Академический Документы
Профессиональный Документы
Культура Документы
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: This study assessed the feasibility of pervious concrete pavement (PCP) followed by a bamboo bioreten-
Received 19 September 2016 tion basin (BBB) with Dracaena sanderiana for urban stormwater volume control and water quality
Received in revised form 26 November 2016 enhancement. Two PCPs (nPCP and tPCP) having a permeability of 4.5 mm/s were prepared with opti-
Accepted 29 November 2016
mized mix designs. A control impervious concrete pavement (CP) was also prepared. Results showed that
Available online 7 December 2016
both PCP-BBB’s outperformed CP-BBB not only in runoff volume reduction, but also in pollutant removal;
both PCPs removed a considerably high amount of pollutants, especially in fecal coliform and phosphate
Keywords:
removal. tPCP-BBB improved water quality better than nPCP-BBB in terms of chemical oxygen demand
Dracaena sanderiana
Fly ash
and pH. Additionally, D. sanderiana grew healthier in tPCP-BBB than in nPCP-BBB. Production costs for
Pervious concrete tPCP and CP were similar, whereas nPCP production was twice as expensive.
Stormwater runoff Ó 2016 Elsevier Ltd. All rights reserved.
Water quality
http://dx.doi.org/10.1016/j.conbuildmat.2016.11.134
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
162 V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169
Evidently, there is a need to control the volume and water qual- of 180–600 m2/g. The coarse aggregates used were limestone grav-
ity of NPS pollutant-containing urban stormwater runoff. The use els purchased from a local hardware store. Prior to use, they were
of pervious pavement is among the best management practices sieved to collect the sizes in the range of 4.75–12.5 mm. The mass
(BMPs) for the control of stormwater runoff. Pervious pavement ratio of coarse aggregates to binder was fixed at 4:1 for the produc-
reduces the need for water detention ponds, permits groundwater tion of PCPs and CP. The binder is defined in this study as the total
recharge, decreases or eliminates pollutants from runoff, and amount of Portland cement and FA. Fine aggregates (sands) were
improves water quality [9]. used only for production of CP in the current study.
A bioretention basin is a landscaped depression where Miniature bamboo species (D. sanderiana) were used in BBB to
stormwater runoff is collected and the pollutants are captured accommodate the lab-scale experiment. They were purchased at
and biodegraded by biological media such as grass and plants a local store. The same coarse aggregates aforementioned were
[10]. It is also one of the BMPs for stormwater management. Bam- used as a supporting medium for D. sanderiana.
boos can be utilized as biological agents in the bioretention basins Tap water was used for rainfall simulation. Treated wastewater
due to their unique water retention properties. They are capable of effluent (TWE), fertilizer solution, and coolant solution were used
absorbing the most water among the agricultural plants whilst as the sources of NPS pollutants of fecal coliform (FC), phosphate
requiring little to no irrigation [11]. Therefore, they are a good, (PO4 -P), and automobile-related COD pollutant, respectively. Trea-
arguably the best, candidate biological agents for stormwater vol- ted wastewater effluent was sampled at a local wastewater treat-
ume and water quality control in locations where intense storm is ment prior to the chlorination process to ensure FC presence in
seasonal, such as in Puerto Rico. it. A phosphorus-rich commercial fertilizer was dissolved in deion-
In this study, a combination of two BMPs, a pervious concrete ized water to make desired PO4 -P concentrations. A commercial
pavement (PCP) and a bamboo bioretention basin (BBB), was antifreeze/coolant solution was diluted with deionized water to
assessed for their potential for volume control and water quality make a desired COD concentration. Table 2 shows physiochemical
enhancement of NPS pollutant-containing urban stormwater run- characteristics of rainwater and three NPS waters.
off. Two statistically optimized PCPs (namely, nPCP and tPCP)
and BBB with Dracaena sanderiana were evaluated for the reduc- 2.2. Production of optimized pervious concrete pavement
tion of runoff volume and NPS pollution. A control impervious con-
crete pavement (CP) in combination with BBB (i.e., CP-BBB) was tPCP was optimized for 7-day compressive strength and perme-
also tested in parallel to PCP-BBB’s. A preliminary production cost ability by Response Surface Methodology (RSM) in a two-factor,
analysis was done for PCPs and CP and potential influence of PCP’s two-level (22) CCD (Table 3). A mechanical mixer was used to pre-
on the growth and health of D. sanderiana was evaluated as well. pare the concrete specimens in accordance to the ASTM C192 [16].
For tPCP optimization, mixtures in triplicate were cast in a cylindri-
cal plastic mold (10 cm in diameter 20 cm in height). The stan-
2. Materials and method dard rodding consolidation method was used for compaction of
each specimen in accordance to the ASTM C192 [16]. The speci-
2.1. Materials mens in the mold were immediately put in an individual airtight
plastic bag to minimize moisture loss. After a 24-h curing under
The main materials used for the PCPs production were fly ash ambient environment (20–30 °C, unless otherwise specified), spec-
(FA), Portland cements, nano-sized silica (nanoSiO2), and coarse imens were demolded and further cured for 7 days in lime-
aggregates. FA was obtained from a local coal-fueled power plant saturated water under ambient environment. The compressive
in Puerto Rico. Type IP Portland cement was used for nPCP produc- strength of 7-day cured tPCP specimens was tested in triplicate
tion, whereas Type GU Portland cement was used for the produc- in accordance to ASTM C39 [17]. The permeability of 7-day cured
tion of tPCP and the control CP. Both cements comply with ASTM tPCP specimens was tested by a constant head method modified
C595 [12]. The physiochemical characteristics of Portland cements from ASTM D2434 [18].
and FA are shown in Table 1. For construction of PCPs and CP used in this study in combina-
A commercial nanoSiO2 was purchased and used as received. It tion with BBB, the optimized tPCP mix was cast in a rectangular
has an average particle size of 20–30 nm and a specific surface area wooden mold (9 cm 20 cm 5 cm). The same specimen prepara-
tion and curing procedures that used for tPCP optimization afore-
Table 1 mentioned was used. After a 24-h curing under ambient
Physiochemical properties of Portland cements and fly ash used in the study. environment, the specimens were demolded. nPCP and CP were
also cast in rectangular wooden molds. nPCP was made with (by
Properties Cement type IP Cement type GU Fly ash
wt%) liquid-to-binder at 50%, FA-to-binder at 60% and nanoSiO2-
Mineralogical composition (% wt)
to-binder at 0.04%, with sodium hydroxide (NaOH) solution at
SiO2 27.14 19.80 30.84
Al2O3 6.68 5.10 9.93 1.71 M as the liquid [19]. The control impervious concrete pave-
Fe2O3 3.71 3.10 5.01 ment (CP) was produced with (by vol%) Portland cement (Type
CaO 55.47 67.3 39.61 GU) at 11%, water at 16%, coarse aggregates at 41% and fine aggre-
MgO 1.62 0.8 0.35 gates (sand) at 26%. After a 24-h curing in an airtight plastic bag,
K2O 0.48 – 1.01
nPCP specimen was demolded and cured for 7 days in the same
Na2O 0.59 – 0.90
SO3 3.48 2.7 11.43 bag [19], whereas CP specimen was demolded and cured for 7 days
TiO2 0.32 – 0.45 in lime-saturated water. Like tPCP, both nPCP and CP were cured
P2 O5 0.11 – 0.11 under ambient environment.
Loss-on-Ignition (% wt)a 5.52 6.8 7.62
Blaine (m2/kg)b 554 488 441
Fineness (% wt)c 92.6 92.5 73.7
2.3. Construction of PCP-BBB and CP-BBB systems
a
The weight loss of the sample due to heating at 900–1000 °C (1650–1830°F) PCP-BBB systems were reconstructed in plastic containers
until a constant weight is obtained (ASTM C114) [13].
b
A measurement of the surface area of the sample, that is referred to as a fine-
(28 cm 17 cm 14 cm) (Fig. A1 in Appendix A of supplementary
ness measure (ASTM C204) [14]. data). Each BBB had 10 stalks of D. sanderiana. Light was provided
c
Wet sieve percentage passing the No. 325 (45 lm) sieve (ASTM C430) [15]. for 8 h per day to the bamboos with a 20 W aquarium light that
V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169 163
Table 2
Characteristics of rainwater, fecal coliform (FC) water, phosphate (PO4 -P) water, and chemical oxygen demand (COD) water.
Table 3
Factors and levels in 22 central composite design for optimization of typical pervious concrete (tPCP).
Factorsa Levelsb
( )axial Low Center High (+)axial
W/B 36 36.6 38 39.4 40
FA/B 10 15.9 30 44.1 50
a
W: water, B: binder, FA: fly ash.
b
Levels: the percentage to the total mass of the binders (FA + cement). Cement: Type GU.
was mounted at approximately 90 cm above the top of the bam- 2.5. Water quality analysis
boos. Table A1 (Appendix A of supplementary data) shows the
design specification of the completed systems used for the FC analysis was done by a membrane filtration technique with a
experiment. 0.45-lm cellulose ester membrane. The filtered membrane was
The CP compartment had storm catchment drains on the sur- placed in Petri dishes containing the HACH m-FC broth and incu-
face of the pavement that drained runoff water to BBB. In compar- bated for 24 h at 44.5 °C. Blue colonies were considered FC colonies
ison, the PCP compartment was equipped with underdrain pipes at that were expressed as colony-forming units (CFU) per 100 mL of
the bottom of PCP, through which the captured and stored water water sample. A Shimadzu Prominence IC system (Kyoto, Japan)
was drained to BBB (Fig. A2 in Appendix A of supplementary data). was used for PO4 -P analysis. COD concentration was measured
For 6 min, 1050 mL of precipitation (tap water) was applied to the by a reactor digestion method (HACH Method 8000). pH was mea-
PCP and CP compartments by a rain simulation bucket positioned sured by submerging the Orion 9157BNMD pH probe into the liq-
60 cm above the surface of the pavement (Fig. A1 in Appendix A uid samples. Leaf chlorophyll intensity of D. sanderiana was
of supplementary data). The rain simulation bucket had 15 capil- monitored with a chlorophyll meter (SPAD-502, Konica Minolta
lary tubes installed on the bottom of the bin, forming droplets of Sensing, Inc., Osaka, Japan).
‘‘rain” and creating reproducible precipitation events. A total of
450 mL of NPS pollutants (150 mL each) was manually added to
2.6. Cost analysis
the rain simulation bucket to start precipitation. Therefore, the
total volume of stormwater applied to each system was 1500 mL
Production cost of two PCPs and CP was calculated as a means
in which the concentration of each NPS pollutant was diluted 10
of a preliminary cost comparison. It should be noted that only
times.
the materials used were considered in the preliminary cost analy-
Considering the surface area of the pavement (180.7 cm2), an
sis (Table A2 in Appendix A of supplementary data).
application of 1500 mL was equivalent to a precipitation depth of
83 mm. This precipitation depth corresponds to the 25-year and
10-year storms for a duration of 1 h in Mayagüez and El Yunque, 3. Results and discussion
Puerto Rico, respectively [20].
3.1. Optimum tPCP mix
2.4. Operation of PCP- and CP-BBB systems Table 4 contains the independent variables (i.e., the levels of
each factor) and the corresponding response variables of the 7-
The experiment consisted of four cycles. In Cycle I, the systems day compressive strength and permeability of the tPCP mixes.
received only rainwater without NPS pollutants. In other words, a The D-optimal function in Minitab 17 was used to optimize the
total of 1500 mL of rainwater was applied in Cycle I. Following tPCP mix design for the maximum achievable compressive
Cycle I, Cycle II started by applying 1050 mL of rainwater with strength and a targeted permeability of 4.5 mm/s simultaneously.
450 mL of three NPS pollutant waters (150 mL each). Cycle II was The reason for targeting the permeability of tPCP to 4.5 mm/s
repeated two more times (i.e., Cycles III and IV). In each cycle, run- was because nPCP had the same permeability at 4.5 mm/s [19].
off volume was measured after the application of rainwater and This was done so that both nPCP and tPCP would have at least
NPS pollutant water. Sampling (40 mL) for water quality analysis one common characteristics despite the different materials used.
was done after 1, 2, 6, and 7 days in each cycle. After 7 days in each Results showed that the optimized tPCP for the maximum achiev-
cycle, leaf chlorophyll of D. sanderiana was monitored and then the able compressive strength of 15.1 MPa and the targeted permeabil-
systems were decanted for the next cycle. At the end of Cycle IV, ity of 4.5 mm/s could be found with a mix design of FA/B at 21.4%
changes in the weight and root growth of D. sanderiana among and W/B at 36.8% (Fig. 1). The predicted maximum compressive
the systems were compared. strength of tPCP of 15.1 MPa, therefore, fell into a typical compres-
164 V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169
Table 4
Matrix of 22 central composite design and the measured response variables.
Run order Independent variables (wt%) Response variables (average ± standard deviation, n = 3)
FA/B W/B Compressive strength (MPa) Permeability (mm/s)
1 30.0 38.0 11.74 ± 4.33 5.10 ± 1.49
2 30.0 38.0 13.38 ± 2.18 5.11 ± 2.08
3 15.9 36.6 17.72 ± 2.72 3.59 ± 0.44
4 30.0 38.0 12.78 ± 1.17 4.64 ± 0.13
5 15.9 39.4 21.68 ± 1.16 2.36 ± 0.40
6 44.1 36.6 6.52 ± 0.98 6.36 ± 0.91
7 44.1 39.4 12.30 ± 0.66 3.52 ± 1.37
8 30.0 38.0 14.13 ± 1.47 4.73 ± 0.81
9 30.0 36.0 11.45 ± 3.51 6.54 ± 1.05
10 50.0 38.0 7.49 ± 0.44 8.28 ± 1.24
11 30.0 38.0 15.38 ± 1.32 4.38 ± 0.50
12 30.0 38.0 14.15 ± 0.88 4.33 ± 0.85
13 10.0 38.0 17.86 ± 2.22 3.58 ± 0.93
14 30.0 40.0 14.39 ± 3.83 4.19 ± 0.34
Fig. 1. RSM optimization plot of tPCP mix design, showing the effect of each factor (columns) on the responses and composite desirability (rows). The vertical red lines
represent the optimum factor level settings, corresponding to the numbers in red. The horizontal blue lines and numbers represent the predicted responses for the optimum
factor levels in red. Individual desirability (d) assesses how well the factor level settings optimize a single response, whereas composite desirability (D), which is a geometric
mean of the ‘‘d”, identifies how well the factor level settings optimize a set of responses overall. Desirability (d and D) has a range of zero to one. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
sive strength of Portland cement pervious concrete of 2.8–28 MPa partment alone (Table 5). However, with an aid of PCPs, stormwa-
[21]. The permeability of tPCP of 4.5 mm/s was also in good agree- ter runoff was captured, infiltrated and stored in the storage gravel
ment with a reported value of 1.4–12.3 mm/s [21]. layer, resulting in 2.7-times and 3.6-times more runoff volume
reduction by tPCP-BBB and nPCP-BBB, respectively, than CP-BBB.
3.2. Runoff and retention volumes This demonstrated that a single green infrastructure (GI) may not
be capable of effectively managing urban stormwater runoffs.
Bioretention basins are typically incapable of treating the peak Instead, integrated approach of GIs is needed to control them, as
flow rates of high-intensity storm event due to their relatively reported by Liu et al. [23] and Mentens et al. [24]. In addition, a
small areal ratio to storm catchment of 5–8% [22]. Approximately smaller footprint of a bioretention basin would be required as long
one thirds of the stormwater was found retained in the BBB com- as stormwater runoff is partially managed by PCPs [21]. A minor
V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169 165
Table 5
Volume (mL) of stormwater runoff from the systems and retention (vol%) of stormwater in the systems.
difference in runoff volumes and retention capacities between the gravels in the current study. FC reduction was also achieved in
two PCP-BBB’s was most likely due to slightly dissimilar locations BBBs, with BBBs after PCPs being better than BBB after CP. It is con-
of underdrain and outflow pipes and also to dissimilar porosities of strued that FC were retained within the supporting media (i.e.,
the storage gravel layers (Table A1 in Appendix A of supplementary gravels) of D. sanderiana due to adsorption and straining, as
data). reported by Rusciano and Obropta [26].
Highly alkaline infiltrated water from a PCP can be detrimental
3.3. Water quality performance to soil and groundwater environment [27]. In this current study,
neither system compartments produced alkaline water, except
Fig. 2 shows the trend of FC reduction by the system compart- for nPCP in Cycles (I) and (II) where pH was 8.5 or greater
ments. It should be noted that the FC concentrations in TWE col- (Fig. 3). However, pHs decreased to neural range (i.e., 6.5–8.5) in
lected each week varied significantly (Table 2). As such, the the BBB compartment after nPCP. A further decrease of pH was also
initial FC concentrations applied to the systems at Cycles II – IV found in the BBB compartment after tPCP.
varied between 150 and 10,000 CFU/100 mL. Nevertheless, nPCP As shown in Fig. 4, both PCPs and D. sanderiana were very effec-
was slightly better than tPCP in FC reduction, regardless of the ini- tive in PO4 -P reduction. This implies that if PCP is preceded BBB, a
tial FC concentration. One of the main mechanisms of a better FC significant enhancement of water quality with respect to PO4 -P
removal could be the elevated pH caused by nPCP (Fig. 3). The pre- can be achieved in comparison to the case of BBB alone. In the pre-
vious research [19] documented a greater FC removal achieved at a vious studies, PO4 -P removal by PCPs were attributed to precipita-
higher pH caused by nPCP. An and Kampbell [25] also reported an tion of PO4 -P in alkaline infiltrated water [19,28]. But, this was not
increased FC mortality with the increase of pH. However, this was the same in the current study where the infiltrated water had a pH
not always true for other system compartments (i.e., tPCP and in the neutral range in most cases. Having said that, adsorption of
BBB’s) where pH was at neutral range (6.5–8.5). Although further PO4 -P onto PCPs and/or storage gravels might be responsible for
investigation is needed to elucidate this phenomenon, FC reduction PO4 -P reduction in PCP compartments.
might also be attributed to adsorption onto PCPs and/or storage
Fig. 2. Reduction of fecal coliform (FC) among the system compartments. The initial FC concentrations in the influent of the Cycles I, II, III and IV were 0, 4700, 10,000 and
150 CFU/100 mL, respectively.
166 V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169
A greater COD reduction was achieved in the BBB compart- At the current moment, it is unclear how D. sanderiana was capable
ments in PCP compartments (Fig. 5). In fact, D. sanderiana alone of reducing COD concentration. However, phytodegradation of
(i.e., in CP-BBB) were able to reduce COD between 55% and 93% COD pollutants within the tissues or in the root zone and/or phy-
despite a high initial COD loading (450 mg/L) to them. Between toextraction accumulating COD pollutant into biomass would most
the PCPs, tPCP was generally better than nPCP in COD reduction. likely be two main mechanisms responsible for COD reduction by
V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169 167
D. sanderiana [29]. Further study is therefore warranted to eluci- However, it should be noted that the preliminary production
date this phenomenon. cost only considered materials and associated cost. If construction
method had been considered as a project cost, the production of
PCP cost would have been different from what was shown in the
3.4. Growth and health of D. sanderiana
preliminary cost analysis. For example, curing of nPCP was less
costly than that of tPCP as nPCP was cured under atmospheric
Leaf chlorophyll intensity was monitored on the same D. sande-
environment but tPCP was cured in lime-saturated water.
riana leaves at the end of each cycle and the mass of D. sanderiana
In addition, nPCP contained 60% of FA, compared to 21.4% in
was measured before and after the experiment, as shown in Fig. 6.
tPCP. A more FA utilization means a less amount of FA disposal
Physical appearance of the leaves and roots of D. sanderiana at the
in landfills, leading to cost reduction in solid waste management.
end of Cycle IV is shown in Fig. A3 (Appendix A of supplementary
As such, if environmental cost had been taken into consideration,
data). In general, physical appearance was related to the leaf
the production cost of nPCP could have been lowered significantly.
chlorophyll content and mass of D. sanderiana.
Likewise, the production cost of tPCP could have been lower than
It was apparent that tPCP played a neutral, if not positive, role in
that of CP if the environmental benefit that tPCP generates, such
the growth and health of D. sanderiana (Fig. 6). In contrast, nPCP
as reduction of stormwater runoff volume and bioretention basin
produced a detrimental impact on the growth and health of D.
area, had been accounted as a project cost.
sanderiana. This could be attributed mainly to a high pH in infiltrat-
ing water through nPCP that discharged to BBB.
3.6. Practical guides
3.5. Production cost of PCPs
Although tPCP could reach a 7-day compressive strength of
A preliminary cost analysis showed that the production cost of 15.1 MPa (or 2190 psi) (Fig. 1), it is still not strong enough to be
nPCP ($661/m3) was twice as expensive than tPCP ($336/m3). The used for high-traffic volume roads. Therefore, it is recommended
production cost of tPCP was similar to that of CP ($298/m3), as that tPCP be used for low-volume roads such as parking areas in
shown in Table A3 (Appendix A of supplementary data). Despite combination with a bioretention basin as an integrated GI for
the three-time more substitution of no-cost FA in nPCP (i.e., stormwater runoff control. Knowing a slow strength development
300 kg/m3) than in tPCP (i.e., 107 kg/m3), the production cost was of FA-added concrete [30,31], it is also recommended that tPCP be
almost twice more expensive for nPCP than tPCP. Although expen- provided with an extended period of moist-controlled curing
sive, nano-sized materials such as nanoSiO2 have been incorpo- beyond the conventional curing period of 7 days.
rated in concrete matrix in order to overcome a major drawback It should be also noted that the optimum FA substitution of
of slow strength development due to FA substitution [30,31]. How- 21.4% for tPCP is not best suited for all applications. For example,
ever, the increase in the production of nPCP in the current study a new mixture proportioning should be done to find an optimum
was due to the costly alkaline activation for geopolymerization of amount of FA when chemical admixtures, such as hydration con-
nPCP matrix, not due to the nanoSiO2 addition. As a matter of fact, trolling agents, water reducing agents and viscosity modifying
the cost of nanoSiO2 addition was offset by the high-volume FA agents, are used to accommodate field conditions and to provide
substitution (i.e., 60%) in nPCP production. additional cementing properties.
168 V. Hwang et al. / Construction and Building Materials 132 (2017) 161–169
Fig. 6. Leaf chlorophyll content of D. sanderiana in the systems after each Cycle (a) and the mass of D. sanderiana in the systems before and after the experiment (b).