influence of reinforcement spacers on mass transport ... · 51 66 [12], and german practice in din...
TRANSCRIPT
Cement & Concrete Research, 2016 (Accepted)
1
Influence of reinforcement spacers on mass transport properties and durability 1
of concrete structures 2
S. Alzyoud, H.S. Wong* & N.R. Buenfeld 3
Concrete Durability Group, Department of Civil and Environmental Engineering, Imperial College London, 4
SW7 2AZ, UK 5
Abstract 6
Spacers are ubiquitous in reinforced concrete, but their influence on durability is unclear. This paper presents 7
the first study on the effects of spacers on mass transport and microstructure of concrete. Samples with 8
different spacers, cover depths, aggregate sizes, curing ages and conditioning were subjected to diffusion, 9
permeation, absorption and chloride penetration, and to μXRF, BSE microscopy and image analysis. Results 10
show that spacers increase transport in all cases, the magnitude depending on spacer type and transport 11
mechanism. Plastic spacers produced the largest increase, followed by cementitious spacers and then steel 12
chairs. The negative effect is due to a porous spacer-concrete interface that spans the cover where 13
preferential transport occurs. Spacers may seem low value, small and inconsequential, but because they are 14
placed every ≤ 1 m along rebars, their overall effect on ingress of external media is significant. This is not 15
currently recognised by standards or by most practitioners. 16
Keywords: Durability (C); Interfacial transition zone (B); Microstructure (B); Transport properties (C); 17
Spacers 18
19
1. Introduction 20
Spacers are essential components in reinforced concrete structures. Their function is to secure steel 21
reinforcements in the correct position within the formwork to prevent movement prior to and during 22
concreting so that the required cover is obtained in the finished structure. The size of spacer determines the 23
size of the cover depth to reinforcement, which in structural design, is defined according to the severity of 24
exposure environment, required durability and fire resistance. Achieving adequate depth and quality of 25
concrete cover is critical because it protects embedded steel reinforcement from the external environment. It 26
is well-known that inadequate cover is the major factor causing premature corrosion of reinforcement, the 27
principal form of degradation of concrete structures. In structural design, it is assumed that achieving the 28
specified cover ensures that the as-built structure achieves the expected design performance in terms of 29
durability, fire resistance and serviceability (crack width). 30
Spacers are made of plastic, metal or cementitious materials, and are available in various sizes and 31
shapes (see Fig.1). In this paper, we will use a generic term “spacers”, but recognising that other terms may 32
be prevalent elsewhere, e.g. bar supports, wire chairs, bolsters, continuous runners, dowels etc. Although 33
many types of spacers are available commercially, they generally fall into one of six categories: a) plastic 34
spacers with integral clip-on action for horizontal rebars of 20 mm or less, b) plastic end spacers that fit ends 35
of rebar for end cover, c) plastic wheel/circular spacers for vertical rebars in columns and walls, d) 36
cementitious block spacers for bar size > 20 mm in heavily-reinforced sections; e) continuous line spacers 37
that are either cementitious or plastic, of constant cross-section in typically1 m lengths to support several 38
bars; and f) steel wire chairs that may be single, continuous or circular, to support the top horizontal rebar 39
from lower rebar or to separate layers of vertical rebars. Of all the available types, plastic spacers are the 40
most popular because they are cost effective and they do not need to be wire tied to rebar, which is labour 41
intensive. For further details on spacer types, readers can refer to refs. [1-6]. 42
The first comprehensive guidance on spacers was probably Concrete Society Report CS 101 in 1989 43
[7]. Prior to that, it was not uncommon to use any material available on-site including bricks, tiles, broken 44
concrete, and timber pieces to support reinforcement [4]. Another practice was to use site-made cement 45
mortar blocks as spacers, but this practice was not stopped in BS 8110-1 [8] because of poor quality control. 46
At present, spacers in the UK should be factory manufactured, conform to BS 7973-1 [9] and placed in 47
* Corresponding author. E-mail: [email protected]. Phone: + 44 20 7594 5956
Cement & Concrete Research, 2016 (Accepted)
2
accordance with BS 7973-2 [10]. The general rule is that spacers should be fixed to reinforcing bars at a 48
spacing not exceeding 50d or 1000 mm, where d is the bar size, and in staggered rows for parallel bars. 49
Other similar recommendations are available for North American practice e.g. ACI 315-99 [11] and ACI SP-50
66 [12], and German practice in DIN EN 13670 [13]. 51
All spacers (except some wire chairs for top rebars) work on the principle of supporting reinforcement 52
from the nearest exposed surface, i.e. from the formwork or blinding. As such, spacers must interrupt the 53
concrete cover and replace a portion of the concrete in the cover zone. Inevitably, they form a link between 54
reinforcement and the external surface, and present a possibility of compromising the effectiveness of the 55
cover to protect embedded reinforcement (see Fig. 2). It is reasonable to suspect that the presence of spacers 56
potentially facilitates ingress of aggressive agents such as water, chloride, CO2 and oxygen either through the 57
spacer itself or its interface with concrete. If true, then spacers would act as weak links and accelerate 58
deterioration. Furthermore, spacers are placed every metre or less along the reinforcement and are left 59
permanently in the structure, and so a structure contains thousands of spacers and their combined effect 60
could be significant. Indeed, several reports and field investigations have observed a link between spacers 61
and reinforcement corrosion. Examples of such reports include refs. [1, 14, 15, 16, 17, 18, 19, 20, 21, 22] and 62
a common observation was that local rebar corrosion occurs at spacer locations. This was assumed to be due 63
to either poor quality of the spacer (itself being highly porous), or the concrete near spacers (poor 64
compaction) or the interface between concrete and spacer (e.g. debonding). 65
Despite concerns expressed over the effect of spacers on durability, to the best of our knowledge, there 66
has been no systematic or fundamental research carried out on this issue. Therefore, the present work seeks 67
to redress this by establishing the effect of several spacer types on mass transport properties and 68
microstructure of concrete. The overall aim of this study is to enhance the understanding of how spacers 69
influence the durability of concrete structures. 70
71
2. Experimental 72
2.1 Spacers 73
Samples of cementitious, plastic and steel spacers were requested from all leading manufacturers and 74
distributors in the UK in order to explore the available range of products. Approximately 110 types of 75
spacers were obtained and they varied in terms of material, height (cover size) and shape, as shown in Fig. 1. 76
From this, 7 spacer types representing the most commonly used in construction, were selected for testing. 77
Details of the selected spacers are given in Table 1. Bar or line spacers were rejected because these would 78
require the preparation and testing of large concrete elements, which is impractical and probably unnecessary 79
for the purpose of study. The selected steel spacers were continuous lattice chairs and these were sectioned at 80
nodes into 100 mm long segments. The spacers used in this study are indicated in Fig. 1. 81
The porosity of cementitious spacers was obtained by measuring the mass difference from a vacuum 82
saturated-surface dry condition to 105°C oven-dried condition, divided by the spacer volume. Plastic spacers 83
contain openings that might trap aggregate particles and block the movement of fresh concrete. Therefore, 84
the openings were measured to ensure that the selected spacer had sufficiently large openings to allow the 85
largest aggregate particles to nestle in them. Some of the plastic spacers were modified by either grinding 86
using a 120-grit size SiC paper or by scoring four 1 mm deep notches on the main flange. The purpose of this 87
is to increase the surface roughness of the spacer in order to improve its adhesion and bond to the concrete 88
matrix. Prior to use, the plastic and cementitious spacers were cleaned and dried, while the steel chair 89
segments were sand blasted to remove any rusting from the surface. The spacers were stored in the 90
laboratory to avoid any moisture or temperature variations. The ‘volume fraction’ in Table 1 denotes the 91
fraction of the disc test sample (Section 2.3) that is occupied by the spacer. 92
93
2.2 Concrete materials and mix proportions 94
Ordinary Portland cement CEM I was used for all mixes. Its oxide composition was 63.4% CaO, 95
20.8% SiO2, 5.4% Al2O3, 2.4% Fe2O3, 1.5% MgO, 0.3% Na2O, 0.7% K2O, 2.9% SO3 and < 0.1% Cl. The 96
calculated Bogue composition was 53.1% C3S, 19.1% C2S, 10.8% C3A and 7.2% C4AF. The loss on ignition, 97
specific gravity and fineness of the cement were 2.1%, 3.06 and 291 m2/kg respectively. Tap water was used 98
as batch water. Thames Valley sand and gravel were used as fine aggregates and coarse aggregates 99
Cement & Concrete Research, 2016 (Accepted)
3
respectively. The maximum particle size was 5 mm for sand. For gravel, the maximum particle size was 100
either 10 or 20 mm. The particle size distributions of the fine and coarse aggregates are shown in Fig. 3. The 101
sieve analysis show that the sand complied with BS 882:1992 medium grading, while the gravel complied 102
with BS EN 12620:2002+A1 overall grading. The specific gravity, moisture content and absorption of the 103
aggregates are given in Table 2. 104
Two concrete mixes with free water/cement (w/c) ratio of 0.4 were prepared according to the 105
proportions given in Table 3. The mixes were designed to the absolute volume method and batch water was 106
adjusted to account for aggregate absorption so that the target free w/c ratio was achieved. The main variable 107
in the mixes was the maximum size of aggregate (MSA) which was either 10 or 20 mm. The total aggregate 108
content was fixed at 70%. Therefore, the cement paste fraction and hence total porosity were the same for all 109
samples allowing meaningful comparison. The sand to total aggregate mass fraction was 0.4. 110
111
2.3 Test samples 112
140 cylindrical samples were prepared in total, as summarised in Table 4. The cylindrical samples 113
have a diameter of 100 mm and thickness of either 25 mm or 50 mm, depending on the cover size produced 114
by the particular spacer. Samples from Mix C10 were cast directly in 100 mm diameter steel moulds as 115
shown in Fig. 4, while samples from Mix C20 were cast in large slabs (1500 ×600×50 mm) and then cored 116
to obtain the required size. The spacers were placed at 150 mm centres in the slab. The latter procedure was 117
carried out to ensure that a proper placement and compaction of the concrete could be achieved with Mix 118
C20, which contained larger aggregate particles (MSA 20 mm). A Hilti DD 120 diamond corer was used to 119
extract the 100 mm diameter cores from the slab at the end of curing. 120
It is absolutely critical to ensure that the spacers are well secured onto the mould or formwork to 121
prevent any relative movement when the fresh concrete is placed and compacted. This is because 122
displacement of the spacer may prevent a complete compaction of the concrete or may induce a negative 123
effect on the microstructure of the hardened concrete. In real structures, the weight of the reinforcement 124
should be sufficient to hold spacers in place during concreting. However, samples prepared in the laboratory 125
are relatively small and so do not contain sufficient reinforcement weight to achieve this. Therefore, several 126
timber pieces bolted tightly onto the base plate of the mould were used to clamp the spacer and rebar (high-127
yield steel, deformed, 12 mm) in place prior to concreting, following the assembly shown in Fig. 4. This 128
worked very well and checks showed no relative movement between the spacer and mould when the entire 129
assembly was vibrated. 130
131
2.4 Mixing, curing and conditioning 132
All materials were batched by weight. Cement and aggregates were first dry mixed in a 30-liter 133
capacity pan mixer for about 30s. Water was then added and mixed for a further 3 min. The slump values for 134
all mixes were in the range of 90 to 100 mm. A vibrating table with adjustable intensity was used for 135
compaction. The intensity of the vibrating table was adjusted to suit workability of the fresh concrete mix. 136
The 25 mm thick cylindrical samples were compacted in one layer while the 50 mm thick samples were 137
compacted in two layers. Each layer was vibrated until no significant release of air bubbles was observed. It 138
should be noted that all of the mixes were easily compacted and showed no indication of segregation, 139
bleeding or excessive voidage in the cross-section (Fig. 5). The freshly compacted samples were covered 140
with plastic sheet and wet hessian at room temperature for the first 24 hours and then de-molded. 141
Samples for oxygen diffusion, oxygen permeation, water absorption and microscopy were sealed 142
cured by wrapping in at least 6 layers of cling film, then sealed in polythene bags at 20°C for 3 and 28 days. 143
A short curing of 3 days was carried out to replicate typical site practice in addition to the “standard” 28-day 144
curing. Periodic checks by weighing found insignificant mass loss showing the effectiveness of the sealed 145
curing. At the end of curing, samples were unwrapped and labelled. Samples subjected to chloride diffusion 146
were cured in a fog room at 100% RH and 20˚C for 28 days. This was to ensure that the samples were well-147
hydrated before exposure to chloride solution. 148
Samples were conditioned using one of three regimes prior to transport testing by drying to “constant” 149
mass, taken as when the mass loss was less than 0.01% per day. The purpose of conditioning to equilibrium 150
is to ensure that the measured transport properties are not influenced by variation in moisture content. 151
Cement & Concrete Research, 2016 (Accepted)
4
However, samples were subjected to three conditioning regimes so that the influence of spacers under 152
different exposure environments can be examined. The conditioning regimes were: a) 20°C, 75% RH, b) 153
20°C, 55% RH, and c) 50°C oven drying to replicate hot weather condition. Drying at 20°C was carried out 154
in closed chambers over saturated NaCl or Mg(NO3)2 to achieve the required RH of 75% and 55% 155
respectively. The chambers contained fans to generate circulating air and soda lime to prevent carbonation. 156
The conditioning time required to reach constant mass ranged between 2 and 6 months, depending on drying 157
rate and sample thickness. Samples conditioned at 50 °C were cooled to room temperature in a vacuum 158
desiccator for 24 h to prevent moisture from re-entering the samples during cooling prior to testing. Samples 159
for chloride diffusion were conditioned at 20°C, 55% RH for 28 days, similar to the procedure described in 160
the AASHTO T259 salt ponding test. 161
162
2.5 Oxygen diffusion, oxygen permeation and water absorption 163
Samples from Series I, II and III (Table 4) were tested following the sequence of oxygen diffusion, 164
oxygen permeation and water absorption in three replicates. The same sample can be used for all three tests 165
because the results of each test are not influenced by the previous test. Details of the tests are given in Wong 166
et al. [23] and summarised here. For oxygen diffusion and permeation, the sample was fitted into a silicone 167
rubber ring in a steel cell. The curved side of the sample was sealed by applying 15 kN compression on the 168
silicone rubber ring, which expands and grips the sample at a lateral confining pressure of ~ 0.6 MPa. This 169
prevents leakage through the circumference as shown in Wu et al. [24]. Oxygen diffusivity was determined 170
by exposing the two opposite flat faces of the sample to oxygen and nitrogen at equal pressure. The oxygen 171
and nitrogen diffuse in opposite directions through the sample, and the oxygen concentration in the outflow 172
stream was measured at steady-state flow using a zirconia analyser to calculate diffusivity. 173
Oxygen permeability was determined by applying a gas pressure of 0.05, 0.15 and 0.25 MPa above 174
atmospheric, and measuring the steady-state outflow rates at each applied pressure. The apparent 175
permeability at each pressure was calculated according to Darcy’s law for compressible fluids and the 176
intrinsic permeability was determined by applying Klinkenberg’s correction for gas slippage. The recorded 177
mass of the discs before and after both tests showed negligible change. Water absorption was carried out by 178
placing the sample on two plastic strips in a tray of water to a depth of 2-3 mm, and monitoring the amount 179
of water uptake with time until full saturation was achieved. Mass measurement was done using an 180
electronic balance accurate to 0.01 g. The sorptivity coefficient was obtained from the slope of the regression 181
line of absorbed water per unit flow area against square-root time according to the classical unsaturated flow 182
theory [Hall, 1977]. The best-fit regression line was drawn across at least 10 readings taken during the first 7 183
hours of measurement. The coefficients of regression of the least squares fit for the Klinkenberg correction 184
and sorptivity determination were always greater than 0.985. 185
186
2.6 Chloride diffusion 187
Samples from Series IV (Table 4) were subjected to chloride penetration. Immediately after the 188
predetermined curing and conditioning regime, the curved sides of the samples were sealed with two layers 189
of waterproof adhesive tape to ensure unidirectional chloride ingress. The samples were then placed in a 190
shallow tray containing 3% by mass sodium chloride solution for 90 days. The depth of the solution was 191
about 5 mm above the exposed surface of the sample and the solution was refreshed periodically. The tray 192
was covered to minimise evaporation. At the end of the exposure, samples were sectioned in half using a 193
diamond abrasive cutter and analysed with Orbis PC micro X-ray fluorescence (μXRF) to determine the 194
depth and spatial distribution of chloride penetration. 195
Each sample was analysed to obtain three chloride profiles as shown in Fig 5: a) within the 196
cementitious spacer or concrete nestled in plastic spacer, b) along the spacer-concrete interface and c) near 197
the sample edge away from spacer. Each profile consists of 50 spot analyses carried out at every 1 mm 198
spacing to produce a detailed chloride concentration distribution from the exposed surface through the entire 199
thickness of the sample. The positions of the spots were carefully selected with the help of a high 200
magnification camera (75x) to ensure that the analysis avoided large aggregate particles or air voids that 201
would otherwise produce discontinuities (gaps) in the generated profiles. Prior to analysis, a parametric study 202
was conducted to determine the optimal operating conditions of the μXRF for detecting and measuring 203
chloride. The final analysis was carried out using a 30 μm beam size, 25 kV beam voltage, 12.8 μs, amplifier 204
Cement & Concrete Research, 2016 (Accepted)
5
time constant and 500 ms dwell time. A dead time of ~40% was achieved with this configuration. A 25 μm 205
thick Al filter was used to enhanced the sensitivity for chloride. 206
Subsequently, the collected spectra were analysed to indentify all detectable elements using the 207
ORBIS Vision (v. 2.0) software. False peaks (sum and escape) were accounted and removed prior to 208
quantitative analysis. However, the spectrum contains several peaks partially obscured by the background 209
signal. Therefore, the signal-to-noise ratio of spectral peaks method [25] was used for background 210
subtraction. Finally, the collected signal for chlorine was converted to % wt of cement by referring to a 211
calibration curve obtained by carrying out the same spot analysis on a series of cement pastes at the same 212
w/c ratio and curing age that contained known amounts of admixed chloride [26]. The non-steady state 213
chloride diffusion coefficient was then estimated by fitting the measured chloride profiles to Fick’s second 214
law of diffusion. Microsoft Excel Solver was used to calculate the best-fit values for the diffusion coefficient 215
and surface chloride concentration that minimised the sum of the squares of residuals. The coefficients of 216
regression of the best-fit lines (R2) were above 0.9 for all cases. 217
218
2.7 Fluorescent epoxy impregnation 219
Another method used to examine the influence of spacers on transport properties was to pressure 220
impregnate samples from the exposed flat surface with fluorescein dyed epoxy. The cross-sections of the 221
samples were then examined to measure the depth and distribution of the intruded epoxy. Selected samples 222
from Series II & III (Table 4) were placed in a cell similar to the one used for oxygen permeability and 223
sealed to prevent side leakage. Fluorescein dyed epoxy was then poured onto the top exposed surface and 224
compressed air at 0.7 MPa above atmospheric pressure was applied and maintained for 6 hours to force the 225
epoxy into the sample. The impregnated sample was allowed to harden at room temperature for 2 days and 226
then sectioned with a diamond abrasive cutter in half from the centre. Each cross-section was flat ground 227
with 120-grit SiC paper. Fluorescence imaging was carried out using a digital SLR camera under 15W 228
ultraviolet lamp placed 35 mm above the sample to provide uniform illumination and fluorescence. Images 229
of the entire cross-section (100 × 50 mm) were captured at 2177 × 1085 pixel resolution in TIFF in a dark 230
room to avoid stray light reflection and to achieve a good contrast. The camera was operated at small 231
aperture to increase the depth of field and slow shutter speed to achieve adequate exposure. Full details of 232
the procedures are given in Wu et al. [24, 27]. 233
Image analysis was carried out using ImageJ to measure the area fraction and depth of epoxy 234
impregnation on three replicate samples. The areas intruded with epoxy appear bright green due to 235
fluorescence and these were segmented using colour thresholding based on hue, saturation and brightness 236
thresholding to generate a binary image. For all samples, the hue threshold range was set at 60-180 which 237
corresponds to the green colour angle on the hue circle. The entire brightness and saturation histograms (0-238
255) were utilised so that all shades of the green colour were included. The same thresholding settings were 239
applied to all samples. The accuracy of the segmentation was checked by cross-referencing the binary 240
images with the original fluorescence images. 241
242
2.8 Backscattered electron imaging 243
A field-emission scanning electron microscope operated in the backscattered electron (BSE) mode 244
was used to characterise the microstructure of the spacer-concrete interface in greater detail. Block samples 245
(40 × 20 × 10 mm) were extracted from cross-sections prepared as described in Section 2.7, impregnated 246
with ultra-low viscosity epoxy resin under vacuum and then pressurised to ensure full impregnation. The 247
blocks were then ground and polished at successively finer grades (68 µm, 30 µm, 18 µm, 14 µm, 9 µm, 6 248
µm, 3 µm, 1µm, and 0.25 µm) to achieve a flat and well-polished surface, then carbon coated with an 249
evaporative coater. BSE images were collected at 10 kV accelerating voltage and 10 mm working distance. 250
Fifty images were collected per sample around the spacer-concrete interface for samples containing 251
plastic and cementitious spacers. Thirty-six images per sample were collected for those containing steel 252
spacer; the decrease in number of images captured is due to the small size of steel spacer (5 mm wire), which 253
limits the available area for observation. All images were captured at 500× magnification, digitised to 2560 254
x 2048 pixels at pixel spacing of 0.094 μm, giving a field of view of 240 × 192 μm. These settings were 255
chosen to obtain a good sampling area and resolution to characterise the phases of interest. To ensure random 256
Cement & Concrete Research, 2016 (Accepted)
6
and unbiased sampling, images were captured at ~ every 1 mm distance along the spacer-concrete interface. 257
To isolate the effect of aggregates, images were selected such that they were located at least 50 μm away 258
from the nearest aggregate particle to avoid sampling the aggregate-paste “interfacial transition zone” (ITZ). 259
If an image contained aggregate particle close to the spacer interface, then the image was replaced by another 260
within the adjacent area. Brightness and contrast settings were calibrated to achieve a brightness histogram 261
that utilised the entire dynamic range (0-255). 262
Quantitative image analysis was carried out to examine the microstructure gradients at the spacer-263
concrete interface. First, the spacer edge was carefully traced on an enlarged image and segmented using 264
ImageJ. This was relatively straightforward because all spacers are clearly visible and can be accurately 265
segmented. Then, the pores and any cracks were segmented using the overflow method [28] by defining the 266
upper threshold level from the inflection point of the cumulative brightness histogram of the BSE image. The 267
unreacted cement particles were segmented using the minimum grey value between peaks for hydration 268
products and the unreacted cement as the lower threshold value. Finally, the spatial distribution of segmented 269
porosity and unreacted cement from the spacer-concrete interface were measured using the Euclidean 270
Distance Mapping approach [29] at a resolution of one pixel strip width. 271
272
3. Results 273
3.1 Sample density and compaction 274
It was absolutely critical to ensure that the samples were properly compacted so that any observed 275
effects of the spacers on mass transport properties are not attributable to excessive voidage from poor 276
compaction. As mentioned previously, the spacers were tightly secured to the sample mould to prevent 277
movement during concrete placement and compaction. However, the samples are relatively small and the 278
spacer may occupy up to ~25% of the sample volume (Table 1). Furthermore, plastic spacers contain small 279
openings that may create problems during concrete placement and compaction. To check this, the density of 280
the hardened concrete sample was measured and compared against its theoretical density, calculated from the 281
mix design and spacer density. Measurements were made on three replicate samples. The difference between 282
measured and theoretical density was found to be very small and ranged between 0.03% and 0.5%, with an 283
average difference of 0.3%. Furthermore, all of the examined cross-sections show that the samples were 284
indeed well compacted with no sign of excessive voidage or segregation. Some examples of sample cross-285
sections are shown in Figs. 2, 5 & 10. 286
287
3.2 Oxygen diffusivity, oxygen permeability and water sorptivity 288
The measured oxygen diffusivity, oxygen permeability and water sorptivity coefficients are presented 289
in Figs. 6-8. Each data point represents the average of measurements on three replicate samples and the error 290
bar indicates standard error 𝜎 √𝑛⁄ , where σ is the standard deviation and n is the number of replicates. The 291
results show that increasing curing age from 3 to 28 days caused a reduction in transport properties in all 292
cases, as expected. Transport properties decreased by about 7% to 37% when comparing the data obtained at 293
3d curing (Fig. 6a) to that at 28d curing (Fig. 6b) for the same sample type and thickness. It is also evident 294
that the height of spacer (sample thickness) has a huge influence on the measured permeability, but not on 295
diffusivity and sorptivity. The 25 mm thick samples (t/MSA = 2.5) consistently gave much higher 296
permeability, by about 38% to 52%, compared to the corresponding 50 mm thick samples (t/MSA = 5) for 297
the same mix, spacer type and curing age. This is due to drying-induced surface microcracking and the 298
related size effects on permeability as reported in Wu et al. [27]. 299
Interestingly, there is a clear and consistent trend that samples containing spacers have higher 300
transport properties than their respective control samples. Overall, the increase in transport properties as a 301
result of spacers ranged between 10% and 300%. The average increases in oxygen diffusivity, oxygen 302
permeability and water sorptivity were 57%, 138% and 27% respectively. However, the increase in transport 303
property is dependent on spacer type. Samples containing plastic spacers consistently gave the highest 304
diffusivity, permeability and sorptivity coefficients followed by samples with cementitious spacers, then 305
steel spacers. The control samples had the lowest transport coefficients in all cases. For example, the average 306
increases in diffusivity, permeability and sorptivity for samples containing plastic spacers were 85%, 193% 307
and 41% respectively. For samples containing cementitious spacers, the average increases in diffusivity, 308
Cement & Concrete Research, 2016 (Accepted)
7
permeability and sorptivity were 37%, 125% and 18% respectively. For samples with steel spacers, the 309
average increases in diffusivity, permeability and sorptivity were 25%, 36% and 10% respectively. 310
It should be noted here that the percentage differences in transport properties between samples with 311
and without spacer presented in this paper were obtained from 100 mm diameter cylindrical test sample with 312
a centrally placed spacer. Furthermore, the spacer occupies between 6 and 25% vol. of the test sample (Table 313
1) depending on spacer type. Therefore, it is expected that the percentage influence of a particular spacer on 314
the average transport property depends on the size of the test sample relative to the spacer. For example, it is 315
expected that the measured increase in transport property for samples containing spacers would be higher if a 316
smaller test sample were used, and vice-versa. 317
Fig. 7 & Fig. 8 show that conditioning regime prior to testing has a major influence on the measured 318
transport properties. Increasing the severity of drying increases the accessible porosity, therefore oven drying 319
at 50 °C consistently produced the highest transport coefficients, followed by room temperature (20°C) 320
drying at 55% RH and 75% RH. For samples conditioned at 50°C, the average increases in diffusivity, 321
permeability and sorptivity of samples containing plastic spacers relative to the control were 102%, 162% 322
and 44% respectively. For samples containing cementitious spacers, the corresponding values were 87%, 323
96% and 25% respectively. In contrast, samples containing steel spacers showed modest increases in 324
diffusivity, permeability and sorptivity of 21%, 22% and 9% respectively. 325
It is important to note that the effect of spacers on transport is also evident in samples that were 326
subjected to gentle drying at 20°C, 75% RH. For example, the average increases in diffusivity, permeability 327
and sorptivity for samples containing plastic spacers relative to the control after 20°C, 75% RH drying were 328
52%, 243% and 37% respectively. The corresponding values for samples containing cementitious spacers 329
were 57%, 107% and 19% respectively, and for samples containing steel spacers, 25%, 31% and 8% 330
respectively. The results in Fig. 7 also show that samples containing the modified plastic spacer PS(a) and 331
PS (b) gave a noticeable improvement in transport properties. For example, the average decreases in 332
diffusivity, permeability and sorptivity relative to the samples containing unmodified plastic spacer (PS) 333
after conditioning at 50°C were 14%, 38% and 11% respectively. 334
Cementitious spacers are inherently porous, and this may be an important factor. Therefore additional 335
testing was carried out to determine the sorptivity and total water absorption of the cementitious spacer. The 336
results show that the average sorptivity of the cementitious spacer was 31.9 g/m2.min0.5, which was ~39% 337
that of the control sample (C10-Co50, 3d-cured) after conditioning at 20°C, 55%RH. It was also observed 338
that the absorption of the cementitious spacer to saturation was only 18% of the control sample. These results 339
suggest that the observed increase in transport properties of samples containing cementitious spacers is due 340
to factors other than the porosity of the spacer itself. 341
342
3.3 Chloride penetration 343
Chloride concentration profiles measured at 1 mm increments along the spacer-concrete interface, at 344
the centre and edge of the sample are shown in Fig. 9. For samples containing plastic spacers, the highest 345
chloride concentration and penetration depth occurred along the spacer-concrete interface and at the centre, 346
i.e. concrete nestled within the spacer. The lowest chloride concentration profile was obtained near the edge 347
of sample away from the spacer, and this was very similar in magnitude to that obtained from the control 348
sample. For samples containing cementitious spacers, the highest chloride concentration profile also 349
occurred along the spacer-concrete interface. In contrast, the lowest chloride concentration profile was 350
obtained from the centre of the sample, i.e. within the spacer. This is consistent with the fact that the porosity 351
and transport property of the cementitious spacer are much lower than those of the concrete. 352
The average non-steady state diffusion coefficient Deff and surface chloride concentration Cs are 353
presented in Table 5. It can be seen that the chloride diffusion coefficient and surface chloride concentration 354
for samples with spacers were highest at the spacer-concrete interface, for both plastic and cementitious 355
spacers. The lowest values were found at the centre of the cementitious spacer. Interestingly, the values 356
obtained at the edge of the samples containing spacers were very similar to those measured for the control 357
samples without spacers. Plastic spacers produced a greater effect on chloride penetration compared to 358
cementitious spacers. At the spacer-concrete interface, the chloride diffusion coefficient and surface chloride 359
concentration were higher than the control sample by 33% and 203% respectively. 360
Cement & Concrete Research, 2016 (Accepted)
8
361
3.4 Epoxy impregnation depth 362
Fig. 10 shows cross-sections of samples impregnated with fluorescent epoxy resin from the bottom 363
exposed surface and their respective binary images to highlight the epoxy intruded areas. The measured area 364
fraction and epoxy impregnation for samples with different spacer types, MSA and conditioning regimes are 365
presented in Fig. 11. The results are the average of three replicates and error bars shown on the figure 366
represent +/- one standard error of the average. It can be seen that the presence of spacers increases the 367
amount and depth of epoxy impregnation. Samples containing plastic spacers consistently showed highest 368
epoxy penetration followed by samples with cementitious spacers and then steel spacers. The control 369
samples had the lowest amount of epoxy penetration. 370
It can also be seen that the amount of epoxy intrusion increased with increase in severity of drying, as 371
expected. These trends are similar to those observed for mass transport properties presented in the preceding 372
sections. When conditioned at 50°C, samples with plastic spacer showed epoxy impregnation that almost 373
reached the full thickness of the sample (50 mm). The area fraction and depth of impregnation were greater 374
than those of the control sample by 127% and 273% respectively. Fig. 10d shows that most of the epoxy 375
intrusion occurred along the spacer-concrete interface and through the concrete nestled inside the plastic 376
spacer. The presence of steel spacer had the least effect on the area and depth of epoxy impregnation, 377
increasing by only 22% and 6% relative to the control on average. 378
379
3.5 Microstructure of the spacer-concrete interface 380
The distributions of unreacted cement and detectable porosity, measured at 0.1 μm intervals up to a 381
distance of 80 μm from the spacer boundary, are shown in Fig. 12 (a) and (b) respectively. Results are the 382
average of fifty frames (thirty-six for steel spacer) and expressed as the area percentage of the cement paste 383
area. The results are compared to the microstructural gradients measured at the aggregate-paste interfacial 384
transition zone (ITZ) and bulk paste of the control sample. It can be seen that the unreacted cement fraction 385
increased steadily from < 0.5% at the spacer boundary to around 14% at the “bulk” cement paste region and 386
remained relatively constant beyond 50 μm away from the spacer boundary, for all spacer types. In contrast, 387
the detectable porosity was highest at the spacer boundary, and decreased with increasing distance from the 388
spacer boundary. Beyond about 50 μm from the spacer boundary, the porosity for all sample types was 389
relatively stable and achieved a similar value to that of the bulk paste porosity. The porosity at the spacer-390
concrete interface is approximately 3 to 6 times that of the bulk paste region, depending on spacer type. 391
It is evident that plastic spacers had the greatest effect on the microstructure of the spacer-concrete 392
interface. Nevertheless, all spacer types produced strong microstructure gradients that are reminiscent of the 393
aggregate-paste interfacial transition zone. Measurements made on the control sample along the bulk cement 394
paste region showed no significant gradients in unreacted cement and porosity, as expected. When all of the 395
acquired BSE images of the spacer-concrete interface and their individual gradients where compared, it was 396
observed that the microstructure is heterogeneous and spatially variable. Although the averaged porosity and 397
unreacted cement distributions displayed clear trends, the individual result from each location was quite 398
variable. The microstructure at some locations was highly porous with a visible bond crack forming between 399
the spacer and concrete, while others were seemingly unaffected by the presence of spacers. Furthermore, 400
fine microcracks of ~4 to 6 μm width were detected on the exposed face, along the spacer-concrete interface 401
in some of the samples containing plastic and cementitious spacers. Several example BSE images of highly 402
porous spacer-concrete interface are shown in Figs. 12 (c) to (d). 403
404
4. Discussion 405
Mass transport properties 406
Spacers are either non-porous/impermeable (e.g. plastic and steel spacers), or have lower porosity 407
(cementitious spacers) to that of the concrete around it. In samples containing spacers, the spacer occupies 408
and replaces a substantial volume of the concrete, ~25% for samples containing cementitious spacers and 409
~14% for samples containing plastic spacers. As such, the spacers should act as obstacles to transport by 410
reducing the cross-sectional area available for flow and this should decrease overall transport. However, the 411
Cement & Concrete Research, 2016 (Accepted)
9
results from this study collectively show that the inclusion of spacers always increases overall transport. The 412
spacers facilitate transport to gas, chloride ions and water, regardless of the mechanism, i.e. under 413
concentration gradient, pressure gradient or capillary suction. 414
The magnitude of increase in transport mainly depends on spacer type and transport mechanism. 415
Curing age, conditioning regime and thickness of the sample or spacer did not produce a huge impact on 416
how spacers influence transport properties. In other words, the negative effect of spacers was observed for all 417
curing ages and conditioning regimes, even for thick samples (50 mm) that were reasonably well cured (28 418
day) and subjected to very mild drying at 20°C, 75% RH. However, plastic spacers consistently produced the 419
largest increase in transport, followed by cementitious spacers and steel spacers. In terms of transport 420
mechanism, permeability was affected the most, with increases of up to 300% observed in some samples. 421
Unfortunately, it is not possible to compare our results with other studies since there are no 422
precedents. However, it is interesting to note that the magnitude of increase in transport due to spacers is of 423
the same order of magnitude as that caused by drying-induced microcracking or air voids [24, 27, 30]. The 424
transport properties of the control samples (no spacers) obtained in this study are in good agreement with 425
available experimental data from samples of similar mix proportions, curing age and conditioning regime. 426
The measured chloride diffusion coefficients and surface concentrations are slightly higher than those 427
reported in earlier studies such as [31, 32, 33]. This could be due to the samples being dried prior to exposure 428
to salt solution. Samples from Mix C20 with maximum aggregate particle size of 20 mm were cored from a 429
larger slab to ensure proper placement and compaction, but coring could be damaging and this may have an 430
effect on the measured transport properties. However, the results from the cored samples were not hugely 431
different to those from cast samples with smaller aggregate particle sizes (10 mm). This indicates any 432
damage caused by coring is likely to be minimal. 433
Microstructure of the spacer-concrete interface 434
The results clearly show that inclusion of a spacer disturbs concrete microstructure by decreasing the 435
cement content and increasing the porosity at the vicinity of spacer. The deficiency in cement content at the 436
spacer-concrete interface is due to difficulties in packing cement particles close to the spacer surface, i.e. the 437
“wall effect”. The increased porosity can be explained by higher initial water content at the interface due to 438
the wall effect and bleeding along the spacer surface or entrapment of bleed water beneath horizontal 439
surfaces of the plastic spacer. Furthermore, debonding and microcracking occurred at the interface, and this 440
is probably due to differential volumetric changes (drying shrinkage or thermal) or relative movements 441
between spacer and concrete. 442
For all cases, the zone around the spacer with affected microstructure extends to around 50 μm from 443
the spacer boundary, with respect to both unreacted cement and porosity gradients. We note that the 444
characteristics of these gradients and the width of the affected zone are similar to those of the aggregate-445
cement paste “interfacial transition zone” [34, 35, 36, 37], as well the steel-concrete interface [38, 39] and air 446
void-paste interface [30]. The width of the affected zone should be related to the size of cement particles and 447
so it is expected to decrease if finer cement particles or supplementary cementitious materials such as 448
microsilica are used. However, this needs to be verified. 449
The results show that a zone of weakness exists in the concrete near spacers that has higher 450
water/binder ratio and porosity compared to the concrete located farther away. The increase in transport 451
properties recorded for samples with spacers is caused by the porous spacer-concrete interface. This is also 452
confirmed by the fluorescent epoxy impregnation (Fig. 10 & 11) and μXRF results (Fig. 9) showing a 453
preferential transport path along the spacer-concrete interface. The observed increase in surface chloride 454
content is most probably caused by the increase in the porosity at the spacer-concrete interface. The 455
performance of samples with plastic spacers was the worst across all properties measured. This is probably 456
due to a weak bond between plastic and concrete. There is also a greater mismatch in material properties in 457
terms of drying shrinkage and thermal expansion/contraction. In addition, samples with plastic spacers have 458
the highest amount of spacer-concrete interface; the surface area of a plastic spacer is 3-4 times that of a 459
cementitious spacer and 23 times that of a steel spacer (Table 1). 460
Comparison to aggregate-paste ITZ 461
Although the microstructure gradients at the spacer-concrete interface and aggregate-paste ITZ are 462
similar, their effect on the overall transport property and durability is probably very different. This is because 463
the spacer-concrete interface forms a continuous link that spans the full cover and so directly exposes 464
Cement & Concrete Research, 2016 (Accepted)
10
embedded reinforcing steel to external aggressive species. The ITZ on the other hand, envelopes aggregate 465
particles that are discontinuous. At high aggregate contents, the ITZ fraction may increase to the point where 466
adjacent ITZs overlap and interconnect. However, the effect of this is balanced by reduction in cement paste 467
content and total porosity of the concrete, and the increase in cement paste tortuosity. Furthermore, 468
conservation of water in the system dictates that increasing ITZ fraction will decrease porosity of the bulk 469
paste and therefore, its transport property. Indeed, experimental and numerical studies have shown that the 470
net effect of ITZ on bulk transport properties is small, even though the ITZs themselves have a greater local 471
porosity (and transport property) and are interconnected [e.g. 31, 40, 41, 42, 43, 44, 27]. Therefore, the effect 472
of the ITZ is clearly different to that of the spacer-concrete interface despite both having similar 473
microstructure characteristics in terms of porosity and unreacted cement gradients. 474
Implications 475
It is well-known that poor quality and/or inadequate concrete cover thickness is a major factor causing 476
premature degradation in concrete structures worldwide resulting in huge repair and replacement costs. 477
Spacers may seem low value, small and inconsequential, but because they are placed every metre or less 478
along the length of steel reinforcement and are left permanently in the cover zone, their overall effect could 479
be significant. The results from this study suggest that for any concrete structure, the locations where 480
reinforcements are supported by spacers would experience a greater exposure to external environment and 481
ingress of aggressive species. The assumed protection afforded by the design concrete cover is therefore not 482
achieved at the location of spacer. For instance, if a spacer increases diffusivity by a factor of four, the effect 483
of this is broadly similar to reducing the local cover by half. This leads to onset of carbonation-induced or 484
chloride-induced corrosion in quarter of the expected time, compromising long-term durability. 485
At present, it seems there is very little emphasis on the potential effects of spacers on long-term 486
durability, whether in research, design or construction. There is no guidance concerning how one could 487
reduce, prevent or mitigate the negative effects of spacers. It should be noted that codes of practice do 488
recognise that spacers must be durable and should not cause corrosion of reinforcement, spalling of concrete 489
cover, cracking or allow free passage of moisture, e.g. BS 7973-1:2001 [9] and BS EN 13670: 2010 [45]. 490
However, there are no recommendations on performance testing to ensure the quality of spacers or the 491
concrete placed near spacers or the bond between spacer and concrete. There is very little differentiation 492
between spacer types in terms of how their material, shape or execution may influence long-term durability. 493
For example, in BS 7973-1:2001, spacers are classified mainly according to their load capacity. The 494
emphases in current standards are on fixity, stability and load bearing capacity of the spacer, as well as their 495
impact on aesthetic appearance of the finished structure, rather than durability. It is also not common practice 496
for designers to specify spacers or include them in drawings or contract documents. Very often, the 497
responsibility for selecting spacer type to be used is left to the contractor. If guidelines are not provided, then 498
contractors may base their decision on cost, which may not be in the best interests of the durability of the 499
structure. 500
Single spacers were used in this study for the sake of convenience because it allowed testing to be 501
carried out on relatively small samples. However, continuous/bar/line spacers are commonly used, for 502
example in bridge decks and large slabs, to support several bars per spacer. These continuous spacers are 503
attractive because they reduce bar fixing time, but they create a large spacer-concrete interface area 504
compared to that of single cover spacers and thus could be more problematic in terms of cracking and effect 505
on durability. Furthermore, the samples in this study were tested in constant/controlled environment and 506
under non-loaded conditions. Real structures are subjected to moisture, temperature and load cycles. These 507
factors will increase the likelihood of debonding and cracking at the spacer-concrete interface, thus 508
exacerbating the effect of spacers on transport and microstructure. 509
510
5. Conclusion 511
The effect of reinforcement spacers on mass transport and microstructure of concrete was investigated via an 512
experimental programme involving 140 samples. Tests were carried out on 100 mm diameter samples with a 513
centrally placed spacer that is either plastic, cementitious or steel wire chair. Other test variables include 514
cover depth (25 and 50 mm), maximum aggregate particle size (10 and 20 mm), curing age (3 and 28 days) 515
and conditioning regime (20°C/75% RH, 20°C/55% RH and 50°C oven). The main conclusions are as 516
follows: 517
Cement & Concrete Research, 2016 (Accepted)
11
a) Presence of spacers always increases transport properties despite the spacers themselves containing little 518
or no porosity, and the concrete being fully compacted. Spacers increase penetration of gas, chloride ions 519
and water, regardless of the mechanism, i.e. permeation, diffusion or capillary absorption. The magnitude 520
of increase in transport depends on spacer type and transport mechanism. 521
b) Permeability was affected the most, with increases of up to 300% observed for a 100mm diameter 522
concrete sample with a centrally placed spacer. The average increase in oxygen diffusivity, oxygen 523
permeability and water sorptivity for all samples containing spacers were 57%, 138% and 27% 524
respectively. At the spacer-concrete interface, the chloride diffusion coefficient and surface chloride 525
concentration were higher than the control by 33% and 203% respectively. 526
c) The negative effect of spacers on transport was observed for all curing ages and conditioning regimes, 527
even for thick samples (50 mm) that were well cured (28 day) and subjected to very mild drying at 20°C, 528
75% RH. 529
d) The spacer-concrete interface contains higher porosity, lower cement content, and therefore higher 530
water/binder ratio compared to the concrete farther away from the spacer. Debonding and microcracking 531
were also observed at the spacer-concrete interface. The width of the disturbed microstructure is around 532
50 μm from the spacer. The porosity at the spacer-concrete interface is 3 to 6 times that of the bulk paste 533
region depending on spacer type. 534
e) The increase in transport recorded for samples with spacers is caused by the porous spacer-concrete 535
interface, which forms a continuous link spanning the full cover depth and provides an easy path for 536
ingress of aggressive species to reach embedded reinforcement. This is confirmed by fluorescent epoxy 537
impregnation and μXRF showing a preferential transport path along the spacer-concrete interface. The 538
increase in surface chloride concentration is linked to the higher porosity at the spacer-concrete interface. 539
f) Plastic spacers consistently produced the largest increase in transport and the highest porosity at the 540
spacer-concrete interface, followed by cementitious spacers and steel spacers. The performance of 541
samples with plastic spacers was the worst across all properties measured. This is probably due to a weak 542
bond between plastic and concrete. There is also a greater mismatch in material properties in terms of 543
drying shrinkage and thermal expansion/contraction. 544
g) Spacers may seem low value, small and inconsequential, but because they are placed every 1 m or less 545
along the length of steel reinforcing bar and are left permanently in the cover zone, their overall effect on 546
the penetration of external media into the cover zone of concrete structures is significant. This is not 547
currently recognised by codes of practice or by most practitioners. 548
549
Acknowledgements 550
We would like to thank Mr. Andrew Morris for his help with the laboratory work. 551
552
References 553
[1] Portland Cement Association (1981), Choosing the right bar support in aggressive environments, 554
Concrete Technology Today, September, 3. 555
[2] Concrete Society (1994), Spacers for reinforcement, Concrete, May/June, 27-30. 556
[3] King, E. S. & Dakin, J. M. (2001), Specifying, detailing and achieving cover to reinforcement, CIRIA 557
C568, Construction Industry Research and Information Association, London, 90 pp. 558
[4] Shaw, C. (2007), Cover to reinforcement: getting it right. The Structural Engineer, February, 31-35. 559
[5] Barnes, R. (2012), Spacers and visual concrete, Concrete Advice No. 43, Concrete Society. 560
[6] CRSI (2011), Placing Reinforcing Bars: Recommended Practices, 9th Edition, Concrete Reinforcing 561
Steel Institute, 288 pp. 562
[7] Concrete Society (1989), Spacers for reinforced concrete, Standard provisions for achieving cover 563
encircling reinforcing steel, Concrete Society Report CS 101, Crowthorne. 564
Cement & Concrete Research, 2016 (Accepted)
12
[8] BS 8110-1 (1997), Structural use of concrete. Part 1: Code of practice for reinforced concrete. British 565
Standard Institution, London, 1997. 566
[9] BS 7973-1: 2001, Spacers and chairs for steel reinforcement and their specification. Part 1: Product 567
performance requirements, British Standards Institution. 568
[10] BS 7973-2:2001, Spacers and chairs for steel reinforcement and their specification. Part 2: Fixing and 569
application of spacers and chairs and tying of reinforcement, British Standards Institution. 570
[11] ACI 315-99 (1999), Details and detailing of concrete reinforcement, American Concrete Institute, 44 571
pp. 572
[12] ACI SP-66 (2004), ACI Detailing Manual, American Concrete Institute, 219 pp. 573
[13] DIN EN 13670:2011-03, Execution of concrete structures, Deutsches Institut für Normung. 574
[14] Levitt, M. & Herbert, M.R. (1970), Performance of spacers in reinforced concrete. Civil Engineering, 575
August, 1-5. 576
[15] Jahren, P. (1994), Carat - a development programme for new reinforcement chairs, Betong - Industrien 577
2/94, Oslo, 25-27. 578
[16] Kawahigashi, T., Kuzume, K. & Miyagawa, T. (1999), Deterioration process and estimation of 579
durability of reinforced concrete beams in long-term exposure to marine environment, Proceedings of 580
Japan Society of Civil Engineers, 71-84. 581
[17] Rostam, S. (1996), High performance concrete cover - why it is needed, and how to achieve it in 582
practice, Construction and Building Materials, 10 (5) 407-421. 583
[18] Vik, O. T. K. (2002), The role of bar supports in combating corrosion in reinforced concrete, In: Brett, 584
P.H., Banthia, N., Buckland, P.J., Sixth International Conference on Short and Medium Span Bridges, 585
Canadian Society for Civil Engineering, Vancouver, Canada. 586
[19] Kenai, S. & Bahar, R. (2003), Evaluation and repair of Algiers new airport building, Cem. Concr. 587
Compos., 25, 633-641. 588
[20] Shaw, P., Materialrőntgen, A. & Kutti, T. (2003), Field measurement and experience of chloride 589
induced corrosion of reinforcement in submerged structures, AB Fardig Betong, Goteborg, Sweden. 590
[21] Tang, L. & Utgenannt, P. (2009), A field study of critical chloride content in reinforced concrete with 591
blended binder, Materials and Corrosion, 60, 617-622. 592
[22] Poulsen, E. & Mejlbro, L. (2010), Diffusion of Chloride in Concrete: Theory and Application, CRC 593
Press, 480 pp. 594
[23] Wong, H.S., Buenfeld, N.R., Hill, J., Harris, A.W. (2007), Mass transport properties of mature 595
wasteform grouts, Adv. Cem. Res., 19, 35-46. 596
[24] Wu, Z., Wong, H.S., & Buenfeld, N.R. (2014), Effect of confining pressure and microcracks on mass 597
transport properties of concrete, Adv. Appl. Ceram., 113 (8) 485-495. 598
[25] Ernst, T., Berman, T., Buscaglia, J., Eckert‐Lumsdon, T., Hanlon, C., Olsson, K., Palenik, C., Ryland, 599
S., Trejos, T. & Valadez, M. (2014), Signal‐to‐noise ratios in forensic glass analysis by micro X‐ray 600
fluorescence spectrometry, X‐Ray Spectrometry, 43, 13-21. 601
[26] Wahid, F. A. (2016), Characterising concrete using micro X-ray fluorescence, Imperial College 602
London, PhD thesis (in preparation). 603
[27] Wu, Z., Wong, H. S. & Buenfeld, N. R. (2015), Influence of drying-induced microcracking and related 604
size effects on mass transport properties of concrete, Cem. Concr. Res., 68, 35-48. 605
[28] Wong, H. S., Head, M. K. & Buenfeld, N. R. (2006), Pore segmentation of cement-based materials 606
from backscattered electron images. Cem. Concr. Res., 36, 1083-1090. 607
[29] Wong, H. S. & Buenfeld, N. R. (2006), Euclidean distance mapping for computing microstructural 608
gradients at interfaces in composite materials, Cem. Concr. Res., 36, 1091-1097. 609
Cement & Concrete Research, 2016 (Accepted)
13
[30] Wong, H. S., Pappas, A. M., Zimmerman, R. W. & Buenfeld, N. R. (2011), Effect of entrained air 610
voids on the microstructure and mass transport properties of concrete, Cem. Concr. Res., 41, 1067-611
1077. 612
[31] Buenfeld, N. R. & Okundi, E. (1998), Effect of cement content on transport in concrete, Magazine of 613
Concrete Research, 50, 339-351. 614
[32] Nokken, M., Boddy, A., Hooton, R. D. & Thomas, M. D. A. (2006), Time dependent diffusion in 615
concrete: three laboratory studies, Cem. Concr. Res., 36, 200-207. 616
[33] Yang, C. C. (2005), A comparison of transport properties for concrete using the ponding test and the 617
accelerated chloride migration test, Materials and Structures, 38, 313-320. 618
[34] Larbi, J.A. (1993), Microstructure of the interfacial transition zone around aggregate particles in 619
concrete, Heron, 38 (1) 5-69. 620
[35] Ollivier, J. P., Maso, J. C. & Bourdette, B. (1995), Interfacial transition zone in concrete, Advanced 621
Cement Based Materials, 2, 30-38. 622
[36] Scrivener, K. L., Bentur, A. & Pratt, P. L. (1988), Quantitative characterization of the transition zone 623
in high strength concretes, Advances in Cement Research, 1, 230-237. 624
[37] Scrivener, K. L., Crumbie, A. K. & Laugesen, P. (2004), The interfacial transition zone (ITZ) between 625
cement paste and aggregate in concrete, Interface Science, 12, 411-421. 626
[38] Glass, G.K., Yang, R., Dickhaus, T., Buenfeld, N.R. (2001), Backscattered electron imaging of the 627
steel-concrete interface, Corrosion Science, 43 (4) 605-610. 628
[39] Horne, A.T., Richardson, I.G., Brydson, R.M.D. (2007), Quantitative analysis of the microstructure of 629
interfaces in steel reinforced concrete, Cem. Concr. Res., 37 (12) 1613-1623. 630
[40] Bentz, D.P., Garboczi, E.J. & Lagergren, E.S. (1998), Multi-scale microstructural modelling of 631
concrete diffusivity: Identification of significant variables, Cement, Concrete and Aggregates, 20, 129-632
139. 633
[41] Carcasses, M., Petit, J.Y. & Olliver, J.P. (1998), Gas permeability of mortars in relation with the 634
microstructure of interfacial transition zone (ITZ), in: A. Katz, A. Bentur. M. Alexander, G. Arliguie 635
(Eds.), Interfacial Transition Zone in Cementitious Composites, RILEM Proc 35, E & FN Spon, 85-636
92. 637
[42] Delagrave, A., Marchand, J. & Pigeon, M. (1998), Influence of microstructure on tritiated water 638
diffusivity of mortars, Adv. Cem. Based Mater., 7, 60-65. 639
[43] Wong, H. S., Zobel, M., Buenfeld, N. R. & Zimmerman, R. W. (2009), Influence of the interfacial 640
transition zone and microcracking on the diffusivity, permeability and sorptivity of cement-based 641
materials after drying, Mag. Concr. Res., 61, 571-589. 642
[44] Abyaneh, S. D., Wong, H. S. & Buenfeld, N. R. (2013), Modelling the diffusivity of mortar and 643
concrete using a three-dimensional mesostructure with several aggregate shapes, Computational 644
Materials Science, 78, 63-73. 645
[45] BS EN 13670: 2009, Execution of concrete structures, British Standards Institution. 646
[46] BS 4482: 2005, Steel wire for the reinforcement of concrete products - specification, British Standards 647
Institution. 648
649
650
651
652
653
654
655
Cement & Concrete Research, 2016 (Accepted)
14
656
657
Table 1 Properties of spacers used in this study 658
Spacer
ID*
Cover size
(mm)
Porosity
(%)
Surface area
(m2)
Vol. fraction
(%) ** Notes
CS-25 25 15.9 0.00472 25.6 Cementitious spacer made of polypropylene
fibre-reinforced Portland cement mortar
containing 50% GGBS replacement at 0.35
w/cm ratio. With centre hole for tie wire. CS-50 50 16.2 0.00963 25.1
PS-25 25 0 0.00165 13.2 Clip-on “A” spacer made of recycled plastic.
Smooth surface. PS-50 50 0 0.00256 14.1
PS-50 (a) 50 0 (0.00256) 14.1 Clip-on “A” spacer made of recycled plastic.
Surface roughened by grinding with 120-grit
SiC paper (a) or by scoring notches on the
main flange (b). PS-50 (b) 50 0 (0.00256) 14.1
SS-50 50 0 0.00041 6.1
Continuous lattice steel wire chairs made of 5
mm diameter cold reduced wire conforming
to BS 4482:2005 [46]. Sectioned to 100 mm
segments.
* CS = cementitious spacer, PS = plastic spacer, SS = steel spacer. 659
** Expressed as volume fraction of the test sample (100 × 50 mm or 100 × 25 mm discs). 660
661
662
Table 2 Specific gravity at saturated-surface dry condition (SSD), moisture content and absorption of 663
the aggregates used. 664
Aggregate type MSA (mm) Specific gravity
(SSD) Moisture content (%)
24-hr moisture
absorption (%)
Sand (5.0) 5 2.60 0.25 0.62
Gravel (10) 10 2.51 0.40-0.43 0.82-0.86
Gravel (20) 20 2.56 0.41-0.46 0.80-0.84
665
666
Table 3 Concrete mix proportions. 667
Mix ID Cement
(kg/m3)
Water
(kg/m3) Free w/c
MSA
(mm)
Sand
(kg/m3)
Gravel
(kg/m3)
Aggregate vol.
fraction (%)
C10 413 172 0.4 10 726 1090 70
C20 413 172 0.4 20 733 1098 70
Cement & Concrete Research, 2016 (Accepted)
15
668
Table 4 Summary of the cylindrical test samples (100 mm). 669
Series Sample ID Mix Spacer Thickness
(mm)
Curing
(days) Conditioning Type
I C10-Co25
C10
None 25
3 & 28
(sealed) 20°C, 55% RH
C10-Co50 None 50
C10-CS25 CS-25 25
C10-CS50 CS-50 50 Cast
C10-PS25 PS-25 25
C10-PS50 PS-50 50
C10-SS50 SS-50 50
II C10-Co50
C10
None
50 28
(sealed)
20°C, 75% RH
20°C, 55% RH
50°C oven
Cast
C10-CS50 CS-50
C10-PS50 PS-50
C10-PS50a PS-50a
C10-PS50b PS-50b
C10-SS50 SS-50
III C20-Co50
C20
None
50 28
(sealed)
20°C, 75% RH
20°C, 55% RH
50°C oven
Cored C20-CS50 CS-50
C20-PS50 PS-50
C20-SS50 SS-50
IV C10-Co50
C10
None
50 28
(100% RH) 20°C, 55% RH Cast C10-CS50 CS-50
C10-PS50 PS-50
670
Table 5 Chloride diffusion coefficient (Deff) and surface chloride concentration (Cs) obtained from 671
fitting chloride profiles to Fick’s second law. 672
Sample ID Spacer Location Cs (%) Deff (×10-12 m2/s)
C10-Co50 None - 0.34 (0.071) 12.6 (0.12)
C10-PS50 Plastic Interface 1.03 (0.057) 16.8 (0.18)
C10-PS50 Plastic Centre 0.76 (0.062) 14.4 (0.31)
C10-PS50 Plastic Edge 0.35 (0.091) 12.7 (0.89)
C10-CS50 Cementitious Interface 0.69 (0.041) 14.1 (0.22)
C10-CS50 Cementitious Centre 0.29 (0.059) 11.7 (0.34)
C10-CS50 Cementitious Edge 0.33 (0.088) 12.7 (0.94)
673
Cement & Concrete Research, 2016 (Accepted)
16
Figure 1 Typical examples of spacers used in reinforced concrete: a) plastic clip-on “A” shaped, b) 674
plastic wheel/circular, c) plastic tower, d) cementitious single spacer, e) cementitious line/bar spacer, f) 675
steel wire chairs and g) steel wire continuous lattices. 676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
(a)
(b)
(c)
(d)
(e)
(f)
(g)
PS-50
PS-25
PS-50a
PS-50b
CS-50
CS-25
SS-50
Cement & Concrete Research, 2016 (Accepted)
17
Figure 2 Cross-section of reinforced concrete showing the placement of steel reinforcement on a 694
cementitious spacer and plastic clip-on “A” spacer to achieve 50 mm cover. Aggressive species (e.g. 695
water, oxygen and chloride) may penetrate through the porous hardened cement paste (A), concrete-696
spacer interface (B) or the porous spacer (C). 697
698
699
700
701
702
Rebar
Cementitious
spacer
10 mm
(Exposed surface) (H2O, O2, Cl2 etc)
A C B
Cover
depth
Rebar
Plastic spacer
10 mm
(Exposed surface) (H2O, O2, Cl2 etc)
C B
Cover
depth
A
Cement & Concrete Research, 2016 (Accepted)
18
Figure 3 Particle size distribution of fine and coarse aggregate (MSA = maximum size of aggregate). 703
704
705
Figure 4 Set-up for preparing cylindrical test samples of 100 mm diameter containing various types of 706
spacers. 707
708
709
710
711
712
713
714
715
0
10
20
30
40
50
60
70
80
90
100
0.01 0.1 1 10 100
Cu
mu
lati
ve
pas
sin
g (%
)
Sieve size (mm)
Sand (MSA:5)
Gravel (MSA: 10)
Gravel (MSA: 20)
Rebar
Cementitious
spacer
Plastic
spacer
Control sample
(no spacer)
Rebar
Steel spacer
Timber
Timber
Cement & Concrete Research, 2016 (Accepted)
19
(a)
(b)
Figure 5 Cross-section of samples containing a) cementitious spacer (CS-50) and b) plastic spacer (PS-716
50). Chloride concentration profiles within the spacer, along the spacer-concrete interface and near 717
the sample edge away from spacer, were measured using micro X-ray fluorescence. Locations of the 718
spot analyses are indicated with crosses (). 719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
Inte
rface
spo
ts
Ed
ge sp
ots
Cen
tre
spo
tsCen
tre sp
ots
Inte
rface
spo
ts
Ed
ge sp
ots
10 mm
10 mm 10 mm
(Surface exposed to chloride) (Surface exposed to chloride)
Cement & Concrete Research, 2016 (Accepted)
20
a) 3-day cured
b) 28-day cured
Figure 6 Effect of spacer type and height on oxygen diffusivity, oxygen permeability and water 741
sorptivity of samples from Series I after curing for 3 and 28 days. Samples were conditioned at 20°C, 742
55% RH prior to transport testing. Notation: Co = control (no spacer), SS = steel spacer, CS = 743
cementitious spacer, PS = plastic spacer. 744
745
746
747
748
749
750
751
752
753
3
4
5
6
7
Co SS CS PS
O2
Dif
fust
ivit
y (x
10
-8m
2 /s)
Sample type
25mm
50mm
0.0
0.4
0.8
1.2
1.6
2.0
Co SS CS PS
O2
Pe
rme
abili
ty
(x1
0-1
6m
2 )
Sample type
25mm
50mm
50
60
70
80
90
100
Co SS CS PS
Sorp
itvi
ty
(g/m
2 .m
in0.
5)
Sampe type
25mm
50mm
2
3
4
5
Co SS CS PS
O2
Dif
fust
ivit
y (x
10
-8m
2/s
)
Sample type
25mm
50mm
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Co SS CS PS
O2
Pe
rme
abili
ty
(x1
0-1
6m
2 )
Sample type
25mm
50mm
50
55
60
65
70
75
80
85
Co SS CS PS
Sorp
itvi
ty
(g/m
2 .m
in0
.5)
Sample type
25mm
50mm
Cement & Concrete Research, 2016 (Accepted)
21
754
Figure 7 Effect of spacer type and drying regime on oxygen diffusivity, oxygen permeability and water 755
sorptivity of concretes from Series II (10 mm maximum aggregate size, 50 mm thick, 28 day cured). 756
757
0
2
4
6
8
10
12
14
16
18
Co SS CS PS PS(a) PS(b)
O2
dif
fust
ivit
y (x
10
-8m
2 /s)
Sample type
20 C 75% RH
20 C 55% RH
50 C
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Co SS CS PS PS(a) PS(b)
O2
pe
rme
abili
ty (
x 1
0-1
6m
2 )
Sample type
20 C 75% RH
20 C 55% RH
50 C
50
60
70
80
90
100
110
Co SS CS PS PS(a) PS(b)
Sorp
itvi
ty
(g/m
2 .m
in0
.5)
Sample type
20 C 75% RH
20 C 55% RH
50 C
Cement & Concrete Research, 2016 (Accepted)
22
Figure 8 Effect of spacer type and drying regime on oxygen diffusivity, oxygen permeability and water 758
sorptivity of concretes from Series III (20 mm maximum aggregate size, 50 mm thick, 28 day cured). 759
760
761
762
a) Plastic spacer
b) Cementitious spacer
Figure 9 Chloride penetration profiles of samples from Series IV measured within the spacer (centre), 763
along the spacer-concrete interface and near the edge of sample at 1 mm spacing as shown in Fig. 5. 764
Results are compared to the control sample (no spacer). 765
766
767
768
769
770
0
2
4
6
8
10
12
14
16
18
Co SS CS PS
O2
dif
fust
ivit
y (x
10
-8m
2 /s)
Sample type
20 C 75% RH
20 C 55% RH
50 C
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Co SS CS PS
O2
pe
rme
abili
ty (
x10
-16
m2 )
Sample type
20 C 75% RH
20 C 55% RH
50 C
50
60
70
80
90
100
110
120
130
140
Co SS CS PS
Sorp
itvi
ty
(g/m
2.m
in0.
5)
Sample type
20 C 75% RH
20 C 55% RH
50 C
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20 25 30 35 40 45 50
% C
l b
y m
as
s o
f c
em
en
t
Depth (mm)
C10-PS50-Interface
C10-PS50-Edge
C10-PS50-Centre
C10-Co50
0.0
0.1
0.2
0.3
0.4
0.5
0 5 10 15 20 25 30 35 40 45 50
% C
l b
y m
as
s o
f c
em
en
t
Depth (mm)
C10-CS50-Interface
C10-CS50-Edge
C10-CS50-Centre
C10-Co50
Cement & Concrete Research, 2016 (Accepted)
23
a) Control sample (no spacer)
b) Sample with steel spacer
c) Sample with cementitious spacer
d) Sample with plastic spacer
Figure 10 Cross-sections of fluorescent epoxy impregnated samples imaged under UV light and their 771
respective binary images showing the epoxy intruded areas. Samples are a) C10-Co50-28d-50°C, b) 772
C10-SS50-28d-50°C, c) C10-CS50-28d-50°C and d) C10-PS50-28d-50°C. 773
774
775
776
777
778
779
10 mm
10 mm
10 mm
10 mm
Cement & Concrete Research, 2016 (Accepted)
24
(a)
(b)
Figure 11 Effect of spacer and conditioning regime on a) area fraction and b) maximum depth of 780
epoxy penetration. Samples are from Series II & III. 781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
2
3
4
5
6
7
8
Co SS CS PS
Epo
xy im
pre
nat
ed
are
a (
%)
Sample type
20 C 75% RH (C10)
20 C 55% RH (C10)
20 C 55% RH (C20)
50 C (C10)
0
10
20
30
40
50
Co SS CS PS
Max
imp
regn
atio
n d
ep
th (
mm
)
Sample type
20 C 75% RH (C10)
20 C 55% RH (C10)
20 C 55% RH (C20)
50 C (C10)
Cement & Concrete Research, 2016 (Accepted)
25
(a)
(b)
(c)
(d)
(e)
Figure 12 Effect of spacers on the average distribution of unreacted cement (a) and porosity (b) from 806
the spacer-concrete interface. Results are compared to that of the aggregate-paste interface (ITZ) and 807
bulk paste of the control sample. Figures (c) to (d) show example BSE images of very porous spacer-808
concrete interface. Samples are from Series II & III that were 28-day cured and conditioned at 20°C, 809
55%RH. 810
811
812
813
814
0
2
4
6
8
10
12
14
16
0 10 20 30 40 50 60 70 80
An
hyd
rou
s ce
me
nt
(%
)
Distance from interface (μm)
CS-50
PS-50
SS-50
Co (bulk paste)
Co (Aggregate ITZ)0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80
De
tect
able
po
rosi
ty (
%)
Distance from interface (μm)
CS-50
PS-50
SS-50
Co (bulk paste)
Co (Aggregate ITZ)
50 μm 50 μm 50 μm
Pla
stic
sp
ace
r
Cem
enti
tio
us
spa
cer
Ste
el s
pa
cer