.2
6.4
61
61
63
63
63
65
Introduction
Analytical model
Service-life modelling
Definition of limit state
Service life
Discussion
CONCLUSIONS
67
RESEARCH NEEDS
69
REFERENCES
71
APPENDICES
PAGE
APPENDIX A:
75
APPENDIX B
77
APPENDIX C
79
APPENDIX D
81
APPENDIX E
83
APPENDIX F
95
XIII
XIV
NOTATIONS
Latin letters
A
Ac
fb
fs
Fmean
F
h
Ie
l
le
lec
M1
M2
N
r
Rel
Rc
Ra
Rs
RH
S
Sc
T1
T2
t
tc
Ue
w
x
Greek letters
XV
Chapter 1 - Introduction
INTRODUCTION
In the late 80:s the technique to reinforce sprayed concrete with steel fibres became
common practice. The construction of an underground deposit for waste materials
from the nuclear power plant in Forsmark, Sweden and the railroad tunnels along
Grdingebanan, Sweden were two of the first major projects. Great advantages by
decreased amount of work needed for placing mesh reinforcement was most obvious
in rock strengthening applications. Today steel fibre reinforced sprayed concrete is
common practice for permanent linings in underground construction in Scandinavia.
It also occurs in repair and strengthening of concrete structures.
1.1
Service-life requirements
Except from advantages during construction, fibres are used in two major purposes.
Reduction of crack widths and achieval of a ductile post-crack behaviour. Where
fibres are used in the last mentioned application design criterias can be found in
Holmgren (1992). Sprayed concrete is most commonly applied in relatively thin
layers (typically 50-100 mm) and the concrete quality is high. This will give a high
degree of shrinkage due to drying out. High air speeds, in e.g. tunnel applications,
due to ventilation, further increase this effect. Some types of accelerators also give an
increased shrinkage (Manns & Neubert, 1992). All this together significantly
increases the risk for shrinkage cracks (crack width= 0.1-0.5 mm). Loads from
1
Chapter 1 - Introduction
movements in the substrate being sprayed on is another possible, but less common,
source to cracks. These cracks could be wider. A positive property of sprayed
concrete is that it is always applied on another substrate and due to the limited
thickness the restraint also limits the crack width.
1.3
One could suspect that the relatively thin steel fibres would discontinue to carry load
relatively fast due to decrease of fibre diameter caused by corrosion. Especially in
cracked concrete. The excellent performance of fibres against corrosion in
homogenous concrete has been shown as early as 1966 by Shroff (1966). Previous
investigations on cracked concrete are presented in chapter 3. It seems as steel fibres
corrode at a lower rate than conventional reinforcement at the same conditions. The
parameters giving this positive property is not very well known. It is also possible
that the rate of degradation is not linear. An increase in the rate of corrosion will give
a rapid break down of the thin fibres and by this a reduced load bearing capacity.
1.4
Research significance
With knowledge that makes service life estimations possible, the confidence for and
use of steel fibre reinforced concrete can increase. Knowledge of important
parameters ruling initiation and propagation of cracked steel fibre reinforced concrete
can also be used in design e.g. when choosing type and amount of fibres or in
definition of an extra cover without fibres. It can further be used to assure a certain
service life of old structures when it concerns acceptable crack widths or chloride
contents in different environments.
1.5
2.1
Introduction
Inititally a general inspection was performed to control the general status of the
structures. The results from this first inspection then determined whether a detailed
inspection with sampling etc. should be made.
2.2.1 Collection of object information
A general inspection can be a good and relatively cheap method to estimate the
condition of a sprayed concrete structure. Some simple field methods are:
G
location on the structure also give imporant input. Change in colour etc. is another
useful observation. An ocular inspection is essential to create a complete picture
of the condition of the structure. Single test results is most commonly not relevant
without this type of inspection.
G
Crack width measurment. Can give information about the reason to the cracking.
Hammer tapping. Used to locate areas with loss of bond strength. It can also be
used to find areas with sprayed concrete of low quality.
Photo and video documentation. A good support for the recollection back at the
office. This can prevent costly extra visits to the object. Important to take pictures
in all scales, both close-ups and overall views.
A detailed inspection needs more equipment like a drilling machine for cores, a
portable generator and a pump for cooling water to the drilling. On the other hand a
more accurate and definite evaluation will be the result. All methods presented below
depends on the possibility to drill out cores.
G
With drilling equipment in field it is possible to take samples for further investigation
in lab.
G
Chloride profile. Drilling debris is collected at different depths from the exposed
surface. The debris is then tested regarding the chloride content accordring to e.g.
the RCT-method (Rapid Chloride Test) developed by Germann & Hansen (1991).
Compressive strength. From concrete cores samples can be sawn for testing of
compressive strength. Except from the actual compressive strength of the structure
this will also give an indication of the overall quality (low compressive strength =
low quality, more porous and higher permeability).
Fiber content. By crushing of concrete cores the actual fibre content can be
controlled
Age. The primary criteria used taking into consideration the type of spraying
method. The dry-mix spraying method has changed to a limited amount during the
decades. Therefore structures constructed with dry-mix spraying usually are older.
Structures with wet-mix spraying combined with steel fibres are most commonly
younger.
5
Structure type. A mix of underground structures and concrete repairs was wanted.
Environment. Structures from as many categories as possible was the goal. Both
rock cavities with relatively constant temperarure and humidity and concrete
bridge repairs with outdoor conditions and splash from thaw salts are included.
A total of 16 different structures where examined during the inspections (see Table
2.1-Table 2.3). In Table 2.4 the structures are divided according to exposure
environment and severeness of freeze-thaw exposure and corrosivity. Class A
represents the corrosivity and class B the freeze-thaw action. The numbers show the
degree of severeness with a maximum at 4.
Table 2.1
No. Site
Structure
Method
Reinforcement
Constructed
Wet-mix
Steel fibres
1985
Stornorrfors, Ume
Viskan, Varberg
Road tunnel
Dry-mix
Steel fibres
1980
Skogby, Halmstad
Dry-mix
Steel fibres
1985
Eugenia, Stockholm
Road tunnel
Dry-mix
Bars
1990
Umluspen, Storuman
Ventilation tunnel
Wet-mix
Steel fibres
1990
Graversfors, Norrkping
Dry-mix
Mesh
1960
d, Kramfors
Dry-mix
Bars
1957
power plant
Table 2.2
No. Site
Structure
Method
Reinforcement
Constructed
Nmforsen, Nsker
Spillway
Dry-mix
Polypropylen fibres
1990
Bridge column
Dry-mix
Bars
1991
10
Stadsforsen,
Spillway
Dry-mix
Steel fibres
1989
Bispgrden
11
Bridge column
Dry-mix
Bars
1990
12
Viaduct
Dry-mix
1990
13
Hlleforsen,
Spillway
Dry-mix
Steel fibres
1990
Crane foundation
Dry-mix
1991
Bispgrden
14
Forshuvud, Borlnge
Table 2.3
Structure
Method
Reinforcement
Constructed
15
Road culvert
Dry-mix
Steel fibres
1991
16
Road culvert
Dry-mix
Bars
1991
Table 2.4
1.
A2
A3
A4
Stornorrfors
B1
Freeze-thaw classification
(steel fibres)
-
B2
5.
Umluspen
3.
Skogby
7.
14. Forshuvud
-
6.
Graversfors
10.
Stadsforsen
8.
Nmforsen
4.
Eugenia
13. Hlleforsen
B3
2. Viskan
9. Road E4,deshg
11. Road E20,Lerum
B4
(bridge column)
12. Road
E4,Norrkping
The Stornorrfors hydropower plant is situated some 15 km west of Ume along the
Ume river. Mainly this underground plant was constructed in 1958. Extra space for
the fourth generator was made in 1985 and in connection to this the headrace tunnels
were strengthened with steel fibre reinforced sprayed concrete.
Examined
Figure 2.2 Skeleton sketch over headrace tunnel for unit G4, Stornorrfors hydro
power plant.
The climatic conditions in the area above water give a relatively constant temperature
(8-10C) and a high level of humidity. Only limited leakage through cracks and other
defects could be seen.
Generally the sprayed concrete seems to be in good condition with few cracks above
water level. To examine steel fibres crossing a crack, core drilling was performed
from a rubber boat in headrace tunnel G3. Cores at levels 0.1m, 0.5 m and 1.25 m
above the mean water level were taken out. Very few fibres were crossing the crack
when comparing it to the added amount of fibres (70 kg/m3) in the sprayed concrete.
No signs of corrosion could be found on fibres crossing the crack, only on fibres
exposed to the surface. A thin layer of cement paste (1 mm) seem to be enough to
protect the fibres from corroding in this environment.
The sprayed concrete thickness was measured to 70-150 mm. Depth of carbonation
was controlled and the results are presented in Table 2.5.
Table 2.5
Height above
Carbonation
depth
(mm)
C1
1.25
1-10
C2
0.5
C3
0.1
Lime deposits could be found in the crack in sample C1. Close to the surface the
crack was completely healed. In sample C2 lime deposits could be found at a depth of
10 mm and downwards in the crack. Sample C3 contains no crack but was split in
two pieces to study the fibres. The amount of fibres was much larger than in the other
samples.
A single test of compressive strength was performed on one core from the bond
strength tests. It was measured to be 109 MPa. The amount of fibres was measured
to 70 kg/m3. The fibres were of type Hrle (low carbon steel and hooked ends, l=
30 mm, =0.6 mm). According to available documentation the cement content is 450
kg/m3, maximum aggregate size 8 mm and water glass accelerator was used in wet
areas.
G
Discussion:
In general the examined part of the structure is in good condition. Steel fibre
corrosion is located to fibres exposed to the surface and after 10 years of use no
indications of initiation of corrosion on fibres crossing cracks could be seen.
A question about crack localization could be rised since a low amount of fibres could
be found crossing the crack, but a sufficient amount was available close to the crack.
If cracks propagate in areas with low amount of fibres this focus on the need for
adequate distribution of fibres in the sprayed concrete. The load-bearing capacity in
the cracked state is depending on the presence of fibres crossing the crack.
Only one result on compressive strength is too little but the remarkably high strength
is still an indication of very high quality of the sprayed concrete.
The existing steel sheet road culvert with 2.5 m diameter was built in the early 60:s
and was heavily corroded before sprayed concrete was applied. Especially in the
water line. Access to the road culvert is via a manhole which means that there is no
direct access to outdoor conditions. Even during the winter water is pouring trough
and therefore there is a constant humid climate. Above the road culvert there is 10 m
of road fill. The sprayed concrete is reinforced with steel fibres of type low carbon
steel and hooked ends (Dramix 40/0.5) according to available documentation. The
final layer of sprayed concrete is made without fibres. Tests made during construction
shows a compressive strength of 79 MPa (cubes sawn from sprayed panels).
Figure 2.3 Crack (left) and core drilling (right) in road culvert with steel fiber
reinforcement under road E20, Lerum.
Cracks with various widths (0.5-2 mm) where visible radially all around the crosssection. In several of the cracks red deposits from corrosion of the corrugated steel
could be seen. Core drilling was performed at 0.1, 0.55 and 1.15 m above actual
water level (high water level). The high water level and water speed made drilling
difficult and extra safety precautions had to be taken (see Figure 2.3). At levels 0.1
and 1.15 m the core was crossing a crack. In the upper core only one single fibre
crossed the crack, and this was unaffected by corrosion. Slight surface corrosion
could be found on fibres crossing the crack in the lower core. The underlying steel
sheet thickness was 5 mm in the upper hole and only 1 mm remaining in the lower
due to severe corrosion.
G
Discussion:
It is difficult to state the origin of the cracks. Cracks due to shrinkage is most
commonly not as wide as 1-2 mm and the number of cracks should also be larger.
The low content of fibres crossing the crack could however be the reason to large
crack widths. Redistribution of structural load due to damages of corrosion in the
steel sheets is another possible explanation. In spite of the high level of humidity in
combination with cracks the corrosion on fibres was limited.
10
The Stadsforsen power plant is situated along the river Indal 4 km east of Bispgrden
and it was constructed in 1939 and 1952. The dam has five spillways (see Figure 2.4)
and they where repaired in 1989 with sprayed concrete. The repair works were
performed by removing damaged concrete with hydro demolition. Accurate cleaning
and pre-wetting of the surface to a humid but slightly absorbing surface was
prescribed. The surfaces where steel trowelled and they were also vacuum treated in
the lower parts. Water curing five days after spraying was prescribed.
Figure 2.4 View of spillways (left) and detail of spillway C (right) at the Stadsforsen
power plant.
Spillway C was examined. Steel fibres exposed at the surface are corroded. The joints
between the different spraying stages can be seen at the surface. In the most left part
there is some map cracking but otherwise there are very few cracks. One single crack
could be found in the center of the spillway and a core was drilled across the crack. In
Figure 2.5 a sketch over the spillway and the positions for testing of bond-strength
and core drilling are shown.
4m
B2
4m
S1
B1
B3
S2
20 m
Figure 2.5 Sketch over spillway C and positions for testing at the Stadsforsen
power plant, Bispgrden.
Core S1 was taken out in the joint between two sections and the joint had been coated
with bitumen. Parts with inhomogenous concrete could be seen in connection to the
11
joint. In this part partly corroded fibres was found. No signs of fibre corrosion could
be seen in core S2 taken across the crack. Lime deposits was present outside the
crack and in the crack leached concrete was visible in the upper 2 cm.
The compressive strength of one sample was 129 MPa and compared to measures
during construction (85 MPa) this is a large increase. The steel fibre content was now
measured to 50 kg/m3.
G
Discussion:
In general the spillway is in a good condition without larger damages. The area with
low quality sprayed concrete is most probably due to trapped rebound not being
removed before spraying. Except from this area the compressive strength is
remarkably high which indicates a very high concrete quality. In the examined crack
(not joint), no signs of corrosion was visible.
2.4.3 Structures exposed in severe conditions
G
These 16 year old road tunnels are situated along road E6 approximately 10 km north
of Varberg. Sprayed concrete is mainly used in the ceiling but also along drains and
in the tunnel mouth down to the ground. The inspections where concentrated to the
northbound lane.
Figure 2.6 South mouth of the Viskan tunnel along road E6, Varberg.
Core drilling across a crack was made to check corrosion on steel fibres. The core
was taken out from the top of a drain. The thickness of the sprayed concrete was 2025 mm and farthest out a layer without fibres was applied. In some parts a welded
mesh was visible in the crack but no fibres could be seen across the crack. The mesh
was not corroded in the crack. The chloride content in the concrete surrounding the
crack was measured3 and is presented in Figure 2.7 (In the calculations a cement
content of 450 kg/m has been assumed).
12
1
0.15m above road
0.85m above road
1.45m above road
0,8
0,6
0,4
0,2
0
Surface
1.5cm
Depth from surface
3cm
Figure 2.7 Chloride profile for sprayed concrete in the Viskan tunnel along road E6,
Varberg.
G
Discussion:
The thin sprayed concrete layer over the drain combined with drying shrinkage is
most probably the reason for cracking. The lack of fibres in the investigated part of
the crack points to that this is another reason for localization. Measured chloride
contents are quite high in the lowest core and combined with the crack width 0.2 mm
one could have expected high rates of corrosion on fibres or welded mesh.
2.5 Conclusions
In all the inspected structures a low amount of fibres crossed the cracks when
checked after core drilling. A short distance from the crack the amount of fibres was
sufficient. This points out that cracking in the inspected structures has ocurred where
the least amount of fibres are present.
After 5-15 years of exposure no indications of severe corrosion in cracks could be
seen. The type of environmental exposure does not matter, even with high amount of
chlorides the amount of corrosion was limited. There was also little influence of the
crack width (0.2 2 mm). One should remember that the amount of fibres was very
low in all the inspected structures (no fibres but mesh in the Viskan tunnel).
Generally the concrete quality was very high. More general corrosion only took place
in an area with deficient quality due to rebound being sprayed in.
13
3.1
Corrosion in general
Fe Fe 2 + + 2e
1
O 2 + H 2 O + 2e 2OH
2
1
Fe + O 2 + H 2 O + 2e Fe 2 + + 2e + 2OH
2
(3.1)
(3.2)
(3.3)
14
corrosion. Addition of chlorides increase the potential and corrosion can occur at a
higher value of pH.
Table 3.1
Electrode potentials for metals in sea water at 20C after Mattson (1992)
Metal
Gold
+0,42
Silver
+0,19
Figure 3.1
+0,09
Cupper
+0,02
Tin
-0,26
1
-0,29
Led
-0,31
Steel
-0,46
Cadmium
-0,49
Aluminum
-0,51
Galvanized steel
-0,81
Zinc
-0,86
Magnesium
-1,36
In the passive state the metal surface has a thin coating which retards the
reaction. In a active state, like in a corrosion pit, there is no coating.
16
3.2
In homogenous concrete steel and e.g. reinforcement bars are well protected against
corrosion. The high pH and presence of calcium-hydroxide will create a stable
passifying oxide layer on the steel that prevents further corrosion.
The most commonly refered corrosion model is the one by Tuutti (1982) regarding
conventional reinforcement in homogenous concrete. In the model the corrosion
process is divided into two parts (see Figure 3.2). Part one, called time to intitiation,
describes the time for the protection given by concrete cover to weaken. During this
time conditions for active corrosion is created. Part two, called propagation, describes
ongoing corrosion of the reinforcement.
Penetration towards
reinforcement.
Acceptable depth
Depth of corrosion
RH
O2
CO2, Cl
Time
IInititiation
Propagation
Figure 3.2
Depassiviation of the reinforcement and break down of the layer can occur due to
ingress of chlorides and by carbonation of the concrete. Chlorides will also act as a
catalyst on the corrosion process by a local acidification at the corrosion pit.
Carbonation gives a lowered pH by a reaction between carbon dioxide from the air
and the calcium hydroxide.
3.2.1 Initiation by carbonation
Ingress of CO2 from the air will react with the calcium hydroxide in the concrete
under formation of calcium carbonate and a reduction of pH to approximately 9. The
reaction is shown in equation (3.4). The coating making the reinforcement passive is
degraded when when the carbonation front reaches the reinforcement. At this stage
corrosion is initiated, regarded that other parameters are fulfilled (sufficient amount
of electrolyte etc.). The rate of carbonation is ruled by a number of parameters and
17
some of them are shown in Table 3.2. A schematic sketch showing the initiation and
propagation due to carbonation is shown in Figure 3.3.
Ca(OH)2 + CO2 CaCO3 + H 2O
Table 3.2
(3.4)
* Curing at construction
Unaffected
CO2
CO2
pH> 9
pH>12.5
Passive
Figure 3.3
Passive
Active corrosion
combined with the structure not being sheltered from rain can make the chloride
concentration decrease in the outer parts.
When the chloride concentration reaches a critical level, called the threshold value,
corrosion will be initiated. The chloride ingress and its influence on propagation
depends on a number of factors and some of them are presented in Table 3.3.
Table 3.3
* Threshold values
As can be seen in Table 3.3 chlorides can bind chemically to the cement paste.
Chemical composition (mainly C3A, C4AF, alkali content) of the cement is the main
factor affecting the ability to bind chlorides according to Byfors (1990). Other cat
ions like OH-, SO42-, CO32- also compete about spots where chlorides can bind. An
increased C3A content and decreased alkali content both give an increased chloride
binding. Addition of mineral additive materials like e.g. silika and flyash increases
the chloride binding. It is only the content of free chlorides in the pore solution (not
bound) that influences the corrosive process.
As mentioned above the attack of free chlorides on the reinforcement is connected to
the amount of other cat ions (mainly hydroxide ions) in the pore solution. The critical
ratio between these two when corrosion is initiated is described by Hausmann (1967)
as the threshold value (see equation 3.5) Relevant figures on threshold values vary
between authors according to Sandberg (1998). In previous investigations the
difference between total amount and the content of free chlorides in the pore solution
is not always made. The choice of method for determination of chloride content can
also influence the result. Sandberg (1998) means that there is a particulary large
difference in determining the content of free chlorides in concretes with a w/c-ratio
below 0.45. If the total amount of chlorides is used, the thresholdvalue can vary
between 0.17-2.2 % according to Sandberg (1998).
[Cl ] (mol/l)
[OH ] (ekv/l)
(3.5)
with the iron ions. Sandberg (1998) furthermore means that the complex formed will
be degraded with access to oxygen and this will result in a local decrease of pH
which further increase the corrosion process. Another effect of chlorides in concrete
is that the electrical conductivity in the concrete increases and this facilitates the
anode-/cathode-reaction especially in macro-cell corrosion (see part 3.3.2).
The rate of corrosion after initiation depends on available amount of free chlorides,
the humidity conditions and access to oxygen. A schematic sketch of the process can
be seen in Figure 3.4.
Unaffected
Carbonation front
-
Cl
O2
-
Conc. free Cl
Cl
Conc. free Cl
pH>12.5
"Safe"
concentration
Passive
Figure 3.4
3.3
Passive
"Harmful"
concentration
Active corrosion
There are many suggestions of critical crack widths with regard to corrosion. The
background material can sometimes be questioned since the effect of the concrete
composition and the exact exposure conditions can vary very much. According to
Raupach (1996) cracks mainly affect the initiation of corrosion. As soon as corrosion
has been activated other parameters are more important (micro-chlimate, chloride
binding capacity, anode-/cathode-area etc).
20
The rate of corrosion in cracks can also be decreased by the formation of corrosion
products according to Tuutti (1982) since corrosion products prevent access to e.g.
oxygen. Alonso & al. (1998) support this hypothesis by showing results from longtime exposure where a dry environment can make the anode passive again. This as an
effect of accumulated iron-oxides on the anode, the electrical conductivity in the ironoxides, a redox process of iron-oxide (Fe2+ / Fe3+) and access to oxygen.
3.3.2 Anode-/catode-area ratio
Cracks give rise to so-called macro-cell corrosion which means distinctly separated
anode- and cathode-areas up to several centimeters according to Raupach (1996). The
principles for the reaction is shown in Figure 3.5. Raupach (1996) also presents a
simplified model of an electrical circuit to illustrate the reaction (see Figure 3.6). It
can be seen that important factors are the electrical resistance from the oxide layer in
the transition zone between bar and concrete as well as in the concrete and in the
corrosion products.
Figure 3.5
Figure 3.6
21
The anode reaction is visible due to the formation of corrosion products, while the
cathode reaction is more complex to demonstrate. By measuring the current between
anode and cathode the importance of the ratio between them can be shown. In Figure
3.7 measured out anode- and cathode-currents are shown for a specimen with a crack
and with separated cathode areas.
Figure 3.7
At limited crack widths it is also important to take into consideration the ability for
self-healing. Under humid condition with limited flow of water through a crack
combined with small crack widths, self-healing can occur. Self-healing is a reaction
when carbon dioxide in the water dissolve calcium ions out from the calcium
hydroxide content in the concrete. At low flow of water, the water becomes
supersaturated with calcium- and bicarbonate ions. Calcite crystals deposit along the
crack walls and the crack can slowly be sealed. Concrete quality, crack movements
(static-/dynamic), type of water (pH etc.) are other parameters of importance for selfhealing to occur or not. In Table 3.4 some figures are given on acceptable crack
widths in relation to water pressure and structural dimensions.
The influence of self-healing has been tested by Schiessl and Brauer (1996). The
authors mean that if the water flowing through a crack has a high chloride content
(>0,5 M-% /kg cement) the corrosion process continues in spite of self-healing of the
crack. Otherwise the corrosion process stops.
22
Table 3.4
3.4
(m/m)
(mm)
< 2,5
< 0,2
<5
< 0,15
< 10
< 0,10
< 20
< 0,05
Normally hot-rolled ribbed steel bars are made of steel qualities SIS 2164, 2165,
2167 or 2168. The yield stress for these types of steels are in the region of 600 MPa.
In a normal low carbon steel the carbon content is approximately 1.3%. According to
Mattson (1993) these low levels does not affect the corrosion process.
Stainless steels have alloys with mainly chromium at a content of more than 12%
according to Mattsson (1993). Other alloying materials can be molybdenum, nickel,
cuper and manganese. Even stainless steels can be attacked by corrosion at
unfavourable conditions. Stainless steels are not dealt with in this thesis. A common
materials composition for normal low carbon steels is shown in Table 3.5
Table 3.5
3.5
Si
Mn
(%)
(%)
(%)
(%)
(%)
0.28
0.6
1.6
0.06
0.05
Most of the previous work presented in the area deals with SFRC (Steel Fibre
Reinforced Concrete) exposed to a marine environment. One of these is a compilation
by Hoff (1987). Mainly corrosion of steel fibres in uncracked concrete has been
investigated, but there are a few dealing with cracked SFRC. A short summary of the
four most important ones is presented below. In Table 3.6 a compilation of the data
from these investigations is presented.
3.5.1 IBAC, Aachen, Germany, Schiessl & Weydert (1998)
The scope of the investigations was to study corrosion mechanisms mainly in cracked
SFRC. A few sprayed concrete samples are also included. Both initiation by ingress
of chlorides and by carbonation was dealt with. The specimens were beams sawn out
23
of larger slabs. The crack widths ranged from 0.05-0.4 mm divided in two test
rounds. After exposure the beams were evaluated by dismembering into small plates
around the cracked area. Chloride profiles were created both from the exposed
surface and perpendicular to the crack surface.
In this investigation no significant correlation between crack width and chloride
content could be found. The results of the chloride measurements showed a
decreasing chloride content with increasing distance from the crack opening. The
extent of corrosion also decreased in the same way. Schiessl & Weydert (1998)
suggest that no critical crack width can be stated under which corrosion does not
occur. The reason for this should be that chlorides still can penetrate very thin cracks.
A requirement for corrosion is access to oxygen that can be limited in thinner cracks,
and the positive effect of the alkaline environment in the crack should also be
emphasized in thinner cracks.
The authors also claim that chloride initiated corrosion never will be repassivated in a
crack. Therefore no service life, in the sense that fibres carry load by crossing cracks,
above 10 years for cracked SFRC can be expected.
In specimen exposed to accelerated carbonation the crack walls were carbonated and
the fibres therefore not fully protected against corrosion. In spite of this the fibres did
not corrode to a great extent. The authors suggest that this would be due to lack of
humidity. Therefore carbonation in cracks is of minor importance, but the state of
humidity is the ruling factor. The tests performed also included galvanised fibres that
gave a delay in initiation of corrosion but no full protection. After some longer
exposure these fibres also corroded.
3.5.2 University of Michigan, USA, Kosa (1988)
Kosa (1988) tried to determine the rate of corrosion in SFRC and how corrosion
affects load carrying capacity and ductility. Only a small number of the specimen was
cracked prior to exposure. Apart from the cracked specimen pre-carbonated, high
permeability and pre-corroded fibres cast in concrete specimen were tested.
Three types of flexural load setups were tested:
1. Small specimen (seeTable 3.6) loaded to 3 mm deflection.
2. Larger specimen loaded to 3 mm deflection.
3. Larger specimen loaded over the ultimate load before exposure.
When testing properties in flexural load, the specimen size used in Kosas (1988)
investigation has to be questioned. The relatively thin beams used should give a
large scatter in the results. Average crack widths were 0.12; 0.20 and 0.27 mm for the
three different specimen types.
24
The investigation only gives limited amount of information about the rate of
corrosion but shows instead clearly that the decrease of fibre diameter is vital for the
change of load carrying capacity. The increased bond between a corroded fibre and
the concrete is of minor importance. A decreased fibre diameter especially affects the
residual strength but also the ultimate load carrying capacity. A rough estimation
shows that a decrease of average fibre diameter with 20% gives a 10% decrease of
the load carrying capacity and a 25% decrease of toughness (I5) (I5 is a measure of the
ductility or work the cracked sample perform up to a deflection of 3.5 times the
deflection when the first crack appears). The fracture type turning from being a bond
failure between fibre and concrete to a tensile failure of the fibre explains the
phenomenon.
When it concerns the influence of crack width on corrosion and the following change
of load carrying capacity, Kosa (1988) suggests a critical crack width of 0.15 mm.
Below 0.15 mm no change of load carrying capacity can be seen with the exposure
conditions (see Table 3.6) used in the investigation. Kosa (1988) also presents an
analytical model to determine the influence of fibre corrosion in cracked concrete on
residual strength.
3.5.3 University of Aberdeen, Great Britain, Mangat & Gurusamy (1985,
1987a,b)
The first investigation by Mangat & Gurusamy (1985) aimed at defining the chloride
diffusion in SFRC exposed to a marine environment. The specimens were exposed to
a simulated tidal marine situation by spraying with seawater in the laboratory.
Evaluations of the tests were performed by measuring the chloride penetration both
from the exposed surface as well as 10 mm perpendicular to the crack surface. No
significant difference between SFRC and plain concrete specimen could be found. It
could also be stated that the chloride diffusion coefficient decreases with time both
due to continued hydration and by deposit of a brucit-like material formed from ions
in the sea water. The chloride concentrations increased with increasing crack width.
At crack widths below 0.2 mm there was no difference from uncracked specimen. At
crack width above 0.5 mm the effect is significant. The authors also found that most
of the ingress of chlorides took place during the first 3 months.
In another article of Mangat (1987a) results of tests on residual strength are
presented. At crack widths below 0.2 mm there was an increase of residual strength
compared to uncracked samples. Probable explanations given by the author are
autogenous healing and/or increased anchorage of the fibres.
Mangat (1987a) also tested the effect of different steel qualities. Low carbon fibres
did not show any corrosion below crack widths of 0.24 mm. Melt extract
(ME)(stainless steel) fibres showed corrosion above crack widths of 0.94 mm.
Mangat (1987a) points out that the resistance for corrosion on stainless steel fibres is
25
due to how stable the passive oxide layer on the fibres are. The stability of the oxide
layer depends on e.g. the chromium content (above 12 % is good).
With the exposure method that Mangat (1987a) used autogenous healing seemed
possible for crack widths below 0.5 mm. For larger crack widths the material was
washed away from the crack walls. Taking both corrosion and effect on residual
strength into consideration Mangat (1987a) recommends a maximum crack width for
ME fibres of 0.2 mm in marine environment. Corresponding crack width for low
carbon fibres is 0.15 mm.
In a paper Mangat (1987b) presents results from pore squeezing aiming to determine
the content of free chlorides in the exposed samples. No difference in the amount of
bound chlorides between the concrete with or without fibres could be seen. It could
also be stated that the amount of free chlorides was higher in the samples exposed in
laboratory environment (actual sea water was used). Mangat (1987b) explains this
with increased concentration due to evaporation. Mangat (1987b) also noted that the
chloride concentration decrease close to the surface and explains this with either
carbonation or a reaction with the hydroxide ions.
3.5.4 University of Surrey, Great Britain, Hannant & Edgington (1975, 1976)
The aim of the investigation was to determine the effect of steel fibre corrosion in
cracks on the remaining load carrying capacity. Experiments were not accelerated.
Samples without cracks are also exposed but not dealt with here. Cracking of the
beams were performed 8 days after concreting. Some of the samples had their crack
sealed before exposure. The authors point out that those beams showing most
damage at pre-cracking was chosen to be sealed. This must be regarded when
evaluating the results. Evaluation of the load carrying capacity was made before
exposure and after different exposure lengths.
A relatively large scatter in ultimate load is explained with uneven fibre distribution.
The residual strength generally increased after exposure though corrosion was
initiated. Carbonation of the crack surfaces occurred and the depth of the area with
corroded fibres increased with increased time of exposure. The sealed specimen
showed unchanged or slightly increased load carrying capacity. Hannant et.al (1975)
state that residual strength is no efficient tool to determine the rate of corrosion. This
due to the increased load carrying capacity in spite of different degree of corrosion
attack.
In a second article Hannant (1976) presents results after another 8 months of
exposure. The trend is the same as earlier. In addition a model to estimate the bond
strength between the fibre and the concrete with the results from flexural loading as
base is presented. Hannant (1976) suggests that the increased bond possibly can be
derived from shrinkage around the fibres. The hypothesis must be questioned since
shrinkage should occur evenly in the cement paste and therefore should give lowered
26
bond strength instead. From the model it could be stated that the fibre diameter has to
be reduced with 77% before turning from bond failure to tensile fibre failure.
3.6
Discussion
This thesis has a focus on steel fibres in cracked sprayed concrete. This differs from
the presented previous works where most of the applications are of ordinary concrete
in marine environment.
In commonly reinforced concrete structures the cracked state with initiated corrosion
is considered as the limit for service-life. For steel fibres this is especially
unfavourable since steel fibre reinforced sprayed concrete structures use the cracked
state in serviceability limit state. Design of rock strengthening structures is one such
example of application. A further refinement is therefore needed in a service life
model for steel fibre reinforced sprayed concrete.
The usage of stainless steel fibres as a solution to ascertain a long time resistance
against corrosion is not motivated due to the extra high material costs. Especially as
long as the resistance against corrosion in cracks has not been proved to be a
durability problem.
One possible influence from the sprayed concrete technique on the corrosion process
could be the usage of accelerators. Alkali silicate based ones were most commonly
used before and addition of alkali to the concrete lowers the possibility for chloride
binding which is a negative effect. More commonly today is the usage of so called
alkali-free accelerators. Many of these accelerators have a very low pH (2.5-3.5)
which also could influence the resistance to corrosion in a negative manner. On the
other hand some of them have a calcium-aluminate-part which on the other hand will
increase the chloride binding. Common for both type of accelerators are that the
amounts added is in the range of 5-8% of the weight of cement which is very low.
Silica fume is also commonly used in sprayed concrete mix-design. As previously
mentioned the silica fume increases the chloride binding.
Splashing from de-icing salts gives a type of exposure which should be more severe
than the marine exposure. This is so because periods with no addition of chlorides
combined with precipitation will significantly decrease the possibilities for selfhealing in comparison to the marine applications.
27
CRACK
BEAM SIZE
EXPOSURE
EVALUATION
w/c
cement
3
IBAC,
0,5
Schiessl &
(0,4)
types
(kg/m )
(no.)
350
/L
(mm)
content
(mm)
(mm)
type
load
(kg/m )
60
0,2-0,4
150x100x700
acc1
(30-120)
Weydert
(1998)
acc
0,05-0,4
1%NaCl;
cycle
length
(days)
(months)
7;7;14
14-17
parameters
flexural load
water;
geometrical
air
chloride cont.
3%CO2;
28;90
18
carb. depth
electrode
air
potential
University of
0,42
0,5/30
156
0,5 / 1
75x37,5x300
acc
3,5%NaCl
28
Michigan,
&
Kosa
75x12,5x450
50C;air
3;3
2,6 & 9
flexural load
geometrical
(1988)
University of
Aberdeen,
Mangat &
Gurusamy
0,4
590
435
(+155
norm2
spray with
2x
187
sea water,
(0,25;0,25)
0,6/40
165
air
0,25/25
110
0,51/25
225
0,48/28
0,07-1,08
100x100x500
2; 30
flexural load
energy absorb.
ocular
PFA3)
(1985;1987a,
1987b)
University of
Surrey,
Hannant &
Edington
(1975,1976)
480
0,5/50
0,1-0,3
100x100x500
norm
air,
marine
145,145-
1, 11, 19
flexural load
geometrical
WIDTH
FIBRE
Table 3.6
CONCRETE
4.1
Background
Scope
The purpose with the field exposure tests is to study steel fibre corrosion in real
environments with controlled material properties during exposure with normal rate
(not accelerated). The goal with the field exposure tests are:
Define the time to initiation and the rate of propagation for corrosion in cracked
steel fibre reinforced sprayed conrete exposed in field.
Investigate the influence of some relevant material parameters on time to initiation
and/or rate of corrosion.
Collect reference data for definition of critical crack width for initiation.
Create a reference for accelerated exposures in laboratory.
4.3
Methodology
The field exposure tests were started in September 1997 and evaluations have been
made in the autumn of 1998 and the winter of 2000. The purpose with the evaluations
is to examine the status of fibres crossing cracks after different time of exposure.
A great number of parameters could be of interest for testing in the field exposure
tests. In Table 4.1 some parameters are listed. Due to the methodology with
destructive evaluation of the samples after exposure only a limited number of
parameters could be choosen (see Figure 4.1). Since the goal is to register corrosion
data after different time of exposure the number of samples increase greatly for each
parameter added to the matrix. After validation of the laboratory exposure tests more
parameters could be included there for faster evaluation than in the field exposure
tests.
29
Table 4.1
METHOD
Wet-mix
Crack width
Dry-mix
Crack depth
CONCRETE
EXTERNAL
Mix
Fibers
Accelerator
Additions
w/c-ratio
Length
water glass
Silica fume
Humidity
Cement
Diameter
alkali free
Limestone
Temp.
(amount,
filler
type)
Size
Dosage
Dosage
Dosage
Chlorides
Steel grade
Load (static)
Production
Air pollution
type (cold
(NOx, SOx)
drawn, cut,
chopped)
Coating
Four different concrete mix-types are used. The wet-mix sprayed concrete with
30 mm fibres and usage of accelerator (WA30) is the main mix used in all
combinations of exposure type and crack widths. The mixes and the abbrevations are
presented in Table 4.2.
30
Table 4.2
Dry-mix
Accelerator
Dramix 30/0.5
Dramix 40/0.5
WA30
W30
WA40
D30
WA40
W30
D30
0.42
0.42
0.42
0.3
510
510
510
500
1202
1202
1202
815
286
260
298
298
298
138
1.4
1.4
1.4
Accelerator
(%/kg C)
3.5
3.5
Fibre
(kg/m )
70
70
70
65
w/c
Cement
Aggregate 0-8 mm
(kg/m )
Aggregate 4-8 mm
(kg/m )
Aggregate 2-5 mm
(kg/m )
Aggregate 0-1 mm
1)
(kg/m )
(kg/m )
Approximated by measuring the amount of water required during spraying (see appendix A)
The longer fibres (Dramix 40/0.5) are tested to study the influence of the
anode/cathode area on the rate of corrosion. Research results by e.g. Raupach (1992)
and Okada & al (1980) show that this is an important parameter for conventional
reinforcement. This is valid under conditions where the electrical resistance of the
concrete matrix surrounding a reinforcement bar are similar.
A sodium silicate based accelerator was used in production of mixes WA30 &
WA40. For the time of producing the samples and also previously this was the most
common type in Sweden. The effect of it and the influence on other concrete
properties is well documented. The target amount added to the mixes was 3-5 % by
weight of cement. According to Burge (1984) this reduces the compressive strength
with 12-20 %. Too high addition would give a coarser matrix and a further decrease
in compressive strength. The increased coarseness of the concrete matrix should give
an increased permeability and larger ingress of eg. chlorides. An increase in rate of
corrosion rate is therefore suspected. This could possibly be counteracted by the high
pH (10.5-11.5) since alkali silicates is commonly used for corrosion protection in
water plants. The influence on the total concrete pore water pH is not expected to be
high and therefore this effect should not be dominating.
A sketch over the test program is shown in Figure 4.1.
31
Method
Concrete
Fibre
length
Crack
width
Exposure
0.5 mm
Lab
0.0 mm
Time
Analyse
method
1 year
Rv 40
0.1 mm
30 mm
2.5 year
residual
strength
geometrical
River Dal
0.5 mm
Eugenia
tunnel
Normal
1.0 mm
?
Rv 40
Wet-mix
40 mm
0.5 mm
River Dal
Normal - acc.
Dry-mix
Figure 4.1
Normal
0.1 mm
Rv 40
0.5 mm
River Dal
0.1 mm
Rv 40
0.5 mm
River Dal
30 mm
30 mm
The concrete was firstly sprayed as large slabs (2*1.2*0.15 m). The purpose of
spraying the large slabs was to reduce the amount of rebound and recieve a more
homogenous composition. At the end 11 large slabs were sprayed. From the slabs
beams were sawn with the dimension 75*125*500 mm. All concrete was sprayed
using a rotor machine of type ALIVA 262 for both wet- and dry-mix spraying (see
Figure 4.2). When adding accelerators a pump of type ALIVA 403 was used. In
Appendix A all data from mixing and spraying are shown. Some problems with
addition of accelerator occured during spraying due to stoppage in the nozzle and
differences in coefficient of fullness in the rotor (Average 56 %, standard deviation
11 %). The average addition of accelerator was 4.4 % by weight of cement with
1.7 % standard deviation.
After storing under humid conditions for approximately 56 days the beams were
cracked to the desired crack width. Cracking was obtained by performing a four-point
load set up to create a flexural crack. In general the flexural test is in accordance with
the ASTM C1018 test (other beam dimensions and rate of deflection 0.25 mm/min).
To reach the desired crack width after unloading the elastic deformations had to be
considered by exceeding the crack width before unloading. To obtain a crack width
(w) of 0.1 mm after unloading the beam was subjected to a deflection () of 0.4 mm.
32
Corresponding values for 0.5 and 1.0 mm crack width are 1.0 and 1.7 mm. For every
combination of parameters and occasion for evaluation there are three samples
available. This gives a total of 348 beams exposed in the tests. To minimize the risk
for systematic errors the choice of beams from every type of mixture, different crack
width and exposures site was made by random selection.
F
75
w
150
Figure 4.2
150
150
Spraying equipment ALIVA 262 (left). Setup for flexural load test (right).
Two major climatic factors ruling the rate of corrosion are relative humidity and
presence of chlorides. In giving the field exposure tests relevance to actual situations
where steel fibre reinforced sprayed concrete is commonly used the choice of
exposure environment had to be made carefully. In the field exposure tests three
different sites were choosen together with the control site in the laboratory (20C och
65% RH).
RV40. National Road 40 close to Bors, outdoor along motor highway.
DAL. River Dal at lvkarleby, outdoor with specimen partly immersed.
EUG. Eugenia tunnel, road tunnel in Stockholm.
Details and examples are shown in Table 4.4 and Figure 4.3.
33
Table 4.4.
Exposure environments.
Location
Type of exposure
Typical structure
Eugenia tunnel,
Humid
Stockholm
Chlorides
Sheltered from rain
Acidifying gases
Humid
Bors
Concrete repairs
Rain
Figure 4.3
River Dal,
Humid
Intake channel
lvkarleby
Rain
Intake tunnel
Frame for field exposure tests at Rv40 (left) and map over exposure
sites (right).
To make it possible to use the results from the field exposure tests in a service life
model and for correlation to accelerated exposures a follow up of the actual exposure
environment is needed. At the Rv40-site a number of other tests in other Swedish
research projects are also running. In Table 4.5 the instrumentation and type of
climatic follow up for all the sites are presented.
34
Table 4.5
air temperature
water temperature
relative humidity
Originally there were supposed to be four occasions for evaluation of the field
exposure tests. Three beams of every combination of parameters were supposed to be
examinated. During the first year of exposure the technique to evaluate changed and
only two beams of every combination were used. The major purpose is to extend the
time for exposure and give further opportunities for evaluation. In the following the
techniques for evaluation are described. One of the beams is used to measure the
residual strength after exposure, the depth of area with corroded fibres and the
chloride contents. The other beam is used for evaluation of the corrosion on single
fibres.
G
Residual strength
Steel fibres are used in the design process for e.g. permanent linings in underground
construction to achieve sufficient post-crack behaviour of the sprayed concrete. The
function of e.g. a system with steel fibre reinforced sprayed concrete combined with
bolts is fully dependent on obtained residual strength. Therefore the influence of
corrosion on the long-term residual strength is of interest and also studied here.
Ductility in concrete with steel fibres is obtained by the pull-out resistance occuring
between fibre and concrete matrix. Active forces are bond strength and for most
fibres also friction due to hooked ends or corrugated shapes of the fibre. Steel fibres
most commonly have high tensile strength (typically >1000 MPa) to utilize the
ductility from interaction between steel and concrete. If corrosion has been initiated
the fibre cross section decreases locally in the crack region. This should give a
change from a ductile pull-out from the matrix to a brittle fibre failure.
By comparing the residual strength levels at initital flexural cracking and after
different time of exposure this effect can be investigated. By re-loading the beams
with flexural load up to 5 mm deflection after exposure it was possible to check the
change in residual strength. The residual strength levels in the re-load are compared
with the levels when the initial flexural load was interrupted.
35
Removal of fibres
To check if corrosion has been initiated (or possibly the degree of corrosion on fibres
crossing cracks) the fibres have to be removed from the concrete matrix. Crushing of
the concrete plates would give a risk for steel fibre failure and ruined possibilities for
evaluation. This is valid especially for fibres showing great extent of corrosion.
Therefore the small plates (see Figure 4.4) where subjected to repeated freezing and
thawing. The plates where sawn on different levels from the crack mouth from the
beams not subjected to continued flexural load. To secure a failure during freezing
and thawing the plates where first dried out completely for 24 h in 200 C. After
drying they were put in a vessel for vacuum treatment. For 3 days they were exposed
to 98 % vacuum. Before normal pressure was restored the vessel was filled with tap
water. When normal pressure was obtained in the vessel the water will be sucked into
the plates due to the gradient in air pressure in the plates. The gradient will lead to
that the plates recieve a very high degree of saturation which gives a total degradation
when exposed to freezing (Fagerlund, 1994) (see Figure 4.5).
20
75
5
12
80
500
B Scrit
Sact
+
0
-
Function
Temp.
C
Degree of
saturation, S
Figure 4.4
Figure 4.5
Time
36
After saturation the specimen were exposed to a freeze thaw cycle from +20 C to 30
C to +20 C in 24 hours. The exposure continued for approximately 3 weeks. From the
completely degraded concrete plates the fibres are collected by using a magneto.
Only whole fibers where picked out i.e. only fibres crossing cracks is included in the
evaluation. Unfortunately this also make fibers crossing two plates on different levels
from the crack mouth useless.
One might suspect that there will be a risk for corrosion during the drying or freezing
process. High temperature (200C) makes the plate dry very fast and the time at a
condition when the humidity is sufficient to give rise to corrosion is very short. Since
the plates completely degrades during freezing the water surrounding the plate will
recieve a very high pH (average pH=12.2 from 6 measurements). The plates are also
subjected to sub-zero temperatures more than half the time during freezing, which
will make a potential corrosion process very slow. Due to this there will be no risk for
continued corrosion of the fibres during theese two processes.
G
Corrosion attack
Only fibres crossing the crack are of interest and they are therefore selected by ocular
examination. The fibres selected show corrosion in any part except from the ends. All
other fibres are discarded. To make it possible to measure the loss of fibre diameter,
the fibres are pickled to remove all corrosion products by using Clarks-solution.
The solution consists of antimony trioxide (Sb2O3) (20 g/l) and tin chloride (SnCl2)
(50 g/ml) dissolved in concentrated hydrochloric acid. The solution will make only
the corroded part of the steel fibres to be washed away. After treatment in
Clarks-solution the fibre diameter was measured with a micrometer in the part
crossing the crack. The diameter in the corroded area was compared with the
unaffected part beside the crack. The loss is presented as percent of the original
diameter. Inititally weighing of the fibres to estimate the fibre loss was an alternative.
Measuring the weight of 60 fibres shows however that it is not a realistic alternative
since there is over 1 % variance in the original weight. Loosing e.g. 10 % of the
diameter on 3 mm length would not be detectable (gives a loss of 0.07 % of total
weight). The actual original weight is also not controllable afterwards for a single
fiber.
Measuring the fibre diameter of 50 fibres gives a variance of 0.8 % in original
diameter which is acceptable i.e. a 10 % loss of diameter is detectable with available
equipment.
37
18
16
Amount of fibres
14
12
Average
= 0.507 mm
Std. deviation = 0.004 mm
10
8
6
4
2
0
0.5 0.502 0.505 0.507 0.51 0.512 0.515 0.517 0.52 0.522 0.525
Fibre diameter (mm)
Figure 4.6
G
To find out the amount of chlorides penetrating the concrete and crack, measurements
were made as a chloride profile from the exposed surface. The chloride content was
also measured at different distances from the crack mouth along the crack surface
(Figure 4.7). Drilling was performed with a 5 mm drill to a depth of 2-4 mm
perpendicular to the crack surface. This will make the measurements in the crack
being an average for 2-4 mm from the crack surface. Drilling for the chloride profile
from the exposed surface was made to a depth of 5-10 mm perpendicular to a new
crack. During drilling the debris was collected and solved in hydrochloric acid before
measuring with the RCT-method by Germann and Hansen (1991). The RCT-method
is based on measuring the chloride content in a solution with a chloride selective
membrane electrode.
Carbonation depth was measured by splitting the concrete and detecting the
carbonation depth with a phenophtalein-solution (areas not coloured red are
carbonated).
Existing crackplane
30
15
45
New crackplane
Exposed surface
Figure 4.7
38
4.4
Results
A continous measurement according to 4.3.4 was made and an example of the results
is presented in Figure 4.8 below. A full set of data can be found in Appendix D. The
slightly higher temperatures in the Eugenia tunnel could be explained by heat from
the intence motor traffic. The higher humidity at the River Dal depends on the
nearness to the water. Slightly lower values on RH in the Eugenia tunnel is depending
on the samples being sheltered from rain, but fluctuations still occur since rain affects
the RH in the tunnel by the motor traffic (pulling humid air into the tunnel and road
surface being wet).
25
100
90
20
70
10
RH (%)
Temperature (C)
80
15
5
0
-5
-10
1997-08-23
Figure 4.8
1998-03-11
Date (YYYY:MM:DD)
60
50
40
RV40
30
DAL
20
EUG
10
1998-09-27
0
1997-08-23
RV40
EUG
DAL
1998-03-11
1998-09-27
Date (YYYY:MM:DD)
A typical test result for a thin crack (w= 0.1 mm) is shown in Figure 4.9 where it can
be stated that the residual strength has increased with 15 % for the sample exposed 1
year and 29 % for the sample exposed 2.5 years. Complete results can be found in
Appendix B and C.
The change of residual strength after 1 & 2.5 years of exposure is shown in Figure
4.10. A slightly higher increase of the residual strength can be seen for crack width
0.1 mm. For crack widths 0.5 and 1.0 the increase seem to have stopped or started to
decline after 2.5 years. The influence of type of exposure can also be studied in
Figure 4.10 where it can be seen that the increase of residual strength is lower or even
declines (w= 0.1 mm & w= 1.0 mm) for samples exposed at the River Dal site.
39
12000
Year 1
10000
Year 2.5
8000
Load (N)
+29 %
6000
+15 %
4000
2000
0
0
Figure 4.9
0,5
1,5
2
Deflection (mm)
2,5
w= 1.0 mm
60
50
w= 0.1 mm
40
w= 0.5 mm
30
w= 1.0 mm
RV40
50
Change of residual strength (%)
60
20
10
0
-10
EUG
40
DAL
30
20
10
0
-10
-20
-20
0.5
1.5
2.5
0.5
w= 0.5 mm
1
1.5
2
Time of exposure (years)
2.5
2.5
w= 0.1 mm
60
60
50
RV40
EUG
40
DAL
LAB
30
20
10
0
-10
50
RV40
40
EUG
DAL
30
20
10
0
-10
-20
-20
0.5
1
1.5
2
Time of exposure (years)
2.5
0.5
1
1.5
2
Time of exposure (years)
Figure 4.10 Development of residual strength for samples from mix WA30.
Influence of crack width, all sites (Top left). Influence of exposure site,
w= 1.0 mm (Top right), w= 0.5 mm (Low left), w= 0.1 mm (Low right).
40
In Figure 4.11 the influence of mix type on the residual strength can be studied. The
wet-mix samples with 40 mm fibres and accelerator (WA40) did not recieve any
increase of the residual strength and the wet-mix samples with 30 mm fibres and
accelerator (WA30) seem to increase the residual strength more than other mix types.
No significant influence of spraying method can be seen from the results of residual
strength.
Rv40, w= 0.1 mm
60
60
50
50
WA30
40
W30
30
D30
20
10
0
-10
WA30
W30
40
D30
30
20
10
0
-10
-20
-20
0
0.5
1
1.5
2
Time of exposure (years)
2.5
0.5
1
1.5
2
Time of exposure (years)
2.5
2.5
Rv40, w= 0.5 mm
60
60
50
50
40
WA30
W30
30
D30
WA40
WA30
20
10
0
-10
W30
D30
40
WA40
30
20
10
0
-10
-20
-20
0
0.5
1
1.5
2
Time of exposure (years)
2.5
0.5
1
1.5
2
Time of exposure (years)
Figure 4.11 Development of residual strength for all mix types. River Dal, w=0.1 mm
(Top left). Rv40, w=0.1 mm (Top right), River Dal, w=0.5 mm (Low left),
Rv40, w=0.5 mm (Low right).
Generally the chloride contents were low after year 1 but with a few values at higher
levels. After 2.5 years of exposure the chloride contents were higher than at year 1.
41
From Figure 4.12 and Figure 4.13 it can be seen that the trend for chloride content
after 2.5 years shows a higher concentration closer to the crack mouth.
The chloride concentration is slightly lower for crack width 0.1 mm but no significant
difference between mix types can be seen. No difference in behaviour can be seen
between Eugenia and Rv 40.
0.35
Chloride content (%Cl / kg cement)
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0.3
0.25
0.2
0.15
0.1
0.05
0.000
5
15
25
35
45
15
25
35
45
w= 0.1 mm
w= 0.5 mm
w= 1.0 mm
Data year 1
Trend year 1
w= 0.1 mm
w= 0.5 mm
w= 1.0 mm
Data year 1
Trend year 1
Figure 4.12 Infuence of crack width on chloride content at crack surface. Rv40 (left)
and Eugenia tunnel (Right).
Chloride content (%Cl / kg cement)
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0.000
5
15
25
35
45
WA30
W30
D30
Data year 1
Trend year 1
Figure 4.13 Influence of mix type on chloride content at crack surface for samples
exposed at Rv40. Crack width = 0.5 mm.
42
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
0
10
20
30
40
Depth from exposed surface (mm)
WA40
W30
Data year 1
Trend year 2.5
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
50
10
20
30
40
Depth from exposed surface (mm)
WA30
D30
Trend year 1
50
WA30
Data year 1
Trend year 1
Figure 4.14 Chloride profile at a distance from the exposed surface for samples
after 1 and 2.5 years of exposure. Samples from Rv40 (left). Samples
from Eugenia tunnel (right).
4.4.4 Carbonation
It could be stated that all concrete types were carbonated only a few millimetres. The
outdoor conditions and the short time of exposure is the reason to this. High quality
of the sprayed concrete giving a low permeability to carbondioxied is another
important factor.
4.4.5 Corrosion on fibres
From examination of the crack surface at the beams subjected to continued flexural
load additional information about the amount of corrosion could be found. In Table
4.6 the depth from the crack mouth of the area where corroded fibres could be found
is shown.
The depth of corrosion on single fibres was measured by using a micrometer. In
Figure 4.15 and Figure 4.16 some of the results can be seen.
Table 4.6
Rv40
-98
15
30
5
Eugenia
-00
15
40
60
-98
0
0
0
-00
3
3
0
43
River Dal
-98
0
0
0
-00
2
2
8
Lab
-98
0
-
-00
0
-
w = 0.5 mm
30
30
10
5
WA30 / 1998
W30 / 2000
w= 0.5 mm
w= 0.1 mm
30
25
25
15
10
5
20
15
10
5
W30, 1998
02525
50 5075
Depth from
crack
mouth
(mm)
D30, 2000
50-75
WA30 / 1998
25-50
D30 / 2000
D30 / 1998
0-25
D30, 1998
WA30, 1998
WA30, 2000
20
W30, 2000
Decrease of fibre
diameter (%)
30
WA30 / 2000
Decrease of fibre
diameter (%)
15
W30 / 1998
5075
w=1.0 / 1998
2550
w=0.1 / 1998
0-25
Depth from
crack m outh
(m m )
w=0.5 / 1998
w=0.5 / 2000
w=1.0 / 2000
10
20
WA40 / 2000
15
25
WA40 / 1998
20
WA30 / 2000
Decrease of fibre
diameter (%)
25
w=0.1 / 2000
W A30
Figure 4.15 Decrease of fibre diameter for samples exposed at Rv40. (Top left) Mix
type WA30, different crack widths. (Top right) Crack width w= 0.5 & mix
type WA40, WA30 & W30. (Low left) Crack width w= 0.5 mm and mix
type D30 & WA30. (Low right) Crack width w= 0.1 mm and mix type
WA30, W30 & D30.
DAL
20
15
10
5
Depth from
crack mouth
w= 0.5 mm /
1998
w= 0.1 mm /
1998
30
25
20
15
10
5
0
w= 1.0 mm/
1998
Decrease of fibre
diameter (%)
25
w= 1.0 mm /
1998
Decrease of fibre
diameter (%)
30
w= 0.1 mm/
1998
EUG
Figure 4.16 Decrease of fibre diameter for mix type WA30 with different crack
widths. Eugenia tunnel (left) and The River Dal (right).
44
4.5
Discussion
The expectations after the first year of exposure was solely to verify that no specimen
started to corrode and to test the methods for evaluation. The number of samples
brought back to the lab was also small to make additional future occasions for
evaluation possible. To make accurate statistical treatment of the results the amount
of samples was too small. Another factor not taken into consideration in the
evaluations is that the samples are places on slightly different levels (+/- 250 mm)
above ground (except the River Dal). This will influence the micro climate by
slightly different humidity conditions (lower level samples= higher humidity and
higher degree of exposure to de-icing salts).
4.5.1 Residual strength
The initial increase of the residual strength after one year of exposure is probably
caused by continued hydration of the concrete. Continued hydration results in
increased anchorage strength between fibre and concrete which gives an increased
residual strength. The increase seems to be higher for smaller crack widths and the
effect of self-healing could be an additional reason. Some of the thinner cracks were
completely filled with deposits from self-healing. For samples with ongoing
corrosion of fibres the increase of residual strength must not be expected to continue.
The fibres should instead turn to show a brittle tensile failure instead of a ductile
pullout failure at some critical degree of corrosion. The trend after 2.5 years is that
the increase of residual strength already has slowed down or started to decrease for
all samples except for 0.1 mm crack width. The influence of self-healing seems
therefore to be important. It also seems like that the continued increase for 0.1 mm is
smaller at the River Dal. This could be an effect of conditions with more constant
humidity and without splashing or a lower ability to self-heal due to other water
charateristics. Continous flow of water could also flush the cracks. Ice formation in
the cracks or pressure from ice between the beams could make the crack width vary
over the year and therefore diminish the possibilities for self-healing.
The influence of larger loss on fibre diameter for the 40 mm fibres is also visible on
the residual stregth. The loss in residual strength is larger for theese samples.
4.5.2 Chloride content
The evaluation after year 1 was performed in the autumn. This is the reason for
measuring lower chloride contents closer to the crack mouth after year 1. It is a result
of the samples being exposed to rain and splashing from the road during the summer
period. Chlorides close to the crack mouth are then washed away. This is validated by
the evaluation during winter conditions after 2.5 years of exposure where the chloride
concentration instead is higher closer to the crack mouth. Samples exposed in the
Eugenia tunnel are sheltered from rain but also show this behaviour. One possible
explanation is that the tunnel surface is washed with water applied at high pressure
45
once a year (late summer). It could also be an effect of splashing of rain water from
the road during the summer. High pressure water cleaning is more likely if Figure
4.14 is studied since very high chloride concentrations (higher than at Rv40) can be
seen. The reason should therefore be that no wash-out from rain occurs during the
winter-time in the tunnel.
4.5.3 Extent of corrosion
Surprisingly corrosion had initiated after only one year of exposure at Rv40. No or
very small amounts of corrosion could be seen in samples exposed in the Eugenia
tunnel or at the river Dal. After 2.5 years the corrosion continues to propagate at
Rv40 and is starting in samples with larger crack width in the Eugenia tunnel. The
influence of crack width (0.5 mm and 1.0 mm) seems to be less important after 2.5
years provided corrosion is ongoing. This supports the theory that crack width mainly
affect the time to initiation of corrosion and not the rate of corrosion in the same
manner. Samples with crack width 0.1 mm still show a very low amount of corrosion
due to self-healing.
It can be seen that the corrosive attack on fibres decreases with increased depth down
in the crack. The reason should be decreased crack width and therefore lower access
for oxygen due to higher RH longer periods than in the parts close to the crack
mouth.
From the limited amount of samples it seems like the 40 mm fibres show a more
severe damage from corrosion than the 30 mm fibres. The loss of fibre diameter on
the 40 mm fibres is twice the loss on the 30 mm fibres. This supports the hypothesis
that the effect of the anode to cathode area to be an important factor. No significant
influence of mix-type can be seen.
Uncertanties in the evaluation of loss of fibre diameter could be the identification of
fibres crossing the crack. The choosen methodology makes it impossible to detect
non-corroded fibres crossing the crack region. Therefore the number of fibres behind
the figures on loss of fibre diameter sometimes is very small (2-10 fibres). An
improvement of the methodology could be to grout the cracks with epoxy before
sawing the plates from the beams.
In general the rate of corrosion is much higher than expected in the Rv40
environment. If the rate of corrosion in this environment is constant there will be a
great loss of residual strength in a couple of years for cracks larger than 0.1 mm. No
larger cracks can therefore be accepted in a structure placed in this type of
environment if fibres are used as structural reinforcement.
46
4.6
Conclusions
After 2.5 years of field exposure tests the following preliminary conclusions can be
drawn:
- Possible continued hydration and therefore increased anchorage strength for the
steel fibres gives an initial increase in residual strength after started exposure. The
increase is temporary for crack widths above 0.1 mm since loss of fibre diameter
by corrosion will lower the ductility and reduce this effect. Self-healing of thin
cracks (w=0.1 mm) makes the increase continue or stabilize at the same level.
- The chloride content along the crack surface depends on seasonal variations. High
levels can be measured during the winter (deicing period) and lower after a
summer. The effect of seasonal variations depends on the degree of exposure to
rain. The effect is also most obvious in parts close to the crack mouth.
- Generally the degree of corrosion is limited. A high degree of exposure for
deicing-salts during the winter time gives a larger corrosive attack than the other
exposure environments.The importance of crack width seems to decrease with
time after initiation. Longer fibres corrode more than shorter ones at the same
crack width. Fibre length is therefore more important than crack width.
- The loss of fibre diameter has decreased the residual strength. After 2.5 years of
exposure the only mix-type showing a net loss (from start of exposures) in
residual strength is the wet-mix samples with 40 mm fibres. The other types still
have an increased, or returned to the original, residual strength.
47
5.1
Scope
The main objective with the laboratory exposure tests was to make a relative
comparison between different parameters tested under controlled climatic conditions.
In general there are two major techniques to predict long-time behaviour during
degradation. Either by more accurate measurements of the degrading process or by
increasing the severeness of the exposure. In the following exposure tests the last
mentioned technique has been used. When designing accelerated exposure tests great
care has to be taken to minimize the risk for changing the process to be studied.
The previously described field exposure tests can be used to estimate the degree of
acceleration in the laboratory tests. When establishing a correlation between field and
laboratory conditions other, and more, parameters can easily be tested in laboratory
exposures. The accelerated laboratory exposure tests can then also be used to estimate
long-term behaviour. In this first set of parameters tested, the purpose is solely to
control if the technique choosen in the tests is possible to use for adequate
acceleration of corrosion.
5.2
Methodology
In the following section the methodology for how samples where poured, cured,
cracked, exposed and evaluated is presented.
5.2.1 Concrete composition
This first set of parameters to be tested are choosen mainly to coincide with the tested
parameters in the field exposure tests. Three fibre lengths (35, 70 and 105 mm) and
three crack widths (0.1 , 0.5 , 1.0 mm) are tested. The combination of the different
parameters can be seen in Table 5.1.
Table 5.1
(mm)
Crack width
70
105
0.1
0.5
1.0
In the field exposure test the maximum aggregate size used was 8 mm. To make it
possible to pour with a small distance between the fibres in the laboratory exposures
48
the maximum aggregate size used was 4 mm. The water-cement-aggregate ratio is
still the same. Usage of 4 mm instead of 8 mm leads to a relatively larger number of
smaller aggregate than in the field exposures. On the other hand the relation between
aggregate and cement paste area exposed on the crack surface is the same in both
cases. This should give approximately the same permeability for e.g. chlorides. In
Table 5.2 the mix-composition can be seen. To make pouring possible with remained
location of fibres the lower part was poured with plastic consistency and mechanical
compaction. The upper part was poured with a more fluid consistency and only a
slight compaction by tapping the mould (see also Figure 5.1). The cement used is the
same type as in the field exposure tests (Swedish standard portland cement,
Degerhamn).
Table 5.2
0.45
3
510
1202
292
Cement
(kg/m )
0-4
(kg/m )
0-1
(kg/m )
2.5
In the first stage three samples of every combination of fibre-length, crack width,
exposure type and exposure time were poured. The samples are cylindrical with a
diameter of 57 mm and a length of 170 mm with a nocth in the middle. 37 fibres were
placed in every sample with a 8 mm gap between every fibre. A total of 113 samples
(3 samples of every combination of parameters, exposure envirionment and occasion
for evaluation) where poured. The fibres where made from cutting cold-drawn wire
into desired lengths. Cutting was made by using an equipment for direct cutting of
fibres into concrete mixers. The wire used is the type beeing used in manufacturing of
the Dramix-fibre from Bekaert.
G
Pouring
By pouring the lower part of the samples first and then placing the fibres manually
full control over the location was obtained. Especially for the 35 mm fibres it was
crusial to place the fibres on the notch level (see Figure 5.1) to secure them to cross
the crack. A notch was created by gluing a rubber ring on the inside of the mould.
Demoulding after one day was followed by three days of water curing.
49
Step I
Step II
Step III
Figure 5.1 Moulds and sequence for pouring (left). Placing of fibres (right).
G
Cracking
Cracking was performed 5 days after pouring and to create a crack of desired width
the samples where subjected to uniaxial stress and displacement controlled loading.
In the beginning the samples where cracked using closed-loop equipment with CODtransducers (Crack-Opening-Displacement). This method was too time consuming to
use for the large number of specimen to be cracked. Therefore a simpler method with
controlled stroke displacement speed was used instead. The load was applied with a
rate of deformation of 0.02 mm/s until the desired crack width was obtained. To
make the cracking of samples rational a device with friction grips was also
constructed. The friction grips and the equipment used for cracking is shown in
Figure 5.2.
Figure 5.2 Device with friction grip for uniaxial testing (left). Equipment for
displacement controlled cracking of samples (right).
Since the friction grips have quite small dimensions some rotation occured during
cracking. The crack width was therefore adjusted by hand to the desired value. Since
no full fracture mechanical evaluation was needed this methodology is considered to
be acceptable. The major purpose is solely to create a crack of desired width. A
further development of the equipment could be made to increase the stiffness and
minimize the effect of rotation (see Figure 5.3). From Figure 5.3 it can be seen that
50
the concrete sample is clamped inside a notched steel tube which is welded to a steel
plate. Another effect to take into consideration is the effect of applying a radial
pressure on the sample and the contribution this has on the results. If it works the
time consuming technique to glue concrete on steel plates could be eliminated.
Notched
concrete sample
Clamping device
Notched steel
tube
Steel plate
The accelerated exposure test where started approximately 30 days after pouring of
the samples. Two types of exposure environments are used in the tests and the cycle
is similar for both types.
CHLORIDE - Submerged in water with 3.5% chlorides (3 days) + Air with RH 50%
(4 days).
TAP WATER - Submerged in water (3 days) + Air with RH 50% (4 days).
Control specimen stored continously in air with RH 50% where also made. Ambient
temperature was 20C (+/- 2C) for all environments.
The choice of climate and cycle in the two chambers are made to accelerate the
corrosion process sufficiently. The CHLORIDE environment can be related to a real
situation with splashing of water from de-icing salts and the structure being sheltered
from rain. When it concerns chloride concentration the target is to create as high
concentration as possible inside the samples. On the other hand a too high
concentration in the water will give rise to a risk for formation of salt crystals inside
the cracks which would make the process deviate too much from real conditions. The
TAP WATER environment simulates a situation where the structure is not sheltered
from rain or in connection to water with large fluctuations. A limitation in the tests is
that the automatized control of climate is depending on a 7 day cycle timer
equipment.
In choosing the RH of the air faned down into the storage tank there where two
obvious options depending on the access to climate chambers with 50 or 65% RH.
51
The lower level was selected since the period of drying the samples from completely
soaked to a state where corrosion propagates at maxium speed (RH 80-95%) was
only a couple of days (7-days cycle). There are large uncertanties regarding the
humidity conditions inside the crack. Capillary action, evaporation and supply of
water from the surrounding concrete makes an estimation difficult.
A continous follow up of the climatic conditions inside the storage tank and in the
climate chamber was made with loggers that registered temperature and RH. In
addition the pH and the redox-potential in the water were measured approximately
every two weeks. Addition of tap water was made monthly in both systems (chloride
& tap water) since the continous evaporation by fanning otherwise would increase the
chloride content in the remaining water. The only loss of chlorides is by removal of
samples containing chlorides, but this is considered being insignificant on the total
concentration.
Water
tank
Timer
Magnet
valve
Storage
tank
Fan
In Figure 5.4 the equipment for acceleration of corrosion is shown. The principle is
that the samples are stored in the storage tank all the time. The temperature is
constantly kept at +20C. The sequence to create the climate described above is:
1.
2.
3.
4.
52
The evaluation was made in the same manner as in the field exposure tests. Drying,
vacuum treatment, water saturation and freezing made the fibres easy to free from the
concrete matrix. For details see description in part 4.3.5. Samples where taken out at
70 and 120 days of exposure. Examination of the samples were also made as in the
field exposures with pickling of iron oxide and measurement of fibre diameter in the
corroded area with a micrometer. Samples for further evaluations after longer time of
exposure are available.
5.3
Results
In general, corrosion have been initiated on all samples already after 70 days. In the
following a compilation of the results is described. In Figure 5.6 and Figure 5.7 the
loss of fibre diameter in the crack region is showed. From the figures it can be seen
that samples stored in the CHLORIDE environment corrode more than samples
stored in TAP WATER environment. Another difference is that some of the fibres
(mainly 105 mm:s) stored in the CHLORIDE environment show an oxide layer on
other parts of the fibre length than exposed in the crack region (see Figure 5.5). This
implies formation of micro-cell corrosion on other places along the fibre. On the
other hand the oxide layer on parts not exposed in the crack was very thin and no
severe attack on the fibre could be seen. This effect could be possibly be explained
with the pouring technique. Inhomogenous enclosure by the concrete surrounding the
fibres could make chlorides penetrate the interface between fibre and concrete. It
could also possibly be explained by the small size of the samples which make fibres
with little cover corrode in a similar way as conventional reinforcement. As
preivously mentioned this is also only accentuated in samples with long fibres.
As can be seen from Figure 5.6 the influence of both crack width and fibre length is
obvious. The difference between crack width 0.5 and 1.0 mm is however very small.
The influence of crack width and fibre length is not that accentuated for TAP
WATER samples. An important aspect is to take into consideration the deviation of
results. In Table 5.3 calculated standard deviations are presented. To keep in mind
studying the standard deviations is that the distanced to the crack mouth has not been
considered.
53
100
100
80
80
60
Figure 5.5 Corroded area in crack from accelerated exposure tests in CHLORIDE
(left) and TAP WATER (right) environment. (L= 70 mm, w= 1.0 mm,
70 days of exposure).
0.2 mm
0.5 mm
1.0 mm
40
20
35 mm
70 mm
60
105 mm
40
20
70
Time of exposure (days)
120
70
Time of exposure (days)
120
Figure 5.6 Average loss of fibre diameter in CHLORIDE environment from one
sample (37 fibres). Influence of crack width, L=70 mm (left) and fibre
length, w=0.5 mm (right).
Table 5.3
(mm)
Crack width
35:120
70:70
70:120
105:70
105:120
0.2
9,1
17,2
0.5
6,1
10,6
11,9
14,8
13,2
1.0
10,3
15,7
54
100
80
0.2 mm
Loss of diameter (%)
100
0.5 mm
60
1.0 mm
40
80
60
40
20
20
70
Time of exposure (days)
35 mm
70 mm
105 mm
120
70
Time of exposure (days)
120
Figure 5.7 Average loss of fibre diameter in TAP WATER environment from one
sample (37 fibres). Influence of crack width, L=70 mm (left) and fibre
length, w=0.5 mm (right).
Table 5.4
(mm)
Crack width
35:120
70:70
70:120
105:70
105:120
0.2
6,4
14,3
0.5
12,4
22,9
11,7
13,0
24,3
1.0
12,4
16,9
An example of the measured values on RH can be seen in Figure 5.8. It can be seen
that the conditions in the room seems to be varying a lot. In general the RH-level
during the dry period of the cycle is roughly 65-70% inside the storage tank. A
decrease can be seen as the 50% RH air from the climate chamber successively
lowers the RH in the storage tank. The air surrounding the storage tank seems to be
affected by the humid air being pressed out from the tank.
To control the water quality and possible changes during the exposures measurements
of the pH and redox-potential was made once every two weeks. In Figure 5.9 results
are presented. In the CHLORIDE environment there is a slow acidification where pH
has decrased from 9.7 to 9 during the exposure tests. This can also be seen on the
redox-potential which has been increasing.
55
100
30
80
Climate chamber
RH (%)
Temperature (C)
Storage tank
25
20
15
60
40
Storage tank
20
Climate chamber
10
00-05-31
00-06-10
Date (YY:MM:DD)
0
00-05-31
00-06-20
00-06-10
00-06-20
Date (YY-MM-DD)
Figure 5.8 Measurments of temperature (left) and RH (right) in storage tank and
climate chamber
12
TAP WATER
11
Redox-potential (mV)
CHLORIDE
pH
10
9
8
TAP WATER
CHLORIDE
-40
-80
-120
7
6
00-04-09
00-05-29
00-07-18 00-09-06
Date (YY:MM:DD)
-160
00-04-09
00-10-26
00-05-29
00-07-18
00-09-06
00-10-26
Date (YY:MM:DD)
5.4
There is only a small amount of data available at present but nevertheless an attempt
to correlate the laboratory exposure tests to the field exposure tests (presented in
chapter 4) is made below. With additional data from future evaluations the
correlations could be developed further. At present there are large uncertainties in the
estimation.
There are a couple of differences in the parameters included in the field and
laboratory exposure tests. The shortest fibre length used in the laboratory is 35 mm
and 30 mm in field. Studying the right part of Figure 5.6 it can be seen that an
increase of fibres length from 35 to 70 mm gives an increase in the loss of fibre
diameter from 19 to 33 % after 70 days of exposure. In a rough approximation this
can be used to estimate the loss of fibre diameter a 30 mm fibre would have recieved
in the laboratory exposures. Using linear extrapolation from the results a 30 mm fiber
would have corroded according to equation (5.1).
56
Studying the field exposure tests for 30 mm fibres and w= 0.5 mm exposed at Rv40
give that the plate nearest the crack mouth show a loss of 16% of fibre diameter after
2.5 years of exposure which is fairly similar. Therefore the acceleration factor can be
calculated as in equation 5.2.
30,Cl (70) = 35,Cl (70)
ARV 40 =
t field
tlab
* (l35 l30 ) = 19
365 * 2.5
= 13
70
33 19
* 5 = 17%
70 35
(5.1)
(5.2)
The situation for the Eugenia tunnel is that a fibre loss of 5% is present after 2.5 years
and this gives an acceleration factor roughly 3 times the one compared to Rv40.
AEUG = AEUG * 3 = 13 * 3 = 39
(5.3)
Unfortunately there are no results available from the River Dal at present but a
comparison in a similar way as the chloride environment but for tap water
environment with the results from 1998 can be made. The influence of fibre length is
not as obvious in the tap water experiments. Therefore the estimation of how the
30 mm fibres would have corroded in laboratory conditions is choosen to be similar
as the 35 mm ones which is 15% (from Figure 5.7). Since the loss of fibre diameter is
only 3% in the field exposures which is 1/5 of the estimated loss in laboratory the
acceleration factor will be:
ADAL =
t field
5.5
Discussion
tlab
*5 =
365 * 1
* 5 = 26
70
(5.3)
A question possibly rised regarding the pouring technique is the risk for fibres to get
in direct electrical contact with each other. This a possible scenario especially for
longer fibres. Since macro-cell corrosion will be dominant in cracked concrete this
situation is not necessary a problem. The sum of cathode area with e.g. two fibres
being connected in relation to the combined anode area will still be the same and
should therefore not affect the corrosion process.
In this first study of the technique to accelerate the exposures no measurements of the
amount chlorides penetrated the samples has been made. During further evaluations
this should also be controlled at the time for evaluation of corrosion attack.
57
To take into consideration when studying the standard deviations are that all fibres
from the cross sample cross section are included. That leads to that both fibres
exposed close to the crack mouth and in the center of the sample as well as both
fibres with or without corrosion are included in the deviation. This explains the quite
large scatter. Some of the fibres were broken by corrosion or by the process of
removing them from the degraded concrete. In the calculations all broken fibres are
treated as having lost 90% of the fibre diameter.
It is difficult to estimate the humidity conditions in a concrete crack. The humidity
conditions are also very important in trying to understand the mechanisms ruling
initiation and propagation of corrosion in cracks. On the other hand short time
variations should be of little interest when estimating the service life in a 100 year
perspective. Some kind of average environment is possibly more interesting in
creating a model for prediction of durability.
The decrease of pH in the chloride environment is difficult to explain and the general
level is also higher than in a real situation. Measurements by e.g. Paulsson-Tralla
(1999) show pH in the region of 7 in water from a road surface subjected to de-icing
salts.
If the rough estimations on correlation between field exposure tests and laboratory
exposure tests is used to extrapolate the coming behaviour, the samples with 30 mm
fibres and 0.5 mm crack width will have lost about 20% of the fibre diameter
(according to the results after 120 days) after 4 years in the Rv40 environment.
5.6
Conclusions
Keeping the large uncertanties, depending on limited number of test data, in mind the
following conclusion can be drawn from the first two evaluations of the accelerated
laboratory exposure tests:
- In general the methodology choosen can be used in accelerated exposure of
cracked steel fibres samples. The acceleration factor is roughly estimated as 10
times the environment at Rv40, 40 times the environment at Eugenia tunnel and
25 times the environment at the exposure site River Dal.
- In the chloride environment longer fibres corrode faster than shorter ones. An
approximation is that a change of fibre length with 10 mm should give a change in
loss of diameter in a crack of roughly 5%. In the tap water environment the results
are more contradictory.
- An increased crack width give a larger loss of fibre diameter.
58
LOAD-BEARING CAPACITY
6.1
Introduction
In this part the effect of corrosion on the loadbearing capacity will be studied.
6.2
Analytical model
The corrosion of steel fibres will give a reduced cross section and a change in failure
mode. Commonly the unaffected steel fibres have sufficient tensile strength to be
pulled out from a concrete matrix giving a ductile behaviour of a structure. A reduced
cross section gives a tensile failure of the fibre instead and a more brittle failure.
Kosa (1988) has suggested a simplified model for calculation of the effect of reduced
cross section for fibres. The model takes into consideration the change from bonding
failure of the fibers to tensile failure of the fibers with reduced diameter. Input needed
in the model is the number of fibers in the tensile zone, bond strength, fiber tensile
strength, embedded length of fibers etc. The parameters needed are listed below.
0,2h
C
T1
h
0,8h
cracked concrete
corroded fibers
Figure 6.1
uncorroded state
T2
corroded state
In the model it is assumed that of the fibre length is contributing to the bondstrength.
The strength ratio (S) represents the residual moment capacity in percent of the
original, uncorroded state. To be taken into consideration when using the above
mentioned model is that only straight fibres can be simulated with this model. Usage
of hooked end fibres makes the model more complex since friction between the fibre
and the concrete has to be taken into consideration.
59
l
M 1 = N f b 0,4 h
4
(6.1)
x
(0,8 h x)
M 2 = T1
+ T2 0,8 h
2
2
(6.2)
T1 =
(0,8 h x)
l
N fb
0,8 h
4
(6.3)
T2 =
x
N f mean
0,8 h
(6.4)
l ec
l/2
l
Fmean = f b le dl + f s Ac dl /
l ec
0
2
(6.5)
lec =
(c2 f s )
4 fb
(6.6)
S=
M2
M1
(6.7)
M1,M2
,c
diameter
/ corroded area
embedded length
Fmean
= height of concrete
fb
Ac
(fiber/matrix)
fs
= strength ratio
tc
Sc
60
6.3
Service-life modelling
For conventionally reinforced concrete the limit state when the service life ends is
defined by Sarja & Vesikari (1996) as:
1
2
For steel fibres these two criteria are very unfavourable. The first criterion is not
suitable since the propagation of corrosion is much slower than for conventional
reinforcement, even with chlorides present. Therefore the propagation rate should be
taken into consideration. Fibres are also distributed all over the crack plane.
Therefore the effect of reduced fibre diameter in a part of the crack is not as dramatic
as corrosion on rebars all in one level. Criteria no. 2 is not valid at all since the
amount of corrosion products is too small to exceed the tensile strength of the
concrete.
Taking the propagation period into consideration a suitable limit state is to define an
acceptable reduction in load bearing capacity (i.e. an acceptable/critical strength
ratio Sc).
6.3.2 Service life
c
r
(6.7)
0,8 h x
1,6 h ( M 1 Sc T1
4 l fb
2
tc = 2
1 + 1
r
r
x N l fb
2 fs
63
(6.8)
0,8
0.7
Sc
0,6
0,4
r= 0.03 mm/year
r= 0.09 mm/year
0,2
4.23
r= 0.06 mm/year
0
0
10
12
14
Time (year)
Figure 6.2
In Figure 6.2 a graphical presentation is given of the expected service life with
different corrosion rates and critical strength ratios. If it e.g. is accepted that the
strength ratio is reduced to 70 % this would give a service life of only 4.23 years with
a corrosion rate of 0.06 mm/year. The corrosion rate is taken from the field exposures
where the WA30-mix with a crack width of 1 mm showed an average corrosion of
15 % in the upper 25 mm of the crack. at Rv40. Other assumptions in the calculation
are:
Table 6.1
Value
Unit
0.5
mm
fb
7.8
MPa
30
mm
fs
1250
MPa
25
mm
75
mm
20
64
6.4
Discussion
The use of a model as the one presented is not an attempt to make accurate
estimations of the service-life. It has been presented just to illuminate the need for
taking corrosion in cracks into consideration already in the design stage.
In the presented model only a linear elastic approach is used. Bond-slip behaviour
and/or friction is not considered at all in the model. Since hooked end fibres are most
commonly used in Scandinavian sprayed concrete applications, an analytical model
requires that friction between fibre and concrete during pull-out has to be taken into
consideration. The loss of fibre diameter is also depending on the actual depth down
in the crack and this effect must be included in a refined model. The model does not
either take into consideration the continued hydration and therefore increased
strength or the effect of self-healing.
65
Chapter 7 - Conclusions
CONCLUSIONS
Both from the literature study and the inspections of old sprayed concrete structures it
could be stated that steel fibres are relatively resistant to corrosion even with
chlorides present. The experiences from exposure to de-icing salts is however limited.
The lack of really long time experiences with steel fibre reinforcement also make
estimations in a 100 year perspective uncertain. The usage of stainless steel fibres as
a solution to ensure a long time resistance against corrosion is not motivated due to
the extra high material costs. Especially as long as the resistance against corrosion in
cracks has not been proved to be a durability problem.
To start with, the samples in the field exposures increased their residual strength after
exposure. For the 0.1 mm cracks this effect is still accentuated, most probably due to
self-healing. A connection between the detected corrosion on fibres and a trend
turning from increasing residual strength to a decrease is obvious for crack widths 0.5
and 1.0 mm. The majority of the samples still does not show any loss in residual
strength compared to the estimated behaviour without an exposure. A corrosion rate
of approximately 0.02-0.03 mm/year was measured along the Rv40 for samples with
crack width 0.5 and 1.0 mm (30 mm fibres). The samples with 40 mm fibres have
corrode at a rate almost twice the one for 30 mm fibres. The fibre length and
therefore the anode-/cathode-ratio is of great importance in estimating the rate of
corrosion. The influence of crack width is more unsecure in the field exposures. The
only unambiguous fact is that the samples with crack width 0.1 mm show a smaller
amount of corrosion than samples with larger crack widths. To take into
consideration is that the recieved amount of data still is small and the scatter is quite
large. Future evaluations will reduce the uncertainties. At this stage a preliminary
recommendation could be that cracks wider than 0.1 mm should not be accepted in
the motorway environment.
The concept choosen in the laboratory exposure tests seem to be useful in estimating
the long term behaviour. The rough estimations show that the laboratory exposures
accelerate the motorway environment with 10 times. Studying the influence of fibre
length and crack width it is most obvious in the chloride environment in the same
manner as in the field exposures that longer fibres corrode at a higher rate and with
cracks <0.2 mm the fibres corrode at a lower rate. In the tap water environment the
results are contradictory. To keep in mind is that there is almost no data available
from the River Dal since the amount corrosion is still very limited. The behaviour
could be the same there. If using the suggested correlation to extrapolate linearly,
future problems with corrosion could be the result in the motorway environment.
The simple analytical model used to illuminate the basics of a future service-life
model is probably underestimating the residual strength capacity and will therefore be
further developed. No consideration to friction or bond-slip behaviour is taken. Not
either the effect of self-healing.
66
Chapter 7 - Conclusions
67
RESEARCH NEEDS
The suggested correlation between the laboratory exposure tests and the field
exposure tests should be updated with more future data from the ongoing field
exposures. The prediction that 120 days of laboratory exposures is similar to 4 years
of exposure in the motorway environment would be an interesting follow-up by an
evaluation at that time. With a strengthened correlation extrapolations to predict longterm behaviour will be possible.
Cracks can arise from imposed deformations by e.g. shrinkage. Sprayed concrete is
applied in relatively thin layers and therefore subjected to higher shrinkage than
conventional concrete. The actual crack distribution from shrinkage with or without
fibres should be further investigated. Influence on crack distribution from restraints
by the substrate being sprayed on could be of great importance.
The suggested analytical model is still only an embryo to a service-life model and
need further refinement. Non elastic behaviour like a bond-slip and/or friction should
be considered to estimate the behaviour hooked end fibres show. The loss of fibre
diameter is also depending on the actual depth down in the crack and this effect must
be included in a refined model. The model does not either take into consideration the
continued hydration and therefore increased strength or the effect of self-healing. An
analytical model could also be combined with FE-modelling where the non-linear
behaviour could be included more easily in a service-life prediction.
In all the inspected structures a low amount of fibres crossed the cracks when
checked after core drilling. A short distance from the crack the amount of fibres was
sufficient. This points out that cracking in the inspected structures has ocurred where
the least amount of fibres are present. It also accentuate the need for adequate fibre
distribution when spraying concrete. If cracks anyway occur where there are least
fibres available the function is lost. This area need further investigations and
developments to achieve expected properties of a SFRSC.
Sprayed conrete differs from conventional concrete in e.g. specific material
properties (high quality concrete, usage of accelerators) and production technique
(compaction by spraying) and the possible influences from this on fracture
mechanical properties would be important input in understanding the crack
mechanisms.
68
69
Chapter 9 - References
9.
REFERENCES
Alonso, C.& al. (1998). Effect of protective oxide scales in the macrogalvanic
behaviour of concrete reinforcements. Corrosion science, vol 40, no 8, pp. 13791389.
BBK 94. (1994). Boverkets handbok om betongkonstruktioner. (In Swedish). The
Swedish National Board of Housing, Building and Planning.
BRO 94. (1994). Allmn teknisk beskrivning fr broar. (In Swedish). Swedish road
authorities. publ. 1994: , p. 179.
Byfors, K. (1990). Chloride-initiated reinforcement corrosion - chloride binding.
Swedish Cement and Concrete Institute, report no 1:90.
Fagerlund, G. (1992). Betongkonstruktioners bestndighet-en versikt. (in Swedish)
Cementa, 3:e upplagan.
Fagerlund, G. (1994) Betonghandbok material (in Swedish) AB Svensk Byggtjnst
and Cementa AB, second edition, pp 711-726 & 727-783.
Germann, C. & Hansen, A.J (1991). Rapid chloride test, RCT-metoden Mlning
af betons chloridindhold p byggepladsen. (In Danish) Dansk Beton, nr.1.
Hannant, D.J. & Edgington, J. (1975). Durability of steel fibre concrete. Proc.
Fibre reinforced cement and concrete, RILEM, pp. 159-169.
Hannant, D.J. (1976). Additional data on fibre corrosion in cracked beams and
theoretical treatment of the effect of fibre corrosion on beam load capacity. Proc.
Fibre reinforced cement and concrete, RILEM, vol 2, pp.533-538.
Hoff, G. (1987). Durability of fibre reinforced concrete in a severe marine
environment. Proc. Concrete durability-Katherine and Bryant Mather int. Conf SP100, ACI, pp. 21-38.
Holmgren, J. (1992). Bergfrstrkning med sprutbetong. Vattenfall.
Kobayashi, I.K. (1983) Developments of fibre reinforced concrete in Japan. Cement
and Concrete Composites, 5(1), pp. 27-40.
Kosa, K. (1988). Corrosion of fibre reinforced concrete. PhD thesis, University of
Michigan, Ann Arbor.
70
Chapter 9 - References
71
Chapter 9 - References
72
D30
Accelerator
(mm)
(C)
(% of kg cement)
Comments
7,35
28,93
0,44
95
23,3
3,74
2
3.1
7,35
7,35
27,43
28,75
0,43
0,44
90
105
24,2
23,1
5,94
3,3
Clogging in acc.nozzle
3.2
4
5.1
7,35
7,35
7,35
28,75
28,75
30,33
0,44
0,44
0,46
105
100
100
23,1
22,7
21,1
3,1
4,53
8,24
Clogging in acc.nozzle
5.2
6.1
6.2
7.1
7.2
7,35
7,35
7,35
7,35
7,35
30,33
28,75
28,75
29,25
29,25
0,46
0,44
0,44
0,45
0,45
100
100
100
110
110
21,1
21,3
21,3
22
22
2,56
7,1
8,21
3,61
0,52
7.3
8
9
10.1
10.2
7,35
7,35
7,35
7,35
7,35
29,25
29,25
29,25
29,25
29,25
0,45
0,45
0,45
0,45
0,45
110
110
115
115
115
22
21,8
22
22,1
22,1
5,12
4,46
4,17
3,75
2,69
11
12
13.1
13.2
14.1
14.2
15.1
15.2
7,35
7,35
10,21
10,21
10,21
10,21
10,21
10,21
29,25
29,25
27,78
27,78
29,4
29,4
28,3
28,3
0,45
0,45
0,47
0,47
0,49
0,49
0,47
0,47
100
100
85
85
100
100
110
110
22,6
22,4
23,6
23,6
23,5
23,5
23,2
23,2
4,35
4,02
3,5
3,87
3,65
3,79
3,88
-
15.3
16
17
18
19
20
21
22
10,21
10,21
10,21
10,21
6,54
6,54
6,38
6,38
28,3
29,44
29,4
29,94
28,00
26,09
27,7
29
0,47
0,49
0,49
0,49
0,42
0,40
0,42
0,43
110
95
90
70
100
80
80
65
23,2
19,6
20,5
20,9
21,8
23.1
23.2
24
25.1
25.2
25.3
25.4
6,38
6,38
6,38
6,38
6,38
6,38
6,38
28,5
28,5
28
?
?
?
?
0,43
0,43
0,42
?
?
?
?
85
85
80
?
?
?
?
21,0
21,0
21,2
?
?
?
?
4,66
5,47
7,05
6,76
6,35
7,09
-
26
27
28
29
30
31
3,5
3,5
3,5
3,5
3,5
3,5
12,35
20,3
17,89
16,26
17,89
21,95
0,20
0,30
0,27
0,25
0,27
0,32
32
32.1
32.2
3,5
3,5
3,5
19,51
21,95
21,95
0,29
0,32
0,32
0,45
0,02
0,28
0,04
97
13
22,2
1,1
4,4
1,7
WA40
(kg)
W30
(kg)
10
11
Wet-mix
Dry-mix
Average
Std.dev.
Average
Std.dev.
73
Clogging in acc.nozzle
Clogging in acc.nozzle
Clogging in acc.nozzle
Clogging in acc.nozzle
Clogging in acc.nozzle
Clogging in acc.nozzle
Clogging in acc.nozzle
WA30; w= 0,1 mm
1400
1400
1200
RV40 15B
1000
DAL 16A
800
EUG 19A
DAL 61A
1000
Kraft (kN)
Load (kg)
1200
600
RV40 38A
800
EUG 14B
600
400
400
200
200
0
0
0
WA30; w= 0,5 mm
WA30; w= 1 mm
1400
1400
1200
DAL 35B
1200
1000
RV40 21B
1000
RV40 110B
800
EUG 22A
800
EUG 313A
Kraft (kg)
Kraft (kg)
Deflection (mm)
Deflection (mm)
600
DAL 44B
600
400
400
200
200
0
0
0
Deflection (mm)
Deflection (mm)
W30
W30; w= 0,5 mm
1400
1400
1200
1200
1000
DAL 71A
800
RV40 73A
Load (kg)
Load (kg)
W30; w= 0,1 mm
600
1000
DAL A2A
800
RV40 76A
600
400
400
200
200
0
0
Deflection (mm)
Deflection (mm)
74
10
WA40
WA40; w= 0,5 mm
1400
1200
Load (kg)
1000
DAL C7A
800
RV40 C2A
600
400
200
0
0
Deflection (mm)
D30
D30; w= 0,1 mm
D30; w=0,5 mm
1400
1400
1200
1200
RV40 D6A
800
Load (kg)
Load (kg)
1000
DAL E1B
600
1000
DAL D4A
800
RV40 D4B
600
400
400
200
200
0
0
Deflection (mm)
Deflection (mm)
75
10
WA30, w= 0.1mm
14000
14000
12000
DAL 41A
10000
RV40 413A
8000
EUG 311A
DAL 413B
10000
Load (N)
Load (N)
12000
6000
RV40 48B
6000
4000
4000
2000
2000
EUG 314B
8000
0
0
2
3
Deflection (mm)
2
3
Deflection (mm)
WA30, w= 0.5mm
WA30, w= 1mm
14000
14000
12000
DAL 63A
10000
EUG 36A
8000
Load (N)
Load (N)
RV40 511A
6000
12000
DAL 414B
10000
RV40 412B
8000
4000
4000
2000
2000
0
0
2
3
Deflection (mm)
EUG 26A
6000
Deflection (mm)
W30
W30, w= 0.1mm
W30, w= 0.5 mm
14000
14000
12000
DAL A3A
10000
Load (N)
Load (N)
12000
RV40 A9A
8000
6000
10000
RV40 A4A
8000
DAL A10A
6000
4000
4000
2000
2000
0
0
Deflection (mm)
Deflection (mm)
76
WA40
WA40, w= 0.5mm
14000
12000
Load (N)
10000
8000
DAL C13A
6000
RV40 C6B
4000
2000
0
0
Deflection (mm)
D30
D30, w= 0.1 mm
D30, w= 0.5 mm
14000
14000
12000
RV40 E4B
10000
DAL E9B
Load (N)
Load (N)
12000
8000
6000
DAL D10B
8000
6000
4000
4000
2000
2000
RV40 E10B
10000
0
0
Deflection (mm)
Deflection (mm)
77
10
RH (%)
Temp (C)
15
5
0
-5
-10
-15
1997-08-08
1998-09-12
100
90
80
70
60
50
40
30
20
10
0
1997-08-08
1999-10-17
1998-09-12
1999-10-17
Date (YYYY:MM:DD)
Date (YYYY:MM:DD)
Figure D1
River Dal
25
100
90
80
20
10
RH (%)
Temp. (C)
15
5
0
-5
1998-09-12
1999-10-17
1998-09-12
1999-10-17
Date (YYYY:MM:DD)
Date (YYYY:MM:DD)
Figure D2
40
30
20
10
0
1997-08-08
-10
-15
1997-08-08
70
60
50
Temperature (left) and RH (right) at the River Dal site (sliding average
7d).
78
RV40
25
100
90
80
70
60
50
40
30
20
10
0
1997-08-08
20
10
RF (%)
Temp (C)
15
5
0
-5
-10
-15
1997-08-08
Figure D3
1998-09-12
1999-10-17
Date (YYYY:MM:DD)
1998-09-12
1999-10-17
Date (YYYY:MM:DD)
79
Appendix E Paper I
81
Appendix E Paper I
MATERIALS
Cement
The chemical composition of the cement is the main factor affecting durability of
sprayed concrete. Low C3A-content increases the sulphate resistance and low content
of alkali (K+ & Na+) reduces the risk for alkali aggregate reactions. The resistance
against damage due to freeze-thaw action is also increased since the airpore structure
is more fine and dense (Fagerlund [1]). Some authors claim that low content of C3A
would give a reduced chemical bonding of chlorides and therefore increased rate of
corrosion. This is though refuted by Byfors & Tuutti [2]. Increased cement content
reduces the rebound which gives an increased durability, but at the same time the risk
for cracking due to shrinkage increases when more water is normally needed.
Aggregate
The aggregate should have sufficient strength, be resistant to freeze-thaw action and
not be reactive to alkali. The gradation and the maximum aggregate size also affects
the durability by affecting compaction, permeability, mechanical properties and
rebound. Alkali-aggregate reactions arise from aggregate dissolving in the strongly
alkaline environment the cement paste gives. Figure 1 shows the principle for the
reaction, and if any of the components is missing the reaction will not take place.
Calcium hydroxide
Alkali
Alkali-silica
gel
Reactive
Silica
Sometimes
gives expansion
Water
Figure 1.
Basic sketch for alkali aggregate reactions (Lagerblad & Trgrdh [3])
Reinforcement
Fibres are gradually replacing the conventional net reinforcement due to economy
and performance criteria's. The fibres give effects like reduced plastic shrinkage,
increased toughness at failure and limited crack widths. Glassfibres or other types of
plastic fibres mainly affects the properties of the fresh sprayed concrete whereas steel
fibres more affects the properties of the hardened concrete. According to Opsahl [4]
the risk for debonding caused by shrinkage is reduced with addition of fibres. Other
82
Appendix E Paper I
authors claims this is not the case (Holmgren [5]). Concerning durability
performance is an advantage with fibres compared to conventional net reinforcement
since it is easier to make a homogeneous concrete with fibres.
Additions
Commonly used additions are silica fume, fly ash and ground granulated blast
furnace slag. They are used to achieve good workability, reduced permeability, higher
resistance to leaching and chemical attack and reduced risk for alkali aggregate
reactions. As the additions react with the free content of calciumhydroxide this could
reduce the possibility of "self-repairing" small cracks (Fagerlund [6]). Material
properties are shown in table 1.
Portlandcem.
Silica
Fly ash
Density of
particles [kg/m]
3.12-3.15
2.16
2.35
Bulk dens.
[kg/m]
1.4
0.20-0.22
1.00
Spec. surface
[m/kg]
250-500
18 000-22 000
300-500
SiO2 content
[%]
17-25
88-98
40-55
Table 1. Material data for Portland cement, silica and fly ash (Burge [7])
Silica
Silica is added either with the dry materials or as a slurry. Addition of silica in
combination with a superplasticizer gives a cohesive mix which will give reduced
rebound and better bonding to the sprayed surface (Durand, Mirza & Nguyen [8]).
This also enables increased layer thickness and reduced usage of accelerators
(Wolsiefer & Morgan [9], Fidjestl [10]).
The hardened concrete will receive reduced permeability and therefore increase
durability against freeze-thaw action (Burge [11], Glassgold [12], Mailvaganam &
Samson [13], Morgan & al. [14], Wolsiefer & Morgan [9]). Addition of silica also
gives increased durability against breaking down due to chemical attack and higher
electrical resistivity which reduces the risk for corrosion (Wolsiefer & Morgan [9]).
If damages occur due to freeze-thaw action, in a concrete with silica fume but
without entrained air, the breaking down is very fast compared to ordinary concrete
(figure 2.) (Fagerlund [1]). Addition of silica also gives increased plastic shrinkage at
early age and therefore higher demands on proper curing. Pettersson [15] points out
the risk for a reaction between undispersed silica-gel and the alkaline cement paste.
At normal dosage of well-dispersed silica is the risk for alkali aggregate reactions
instead reduced (Lagerblad & Trgrdh [3]).
83
Appendix E Paper I
100
50
Figure 2.
100
200
300
No of freeze-thaw cycles
Freeze-thaw test with salt solution of concrete with and without silica fume and with no
air entrainment. W= water, C= Portland cement, S= silica fume (Fagerlund [1]).
Fly ash
In the fresh sprayed concrete fly ash has the effect of increased cohesion and reduced
risk for separation. The early strength growth is slower and this behaviour is
accentuated at low temperatures. Fly ash gives increased durability against sulphate
attack, reduced permeability (Mailvaganam & Samson [13]), increased freeze-thaw
resistance and reduced risk for alkali aggregate reactions. High content of fly ash also
gives reduced diffusivity for chloride ions. (Tuutti [16], Johansson, Sundbom &
Woltze [17]).
Variations in the remaining coal content can affect the water demand for the sprayed
concrete and possibly be the reason for the reduction of the air content of approx. 1%
(Fagerlund [1]). To high addition of fly ash gives an increased tixotropic effect,
reduced strength, increased shrinkage at desiccation (Mailvaganam & Samson [13])
and increased carbonation rate (Ljungkrantz, Mller & Petersons [18]).
Slag
GGBS makes the concrete less water demanding and reduces the risk for separation.
Other effects are increased resistance to sulphate attack and reduced chloride
permeability. A temporarily increased resistance to freeze-thaw attack is, in the long
run, replaced with reduced resistance. The reason is the increased water absorption in
the air-pore system. With freeze-thaw load in combination with water containing
chlorides the resistance is due to the amount of added GGBS. Luther, Mikols,
DeMaio & Whitlinger [19] claim that amounts of added GGBS between 35-65%
gives a reduction of the freeze-thaw resistance.
84
Appendix E Paper I
Additives
There are several various additives for sprayed concrete, both dry- and wet-mix.
Their effect could be both on the fresh and the hardened concrete. The most
commonly used additive for wet-mix sprayed concrete is the binder accelerator.
Other types of additives are retarders, air-entraining agents, plasticizers, water
reducers and polymers.
Accelerators
Accelerators are used to achieve early support in rock stabilisation and to make it
possible to spray thicker layers. This gives reduced amount of discontinuities and a
more homogeneous concrete with increased durability. The most commonly used
binding accelerators consists water soluble salts of alkali metals or alkaline-earth
metals. Most known are chlorides, silicates, carbonates and aluminates. Chlorides are
used only to a limited extent because of the increased risk for corrosion. Many
investigations (Burge [11], Manns & Neubert [20]) points out the effect of
accelerators on durability. Alkaline accelerators give a reduction of the compressive
strength at high dosages. Gebler & al. [21] also discovered a connection between
reduced compressive strength and reduced resistance to freeze-thaw load. Probably is
the phenomenon due to increased porosity and increased micro-cracking caused by
the accelerated hydration. This is the reason to why Opsahl [22] observed increased
permeability with the usage of accelerators. Other effects caused by alkaline
accelerators are increased shrinkage with 20-50% both with wet- and dry-mix
sprayed concrete (Manns & Neubert [20]). They also discovered doubled creep with
normal addition of accelerators based on potassium carbonate/aluminate (wet-mix)
and sodium carbonate/aluminate (dry-mix).
Air entraining agents
The purpose with air entraining agents is to create a concrete with small, welldistributed air pores. The air pores will act as expansion tanks for water pressed away
by freezing. In dry-mix sprayed concrete the air entraining agent is added to the dry
materials or in to the mixing water. This leads to difficulties in predicting the air pore
content before spraying. There are many different opinions about the effect of air
entraining agents in dry-mix sprayed concrete. According to ACI 506 R-90 there is
no effect while Durand & Mirza [8] claims there is with reference to their tests.
When using wet-mix sprayed concrete the air entraining agent is added in advance
and the air pore content could therefore be measured before spraying. Generally
known is that the air pore content is reduced with 50% at spraying (Durand & Mirza
[8], Seegebrecht & al. [23], Morgan & al. [14], Schrader & Kaden [24]). To high
addition of air entrainers gives reduced strength.
85
Appendix E Paper I
PERFORMANCE
The performance is very important to achieve high durability and the skill of the
personnel is essential. An optimally composed concrete could easily be wasted with
erroneous performance.
Wet-mix sprayed concrete
High addition of accelerators at wet-mix spraying or variations of added water at drymix spraying reduces the possibility to achieve durable sprayed concrete. Non
homogeneous concrete surrounding reinforcement could cause corrosion. Preparation
and curing should also be correctly performed. To obtain an unobjectionable
performance the staff should be competent and well experienced. The conditions
when it concerns equipment, materials, planning of the site should also be satisfying.
For wet-mix sprayed concrete the advantage is that the concrete is ready mixed and
can be controlled before spraying. Other advantages compared to dry-mix spraying is
high capacity, less dust and reduced wastage. It is difficult to keep the dosage of
accelerator at a constant level and uneven dosage can give differences in strength and
discontinuities. An other disadvantage is that the equipment normally requires more
water-rich concrete, but this can be solved with usage of water reducers or
plasticizers.
Dry-mix sprayed concrete
The main advantage of dry-mix sprayed concrete is the possibility of making sprayed
concrete with low w/c -ratio and that accelerators are not needed. It is also possible to
use pre-bagged material and equipment with relatively low capacity, which can be an
advantage at smaller repair jobs. When using dry-mix sprayed concrete the worker is
adjusting the water content manually and this could make the properties vary
considerably. Uneven distribution of the dry materials will give a variation of the
water content in the sprayed concrete. The wastage is often greater with dry-mix
sprayed concrete.
Preparation & Curing
Sufficient preparation of the surface that is to be sprayed is important. Anything that
could reduce the bond has to be removed. When repairing concrete constructions all
damaged concrete must be removed. Morgan & Neill [25] presents an investigation
of bridges in Canada repaired with sprayed concrete. A common reason to failure
was continued breaking down of the underlying repaired concrete while the repair it
self was unaffected. Other factors reducing durability is reduced bond due to
insufficient cleaning of rock surfaces with eg a film of oil from diesel engine exhaust
or water containing soil or clay. The surface should be damp but with no free water
on the surface for concrete and as dry as possible for rock. For rock surface
strengthening it is therefore important with sufficient drainage of water from the
surface to be sprayed.
86
Appendix E Paper I
When applying new layers of sprayed concrete it is important to remove old curing
agents from the surface (should not be applied between layers) since some agents
reduce bond effectively.
After spraying the young concrete must be protected against drying out to avoid
cracking. Either by water or membrane curing. Some authors question the effect of
some membrane curing agents. Particularly important is the curing of sprayed
concrete containing silica while this gives increased plastic shrinkage (Opsahl [4]).
87
Appendix E Paper I
B Scrit
Sact
+
0
-
Function
Temp.
C
Degree of
saturation, S
0
Figure 3.
Time
88
Appendix E Paper I
Testing of freeze-thaw durability is divided in to two tests, one with water containing
chlorides and one without, usually the resistance is lower to water with chlorides.
The phenomena is probably due to osmotic pressures (pressure obtained by
differences in chloride concentration) occurring between the surface and the pore
water in the concrete. The damages are most commonly surface damages. According
to the Norwegian Concrete Association [31] the most common damage on sprayed
concrete due to freeze-thaw action is debonding from water-bearing rock. This
points-out the importance of sufficient drainage behind sprayed concrete in zones
with low temperatures.
Leaching
Chemical attack could be of eg acids, sulphates or salts. Acids cause surface damages
in the sprayed concrete and reinforcement corrosion. The phenomena can occur when
storing acids or in waste water pipes. Normally the sulphate contents in Swedish soils
are too low to cause a reaction. A concentration at the surface of the sprayed concrete
when water containing sulphate evaporates is though possible. Different kinds of
salts could concentrate at the surface or at the boundary in the sprayed concrete and a
growth of crystals could cause cracks and spalling.
FURTHER RESEARCH
From this investigation it can be said that more knowledge about corrosion of steelfibre-reinforcement in sprayed concrete is needed. Especially the theories about fibre
corrosion in cracked sprayed concrete, but also the effect on load-bearing capacity
with corroded fibres. Not many investigations about freeze-thaw resistance for
sprayed concrete with presence of water containing chlorides are found in this study
and this can be an area for further research. This should be a common situation in
road-tunnels with splashing of water containing thaw salts. To predict or measure the
durability of a structure with sprayed concrete it could be convenient to be able to
measure the permeability in situ. Available methods should be tested.
89
Appendix E Paper I
REFERENCES
1.
2.
3.
4.
5.
6.
7.
BURGE, T A. "Additives and mixtures for shotcrete" Tunnels & tunneling, Jan,
1993.
8.
DURAND, B.; MIRZA, J. & NGUYEN, P. " ASTM C666 (A) Freeze-thaw
durability of air-entrained wet- and dry-mix shotcrete ", Shotcrete for
underground support VI, 1993, pp 188-196.
9.
WOLSIEFER, J Sr. & MORGAN, D R. " Silica fume in shotcrete ", Concrete
Int., pp 34-39, April, 1993.
10. FIDJESTL, P. "Applied silica fume concrete", Concrete Int., Nov, 1993, pp
33-36.
11. BURGE, T A. "Fiber reinforced high-strength shotcrete with condensed silica
fume" ACI report SP 91-57, 1991, pp 1153-1170.
12. GLASSGOLD, L I. "Shotcrete durability: an evaluation", Concrete Int., Aug.,
1989, pp 79-85.
13. MAILVAGANAM, N P. & SAMSON, D. " The role of admixtures in the
effective use of fly ash and silica fume in concrete mixes ", Proceedings from "
Ash - A valuable resource ", 1987, pp 1-18.
14. MORGAN, D R.; KIRKNESS, A J.; McASKILL, N & DUKE, N. " Freezethaw durability of wet-mix and dry-mix shotcretes with silika fume and steel
fibres ", Cement, concrete & aggregates, 1988, pp 96-102.
15. PETTERSSON, K. " Effects of silica fume on alkali-silica expansion in mortar
specimens ", Cement & concrete research, vol 22, no 1, 1992, pp 15-22.
16. TUUTTI, K. " Korrosion p armering ", Marina betongkonstruktioners
livslngd, seminariehandling, Uppsala, 1993, pp 85-101 (in Swedish).
17. JOHANSSON, L.; SUNDBOM, S. & WOLTZE, K. " Permeabilitet - provning
och inverkan p betongs bestndighet ", CBI rapport, 2:89, 1989 (in Swedish).
90
Appendix E Paper I
91
Appendix F Paper II
Summary
As a part of ongoing research in the area of sprayed concrete durability, Vattenfall
Utveckling has accomplished a survey of the status on sprayed concrete structures in
Sweden. The survey is financed by Elforsk (Swedish Electrical Utilities Research and
Development Company) and SveBeFo (Swedish Rock Engineering Research). Included in
the survey are both strengthening of rock and concrete repair, but the paper will put
emphasis on concrete repair of dams. Three objects of spillways repaired with sprayed
concrete will be presented in this paper. Two objects where damages appeared after a few
years of use and one showing no damages. It seems like the difference in durability can be
derived from erroneous design of joints, thin layers of sprayed concrete and difference in
surface treatment after spraying.
Resum
Dans le cadre de la recherche en cours dans le domaine de la longvit du bton projet.
Vattenfall Utveckling a ralis une tude de ltat des structures en bton projet en Sude.
Ltude est finance par Elforsk (Socit Sudoise de Recherche et de Dveloppement des
Utilits Electriques) et SveBeFo (Socit Sudoise de Recherche de lIngnierie des
Roches). Le renforcement des roches et la rparation du bton sont compris dans ltude
mais le compte-rendu soulignera la rparation des barrages. Trois objets de dversoirs
rpars laide de bton projet sont prsents dans le compterendu. Deux o des
dommages se sont manifests aprs quelques annes de service et un ne dmontrant aucuns
dommages. Il semble que la diffrence de longvit soit cause par des erreurs de
conception des raccords, de fines couches de bton projet et des diffrences dans le
traitement de surface aprs pulvrisation.
93
Appendix F Paper II
1.
INTRODUCTION
2.
SELECTION OF OBJECTS
After making inquiries for constructions with sprayed concrete a selection of suitable objects had to
be done. The inquiry resulted in several answers and it would not be practicable to study all of them.
The selection was done with the following criteria.
Age. The primary criterion which also have been governed by the used spraying method. Structures
performed with the dry-mix method could be relatively old since the method it self has been
developed to a limited extent. Whereas structures performed with the wet-mix method was chosen
with an age less then 15 years, since the method and the admixtures are under constant
development.
Type of reinforcement. The objects were divided in the three groups steel fibre reinforcement,
conventional reinforcement and no reinforcement. The study has emphasis on objects with steel
fibre reinforcement.
Climate and environment. The study is aiming to include as many different types as possible. E.g.
repeated freeze-thaw action, one sided water pressure and wear due to running water.
Documentation. Objects with more information (both by personal communication and written
documentation) about preparation and performance is selected in front of others.
Accessibility. Interesting objects with limited or no accessibility has been left out.
94
Appendix F Paper II
3.
For a more thorough inspection of each object a great amount of equipment is needed for e.g. taking
out of test specimen. Therefore it was decided at an early stage that an overarching inspection
should be carried out firstly. For objects showing up something interesting it would then be possible
to go back for a more thorough inspection.
When making an overarching inspection the available test methods are limited but in the following
section some of the methods are described.
3.1
Object documentation
To make an accurate inspection it is of great importance to know something about the history of the
structure. When, how and why the sprayed concrete is applied and preferably something about the
material composition. Anyone that have tried to find out all this information about sprayed concrete
objects know that it is no simple task. The available information is often limited to why and
sometimes when the sprayed concrete is applied. If there are previous test results and reports from
former inspections this can give a great deal of information about the object and a possible time
point for damage appearance.
3.2
When making this type of inspection there are a couple of methods that can be used and these are
e.g. the following.
3.2.1 Ocular inspection
The easiest method to perform but the most difficult to evaluate the results from. Ocular inspection
requires thorough knowledge about the concrete behaviour under different kinds of load from
constructive forces as well as climate. At an ocular inspection cracks and patterns of cracks can be
studied and from this it might be possible to conclude a source for cracking. Deposits from water
leaching through cracks can also be studied. Areas of the sprayed concrete surface diverging in e.g.
colour can give important information.
The ocular inspection together with other types of investigations can give a general picture of the
status of the structure and the reason to this. Separate test results not completed with an ocular
inspection has often little or no relevance.
3.2.2 Crack width
In its simplest form it is carried out with a crack width gauge. The crack width can give a hint about
the reason for appearance.
95
Appendix F Paper II
4.
OBJECTS
Three objects on dam constructions have been subject of investigation and two of them were
inspected on site. All objects are spillways repaired with sprayed concrete. In all cases the reason for
repair has been erosion of concrete and corrosion of reinforcement followed by spalling of concrete.
The two inspected dams are located at the hydroelectric power stations of Nmforsen in the river
ngermanlven and Hlleforsen in the river Indalslven. Both these objects have concrete damages
to a different extent today. The third object is the dam at the hydroelectric power station of
Stadsforsen which also is situated in the river Indalslven. This object does not show any visual
damages today.
4.1
Nmforsen
Appendix F Paper II
15 and 50 mm according to an investigation from October 1990 [2]. The concrete is applied from
the gate and down the spillway in a section of 3-4m width (se figure 1). The sections were divided
with a bulkhead and the space between the sections was filled afterwards.
Joint
9.0m
Sprayed Concrete
Old Concrete
20.5m
Figure 1.
In one of the spillways it was tested to trowel the surface of the sprayed concrete on half the
spillway. The reason for stopping the trowelling was a suspicion that it might reduce the bonding to
the underlying concrete. In the investigation [1] it could be established that the trowelling did not
reduce the bonding strength.
4.1.2 Status today
Spillway A was subject for examination. After two years damages occurred when parts of the
sprayed concrete came loose and were washed away. The damages were concentrated to parts
according to figure 2. Previous damages where repaired with a polymer/silica-composite. Newly
arisen damages have not been repaired yet and therefore it was possible to establish the fact that the
sprayed concrete there was thin (approx. 15mm).
Test points
1.
2.
New damages
Sprayed
Concrete
Figure 2.
Old
Concrete
Old
Repairs
97
Appendix F Paper II
As can be seen the damages are concentrated to the lower part of the spillway and near the joint
between two sections. In figure 2 it can be seen where testing of bonding strength and measure of
sprayed concrete thickness were performed in 1990. The results were 1.3-1.4 MPa and the thickness
was 26mm in point 1 and 15mm in point 2. Damages appeared and were repaired in point 2.
When knocking the surface with a hammer no areas with signs of debonding could be found. Not
even near areas where sprayed concrete is missing. No cracks were visible.
4.2
Hlleforsen
6
9
11
15
Figure 3.
12
18
14
17
10
13
16
98
Appendix F Paper II
Figure 4.
The reason for occurred damages can be many, but an important factor should be the construction of
joints between the different sections (see figure 5.)
ate
fw
o
n
tio
rec
i
D
rflo
cre
on
c
Old
Figure 5.
te
cre
n
co
ed
y
a
r
Sp
te
As can be seen in figure 5 the section (2) applied below the previous upper one (1) is very thin at the
upper end. This part is most likely to come loose and be washed away and water will then continue
to erode the joint between the two sections and cause further damage. This is probably the main
reason to why damages have occurred. Another reason can be poor bonding between the old
concrete and the sprayed concrete and/or between sprayed concrete layers from the beginning. As
mentioned in the beginning this object is not documented and there are uncertainties about the
preparatory work.
99
Appendix F Paper II
4.3
Stadsforsen
5.
DISCUSSION
In this investigation three quite similar sprayed concrete repairs in spillways have been studied. In
spite of the similarity two of the objects show damages and one does not. Why is that?
When trying to explain the difference in durability one must take into consideration the time of the
year for construction. The two damaged structures are performed during wintertime with heated
tents over the spillways. This will of course make it more difficult to produce a good repair with
sprayed concrete.
In both Nmforsen and Hlleforsen damages have arisen in connection to the joints between
different sections of the sprayed concrete. In Nmforsen it seems like the designer/contractor at least
have taken into consideration the risk for damages to occur in the joints. When spraying the
spillways bulkheads were used to divide the sections when spraying, and this gives a possibility to
keep the same thickness of each layer of sprayed concrete. No overlap on previously sprayed
sections that will give a risk for damage. Since the dimensions of the spillways in Nmforsen are
smaller than in Hlleforsen it was also practicable to leave out joints in the horizontal direction
which probably will reduce the damage risk. The joints should also be trowelled to reduce the
unevenness of the surface of the sprayed concrete.
Where the new damages have appeared in Nmforsen the sprayed concrete is very thin and the
probable reason for damage is debonding. Debonding can occur if rebound is not removed when
spraying on an adjacent section. If a delamination should appear the reason for the sprayed concrete
to come loose is most likely due to underpressure caused by running water on the spillway.
100
Appendix F Paper II
The spillways in Stadsforsen are the only ones with a sprayed concrete surface being completely
steel trowelled which will make it fairly smooth with few discontinuities. This will probably reduce
the risk for damages to take root in joints.
6.
CONCLUSIONS
After removal of damaged concrete the surface should be watered to a state of "damp but slightly
absorbing" and no free water should be present on the surface to be sprayed. This will reduce the
risk for capillary action on the fresh sprayed concrete from the old concrete which can cause a
reduced strength due to lack of water for the hydration of cement.
The joints should be designed in a way making them easy to construct with as few discontinuities as
possible. Before spraying an adjacent section the joints should be blasted to increase the bonding
between the different sections.
Rebound from spraying should be collected or blown away by an assistant to the nozzle man. If the
rebound is not removed it can reduce the bonding strength between the old concrete and the sprayed
concrete. The system with a blow man assisting the nozzle man is common in e.g. USA
After spraying trowelling of the surface is preferred since this will make the surface smooth and
reduce the possibilities for a damage to take root.
It is also preferable to avoid thin layers of sprayed concrete which otherwise can be damaged more
easily than thicker layers. If a thin sprayed concrete layer debonds the action of underpressure from
running water will be enough to make the concrete come loose.
7.
REFERENCES
[1]
[2]
Tljsten, Bjrn. Control of sprayed concrete repairs at the Nmforsen power plant
Swedish). Vattenfall Utveckling AB, document 7990002. 1990.
101
(in