Professional Documents
Culture Documents
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: Pervious concrete (PC) mixtures were designed and prepared to measure and study permeability charac-
Received 17 April 2016 teristics at varying head levels using a falling head permeameter. A total of 1092 readings was used to
Received in revised form 3 June 2016 study the permeability properties of eighteen PC mixtures whose porosity was in the range of 1537%,
Accepted 14 July 2016
and permeability in the realm of 0.0763.5 cm/s. The permeability reduced as the head of water
increased, and gradually attained an asymptotic relation with the head. Cement-to-aggregate ratio had
largest contribution in controlling permeability of PC mixtures. Nonlinearity in Darcys law was observed
Keywords:
in respect of permeability of PC mixes, which was modelled using Izbash/power law, and was prominent
Pervious concrete
Permeability
for gradations consisting of larger sized aggregates due to inconsequential tortuous pore structure.
Darcys law Modified Kozeny-Carman equation was fitted for PC gradations to compare the results with Izbash
Izbash law law, which showed good agreement. This study is deemed to assist in understanding the hydrodynamics
Taguchi design of experiments of water flow in pervious concrete, which in turn will aid in rational pervious concrete pavement system
Kozeny-Carman equation designs.
2016 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2016.07.035
0950-0618/ 2016 Elsevier Ltd. All rights reserved.
628 A.K. Chandrappa, K.P. Biligiri / Construction and Building Materials 123 (2016) 627637
been studied using image analyses techniques where the median falling head permeameter, and utilize existing theories/laws in
pore size was found to be 34 mm and mostly followed a normal explaining the features of permeability. The research methodology
distribution [11]. Tortuosity (a dimensionless quantity), which rep- included (Fig. 1):
resents the sinuousness (property indicating twisting and windings
of pores) of internal pore structure was found to be in the range Design and production of eighteen PC mixtures with 6 grada-
of 25 as investigated using X-ray tomography (XRT) [12]. tions and 3 levels of water-to-cement ratio (w/c ratio) and
The permeability characteristics of PC mixes as one of the cement-to-aggregate ratio (c/a ratio)
important properties have been investigated in several studies Determination of permeability of PC mixtures with seven differ-
using either falling head permeameter and/or constant head per- ent head levels
meameter. Permeability being a function of the aforementioned Assessment of the contribution of different mixture variables
pore properties varies in the range of 0.15 cm/s, and is known and relationships between permeability and porosity
to follow either a normal or Weibull distribution depending on Understanding the nonlinearity in different PC mixtures using
the range of permeability values [7,8,12,14]. Furthermore, perme- Darcys law and Kozeny-Carman equation
ability of the in-service PCP is usually determined using ASTM Recommendation of a novel methodology to investigate the
C1701 [15] or the method prescribed by the National Center for hydrodynamics of water flow in porous mixtures which is
Asphalt Technology (NCAT) [16,17]. Although several studies have deemed to assist in rational pervious concrete pavement system
been carried out to determine permeability of PC specimens, very designs
few studies have discussed about the effect of head/hydraulic gra-
dient on the permeability properties of these mixes. Qin et al. [18] 2. Theoretical background
discussed the differences in permeability measured with falling
head and constant head permeameter, and found that those values The flow of the fluid in the porous media is governed by the well
measured with falling head method were normally lower than accepted Darcys law. According to the Darcys law, the flow rate
those obtained from constant head method. Recently, West et al. through a porous media is proportional to the pressure difference/
investigated the effect of head/hydraulic gradient on the Darcys head difference between the two points, and the proportionality
velocity of PC specimens using constant head permeability test constant is called hydraulic conductivity as shown in Eq. (1) [20].
method [19]. The study showed that there existed a nonlinearity
Dh
in the Darcys law that can be modelled using Forcheimmers/ Q K A KiA 1
L
quadratic equation. However, the study considered one type of
PC mixture without giving due attention to the variations in pore where
structure over different types of PC mixtures. Q = flow rate, m3/s
In order to study whether nonlinearity exists in most of the PC K = hydraulic conductivity/permeability, m/s
mixtures, and if it exists to what degree will it be present is yet to Dh = head difference, m
be understood. Further, how this nonlinearity arises and gets influ- L = length of the permeable media, m
enced by the different gradations in PC mixtures needs to be A = cross-sectional area, m2
answered. Therefore, this study investigated the research gap by i = hydraulic gradient
considering a wide variety of PC mixtures with different pore
structure features, which will throw more light on the understand- The Darcys law neglected the inertial and velocity effects;
ing of the hydrodynamics of fluid flow in pervious concrete however, these effects become more pronounced when the
mixtures. Thus, the main objective of this research study was to connected pores in the porous structure turn out to be larger in
investigate the permeability characteristics of PC mixtures using size. The change in the fineness of the porous structure gives rise
Density
Determination of basic properties of
PC specimens (ASTM C1754)
Porosity
to turbulence in the flow of the fluid where the inertial effect ing the hydrodynamics of fluid flow in different PC mixtures, and
(change in momentum) predominates. Owing to this turbulence, would certainly help in the futuristic development of rational
the linear relation between flow rate and the hydraulic gradient thickness designs/procedures pertaining to pervious concrete
is no longer valid, and thus, there exists a nonlinear relation. In pavement systems.
order to take into account the nonlinearity, several forms of
equations were proposed among which, the Forcheimmers 4. Materials and methods
equation and Izbash law/power law are well accepted [21,22].
These equations are shown in Eqs. (2) and (3), respectively. 4.1. Materials
i av bv The aggregates used in this study included four sizes of coarse aggregates with
2
2
maximum diameter of: 6.7, 9.5, 13.2, and 19 mm whose properties were measured
as per IS 2386 [24]. The specific gravity of the aggregates was in the range of 2.72.8
i av b 3 and dry-rodded unit weight in the realm of 1.51.6 kg/l. The air voids in the dry
aggregates were found to be in the range of 4144%. Further, the Los Angeles value
where for aggregates was found to be between 15 and 22% [14]. Ordinary Portland cement
i = hydraulic gradient 53-grade conforming to standard [25] was used as a binding agent. A commercially
v = Darcys velocity, m/s available ligno-sulphate based plasticizer was added at a rate of 0.2% by weight of
cement with gauged water to increase the workability of the fresh PC mixture.
a; b = coefficients
4.2. Experimental matrix
The coefficients a and b are used to interpret the flow regime in
the porous material, which can be laminar, transitional or turbu- In this study, six gradations were used denoted as P1, P2, P3, P4, P5, and P6 that
had individual compositions as shown in Table 1, and the gradation curves are
lent. For example, in Eq. (3), if b = 1, then it reduces to Darcys
shown in Fig. 2. Three levels of water-to-cement ratios by weight (w/c ratio):
law indicating laminar flow concept. Further, in order to study 0.25, 0.3, and 0.35, and three levels of cement-to-aggregate ratios by weight
the influence of the pore sizes and pore structure on permeability (c/a ratio): 0.33, 0.25, and 0.2 were used in order to cover the entire range of mix
properties, Kozeny proposed an equation to predict permeability as variables used in the previous studies [7,14,17]. The full factorial design for the
a function of porosity and specific surface area, which was further above mentioned factors and its levels resulted in 162 PC specimens (6 gradations
x 3 w/c ratio x 3 c/a ratio x 3 replicates). In order to optimize the total number of
modified by Carman as shown in Eq. (4). specimens, Taguchi method of experimental design was applied in this study
3 [26,27]. Taguchi design of experiments relies on specially constructed tables known
; 1
K 4
1 ;
2
C s 2 S2 Table 1
Gradations used in PC mixtures.
where
Gradation ID Aggregate size, mm Gradation parameters
K = hydraulic conductivity/permeability, m/s
* ** ***
; = porosity, % 6.74.75 9.56.7 13.29.5 1913.2 D60, mm D10, mm Cu
90
3. Research significance
80
Pervious concrete is a class of sustainable material whose
Cumulative percent passing, %
Table 2
Taguchi experimental matrix with mix proportions.
Gradation pairs Gradation w/c c/a ratio Aggregate Cement Water, Oven dry density, Porosity,
to be compared ratio kg/m3 kg/m3 kg/m3 kg/m3 [28] (CV%) % [28] (CV%)
P1 0.25 0.33 1462.35 487.45 121.86 1791.66 (0.55) 36.75 (1.80)
P1 0.3 0.25 1561.66 390.41 117.12 1860.20 (0.99) 33.39 (2.86)
P1 0.35 0.2 1627.99 325.59 113.96 1839.83 (0.36) 33.68 (1.57)
P2 0.25 0.33 1445.47 481.82 120.45 1825.61 (0.90) 35.33 (1.90)
P2 0.3 0.25 1542.42 385.6 115.68 1869.54 (0.33) 33.36 (1.00)
P2 0.35 0.2 1607.1 321.42 112.49 1882.48 (1.60) 32.30 (3.31)
P3 0.25 0.25 1598.66 399.66 99.91 1866.88 (0.86) 34.07 (1.90)
P3 0.3 0.2 1659.6 331.92 99.57 1833.67 (1.07) 35.08 (2.16)
P3 0.35 0.33 1376.84 458.94 160.63 2158.88 (0.88) 17.96 (2.92)
P4 0.25 0.2 1685.64 337.12 84.28 1769.37 (0.14) 37.48 (0.58)
P4 0.3 0.33 1411.12 470.37 141.11 2066.89 (0.73) 23.05 (5.82)
P4 0.35 0.25 1515.23 378.8 132.58 1989.43 (1.04) 26.87 (4.58)
P5 0.25 0.25 1594.6 398.65 99.66 1919.40 (0.27) 32.15 (1.31)
P5 0.3 0.2 1655.21 331.04 99.31 1889.06 (1.32) 32.85 (3.60)
P5 0.35 0.33 1373.82 457.94 160.27 2245.57 (1.49) 13.48 (10.51)
P6 0.25 0.2 1691.67 338.33 84.58 2164.71 (0.87) 17.76 (3.99)
P6 0.3 0.33 1415.34 471.78 141.53 2075.83 (0.17) 22.09 (1.01)
P6 0.35 0.25 1520.11 380.02 133 1870.60 (0.17) 33.96 (0.54)
Total mixtures 18
Total samples 54
as orthogonal arrays. These arrays are two-way balanced arrays where within a 4.4. Measurement of permeability
column there exists an equal number of levels of factors and between any two
columns, there exists equal number of all possible combinations. Permeability of PC specimens was measured using falling head permeameter
A mixed factor level Taguchi design of experiments was used where the number shown in Fig. 4. Fig. 4(a) shows the schematic of the permeameter consisting of a
of mixtures for each gradation was reduced to three with a total of nine specimens graduated tube and the PC specimen wrapped with a duct tape. Fig. 4(b) shows
per gradation. The detailed experimental matrix and mix proportions are provided the actual testing where the graduated tube was filled with water, and the PC spec-
in Table 2. imen wrapped with a duct tape to prevent water from flowing off the sides. In order
to prevent water leakage from the junction between the tube and the specimen,
commercially available silicone-acetoxy gel was applied. The specimen was initially
4.3. Measurement of density and porosity saturated by allowing water to pass through it and alternatively switching the valve
on and off until no visible air voids were observed. This step was necessary to
PC specimens of dimensions: 100 mm diameter and 200 mm height were pre- ensure that all the pores in PC specimens were filled with water prior to measuring
pared and cured for 28 days before testing for its properties. A detailed description the permeability.
of sample preparation along with the compaction method can be found elsewhere Permeability test was carried out at different initial head levels, including: 16,
[14]. The density and porosity of all PC specimens were determined according to 21, 26, 31, 39, 44, and 49 cm that corresponded to hydraulic gradients of 0.65,
[28] and their relation is shown in Fig. 3. The oven dried weight was measured after 0.9, 1.15, 1.40, 1.80, 2.05, and 2.30, respectively. For each PC specimen, three
the specimens were conditioned at 50 C for about a week until a constant weight readings of time interval were determined so as to obtain average and standard
was observed between successive cycles. Following this, the PC specimens were deviation of the permeability magnitudes. The water heads were mainly varied to
submerged in water for 12 h. Later, the specimens were tapped on the periphery check the validity of Darcys law, determine the value of saturated permeability
using a rubber mallet and hand for about half-to one-hour consistently until no in different PC mixtures, and study the hydrodynamics of water in PC specimens.
further air voids were observed. The submerged weight of the samples was
determined to calculate the porosity of the PC specimens.
5. Results & analyses
Graduated tube
With water
Graduated tube
PC specimen
Silicone gel
PC specimen
On/Off valve
Silicone gel
(a) (b)
Fig. 4. Permeability test setup for pervious concrete: (a) Schematic; (b) Actual setup.
heads would be rational for the design purpose, which may result (considered within this study) increased, the degree of lubrication
in higher thicknesses but will perform better both as structural and in the pervious concrete mixtures increased, which assisted in
functional layers. Note that a plot of individual values of perme- better densification of the mixture, and hence resulted in
ability versus porosity also exhibited an R2 > 0.88. lower permeability. At lower w/c ratio (i.e., 0.25), the reduced
water content rendered the mix to be harsh that provided
5.2. Effect of mix variables higher resistance for compaction resulting in higher porosity
and permeability.
5.2.1. Influence of gradation Contemporaneously, as c/a ratio increased, the inter-granular
Fig. 6 shows the effect of gradations on the measured perme- void volume reduced due to decreased resistance in compaction
ability, and the Fig. 7 shows the plot of gradation and void content. offered by the reduced amount of aggregates. This resulted in
In Fig. 6(a) that compares two single sized gradations, P1 gradation lower porosity and permeability as shown in Fig. 8(b). However,
made up of 6.74.75 mm sized aggregates had slightly higher per- at lower c/a ratio, reduced coating of the cement paste
meability compared to P2 gradation consisting of 6.79.5 mm and increased resistance for compaction given by larger volume
aggregates. For given w/c ratio and c/a ratio, in P1 gradation, the of aggregates provided higher porosity and permeability of
increased surface area requires more cement paste to coat the the PC mixtures. It has to be noted that the permeability values
aggregates while a small percentage will fill the inter-granular obtained at different w/c and c/a ratios are functions of
voids. But, in the P2 gradation, relatively a small amount of cement compaction method, and magnitudes may change based on the
paste is sufficient to coat the aggregates while larger amounts fill compaction method. However, the trends with respect to w/c
the inter-granular void spaces. Further, it can also be interpreted ratio and c/a ratio are expected to be same as also found in this
that for a given surface area, the degree of interlocking is higher study.
in the case of smaller aggregates that offers more resistance for
compaction resulting in the internal structure with higher amount 5.2.3. Relative contribution of mix variables
of voids. In order to study the relative contribution of each mix vari-
Fig. 6(b) compares one single sized aggregate gradation (9.5 able in defining permeability, ANOVA test was carried out
13.2 mm) with one dense graded pervious concrete mixture. In [26,27]. The relative contribution in percent was found as the
P3 gradation, permeability is higher due to the increased porosity ratio of the sum of squares of a particular mix variable to the
and pore size compared to P5 gradation, which has better packing total sum of squares, as shown in Table 3. In Table 3, df stands
of aggregates resulting in lower permeability. Lastly, Fig. 6(c) com- for degree of freedom; SS stands for sum of squares; variance is
pares two dense graded pervious concrete mixtures with P4 and P6 obtained as the ratio of sum of squares and degree of freedom;
gradations. In P4 gradation, which is a combination of 4.75 F is the F-statistic obtained by dividing the variance with
6.7 mm and 6.79.5 mm in equal proportions, permeability is respect to each variable by error variance; and Fcrit is the crit-
higher due to increased porosity compared to P6 gradation ical value of F obtained from the standard statistics tables. As
(Fig. 7). The voids created between the 6.7 mm aggregates might observed, c/a ratio had the highest contribution in defining
have been too small to accommodate 4.75 mm aggregates result- the variability in permeability followed by w/c ratio, and then
ing in higher permeability compared to P6 gradation, which has by gradation. The contribution of c/a ratio was approximately
a better size distribution. twice the w/c ratio, and 3.5 times that of the gradation. This
behaviour in pervious concrete can be mainly attributed to
5.2.2. Influence of w/c ratio and c/a ratio the absence of fine aggregates where the interlocking between
The effect of w/c ratio and c/a ratio on measured permeability the coarse aggregates controls/dominates most of the other
is shown in Figs. 8(a) and (b), respectively. As the w/c ratio properties.
632 A.K. Chandrappa, K.P. Biligiri / Construction and Building Materials 123 (2016) 627637
R = 0.925
3.50 3.50 3.50
R = 0.929 R = 0.930
Average Permeability, cm/s
5.00
H49
4.50
4.00
y = 7E-06x3.518
R = 0.931
Average Permeability, cm/s
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
(g) Average Porosity, %
Fig. 5. Influence of head on the measured permeability: Initial head of 16 cm vs. Porosity; (b) Initial head of 21 cm vs. Porosity; (c) Initial head of 26 cm vs. Porosity; (d) Initial
head of 31 cm vs. Porosity; (e) Initial head of 39 cm vs. Porosity; (f) Initial head of 44 cm vs. Porosity; (g) Initial head of 49 cm vs. Porosity.
5.3. Effect of hydraulic gradient on permeability select the best fit. It was found that amongs the alternatives; the
power law best fitted the relation as shown in Fig. 9. Each point
5.3.1. Validity of Darcys law in Fig. 9 represents an average of 27 data points since nine different
One of the main purposes of this study was to understand the specimens were prepared for each gradation with three permeabil-
hydrodynamics of water in pervious concrete, and check the valid- ity measurements at each hydraulic gradient. The Figure also
ity of the well-established Darcys law in a wide range of pervious shows that the relation between permeability and hydraulic gradi-
concrete mixtures. In order to accomplish this task, permeability ent is no longer linear, and follows Power law/Izbash law. The
test was carried out at different heads as discussed in the previous relation between permeability and the hydraulic gradient is as
section. For each gradation; linear, quadratic, and power curves shown in Eq. (6).
were fitted to the relationships between permeability and hydrau-
c
lic gradient, and checked against the goodness-of-fit statistics to K ai 6
A.K. Chandrappa, K.P. Biligiri / Construction and Building Materials 123 (2016) 627637 633
2.5
P1 P2
1.5
1.0
0.5
(a) 0.0
0.65 0.9 1.15 1.4 1.8 2.05 2.3
Hydraulic gradient
2.5
P3 P5
Average Permeability, cm/s
2.0
1.5
1.0
0.5
(b) 0.0
0.65 0.9 1.15 1.4 1.8 2.05 2.3
Hydraulic gradient
2.5
P4 P6
Average Permeability, cm/s
2.0
1.5
1.0
0.5
(c) 0.0
0.65 0.9 1.15 1.4 1.8 2.05 2.3
Hydraulic gradient
Fig. 6. Effect of gradation on the measured permeability: (a) comparison of single sized gradations; (b) comparison of single and dense gradation; (c) comparison of dense
gradations.
Now, substituting Eq. (6) in Darcys law and rearranging, the gradation, which was mainly due to the presence of smaller sized
resulting equation will be of the form as presented in Eq. (7): and sinuous pores (tortuous) where the inertial effects were
reduced. This reduced inertial effect will lead to greater dissipation
b
V ai 7 of momentum of water, thus shifting the flow regime more
towards the laminar region indicated by the Izbashs coefficient.
The coefficient b in Eq. (7) indicates the flow regime in the
It can be inferred that for a combination of larger pore size along
material as discussed in Section 2. The coefficient was determined
with decreased sinuousness, flow gains larger momentum giving
for each gradation and tabulated in Table 4. According to [29,30] as
rise to turbulence in the flow of water through PC mixtures.
the coefficient b approaches 0.5, the flow is turbulent and similar
to those observed in rough walled tubes. It was observed that gra-
dations considered in this study had coefficients in the range of 5.3.2. Fitting the Kozeny-Carman (K-C) equation
0.75 and 0.8 indicating the flow is in the transitional regime. P3 The K-C equation that gives a relation between permeability
gradation had the lowest value of coefficient indicating that flow and pore structure features was fitted using nonlinear regression.
was relatively turbulent that is mainly due to the presence of a Since tortuosity and specific surface area were not measured, they
combination of large sized and less amount of sinuous pores as were represented by the parameter A as discussed in Section 2
seen in Fig. 10. The highest value of coefficient was found for P4 and interpreted as follows. The K-C equation for each gradation
634 A.K. Chandrappa, K.P. Biligiri / Construction and Building Materials 123 (2016) 627637
25
1913.2 mm) had higher values of permeability. This indicated
20 that in gradations with smaller sized aggregates, tortuosity and
specific surface area were higher since the parameter A is the
inverse of tortuosity and specific surface area. However, in larger
15
sized aggregate gradations, the pores are larger in size with
reduced tortuosity and the specific surface area, which resulted
10 in higher A values.
This justification was further assessed by cutting a few speci-
5 mens vertically and analyzing those scanned images. The images
were obtained after scanning the vertically cut samples where
the solid was painted in white in a flat-bed scanner at 300 dpi res-
0
P1 P2 P3 P4 P5 P6
olution. The obtained scanned images were initially smoothened
using a median filter with a radius of 2.0 using ImageJ application
Gradation
[31]. Further, the smoothened image was converted to 8-bit image,
Fig. 7. Gradation versus porosity for pervious concrete mixtures. which was then converted to binary images and illustrated in
Fig. 10. In smaller sized aggregate gradations (P1 and P2), the pores
were more tortuous and smaller in size that dissipate the momen-
3.0
tum of water. But in the case of larger sized aggregates, the pores
Average Permeability, cm/s
c/a = 0.33 c/a = 0.25 c/a = 0.25 cating that both the equations captured the features of the internal
2.5 pore structure in an indirect fashion but in a similar manner. If
there had been a very poor correlation, then the above explanation
2.0
should have been irrational.
1.5
1.0
Conclusions and recommendations
0.5
Pervious concrete is a sustainable pavement material that has a
0.0 wide range of environmental benefits and is a promising pavement
(b) H16 H21 H26 H31 H39 H44 H49
material (and system) in the era of increasing urbanization. The
Initial Head, cm
main objective of this study was to investigate the permeability
Fig. 8. Effect of w/c ratio and c/a ratio on measured permeability: (a) w/c ratio vs. characteristics of PC mixtures using falling head permeameter
Initial Head; (b) c/a ratio vs. Initial Head. and utilize the existing theories/laws in explaining the features
Table 3
Contribution of mix variables towards measured permeability.
3.0
P1 P2 P3 P4 P5 P6
2.5
Permeability, cm/s
2.0
1.5
1.0
0.5
y = 1.7344x-0.204 y = 1.6852x-0.217 y = 2.1648x-0.238 y = 1.2669x-0.2 y = 1.9539x-0.23 y = 1.1009x-0.217
R = 0.9998 R = 0.9986 R = 0.9995 R = 1 R = 0.9999 R = 0.9995
0.0
0 0.5 1 1.5 2 2.5
Hydraulic gradient
Fig. 9. Relation between permeability and hydraulic gradient for different PC mixtures.
Table 4 Table 5
Izbash b coefficients for different gradations. Kozeny-Carman equation developed for different gradations.
10.0
9.0 y = 0.305x + 1.094
0.78
0.77
0.76 y = -0.002x + 0.830
0.75 R = 0.783
0.74
0.73
0.72
0.71
0.70
(b) 10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0 26.0 28.0
A-parameter
Fig. 11. Correlations between: (a) weighted aggregate size vs. A parameter; (b) Izbash coefficient vs. A parameter.
[27] A. Joshaghani, A.A. Ramezanianpour, O. Ataei, A. Golroo, Optimizing pervious [29] C.R. Dudgeon, An Experimental Study of the Flow of Water through Coarse
concrete pavement mixture design by using the Taguchi method, Constr. Build. Granular Media, 7, La Houille Blanche, 1966.
Mater. 101 (2015) 317325. [30] J. Mulqueen, The flow of water through gravels, Irish J. Agric. Food Res. 44
[28] ASTM, Standard test method for density and void content of hardened (2005) 8394.
pervious concrete. ASTM C1754-12, American Society for Testing and [31] C. Rueden, C. Dietz, M. Horn, J. Schindelin, B. Northan, M. Berthold, K. Eliceiri,
Materials International, West Conshohocken, PA, USA, 2012. ImageJ Ops [Software]. <http://imagej.net/Ops>, 2015.