Construction and Building Materials 110 (2016) 8997

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Material characterization and hydraulic conductivity modeling

of macroporous recycled-aggregate pervious concrete
Patrick W. Barnhouse, Wil V. Srubar III
Department of Civil, Environmental, and Architectural Engineering, University of Colorado Boulder, ECOT 441 UCB 428, Boulder, CO 80309, USA

h i g h l i g h t s

Macroporous pervious concrete (MPC) with >30% voids was produced and characterized.

Compressive strength of MPC is independent of aggregate type (virgin vs. recycled).

Two binder additives, TiO2 and sand, increased compressive strength of MPC.

A modified CarmanKozeny model more accurately predicts pervious concrete behavior.

a r t i c l e i n f o a b s t r a c t

Article history: The objective of this work was to characterize the mechanical, physical, and hydraulic conductivity prop-
Received 23 September 2015 erties of macroporous (>30% voids) pervious concrete (MPC) containing recycled aggregates and two bin-
Received in revised form 16 December 2015 der additives, namely sand and titanium dioxide (TiO2). While MPC exhibited lower compressive strength
Accepted 2 February 2016
and higher permeability than normal pervious concrete, the experimental data suggest that the high
Available online 10 February 2016
absorption and low specific gravity of recycled aggregates did not compromise the mechanical properties
or permeability of MPC. A modification to the well-known CarmanKozeny hydraulic conductivity model
Keywords:
is proposed to more accurately predict the permeability of both normal and macroporous pervious
Pervious concrete
Recycled aggregate
concrete.
Mechanical properties 2016 Elsevier Ltd. All rights reserved.
Titanium dioxide
Permeability
Modeling

1. Introduction 1.1. Pervious concrete materials

The ability of pervious concrete to reduce the quantity and
improve the quality of stormwater runoff in urban environments
has been exploited in a wide variety of horizontal infrastructure
applications (e.g., pavements). Previous studies have shown that
pervious concrete can significantly reduce stormwater runoff vol-
ume through direct infiltration and can improve the quality of
stormwater by reducing the concentrations of heavy metals (e.g.,
lead, zinc, chromium, copper) that are typically found in stormwa-
ter runoff [1]. Pervious concrete has also been shown to lower the
risk of thermal pollution from urban stormwater and to help
mitigate the urban heat island effect [2], thus contributing to an
overall reduction in health risks to humans, plants, and wildlife
[36].
Corresponding author.
E-mail address: wsrubar@colorado.edu (W.V. Srubar III).
The constituent materials of pervious concrete, including coarse
aggregate, cement, water, and, occasionally, a small amount of fine
aggregate, govern its macroscopic properties. Water-reducing
admixtures are commonly used to maintain sufficient workability
for pervious concrete with low water-to-cement ratios (0.260.40)
[7]. The properties of coarse aggregates impact the performance of
pervious concrete. Kevern et al. [8] found that aggregates with high
absorption or low specific gravity yielded a pervious concrete with
low freezethaw durability. Another study indicated that a small
addition of sand (7% by weight) improved freezethaw resistance
[9]. The best-performing mixture for freezethaw durability
included No. 4 (4.75 mm) aggregates, a 7% by weight replacement
of coarse aggregate with sand, and an air entraining admixture [9].
In this same study, the addition of sand marginally reduced the
permeability of the pervious concrete. Unlike round aggregates,
angular aggregates were found to yield higher actual porosities
compared to the intended design porosities [8]. Finally, increases

http://dx.doi.org/10.1016/j.conbuildmat.2016.02.014
0950-0618/ 2016 Elsevier Ltd. All rights reserved.
90 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997

in bulk volume of aggregate, or the use of larger aggregates, were
also reported to increase the porosity and decrease the compres-
sive strength and modulus of elasticity of pervious concrete [10].
This loss in mechanical properties was attributed to a decrease in
paste content at higher aggregate volumes [10].
The effects of using of recycled concrete aggregates (RCAs) in
pervious concrete have also been investigated. According to the
Federal Highway Administration, RCA is primarily used for aggre-
gate road base. Currently, only eleven states permit the use of
RCA as an aggregate for new concrete projects [11]. While a few
studies have explored the use of RCA in pervious concrete, the body
of literature is not extensive. Nonetheless, multiple authors have
reported that the incorporation of RCA reduces the compressive
strength of normal pervious concrete [1214]. More specifically,
at a 100% replacement of RCA, Berry et al. [12], Gaedicke et al.
[13], and Rizvi et al. [14] have recorded compressive strength
losses of 16%, 15% and 43%, respectively.
In addition to sand, the incorporation of titanium dioxide (TiO2)
in pervious concrete has been studied because of the photocat-
alytic properties of TiO2. A number of studies have shown that
TiO2 can remove noxious pollutants from the environment through
a photo-induced redox reaction [15,16]. It has also been shown
that small additions of TiO2 improve the mechanical strengths of
cement mortar. Chen et al. [17] reported that additions of 5% and
10% TiO2 by weight of cement increased the 28-day strength of
mortar by approximately 1025%. This increase in strength has
been attributed to a nano-filler effect through which the TiO2
nanoparticles provide additional nucleation sites for cement
hydration, thereby increasing both the rate and degree of hydra-
tion [18].
1.2. Beyond horizontal infrastructure
While the vast majority of pervious concrete applications
to-date concern horizontal infrastructure, there is increased
interest in using pervious concrete in vertical infrastructure
applications (e.g., buildings). Historical examples of using pervious
concrete in building walls do exist [7], but these applications were
limited to a few isolated examples in post-World War II Europe.
More recently, pervious concrete has found application as
non-structural components in transportation and building
infrastructure, like sound barriers and reinforced insulation panels
applications that take advantage of its favorable acoustic and
thermal properties [19]. Additional new applications of pervious
concrete include underwater biological habitats that provide
support for a wide variety of marine organisms [20].
Non-structural applications of pervious concrete require differ-
ent properties to suit a wide range of material requirements. For
example, the requirements for durable pervious pavements (e.g.,
high strength, 20% porosity) may not be appropriate in other
applications, like sound barriers or biohabitats, where higher
porosities are favored. While the majority of pervious concrete
materials research completed to-date has been primarily limited
to pavement materials with porosities that range from 15% to
30%, there have been a few isolated studies that reported outlier
samples with porosities greater than 30% [21,22]. To the authors
knowledge, however, no deliberate research has yet been
conducted on pervious concrete with ultra-high porosities greater
than 30%.
incorporating recycled concrete aggregates (RCA) on the material
properties of MPC was also investigated. Prior to mixing, the
physical properties of the RCA were first characterized to better
understand the composition and inherent variability that exists
in reclaimed aggregates.
Due to ultra-high porosities and the incorporation of RCA, low
compressive strengths and higher sensitivities to field-condition
curing (rapid evaporation) were anticipated [12]. Thus, methods
to maintain or improve the mechanical performance of MPC were
explored. The need for an improved paste strength and paste-
aggregate interface in pervious concretes has been previously iden-
tified and achieved through small additions of fines [9]. Therefore,
this study analyzed the impact of both sand and TiO2 additions on
the mechanical, physical, and hydraulic properties of MPC. Mate-
rial property relationships between unit weight and compressive
strength and between porosity and permeability of MPC were com-
pared with the behavior of conventional pervious concrete (<30%
porosity). In addition, the suitability of a well-known hydraulic
conductivity model based on the CarmanKozeny relationship in
predicting the permeability-porosity relationship of MPC was
investigated herein. Finally, a modification to the model was pro-
posed to more accurately predict the hydraulic conductivity of
both normal and macroporous pervious concrete.
2. Materials and methods
2.1. Materials
Type I/II cement, virgin coarse aggregate (pea gravel), and sand was obtained
from a local hardware store. RCA, which consisted of crushed mortar, reclaimed vir-
gin aggregate, crushed masonry, and asphalt was acquired from Allied Recycled
Aggregates in Commerce City, Colorado USA. Both virgin coarse aggregate and
RCA were sieved to yield a uniform No. 4 (4.75 mm) particle size. Rutile TiO2 with
an average particle size of 0.41 lm was acquired from DuPont. MasterGlenium
3400 was obtained for use as a high-range water reducer.
2.2. Experimental methods
This study comprised three experimental phases. In the first phase, plain
cement and TiO2-cement blend mortar samples were formulated and cast to ana-
lyze the impact of TiO2 on the compressive strength at 3-, 7- and 28-days. In the
second phase, the RCA was characterized for composition, gradation, unit weight,
specific gravity, and absorption. In the third phase, a MPC mix was designed and
cylinders were cast to study the impact of RCA, sand, and TiO2 on the mechanical,
physical, and hydraulic properties of MPC.
2.2.1. Mortar preparation and mechanical characterization
Three (3) sets of six (6) mortar cube specimens were cast for each of five (5)
sample formulations according to ASTM C109 [23]. Each set was used for 3-, 7-,
and 28-day compressive strength tests. Table 1 shows the mix designs of the five
(5) sample formulations. The Control sample was proportioned according to ASTM
C109 for portland cement mortars. The remaining four mortar mixes were modified
with binder additives (either cement, C, or TiO2, T). The number (2.5 or 5) in
the mix nomenclature indicates the percent addition of additional cement or TiO2
by weight of cement. Given that both cement and TiO2 additions were treated as
binder components, a total water-to-binder (w/b) ratio is reported in lieu of a typ-
ical water-to-cement (w/c) ratio.
To simulate worst-case field conditions, mortar specimens were exposed to
ambient air during the curing process. One additional set of Control specimens
was cast and placed in a curing room in order to compare ideal conditions to the
simulated field conditions. This additional set of Control specimens was cured for
28 days prior to mechanical testing. Compressive testing of the mortar cubes was
Table 1
Mortar constituents and mix design proportions.

1.3. Research objective Mix Cement, lb [g] Sand, lb [g] Water, lb [g] TiO2, lb [g] w/b
Control 1.10 [500] 3.03 [1375] 0.53 [242] 0.48
The objective of this study was to characterize the mechanical, 2.5C 1.13 [512.5] 3.03 [1375] 0.53 [242] 0.47
physical, and hydraulic properties of macroporous pervious 5C 1.16 [525] 3.03 [1375] 0.53 [242] 0.46
concrete (MPC), which is defined herein as pervious concrete with 2.5T 1.10 [500] 3.03 [1375] 0.53 [242] 0.03 [12.5] 0.47
5T 1.10 [500] 3.03 [1375] 0.53 [242] 0.06 [25] 0.46
a minimum porosity (air void content) of 30%. The effect of
 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
conducted using a 100-kip MTS load frame with a loading rate of 0.0006 in/s
(0.015 mm/s). All specimens were loaded past the peak stress to observe post-
peak behavior.
2.2.2. Virgin and recycled concrete aggregate (RCA) characterization
Along with the virgin coarse aggregate, the physical properties of three (3) can-
didate blends (A, B, C) of RCA were characterized. These blends were pre-mixed by
the RCA supplier and originally designed as different blends for aggregate road base.
Blend A consisted of RCA with particle sizes of 1.500 (40 mm) or less. Blend B
consisted of a well-graded aggregate blend that met Colorado Department of
Transportations Class 6 classification. Class C consisted of concrete crusher fines.
The unit weight and gradation of the virgin and recycled aggregates (A, B, C)
were measured using ASTM C29 [24] and ASTM C136 [25], respectively. ASTM
C29 was also used to measure the unit weight of the selected aggregate size, No.
4 (4.75 mm). The No. 4 (4.75 mm) aggregate of the most promising RCA blend
and virgin aggregate was selected and tested for specific gravity and absorption
according to ASTM C127 [26]. A simple composition analysis for the No. 4
(4.75 mm) of the selected RCA blend was carried out by visual inspection. A small
sample of the aggregate was weighed and each aggregate type within the sample
(e.g., crushed mortar, reclaimed aggregate, asphalt) was segregated and individually
weighed to obtain a percent-weight of each aggregate type. After characterization
of the aggregate samples, the virgin aggregate and selected RCA blend was sieved
to a uniformly graded No. 4 (4.75 mm) particle size for use in the MPC mix
formulations.
2.2.3. Macroporous pervious concrete (MPC) preparation and characterization
Seven (7) MPC sample formulations were designed and cast according to the
mix design proportions presented in Table 2. The National Ready Mixed Concrete
Association (NRMCA) methodology [27] was used to proportion the mixes. The Con-
trol mix was the only mix to use virgin aggregate while the remaining mixes all
contained RCA. RCA in the mix name indicates that only RCA coarse aggregate
was put into the mix while RCAs indicates an addition of sand to the RCA. Sand
was added as a 7% by weight replacement of the coarse aggregate. The TiO2 was
added as a percent addition by weight of cement. The number in the mix name indi-
cates the percent addition of TiO2. A high-range water reducer (HRWR) was added
to each mix at a dosage of 2 oz/cwt (59 mL/cwt).
The MPC cylinders were mixed according to what the authors refer to as a slurry
build-up method. In this method, the cement paste was built-up around the aggre-
gate. The first step of the method combines all aggregates, coarse and fines, with a
small percentage of the batched cement for thirty seconds. A fourth of the cement
(and TiO2, if applicable) and batch water was added and allowed to combine in the
mixer for one minute. This was repeated until all cement and water was added.
Upon the last addition of cement and water, the concrete was mixed for three min-
utes. Lastly, the concrete was allowed to rest for two minutes and then mixed for a
final one minute. To achieve the high porosities required for MPC, light consolida-
tion was used in the casting of cylinders. Concrete was added in two lifts to 400  800
(100 mm  200 mm) cylinder molds, with each lift being rodded five times. The
rodding was applied equally across the area of the cylinder. The MPC was cured
under simulated field conditions. To simulate field curing methods prescribed in
the American Concrete Institute (ACI) Report 522.1-13 [28], the cylinders were left
in their molds for seven days while covered in plastic. The cylinders and plastic
were removed and left exposed to ambient indoor conditions. This curing method
was chosen to best simulate actual field curing conditions.
Compressive testing was performed at 28 days according to ASTM C39 [29].
Compression testing was load-controlled with an average rate of 30 psi/s
(0.21 MPa/s). Moduli of elasticity were calculated by determining the initial slope
of the stressstrain curves. The average unit weight, compressive strength, and
modulus of elasticity were measured for each mix design. Porosity was determined
for each mix according to a modified ASTM C1754 [30]. MPC cylinders were not
oven dried and the air-dry mass was substituted for the oven-dry mass. At least
three cylinders were used in the determination of average porosity. Permeability
of the cylinders was determined with a falling-head permeameter as described in
[9]. Three cylinders of each mix were measured for permeability. Each cylinder
was tested at three different initial head levels.
Table 2
Mix design proportions for macroporous pervious concrete (MPC) cylinders.
Mix Water, lb/yd3 [kg/m3] Cement, lb/yd3 [kg/m3]
Control 222 [132] 823 [489]
RCA 184 [109] 683 [405]
RCAs 184 [109] 683 [405]
RCA + 2.5 184 [109] 683 [405]
RCAs + 2.5 184 [109] 683 [405]
RCA + 5 184 [109] 683 [405]
RCAs + 5 184 [109] 683 [405]
91
3. Results and discussion
3.1. Mortar compressive strength
The 3-, 7- and 28-day compressive strengths for all mortar for-
mulations are given in Fig. 1. Most mortars exhibited a maximum
compressive strength at approximately 7 days. The impact of mois-
ture curing versus ambient-condition curing on mortar compres-
sive strengths is also shown in Fig. 1. The 28-day compressive
strengths of the ambient- and moisture-cured control samples
were 4.8 ksi (33 MPa) and 5.2 ksi (36 MPa), respectively. This
increase, which can be attributed to the ideal conditions of the cur-
ing room in the moisture-curing process, corresponds to a 7.3%
gain in compressive strength. Strength reductions of 3.5% and
0.9% can be seen in Fig. 1 for the 3-day compressive strengths of
the 2.5% and 5% addition of cement formulations. However,
strength gains of 9.1% and 5.2% were observed for the 2.5% and
5% addition of TiO2 formulations, respectively. In a comparison of
the 28-day strengths, 2.5% or 5% additions of cement or TiO2 ame-
liorated the strength losses due to the ambient-curing conditions,
suggesting that the binder additions enhance cement hydration.
The relative percent changes for all sample formulations from
the 3-, 7- and 28-day strength of the Control mortar are given in
Table 3. At three days, no significant changes are seen for the
cement additions, while compressive strength gains can be
observed for both TiO2 additions. This finding suggests, as claimed
by Jayapalan et al. [18], that TiO2 promotes early-strength develop-
ment. At seven days, both cement additions contribute approxi-
mately 6% to compressive strength whereas both TiO2 additions
contribute approximately 8.5%. This result implies that, at this
age, additions of TiO2 are more beneficial than the small change
in w/c ratio with the cement additions. At 28 days, neither of the
cement additions nor the 5% addition of TiO2 exhibited increases
in compressive strength of more than 5% when compared to the
Control mortar. However, the 2.5% addition of TiO2 yielded a 9.7%
strength gain. This increase exhibited by the 2.5T formulation
was greater than the strength enhancement achieved by the ideally
conditioned Cured sample. Additionally, this increase in strength
was similar to the increases in strength for both 5% and 10% addi-
tions of anatase TiO2 reported by Chen et al. [17], suggesting that
2.5% is the max addition of TiO2 needed for strength benefits and
that these benefits can be achieved with either the anatase or rutile
forms of TiO2.
3.2. RCA properties
The unit weight and gradation results for the three candidate
RCA blends are given in Table 4. In lieu of complete gradation
curves, only the percent retained on the No. 4 (4.75 mm) sieve is
reported, as it was the primary quantity of interest from the grada-
tion analysis for the mix design of MPC. The Class A blend had the
largest unit weight but the smallest amount of the desired aggre-
gate size at 14.9% retained. The other blends, Class B and Class C,
Aggregate, lb/yd3 [kg/m3] Sand, lb/yd3 [kg/m3] TiO2, lb/yd3 [kg/m3]
2273 [1350]
2216 [1316]
2071 [1230] 145 [86]
2216 [1316] 17 [10]
2071 [1230] 145 [86] 17 [10]
2216 [1316] 34 [20]
2071 [1230] 145 [86] 34 [20]
92 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997

Fig. 1. Strength development for all mortar sample formulations [1ksi = 6.9 MPa].

Table 3 (1580 kg/m3) and 98.9 lb/ft3 (1580 kg/m3), respectively. Because
Relative compressive strength development of mortars. of the lower yield of the required No. 4 (4.75 mm) aggregate, the
Mix 3-Day (%) 7-Day (%) 28-day (%) Class A blend was not considered in this study for further analysis.
Control 0.0 0.0 0.0
The Class B and Class C blends were sieved to obtain a sufficient
2.5C 3.5 6.0 1.8 amount of No. 4 (4.75 mm) for unit-weight characterization. As the
5C 0.9 5.7 4.8 unit weight results in Table 5 indicate, there was only a minor dif-
2.5T 9.1 8.8 9.7 ference between the aggregates from the different blends. This
5T 5.2 8.5 4.4
finding was also confirmed via a visual inspection of the sieved
Cured 7.3
aggregate, as no appreciable difference was observed between
the samples. Bulk specific gravity and absorption of the RCA were
then measured. The RCA displayed properties that were noticeably
Table 4
Raw recycled concrete aggregate (RCA) characterization. different from the virgin aggregate. The bulk specific gravity of the
RCA (2.16) was much lower than the specific gravity of the virgin
RCA blend Unit weight, lb/ft3 Percentage retained on
aggregate (2.60) and the absorption of the RCA (8.38%) was higher
[kg/m3] No. 4 [4.75 mm] Sieve
than the absorption of the virgin aggregate (1.11%). Both of these
Class A 103.5 [1660] 14.9%
differences were expected due to the porous nature of the RCA
Class B 98.4 [1580] 21.5%
Class C 98.9 [1580] 29.9% compared to the virgin aggregate. Class B and Class C RCA con-
sisted mostly of crushed mortar, as can be seen in Table 6. Crushed
mortar is more porous than gravel, which would consequently
increase the absorption of the material, as observed. The composi-
Table 5
No. 4 Aggregate characterization of Class B and Class C RCA and virgin aggregate.
tion of the RCA is controlled by demolition projects that supply the
raw material. This raw material may be masonry, asphalt,
Aggregate type Unit weight, lb/ft3 Bulk specific gravity Absorption, % reclaimed coarse aggregate, and/or crushed mortar of various pro-
[kg/m3]
portions and strengths. These irregularities inevitably lead to a
RCA Class B 83.3 [1330] 2.16 8.38 high variability in composition and, subsequently, absorption of
Class C 82.4 [1320]
Virgin Gravel 91.3 [1460] 2.60 1.11
the RCA. Because absorption must be considered in mixture pro-
portioning of pervious concrete, an average absorption of 8.38%
was assumed for all RCA used in the study.
Table 6
Recycled aggregate composition of Class B and Class C RCA.
3.3. Mechanical properties of MPC
Constituent material Mass, lb [g] Percent of total (%)
Mortar 0.12 [56] 87.5 During mixing of the MPC samples, the impact of the TiO2 addi-
Asphalt 0.0022 [1] 1.6 tions on the rheology and workability of the mix could be
Reclaimed Virgin Aggregate 0.015 [7] 10.9 observed. For both 2.5% additions (RCA + 2.5, RCAs + 2.5), the
MPC became viscid with large volumes of the material forming a
large cake in the bottom of the mixer. This large cake was easily
contained more of the desired aggregate at 21.5% and 29.9%, broken apart manually, but it remained solid if left undisturbed.
respectively, that was retained on the No. 4 (4.75 mm) sieve. The For the 5% additions (RCA + 5, RCAs + 5), the MPC was highly viscid.
unit weights of these two blends were similar at 98.4 lb/ft3 These results seem to agree with a previous study on the impact of

Table 7
Mechanical properties of MPC sample formulations.

Mix Compressive strength, psi [MPa] Relative compressive strength Modulus of elasticity, ksi [GPa] Relative modulus of elasticity
Control 456 38 [3.1 0.26] 245 29 [1.69 0.20]
RCA 458 31 [3.2 0.22] 1.00 249 25 [1.72 0.17] 1.02
RCAs 542 76 [3.7 0.52] 1.19 284 51 [1.96 0.35] 1.16
RCA + 2.5 488 53 [3.4 0.37] 1.07 228 33 [1.57 0.22] 0.93
RCAs + 2.5 584 35 [4.0 0.24] 1.28 259 29 [1.78 0.20] 1.06
RCA + 5 404 10 [2.8 0.07] 0.89 236 5.7 [1.63 0.04] 0.96
RCAs + 5 306 30 [2.1 0.21] 0.67 218 21 [1.50 0.15] 0.89
 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
Fig. 2. Comparison of MPC cylinder permeability [1 in/s = 2.5 cm/s] and porosity (%). Average data are presented. Error bars indicate one standard deviation.
Fig. 3. Relationship between compressive strength and unit weight for conven-
tional and macroporous pervious concrete (MPC) [1000psi = 6.9 MPa;1pcf =
16.0 kg/m3].
Fig. 4. Relationship between compressive strength and porosity of conventional
and macroporous pervious concrete [1000psi = 6.9 MPa].
TiO2 on cement mortars conducted by Senff et al. [31], who
reported that the addition of TiO2 increased the rheological values
of torque and yield stress while decreasing the spread-on-table of
cement mortars.
The 28-day compressive strengths for all seven (7) MPC mix
designs are shown in Table 7. As expected, the compressive
strengths are low in magnitude compared to traditional concrete
or typical pervious concrete due to the ultra-high porosity of
MPC. In a comparison of virgin aggregate (Control) and RCA MPC
samples, the data indicate that the use of recycled aggregates is
not detrimental to the compressive strength of MPC. This result
implies that the binder, not the aggregate, controls the mechanical
properties of the MPC. An ability to incorporate RCA without com-
promising mechanical properties has positive economic and envi-
ronmental implications. Broader applications of MPC present new
opportunities to reuse construction waste in novel ways and divert
waste from landfills, while reducing the economic and environmen-
tal costs associated with mining and transport of virgin aggregates.
93
Fig. 5. Empirical and theoretical relationship of permeability and porosity for
conventional and macroporous pervious concrete as predicted by the Carman
Kozeny equation [2.5 cm/s = 1in/s].
Table 7 also shows that a 7% addition of sand (RCAs) increased
the strength of MPC and that these relative strengths were higher
than the best-performing TiO2 addition (RCA + 2.5). This suggests
that sand was more effective at increasing compressive strength
than TiO2 when used individually. However, when both sand and
TiO2 were added to the paste (RCAs + 2.5), the MPC displayed a
relative strength increase of 28% compared to the Control mix,
suggesting a synergistic interaction between both additives.
Furthermore, a 2.5% addition of TiO2 increased the strength of
MPC cylinders by 7%, whereas an addition of 5% TiO2 was
detrimental to the compressive strength of the MPC cylinders. This
pattern is similar to the results of the mortar samples where the
2.5% addition of TiO2 yielded strength benefits while the 5%
addition yielded no change compared to the Control (see Table 3).
Similarly, the compressive strength of MPC also decreased with the
addition of 5% TiO2 and sand (RCAs + 5).
The modulus of elasticity for each MPC mix, along with the rel-
ative increases in stiffness for each mix (compared to the Control),
is also shown in Table 7. The values for the elastic moduli were rel-
atively consistent with an overall average of approximately
246 20 ksi (1.70 0.14 GPa). These values lead to a coefficient
of variation of 8.2%, indicating that the modulus of elasticity was
largely independent of the mix design. Similar to the compressive
strength results, there was little difference between the virgin
aggregate Control and the RCA samples. The RCAs samples demon-
strate that a 7% addition of sand increases the modulus of elasticity
of MPC by 16%. Both the 2.5% and 5% additions of TiO2 decreased
the modulus of elasticity by 7% and 4%, respectively, compared to
the Control sample. When sand is added to the 2.5% TiO2 addition
(RCAs + 2.5), the modulus of elasticity increases by 6%, relative to
94 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
Control. This trend is not valid for the 5% addition of TiO2 (RCAs
+ 5), where the addition of sand further decreased the modulus
of elasticity. This reduction in modulus is similar to the reduction
in compressive strength that is also shown in Table 7.
The 5% addition of TiO2 provided little benefit to the mortar
samples and was detrimental to the compressive strength and
modulus of elasticity of the MPC cylinders. Even though this result
was not anticipated, the loss in mechanical properties may be attri-
butable to two possible mechanisms related to the field curing
methods. The first possible mechanism is related to the photo-
chemical reactions that occur with TiO2. Along with the removal
of pollutants through a redox reaction, TiO2 becomes hydrophilic
in the presence of UV radiation. This hydrophilicity has been noted
by multiple authors [32,33]. The difference in curing location,
between the environmental chamber used in a study by Chen,
et al. [17] and the exposed laboratory space used in this study,
could result in different amounts of UV exposure. The light sources
in the exposed laboratory space of this study could have provided
UV radiation to both the mortar and MPC throughout the curing
process. Thus, the TiO2 in the samples could have become hydro-
philic and bound water that was not available for complete or thor-
ough cement hydration. This process would explain why the 5%
TiO2 additions did not suffer early age strength losses as seen in
Fig. 1 and Table 3. At three days, significant time had not yet passed
which would prevent sufficient UV radiation from photo-inducing
hydrophilicity in the TiO2. However, after 28 days of exposure to
UV radiation, the TiO2 may have become hydrophilic. The second
possible mechanism is autogenous shrinkage. In a study completed
by Jayapalan, the replacement of cement with TiO2 in cement-
based materials was found to increase autogenous shrinkage,
which was exacerbated with decreasing TiO2 particle sizes. This
mechanism can be attributed to high capillary stresses that
develop between small TiO2 particles within the paste [34]. This
is similar to the autogenous shrinkage mechanism caused by the
supplementary cementitious material, silica fume [35]. The shrink-
age behavior of the TiO2 could have been exacerbated by the
simulated field curing methods. In the ambient conditions, water
was able to evaporate into the environment, which could, in turn,
further increase capillary stresses. Thus, the 5% addition did not
perform as well as the 2.5% addition. These phenomena could also
explain why the loss of mechanical properties was greater in the
MPC cylinders than the mortar specimens. Given that MPC has a
higher exposed surface area, more TiO2 could have been exposed
to UV radiation and/or excess drying, leading to a lower degree
of hydration and excess internal stresses. The autogenous shrink-
age mechanism is more likely as the specimens were not subjected
to direct UV radiation, but rather laboratory lighting conditions.
3.4. Physical and hydraulic properties of MPC
The average permeability and porosity of each MPC mix is pre-
sented in Fig. 2. All formulations exhibited a permeability of
approximately 2.5 in/s (6.3 cm/s) with the exception of the RCA
sample formulation, which exhibited a permeability of 3.3 in/s
(8.2 cm/s). This result was anticipated given that the RCA mix also
exhibited the highest porosity. As expected, the addition of sand
decreased the permeability of MPC when compared to the MPC
mix with no sand (e.g., RCAs + 5 vs. RCA + 5). The magnitude of
the permeability for all MPC samples was much higher when com-
pared to typical pervious concrete values reported the American
Concrete Institute (ACI) Report 522.1-13. This result was also
expected due to the intended ultra-high porosity of MPC. The
porosity ranged from a low of 35% for the Control and RCAs mixes
to a high of 42% for the RCA mix. The remaining mix formulations
all had measured porosities of approximately 38%.
3.5. MPC property relationships
The relationship between the 28-day compressive strength and
unit weight for all MPC cylinders is given in Fig. 3. These data are
appended to the existing data presented by ACI 522R-10 (from
Mulligan 2005) [36]. Fig. 3 shows that the unit weight of all MPC
cylinders is lower than the range of unit weights previously
recorded in ACI 522R-10. The ACI data does not extend below
105 lb/ft3 (1680 kg/m3), whereas the MPC in this study ranges from
83 to 98 lb/ft3 (13301570 kg/m3). Additionally, the compressive
strengths of MPC are lower, albeit consistent, than the strengths
reported by ACI. This indicates that, for the MPC investigated
herein, the compressive strength is independent of the unit weight
and that a strength plateau exists around 500 psi (3.4 MPa) for the
overall compressive strength versus unit weight relationship of the
MPC samples.
The relationship between average compressive strength and
porosity for each MPC mix design are given in Fig. 4. These data
are also appended to data reported by ACI 522R-10 [37]. As shown
in the figure, data obtained in this study contribute previously
unreported MPC data for porosities that range from 35% to 42%.
A broad range of both porosities and unit weights were found in
the present study. However, the range of compressive strengths
was relatively narrow. For example, the unit weights in Fig. 3
varied from 83 to 98 lb/ft3 (13301570 kg/m3) a difference of
13 lb/ft3 (208 kg/m3) while the compressive strengths ranged
from 260 to 768 psi (1.85.3 MPa) a difference of only 508 psi
(3.5 MPa). It is evident that an increase in unit weight corresponds
to an increase in compressive strength for normal pervious con-
crete (<30% porosity) but this relationship does not necessarily
hold true for MPC. A similar trend of a narrower compressive
strength range is seen by the influence of porosity on compressive
strength shown in Fig. 4. The narrow ranges of compressive
strength in both relationships given in this study indicate that nei-
ther unit weight nor porosity is as important to MPC design as it is
to typical hydraulic design of pervious concrete for horizontal
infrastructure.
The relationship between permeability and porosity of MPC
cylinders is shown in Fig. 5 along with existing data from the ACI
522R-10 report on pervious concrete [37]. Current ACI data contain
some MPC permeability data with porosities that range up to 35%.
However, this study provides additional MPC permeability data for
porosities above 35%. Fig. 5 shows that MPC permeability is higher
than the permeability of normal pervious concrete. The obtained
permeability is also greater than the currently reported permeabil-
ity for 35% porosity pervious concrete. A range of permeabilities for
a given porosity is not unprecedented, even within the existing
data. For example, at an approximate porosity of 28%, the data sug-
gest a permeability range from less than 1.0 cm/s (0.39 in/s) to
greater than 2.0 cm/s (0.79 in/s). These differences can be attribu-
ted to differences in mix design and compaction methods.
Fig. 5 also shows the conventional CarmanKozeny hydraulic
conductivity model, which is commonly used to predict the rela-
tionship between permeability and porosity of pervious concrete
[3840]. The model is given by:
" #
p3
Ka 1
1  p2
where K is the permeability of the pervious concrete (cm/s), p is the
porosity (%), and a is an empirical coefficient. The a coefficient var-
ies from study to study and is used to empirically fit the model to
the data set. For example, Kevern and Schaefer [38] suggest a coef-
ficient of 30, while Montes and Haselbach [39] employ a coefficient
of 18. The variability of the coefficient highlights the inconsistencies
that exist in this empirical modeling approach.
 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
Fig. 6. Relationship of permeability and porosity for conventional and macroporous
pervious concrete as predicted by the proposed model [2.5 cm/s = 1in/s].
As can be seen in Fig. 5, no single a value is capable of modeling
the entire range of the normal and macroporous pervious concrete
data. The value of 18 as used by Montes and Haselbach [39], can
predict permeability for small porosities but is no longer accurate
after approximately 25%. The value used by Kevern and Schaefer
[38], (a = 30), can model the relationship with success up to a
porosity of about 35%. For porosities greater than 35%, the model
predicts permeability values that are too low. If an a value of 45
is selected, the model will over-predict permeability for low
porosities, but may be more accurate at large porosities (e.g.,
>35%). Given these results, the CarmanKozeny model does not
appear sufficient for predicting pervious concrete permeability
Fig. 7. Linear regression for hydraulic conductivity predictions using the conventional CarmanKozeny model with (a) a = 18, (b) a = 30, and (c) a = 45 and (d) the proposed
model with a = 30 [2.5 cm/s = 1in/s].
95
for the entire range of porosities. Thus, a modified version of the
CarmanKozeny model is suggested by the authors:
" #
p3
Ka 2
1  p3
The modified model uses the same variables as the Carman
Kozeny model. The ability of the model to describe the
permeability-porosity relationship for both normal pervious con-
crete and MPC is shown in Fig. 6, where a = 30.
A linear regression analysis was performed to investigate the
goodness-of-fits of the conventional and proposed models. The
results of the analysis are shown in Fig. 7. The ability of the
CarmanKozeny model (Eq. (1)) using (a) a = 15, (b) a = 30, and
(c) a = 45 to predict the hydraulic conductivity of both normal
and macroporous pervious concrete is compared to (d) the pro-
posed model. The results of each model are shown in Fig. 7ad,
respectively. As shown in Fig. 7a and b, the R2 values of the
a = 18 and a = 30 CarmanKozeny models were 3.7 and 0.39,
respectively, suggesting a poor correlation between empirical data
and model predictions for both normal and macroporous pervious
concrete. The negative R2 value indicates that a horizontal line of
zero slope would be a more adequate fit than the conventional
CarmanKozeny model where a = 18. The R2 values of the
CarmanKozeny model with a = 45 and the proposed model with
a = 30 were 0.85 and 0.87, as shown in Fig. 7c and d, respectively.
These results suggest that, while good correlation between
predicted and actual permeability can be achieved using the
conventional CarmanKozeny model (a = 45), the proposed model
is a better predictor of hydraulic conductivity for both normal and
macroporous pervious concrete.
96 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
4. Conclusions
Interest in the development and specification of pervious
concrete with porosities up to 30% has increased in horizontal
infrastructure applications due to the numerous environmental
benefits of permeable pavements. This interest has led to a greater
improvement in design methodologies and better understanding of
the mechanical and physical properties of pervious concrete over
the past few decades. Increased attention is now being given to
novel applications of pervious concrete beyond horizontal infras-
tructure that take advantage of its acoustic, thermal, or physical
(e.g., porosity) properties. Despite growing interest in such applica-
tions, little research has been conducted on pervious concrete that
is intended for these novel applications.
The primary objectives of this work were (a) to characterize the
mechanical and physical properties of macroporous pervious con-
cretes (MPC), defined herein as pervious concrete with porosities
greater than 30%, (b) to investigate the use of sand, titanium
dioxide, and recycled aggregates on the performance of MPC, and
(c) to assess the suitability of conventional models to predict the
hydraulic conductivity of both normal and macroporous pervious
concrete. While this work did not include measures of in-service
durability (e.g., freezethaw resistance, abrasion resistance), fur-
ther studies are essential to fully assess the specific application
potential of MPCs.
The following summarizes the main findings of this work:
1. Compression tests of pure mortar samples demonstrated that
(a) a 2.5% addition of titanium dioxide (TiO2) increased the
strength of cement-based mortars by 9.7% in ambient, simu-
lated field conditions and (b) that these mechanical property
benefits diminished at higher TiO2 dosage levels.
2. Due to higher porosities compared to virgin aggregates, recy-
cled aggregates exhibited lower specific gravities and higher
water absorption compared to gravel. These differences are
important parameters that must be considered in the proper
mix design of pervious concrete.
3. While MPC mechanical properties are lower than the mechani-
cal properties of conventional pervious concrete, the substitu-
tion of virgin aggregates with 100% recycled aggregates did
not compromise the compressive strength or modulus of elas-
ticity of MPC. These results suggest that the use of recycled
aggregates in lieu of virgin coarse aggregate does not compro-
mise the strength of MPC.
4. Contrary to normal pervious concretes with porosities <30%, no
correlations were found between unit weight and compressive
strength or between porosity and compressive strength for
MPCs with porosities >30%.
5. An addition of 2.5% TiO2 and 7% sand increased MPC compressive
strength by 7% and 19%, respectively. Together, the 2.5% TiO2 and
7% sand additions increased the compressive strength of MPC by
28% and increased the modulus of elasticity by 6%; and,
6. A new predictive model based on a well-known CarmanKozeny
relationship is proposed that more accurately predicts the
hydraulic conductivity for both normal and macroporous pervi-
ous concrete in comparison to existing models that, to-date,
have only been employed to predict the porositypermeability
relationship for normal pervious concrete with porosities <30%.
Acknowledgments
This research was made possible by the Department of Civil,
Environmental and Architectural Engineering, the College of Engi-
neering and Applied Sciences, and the Sustainable Infrastructure
Materials Laboratory (SIMLab) at the University of Colorado Boul-
der. The assistance of Mr. Luke Traeger and Mr. Ethan Gormican
at the University of Colorado Boulder and the collaboration with
Mr. Bill Kepler, Mr. Bruce Shulte, and Mr. Fred Travers at the United
States Bureau of Reclamation is gratefully acknowledged. Gratitude
is also given to Mr. Randy Ehmer of DuPont and Mr. Tom Pelo of
BASF for their donation of materials. This work represents the
views of the authors and not necessarily those of the sponsors.
References
[1] A.L. Welker, J.D. Barbis, P.A. Jeffers, A side-by-side comparison of pervious
concrete and porous Asphalt, J. Am. Water Resour. Assoc. 48 (2012) 809819.
[2] L. Haselbach, pervious concrete and mitigation of the urban heat island effect,
in: Transportation Research Board 88th Annual Meeting, 2009.
[3] L. Haselbach, M. Boyer, J.T. Kevern, V.R. Schaefer, Cyclic heat island impacts on
traditional versus pervious concrete pavement systems, Transp. Res. Rec. 2240
(2011) 107115.
[4] P. Starke, P. Gobel, W.G. Coldeway, Urban evaporation rates for water
permeable pavements, Water Sci. Technol. 62 (2010) 11611169.
[5] E.M. Nemirovsky, A.L. Welker, R. Lee, Quantifying evaporation from pervious
concrete systems: methodology and hydrologic perspective, J. Irrig. Drain. Eng.
ASCE 139 (2013) 271277.
[6] P.K. Goel, Water Pollution-Causes, Effects and Control, New Age International,
New Delhi, 2006.
[7] American Concrete Institute (ACI). ACI 522.R-10Report on Pervious Concrete.
Farmington Hills, MI, 2010.
[8] J.T. Kevern, K. Wang, V.R. Schaefer, Effect of coarse aggregate on the freeze-
thaw durability of pervious concrete, J. Mater. Civ. Eng. 22 (2010) 469475,
http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000049.
[9] J.T. Kevern, V.R. Schaefer, K. Wang, M.T. Suleiman, Pervious concrete mixture
proportions for improved freeze-thaw durability, J. ASTM Int. 5 (2008) 101320,
http://dx.doi.org/10.1520/JAI101320.
[10] L.K. Crouch, J. Pitt, R. Hewitt, Aggregate effects on pervious portland cement
concrete static modulus of elasticity, J. Mater. Civ. Eng. 19 (2007) 561568,
http://dx.doi.org/10.1061/(ASCE)0899-1561(2007) 19:7(561).
[11] Federal Highway Administration (FHWA), State of the practice national
review. Transportation Applications of Recycled Concrete Aggregate, 2004.
47p.
[12] B.M. Berry, M.J. Suozzo, I.A. Anderson, M.M. Dewoolkar, Properties of pervious
concrete incorporating recycled concrete aggregate, in: Transportation
Research Board Annual Meeting, 2012.
[13] C. Gaedicke, A. Marines, F. Miankodila, A method for comparing cores and cast
cylinders in virgin and recycled aggregate pervious concrete, Constr. Build.
Mater. 52 (2014) 49503.
[14] R. Rizvi, S. Tighe, V. Henderson, J. Norris, Evaluating the use of recycled
concrete aggregate in pervious concrete pavement, Transp. Res. Rec. 2164
(2010) 132140, http://dx.doi.org/10.3141/2164-17.
[15] S. Asadi, M.M. Hassan, J.T. Kevern, T.D. Rupnow, Development of
photocatalytic pervious concrete pavement for air and storm water
improvements, Transp. Res. Rec. 2290 (2012) 161167, http://dx.doi.org/
10.3141/2290-21.
[16] S. Shen, M. Burton, B. Jobson, L. Haselbach, Pervious concrete with titanium
dioxide as a photocatalyst compound for a greener urban road environment,
Constr. Build. Mater. 35 (2012) 874883, http://dx.doi.org/10.1016/j.conbuildmat.
2012.04.097.
[17] J. Chen, S.C. Kou, C.S. Poon, Hydration and properties of nano-TiO2 blended
cement composites, Cement Concr. Compos. 34 (2012) 642649, http://dx.doi.
org/10.1016/j.cemconcomp.2012.02.009.
[18] A.R. Jayapalan, M.L. Jue, K.E. Kurtis, Nanoparticles and apparent activation
energy of Portland cement, J. Am. Ceram. Soc. 97 (2014) 15341542, http://dx.
doi.org/10.1111/jace.12878.
[19] M. Carsana, F. Tittarelli, L. Bertolini, Use of no-fines concrete as a building
material: strength, durability properties and corrosion protection of
embedded steel, Cem. Concr. Res. 48 (2013) 6473, http://dx.doi.org/
10.1016/j.cemconres.2013.02.006.
[20] M. Tarnai, H. Mizuguchi, S. Hatanaka, H. Katahira, T. Nakazawa, K. Yanagibashi,
Design, Construction and Recent Applications of Porous Concrete in Japan. Our
World in Concrete and Structures, Singapore, 2003.
[21] J.T. Kevern, Climate Change, Energy, Sustainability Pavements (2014) 261273,
http://dx.doi.org/10.1007/978-3-662-44719-2.
[22] J.G. Jang, Y.B. Ahn, H. Souri, H.K. Lee, A novel eco-friendly porous concrete
fabricated with coal ash and geopolymeric binder: heavy metal leaching
characteristics and compressive strength, Constr. Build. Mater. 79 (2015) 173
181, http://dx.doi.org/10.1016/j.conbuildmat.2015.01.058.
[23] American Society of Testing and Materials (ASTM), C109/C109MStandard
Test Method for Compressive Strength of Hydraulic Cement Mortars. West
Conshohocken, PA: ASTM International, 2013.
[24] American Society of Testing and Materials (ASTM), C29/C29MStandard Test
Method for Bulk Density (Unit Weight) and Voids in Aggregate. West
Conshohocken, PA: ASTM International, 2009.
 P.W. Barnhouse, W.V. Srubar III / Construction and Building Materials 110 (2016) 8997
[25] American Society of Testing and Materials (ASTM). C136/C136MStandard
Test Method for Sieve Analysis of Fine and Coarse Aggregates. West
Conshohocken, PA: ASTM International, 2014.
[26] American Society of Testing and Materials (ASTM). C127/C127MStandard
Test Method for Relative Density (Specific Gravity) and Absorption of Coarse
Aggregate. West Conshohocken, PA: ASTM International, 2015.
[27] National Ready Mixed Concrete Association (NRMCA), Pervious Concrete:
Mixture Proportioning Software (computer software), Silver Spring, 2014.
[28] American Concrete Institute (ACI), ACI 522.1-13Specification for Pervious
Concrete Pavement. Farmington Hills, MI, 2013.
[29] American Society of Testing and Materials (ASTM). ASTM C39/C39MStandard
Test Method for Compressive Strength of Cylindrical Concrete Specimens.
West Conshohocken, PA: ASTM International, 2014.
[30] American Society of Testing and Materials (ASTM), C1754/C1754MStandard
Test Method for Density and Void Content of Hardened Pervious Concrete.
West Conshohocken, PA: ASTM International, 2012.
[31] L. Senff, D. Hotza, S. Lucas, V.M. Ferreira, J.A. Labrincha, Effect of nano-SiO2 and
nano-TiO2 addition on the rheological behavior and the hardened properties
of cement mortars, Materials Science and Engineering: A 532 (2012) 354361,
http://dx.doi.org/10.1016/j.msea.2011.10.102.
[32] J. Chen, C.S. Poon, Photocatalytic construction and building materials: From
fundamentals to applications, Build. Environ. 44 (2009) 18991906, http://dx.
doi.org/10.1016/j.buildenv.2009.01.002.
97
[33] M.V. Diamanti, M. Ormellese, M. Pedeferri, Characterization of photocatalytic
and superhydrophilic properties of mortars containing titanium dioxide, Cem.
Concr. Res. 38 (2008) 13491353, http://dx.doi.org/10.1016/j.cemconres.
2008.07.003.
[34] A.R.P. Jayapalan, Properties of cement-based materials in the presence of nano
and microparticle additives, Georgia Inst. Technol. (2013).
[35] B. Bhuvaneshwari, S. Sasmal, Role of nano oxides for improving cementitious
building materials, J. Civil Eng. Sci. (2012) 5258.
[36] A. Mulligan, Attainable Compressive Strength of Pervious Concrete Paving
Systems, University of Central Florida, 2005.
[37] R.C. Meininger, No-fines pervious concrete for paving, Concr. Int. 10 (1988)
2027.
[38] J.T. Kevern, V.R. Schaefer, Mixture proportioning considerations for improved
freeze-thaw durability of pervious concrete, ISCORD (2013) 471481.
[39] F. Montes, L. Haselbach, Measuring hydraulic conductivity in pervious
concrete, Environ. Eng. Sci. 23 (2006) 960969, http://dx.doi.org/10.1089/
ees.2006.23.960.
[40] N. Neithalath, Development and characterization of acoustically efficient
cementitious materials (PhD thesis), Purdue University, 2004.