Accepted Manuscript A practical model for sunlight disinfection of a subtropical maturation pond N.W. Dahl, P.L. Woodfield, C.J. Lemckert, H. Stratton, A. Roiko PII:
S0043-1354(16)30831-4
DOI:
10.1016/j.watres.2016.10.072
Reference:
WR 12470
To appear in:
Water Research
Received Date: 18 August 2016 Revised Date:
26 October 2016
Accepted Date: 27 October 2016
Please cite this article as: Dahl, N.W., Woodfield, P.L., Lemckert, C.J., Stratton, H., Roiko, A., A practical model for sunlight disinfection of a subtropical maturation pond, Water Research (2016), doi: 10.1016/ j.watres.2016.10.072. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
SC
RI PT
ACCEPTED MANUSCRIPT
Progessive die−off of depth distributed E. coli
M AN U
Surface die−off from daytime stratification
0.2 0.3 0.4 0.5
6 5 4 3
Progressive vertical mixing via top down natural convection
0.6 0.7 0.8
12 PM
6 PM
12 AM 6 AM 12 PM Time of day
AC C
EP
0.9 6 AM
TE D
Depth of Water Column (m)
0.1
10 9 8 7
6 PM
12 AM
2
6 AM
1
Surviving E. coli (%)
0
ACCEPTED MANUSCRIPT
A practical model for sunlight disinfection of a subtropical maturation pond
3
a
4
b
Griffith School of Engineering, Griffith University, Gold Coast, QLD, Australia
Smartwater Research Centre, Griffith University, Gold Coast Campus, Edmund Rice
5
Drive, Southport, Queensland 4215, Australia
6
RI PT
N.W. Dahla,∗, P.L. Woodfielda, C.J. Lemckerta,b, H. Strattonb,c, A. Roikob,d
2
c
School of Natural Sciences, Griffith University, Nathan Campus, 170 Kessels Road, Nathan, Queensland 4111, Australia
7
d
School of Medicine, Griffith University, Gold Coast Campus, Parklands Drive, Southport,
8
Queensland 4222, Australia
M AN U
9
SC
1
Abstract
11
Maturation ponds are a type of waste stabilisation pond (WSP) designed to reduce carbon, nutrients and pathogens in
12
the final stages of a WSP wastewater treatment system. In this study, a one-dimensional plug-flow pond model is
13
proposed to predict temperature and E. coli concentration distributions and overall pond disinfection performance.
14
The model accounts for the effects of vertical mixing and ultraviolet light-dependent die-off rate kinetics.
15
Measurements of radiation, wind-speed, humidity and air temperature are recorded for model inputs and good
16
agreement with measured vertical temperature distributions and outlet E. coli concentrations is found in an
17
operational, subtropical maturation pond. Measurements and the model both show a diurnal pattern of stratification
18
during daylight hours and natural convective mixing at night on days corresponding to low wind speeds, strong heat
19
input from solar radiation and clear night skies. In the evenings, the thermal stratification is shown to collapse due to
20
surface energy loss via longwave radiation which triggers top-down natural convective mixing. The disinfection
21
model is found to be sensitive to the choice of die-off kinetics. The diurnal mixing pattern is found to play a vital
22
role in the disinfection process by ensuring that pathogens are regularly transported to the near-surface layer where
23
ultraviolent light penetration is effective. The model proposed in this paper offers clear advantages to pond designers
24
by including geographical specific, time-varying boundary conditions and accounting for the important physical
25
aspects of vertical mixing and sunlight inactivation processes, yet is computationally straightforward.
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
26
Keywords:
E. coli, eddy diffusivity, inactivation, ultraviolet, wastewater
27
29
∗Corresponding
author. Address: Building G09, Room 1.02, Griffith School of Engineering, Gold Coast Campus, Griffith University, QLD
4222, Australia. Tel.: +61 (0)7 5552 7608; fax: +61 (0)7 5552 8065
RI PT
28
30
Email addresses:
[email protected] (N.W. Dahl),
[email protected] (P.L. Woodfield),
[email protected]
31
(C.J. Lemckert),
[email protected] (H. Stratton),
[email protected] (A. Roiko)
SC
32
1. Introduction
34
Maturation ponds make use of sunlight and other naturally occurring phenomena for disinfection of pathogens and
35
reduction of carbon and nutrients in waste stabilisation pond (WSP) wastewater treatment systems (Nelson, 2000;
36
Jemli et al., 2002; Kohn and Nelson, 2007). In such systems, the maturation pond is usually the final WSP in a series
37
of shallow earthen basins with a typically depth of 1-1.5 m, and is exposed to the atmosphere and has a significant
38
effect on the final effluent quality. Predicting and managing the quality of the effluent from these ponds is important
39
due to the widespread use of WSP systems, particularly where recycling of treated wastewater is employed as a
40
strategy for enhancing water use management (Scheierling et al., 2011; Bichai and Smeets, 2013).
41
Inactivation of pathogens and indicator microorganisms present in wastewater has been shown to be affected
42
significantly by sunlight under both laboratory and field conditions (Curtis et al., 1992a,b; Davies-Colley et al., 2000;
43
Bolton et al., 2010; Kadir and Nelson, 2014; Nguyen et al., 2015; Maraccini et al., 2016a,b). Photoinactivation, or
44
inactivation involving sunlight, of pathogens and indicator microorganisms in pond systems can occur via three
45
primary mechanisms: endogenous direct, endogenous indirect, and exogenous indirect (Davies-Colley et al., 1999;
46
2000). Endogenous direct photoinactivation occurs when photons are absorbed by cellular molecules, causing
47
damage to genetic material, such as microbial DNA from ultraviolet-B (UVB) wavelengths (280-320 nm) (Schuch
48
and Menck, 2010).Indirect photoinactivation occurs either within (endogenous) or external (exogenous) to the cell
49
and involves energy or electron transfer to form reactive species that can then cause cell death (Maraccini et al.,
50
2016b). Endogenous indirect photoinactivation is caused mostly by UVB (Maraccini et al., 2016a), while exogenous
51
indirect photoinactivation can involve both the ultraviolet spectra and wavelengths extending to 550 nm (Kadir and
52
Nelson, 2014).
AC C
EP
TE D
M AN U
33
2
ACCEPTED MANUSCRIPT
The efficacy of UV disinfection in the water column depends strongly on a microorganism’s vertical position and
54
proximity to the water surface as sunlight is strongly attenuated in turbid water (Curtis et al., 1994; Davies-Colley et
55
al., 2005). The resulting vertical distribution of inactivation means that die-off in the lower water column is reduced,
56
reducing overall treatment efficiency. Empirical evidence has been reported by Maïga (2009b) identifying higher
57
die-off in shallow microcosms while die-off in microcosms at increasing depth showed little change. Other studies
58
have also reported lower concentrations of coliforms present in the near surface region (Brissaud et al., 2003).
59
Thermal stratification supresses the vertical transport of heat and mass, further complicating the understanding of
60
vertical die-off distributions (Pernica et al., 2013; Yang et al., 2015). Observations from many WSPs, and studies on
61
other shallow water bodies, have shown that heat generation from sunlight attenuation warms the upper water
62
column producing a stable stratification during sunlit hours (Losordo and Piedrahita, 1991; Gu and Stefan, 1995;
63
Brissaud et al., 2003; Badrot-Nico et al., 2009). The physical consequence is to effectively confine microorganisms
64
to stratified layers and limit inactivation to those contained within the surface layer only, where sunlight is present.
65
While stratification may exist during daylight, the onset of natural convection occurs with surface cooling in the late
66
afternoon. In shallow water bodies, like maturation ponds, vertical thermal distributions have been reported to mix
67
completely via this mechanism (Losordo and Piedrahita, 1991). In this context, shallow water bodies refer to depths
68
where stratification and complete mixing occur diurnally, typically begin less than 3 m.
69
Predictions of log unit reduction of target pathogens for the design of new maturation ponds currently use empirical
70
ideal flow models which are very convenient but neglect many of the complex interactions between fluid mixing,
71
stratification and sunlight disinfection (Ashworth and Skinner, 2011). Furthermore, the effect of transient
72
atmospheric boundary conditions is often not included. At the other extreme, computational fluid dynamics (CFD)
73
has been applied to existing ponds to account for spatial and temporal effects giving valuable insight into pond
74
behaviour (Shilton and Harrison, 2003; Badrot-Nico et al. 2009; Alvarado et al., 2012; Passos et al., 2014) but
75
computations can take days, weeks or longer to run. Thus, while CFD models can capture the physics more
76
accurately, the computational expense and expertise required reduces the appeal for design purposes.
77
A solution to overcome the practical shortcomings of utilizing CFD and weaknesses of perfectly-stirred or simple
78
plug-flow reactor models is to develop one-dimensional models to account for the effects of sunlight inactivation and
79
vertical mixing. Further, inclusion of transient boundary conditions for site-specific applications is necessary to
AC C
EP
TE D
M AN U
SC
RI PT
53
3
ACCEPTED MANUSCRIPT
improve on the current models. One-dimensional models have been developed successfully for other water bodies
81
(Dake and Harlem, 1969; Herb and Stefan, 2005; Jacobs et al., 2008). This paper makes a unique contribution by
82
coupling vertical mixing and sunlight inactivation of E. coli in a one-dimensional model of a maturation pond,
83
representing the simplest possible method to account for the spatial and temporal effects. Furthermore, we provide a
84
practical set of field data showing the diurnal mixing pattern, E. coli die-off, and boundary conditions for testing of
85
this type of model.
RI PT
80
86 2. Methods
88
Experimental data were collected from a maturation pond to act as input and validation data for the numerical model.
89
Data included two continuous vertical profiles of temperature, E. coli concentrations and atmospheric conditions.
90
The one-dimensional numerical model was developed by combining published work on UV disinfection kinetics
91
with available fluid mixing and dispersion models for ponds and lakes.
92
2.1 Experimental methods
93
Field data were collected from an operational maturation pond located South East Queensland (SEQ), approximately
94
90 km west of Brisbane. The sewage treatment plant (STP) services a population of approximately 1000 and is
95
located within the Lockyer Valley region of SEQ. Typical climate conditions in this region have average daytime air
96
temperatures of 31 °C in summer and 20 °C in winter, peak solar radiation of 1200 W m-2 (summer) and 600 W m-2
97
(winter) and monthly rainfall averages of 100 mm (summer) and 35 mm (winter).Treatment consists of a screening
98
rack to remove large solid items before entering a primary facultative pond, secondary maturation pond (the focus of
99
this study), two wetland cells, a series of reed beds and is chlorinated before final discharge to an adjacent, privately
AC C
EP
TE D
M AN U
SC
87
100
owned reservoir.
101
The maturation pond layout is presented in Fig. 1a, showing five baffles traversing the width at 30 m in length,
102
inlet/outlet locations and dimensions. Vertical temperature distributions were sampled within the pond at two
103
locations (points 1, 2 in Fig. 1a). Five discrete temperature sensors were used for each vertical distribution along the
104
depth sampling at 5 minute intervals. The locations of the two thermistor chains are shown on this schematic. A
105
thermistor chain is simply a series of temperature sensors spaced along a line, in this case traversing the vertical
4
ACCEPTED MANUSCRIPT depth. Fig. 1b shows the experimental vertical sensor locations in more detail; − for thermistor chains 1 and 2
107
are the temperature sensors located at a depth of 0, 0.1, 0.3, 0.5 and 0.9 metres below the water surface, respectively.
108
Measurements of temperature were recorded over the period of 22 February - 09 March 2015. Meteorological data
109
were recorded on site, taking 15 minute averages of air temperature, wind speed, solar radiation, relative humidity
110
and absolute pressure; wind direction was recorded as the current direction every 15 minutes.
111
Irradiance () profiles were measured on site using a cosine collector (Trios Ramses UV/VIS spectrometer) lowered
112
through the water column, measuring at discrete depths for the purpose of determining radiation attenuation
113
coefficients. The spectral range measured is 280-700 nm.
114
Water samples of 100-500 mL were taken from the maturation pond inlet and outlet, in duplicate, over the period of
115
22 February, 2015 – March 9, 2015 at 8 hour intervals (6 am, 2 pm and 10 pm Australia Eastern Standard Time).
116
Samples were stored in a refrigerated compartment for no longer than 24 hours before being analysed. E. coli was
117
enumerated using the membrane filtration technique (APHA, 1992) with some modifications. Wastewater samples
118
were mixed thoroughly, serially diluted in PBS or sterile distilled water and filtered through 0.45 um sterile mixed
119
cellulose ester membrane filters (Advantec, Tokyo, Japan) using a CombiSart® manifold filtration unit (Sartorius,
120
Gottingen, Germany). The filter was placed on modified mTEC agar plates (BD, Sparks, AR, USA) incubated at
121
35 °C for 2 h followed by an additional incubation at 44.5 °C for 24 h. Single magenta colonies were quantified and
122
reported as E. coli numbers.
123
Duplicate samples were averaged for E. coli counts and Log10 colony-forming units (CFU) per 100 ml-1 are reported
124
below from the inlet and outlet.
EP
TE D
M AN U
SC
RI PT
106
AC C
125 126
2.2 Model Development
127
Conservation equations for energy and concentration of living E. coli are given by:
128
= + +
129
=
(1)
! − " #
(2)
5
ACCEPTED MANUSCRIPT
130 131 132
where is the density (kg m−3), is the specific heat capacity (J kg−1 K−1), $ is the temperature (°C), t is time (s), %
is the vertical distance measured positive from the water surface downwards (m), and are the molecular
is the volumetric heat thermal conductivity and eddy thermal conductivity, respectively (W m−1 K−1), and
133
generation from shortwave penetration (W m−3). C is the concentration (-),
134
concentration (m2 s-1) and kc is the decay rate coefficient (s-1).
135
The effect of turbulent convective heat transport from wind shear and internal currents is modelled by the concept of
136
eddy diffusion:
d d d
RI PT
≤0
SC
= & (01'()*+,) ,
d
>0
M AN U
137
('()*+,) ,
is the eddy diffusion coefficient of the
(3)
138
where keddy(unstable) accounts for buoyancy driven turbulent diffusion (unstable stratification) and keddy(stable) for wind
139
driven mixing:
140
('()*+,) =
141
where Ri is the Richardson number in the form proposed by Sundaram and Rehm (1973):
142
Ri =
3,445
(1,0(6)+)
78.8Ri
?
TE D
:; <' = > d d
(4)
(5)
@A is the coefficient of thermal expansion (K−1), B is the gravitational constant (m s−2) and (1,0(6)+) in Eq. 4 is
144
the neutrally buoyant eddy diffusivity (m2 s−1) (Smith, 1979) given by:
145
(1,0(6)+) =
D
EP
143
E ?<'CD >
F83 ∗
G H3
∗
(6)
where ∗ is the vertical decay in current velocity (m−1) calculated by ∗ = 6JH.KL .
147
Turbulent mixing as described by Eqs 4 to 6 is a function of depth, wind speed and density gradient (Henderson-
148
Sellers, 1984). This kind of one-dimensional diffusion model (Eqs, 1, 4-6) has been applied to large lakes and
149
shallow ponds (e.g. Dake and Harlem, 1969; Sundaram and Rehm, 1973; Henderson-Sellers, 1985; Losordo and
150
Piedrahita, 1991), the shallowest being 20 cm in depth (Jacobs et al., 2008).
151
In the late afternoon, surface cooling creates unstable buoyancy and natural convection is initiated. Thus there is a
152
need to include keddy(unstable) in Eq. 3. The unstable layer is effectively mixed from the surface down (Sundaram and
AC C
146
6
ACCEPTED MANUSCRIPT
Rehm, 1973; Bednarz et al., 2009b). This is illustrated in Fig. 2b by an unstable temperature gradient where the
154
hatched areas show the depth to which the column will mix (Dake and Harlem, 1969). The temperature gradient is
155
approximated by the average temperature (conserving energy and mass) shown by the vertical line separating the
156
two hatched areas. Lumley and Panofsky (1964) noted that free convection is far more efficient as a transport agent
157
than wind-driven mechanical mixing and as such, it is sufficient to simply set (01'()*+,) (see Eq. 6) to a large
RI PT
153
numerical value to achieve the effect shown in Fig. 2b; here we use (01'()*+,) = ⋅ 1 W m-1 K-1 (c.f. Losordo
159
and Piedrahita (1991)).
160
Concentration transport of Eq. 2 (D) is analogous to the transport of heat. Therefore, the vertical transport of
162
concentration is determined through a turbulent or effective Lewis number (OG ) , which relates the thermal
diffusivity + to mass (concentration) diffusivity ( ). For modelling of turbulent transport of passive
M AN U
161
SC
158
163
scalars, turbulent Prandtl, Schmidt and Lewis numbers (OG = /QR ) are typically assigned (or observed to have)
164
similar values in the range from about 0.6 to 1.0 depending on the flow situation (e.g. Reynolds, 1975; Weigand et
165
al., 1997; Chang and Cowen, 2002; Tominaga and Stathopoulos, 2007). The near unity values arise in situations
166
where the passive scalars are transported predominantly via the same turbulent convective mechanism (Tennekes
167
and Lumley, 1972). For Eulerian modelling of dispersion of microorganisms, Elyasi and Taghipour (2006) assumed
a turbulent Schmidt number of 0.7. For the present purpose, OG is assumed to be a constant with a value of unity. In
169
the absence of more precise data for dispersion of E. coli, in Eq. 7 we have assumed the same Lewis number applies
170
to laminar and turbulent diffusion. Thus:
TE D
168
172
where
173
diffusivity (m2 s−1).
174
Concerning boundary conditions for Eq. 2, the concentration flux at the free surface and at the soil-water interface is
175
zero, assuming that in field conditions no losses occur. Adiabatic conditions for Eq. 1 are imposed at the base of the
176
soil domain. Soil properties have been assumed to be VWXY = 1500 kg m-3, VWXY = 1 W m-1 K-1 and = 1000 J kg-1 K-
<"U T(
(7)
is the effective diffusion coefficient for the representative pathogen (m2 s−1) and @ is the effective thermal
AC C
:
EP
=
T(
=
373,445
171
177
1
178
the energy flux comprising of short and longwave radiation and convective heat and mass transfer (Eq. 8):
(Kimball, 1983; Lamoureux et al., 2006; Paaijmans et al., 2008). The water–atmospheric boundary is modelled by
7
ACCEPTED MANUSCRIPT
179
180 181
− +
Z
[8
= (1 − R)\V] + Y] + A + "W^A
(8)
where r is the albedo of the shortwave radiation (-), \ is the fraction of shortwave radiation absorbed at the water surface (-), V] is the measured atmospheric shortwave radiation (W m−2), Y] is the net longwave radiation (W m−2),
A and "W^A are the heat fluxes from mass and heat transfer, respectively (W m−2). The energy flow vectors are
183
defined as positive when providing energy to the water.
184
Incident short-wave radiation (V] ) in Eq. 8 was measured on site, of which a portion is reflected from the water
185
RI PT
182
surface (albedo, r), another portion is absorbed at the surface (\), and the remainder penetrates the column and is
attenuated through absorption. Albedo is calculated using Fresnel’s equation (Eq. 3) where the zenith angle (_ ) is
187
calculated in the simulation as a function of time based on the geographical coordinates and time of day (equations
SC
186
are presented in Kirk (2010, Ch 2) and are based on the work of Spencer (1971)). The refracted angle (_] ) is
189
assumed to follow Snell’s law for pure water (refractive index of 1.33).
190
R=
192 193 194
` abc> (d) 7de )
+
fgc> (d) Hde ) ` fgc> (d) 7de )
(9)
A constant value of \ = 0.4 has been assumed for each of the wavelength bands (Dake and Harlem, 1969); although
our measurements show that \ + R is not constant between UVB, ultraviolet-A (UVA, 320-400 nm) and the
TE D
191
abc> (d) Hde )
M AN U
188
photosynthetically active radiation (PAR) portion of sunlight (400-700 nm). Larger values of 0.55 for \ have been
used by Yang et al. (2015), taking shortwave radiation as extending to 2800 nm; our measurements of V] extend only to 1000 nm and hence have a lower value of \.
196
Net longwave radiation is determined by the Stefan-Boltzmann law applied to a small surface in a large enclosure
EP
195
(i.e. a ‘small pond’ under a ‘large sky’) (Eq. 10) using the emissivity of water (i] ) as 0.96 (e.g. Incropera and
198
DeWitt, 1996; Lamoureux et al., 2006).
199
L Y] = i] j($V3 − $VL )
200
AC C
197
(10)
where σ (5.67 × 10HK W m−2 K−4) is the Stefan-Boltzmann constant, $V is the simulated water surface temperature (K)
201
and $V3 is the effective sky temperature (K). In this study we assume a constant value for $V3 of 0 °C noting the
202
limitation that this will not take into account cloud cover. However, it has been shown to be a reasonable value for
203
clear skies, particularly at night (Pandey, 1995; Pérez-Garcı́a, 2004; Gliah et al., 2011).
8
ACCEPTED MANUSCRIPT
Convective transfer from the surface removes heat through two mechanisms; sensible and latent heat transfer (qconv
205
and qevap respectively in Eq. (8)). Sensible heat transfer is driven by the temperature gradient while the rate of latent
206
heat is driven by the water vapour concentration gradient, both occurring at the free surface. The Chilton-
207
Colburn/Reynolds analogy is used to model these two heat fluxes. This method is applicable to heat transfer from a
208
flat plate in turbulent flow conditions (Incropera and DeWitt, 1996).
209
) Heat generation ( in Eq. 1 is modelled by the penetration of radiation into the water column. Irradiance
210
penetration is calculated by the Lambert-Beer law (Eq. 11), first removing the surface reflection and absorption; η is
211
is then equal to −n/n%. a function of wavelength and assumed to be spatially and temporally constant.
213
SC
= (1 − R)(1 − \)V] G Hop
(11)
where q is the path length correction factor (-) and r is the attenuation coefficient (m−1). Eq. 11 is evaluated
M AN U
212
RI PT
204
214
assuming a proportion of V] is comprised of UVB, UVA and PAR and evaluating the attenuation separately. r was
215
measured on site and these values (as listed in Table 1) have been used in the model.
216
It is assumed that 6.4 % of V] is in the UV band (Maïga et al., 2009a), our measurements further show that UVB is
217
5.6 % of the UV spectrum. q is necessary to account for the refraction angle of the radiation and is calculated by CD
q = (1 − (1.33H sin _ )`) >
219
Three equations from the literature are used to model sunlight inactivation (kc in Eq. 2) of E. coli. Eq. 13 models the
220
endogenous decay of wastewater E. coli as a function of photon fluence (wWW^ , units of m2 Ei−1) (Nguyen et al.,
221
2015). In this simulation, irradiance is converted to einsteins (Ei) in two discrete bands, one centred on UVB (300
(12)
EP
TE D
218
nm) and the other on UVA (360 nm). wWW^ has a value of 3.8 m2 Ei-1. Maïga et al. (2009a) regressed both the UV
223
and sunlight intensity influence against E. coli concentrations to determine Eq. 14 and 15 where either could explain
224
the die-off observed. The two coefficients to I in Eq. 14 and 15 have units of m2 J-1.
225
" = wWW^ ⋅ 0.836 × 10HK y{[`K8 (z, %)z nz
226 227
AC C
222
{[L88
(13)
" = 12.54 × 10H} ⋅ ~
(14)
" = 0.8 × 10H} ⋅
(15)
9
ACCEPTED MANUSCRIPT
228
where λ is the wavelength (nm), ~ is the irradiance of UVB and UVA, and is the irradiance of wavelengths
229
greater than 400 nm.
230 2.3 Numerical Implementation
232
The model has been implemented numerically using the method set out by Patankar (1980). A uniform grid is
233
employed (Fig. 2a) and the discretized algebraic equations are solved using a tridiagonal matrix solution.
234
Linear interpolation between atmospheric data recorded on site is used to calculate the boundary conditions and the
235
water column and soil heights were taken as 0.9 and 0.5 m, respectively. The chosen time step was modelled at 1
236
minute intervals. Grid and time-step size independence was confirmed by comparing the water surface temperature,
237
area-averaged temperature and water surface heat flux at the end of a 12 hr simulation.
238
3. Results
239
Observations of stratification and meteorological data show the diurnal nature of stratification and natural convection.
240
Inlet and outlet E. coli concentrations were measured and log reductions were calculated and compared with
241
simulations.
242
3.1 Diurnal mixing pattern
243
The diurnal mixing pattern recorded by the two vertical thermistor chains is shown clearly in Fig. 3c and d for a
244
seven-day period (Fig. 3c,d), with selected atmospheric variables also graphed; (a) air temperature and wind speed,
245
(b) solar radiation and atmospheric pressure. The wind speeds observed show that during night-time there are
246
generally small velocities, close to zero in magnitude, and greater velocities in the order of 3-4 m s-1 during the day.
247
Observations of solar radiation indicate relatively clear days with intermittent cloud cover. The same pattern of
248
stratification was observed at different locations as illustrated by comparing Fig. 3c and d. Greater stratification and
249
temperatures can be seen in the data recorded by thermistor chain 2 (Fig. 3d) which was located on the west side,
250
opposite thermistor chain 1 (Fig. 3c). The large daytime wind speeds were observed to move from east to west. The
251
cyclic nature of diurnal stratification has also been observed in the same pond during winter (data not shown).
252
Fig. 4 shows a close-up of the measured and simulated temperatures from February 28 to March 1. As can be seen,
253
thermal transport is remarkably different at different times of the day. For example, before 8 am on March 1, the
AC C
EP
TE D
M AN U
SC
RI PT
231
10
ACCEPTED MANUSCRIPT
pond is well mixed with all sensors showing almost the same temperature. As the day progresses, from about 9 am to
255
2 pm the thermal field becomes stably stratified with the highest temperatures near the surface. After 2 pm the water
256
surface starts to cool and unstable temperature gradients cause mixing to gradually progress from top to bottom until,
257
by about 10 pm, the fluid is well mixed vertically with all sensors showing the same temperatures again. These
258
features are clearly evident in both the simulation and the experimental results. Comparison of the interfacial
259
temperature between the experimental and observed data (blue line in Fig. 4a and b) shows that field conditions are
260
much more gradual.
261
In relation to the model, the approximation used for night-time convection (keddy(unstable) in Eq. 3) reproduces the
262
effect shown in the experimental data. The buoyantly unstable temperature gradient decays from the surface in layers
263
(Fig. 4b 4pm-10pm on Feb 28) following the assumption of Fig. 2b. The turbulent nature of this phenomenon has
264
been shown experimentally in Bednarz et al. (2009a,b) and indicates that there is significant mixing during this
265
process.
266
3.2 Disinfection of E. coli
267
E. coli enumeration from the inlet and outlet are presented in Fig. 5a and are graphed on their measured date. Inlet
268
and outlet measurements exhibit diurnal variation and the mean inlet E. coli concentration from samples was
269
1.48 × 10 CFU 100 ml-1. To calculate log reductions, outlet concentrations of E. coli were for t + 16 d, where t is
270
the time the sample was taken from the inlet. These data are presented in Fig. 5b on the sampled outlet date. The
271
reduction is fairly consistent and centred on a mean of 2.1 log units. The daytime samples (square symbols) can
272
clearly be seen to have a higher reduction (lower concentration) to the night and early morning samples, indicating
273
the importance of resolving the vertical distribution.
274
The period of 22 February 2015 – 10 March 2015 has been modelled and the surface concentration for a ‘plug’ of
275
water progressing through the pond is shown in Fig. 6. Results using the three different decay equations (Eqs 13-15)
276
are presented. In addition, two further simulations are presented, both utilizing die-off equation Eq. 13. The two
277
additional simulations show the effect of (1) complete day and night-time vertical mixing by artificially increasing
278
k eddy and (2) daytime transport by molecular diffusion only and night-time natural convection. Experimental E. coli
279
log reductions that were introduced in Fig. 5b have been represented by the symbol in Fig. 6 which corresponds to
280
the case of E. coli entering the pond on 22nd February and the plug of fluid reaching the outlet on the 10th March. For
AC C
EP
TE D
M AN U
SC
RI PT
254
11
ACCEPTED MANUSCRIPT
this case, a reduction in concentration of 2 orders of magnitude is observed. The error bars correspond to the
282
uncertainty in the pond residence time and scatter in the measured concentrations (see Fig. 5b).
283
Comparing the results from the different die-off kinetics shows that the UV dependent equations of Nguyen et al.
284
(2015) and Maïga et al. (2009a) are very similar in magnitude (Eq. 13 and 14). However, the sunlight dependence
285
proposed by Maïga et al. (2009a) shows a far higher reduction (Eq. 15). Eq. 15 should produce similar results to Eq.
286
14 due to the equations being regressed against the same experimental data, and indeed they do produce similar
287
results as the depth tends to zero. However, as the depth increases, the combination of wavelength dependent
288
attenuation and vertical concentration distribution becomes significant and the longer wavelengths produce a greater
289
die-off effect. Considering that there are likely other disinfection mechanisms and influences occurring (i.e. dark
290
inactivation, temperature, pH, DO) which have not been simulated here, the UV dependent results appear quite
291
reasonable.
292
The daytime peaks in surface concentration observed in Fig. 6 can be explained by the vertical distributions of
293
normalized E. coli concentrations which are shown in Fig. 4d. The applied decay rate for Fig. 4d is given by Eq. 13.
294
The distribution is very similar in nature to the thermal distributions. During daytime and for a stable density
295
gradient, it can be seen that the surface concentration is inactivated more quickly than the concentration beneath it.
296
Mixing from wind shear causes the near surface concentrations to mix together, and although inactivation is
297
vertically distributed within this region, die-off appears to be uniform as a result (see red/black line in Fig. 4d). For
298
the case of complete vertical mixing, this effect occurs over the entire water column and as a result, the modelled
299
surface concentration seen in Fig. 6 has increased die-off compared to a stratified water column. At the opposite
300
extreme, vertical transport due to molecular diffusion and natural convection only is observed to produce large
301
surface reductions. However, due to the shallow surface layer daytime inactivation occurs in, the majority of the
302
concentration beneath is not inactivated and mixes to the surface so that the overall die-off is significantly reduced
303
compared to complete mixing. Thus, the daily fluctuations in concentration at the surface sampling point in both the
304
model and experimental results are strong evidence that stratification and sunlight are contributing significantly to
305
the die-off of pathogens.
AC C
EP
TE D
M AN U
SC
RI PT
281
306 307
4. Discussion
12
ACCEPTED MANUSCRIPT
The analysis presented in this study is based upon the empirically derived eddy diffusivity concept, for which the
309
turbulent heat and species transport are dependent upon. Therefore it is valuable to discuss briefly the limitations and
310
assumptions made in the model formulation and their impact on the interpretation of simulation results.
311
4.1. Physical considerations and model effects
312
There are several model assumptions which have been made in the model formulation which, while they greatly
313
simplify the solution, may prove significant to the transport of heat and E. coli. The first is the neglect of stream-wise
314
(i.e. horizontal) mixing. The mixing problem is multi-dimensional while our one-dimensional plug-flow model
315
considers vertical mixing only. The merit of the one-dimensional formulation is that it is the simplest way possible to
316
account for both the effect of sunlight penetration and vertical mixing leading to a practical method of calculation for
317
pond design. However, it makes it impossible to directly account for large scale recirculation (c.f. Badrot-Nico et al.
318
(2009)) and the subsequent dispersion that is occurring. In our simulation we indirectly accounted for stream-wise
319
dispersion by shortening the residence time from that of simple plug flow.
320
Concerning the Eulerian model for E. coli transport and the choice of Lewis number, the actual dispersion behaviour
321
for self-propelled or swimming particles is likely to be significantly more complex than can be captured by Eq. (2)
322
(Wensink et al., 2012). Nevertheless, consistency with the level of complexity in the thermal diffusion model makes
323
Eq. 2 an appropriate choice for the present study in the absence of precise data for E. coli dispersion.
324
Model applicability is primarily limited to ponds which operate close to plug-flow conditions, such as baffled ponds,
325
as stream-wise dispersion is neglected. Although this study is applied in subtropical conditions, the limit of model
326
application to different climatic regions is only limited by freezing conditions which alter dispersion and sunlight
327
penetration.
328
4.2 E. coli reduction effects
329
Here the main assumption in the model is that sunlight-mediated mechanisms (Eq. 13-15) were the primary
330
mechanisms for inactivation with effects of other environmental conditions being of secondary influence (i.e.
331
temperature, pH, predation, sedimentation). Moreover, even amongst the three light-mediated inactivation
332
mechanisms, Eq. 13 is strictly valid only for the endogenous UV inactivation mechanism, while Eqs. 14 and 15 were
333
regressed against field data which should explain the entire sunlight driven die-off observed. E. coli has been
334
reported as being insensitive to the mechanism of direct damage by UVB irradiation and the exogenous inactivation
AC C
EP
TE D
M AN U
SC
RI PT
308
13
ACCEPTED MANUSCRIPT
mechanism non-applicable (Kadir and Nelson, 2014; Nguyen et al., 2015), therefore leaving the majority of die-off
336
to occur through a UV driven endogenous mechanism. Clearly, the results of Eqs. 14 and 15 are very similar (see Fig.
337
6), indicating that the effect of Eq. 14 is interpreted appropriately as an endogenous mechanism. The difference in
338
magnitude between the simulated UV decay results and the measured E. coli concentrations shown in Fig. 6 could
339
potentially be explained by the synergistic effect of elevated pH and DO which would occur throughout the day from
340
algae photosynthesis (Davies-Colley et al., 1999; Kadir and Nelson, 2014).
341
Applying an Eulerian transport model to represent microorganisms imposes the limitation of not being able to
342
account for time dependent inactivation. This is due to the inability to track specific microorganisms and may
343
become important when interpreting the simulation results of Fig. 6 which suggest that constant vertical mixing will
344
produce greater die-off. However, this physically involves microorganisms cycling vertically in the water column
345
and receiving an intermittent sunlight dose. The effect of this cyclic nature has not yet been elucidated in the
346
literature and thus the model results are based on the current knowledge.
347
In addition to the consideration of E. coli cycling in and out of the radiation, the decay equations (Eq. 14) are linear
348
(log reductions) with increasing energy dose. However, it has recently been reported that increasing the energy dose
349
to E. coli does not produce a linear trend in log reductions against UV exposure time, but instead the die-off plateaus,
350
causing no additional inactivation (Kollu and Örmeci, 2012). The experimental evidence for this conclusion was
351
collected under laboratory conditions with UVC irradiation (at 253 nm) so further confirmation may be required for
352
the longer wavelengths occurring in the wastewater environment.
353
Dark inactivation is generally present during field observations and may cause additional reductions overnight and
354
beneath the thermocline during the day (Craggs et al., 2004; Nguyen et al., 2015). This may be countered by repair
355
mechanisms which are believed to occur within WSPs (Sinton et al., 2002). From laboratory results, Bohrerova et al.
356
(2015) did not identify an irradiation threshold at which repair mechanisms would not occur, although the total repair
357
was relatively low. Regrowth, as opposed to repair, showed a 3 log unit increase in concentration over a 48 hr
358
period in which the E. coli was subject to no further inactivation. Kollu and Örmeci (2015) further showed that
359
regrowth was more significant and could prevail for longer periods after being subject to higher UV intensities. They
360
suggested the reason was due to the damaged cells providing nutrients to those surviving and that in nutrient rich
361
wastewater the effect would be even more pronounced.
AC C
EP
TE D
M AN U
SC
RI PT
335
14
ACCEPTED MANUSCRIPT
Sheludchenko et al. (2016) measured E. coli through a baffled maturation pond and reported that the most significant
363
reduction occurred within the first half of the pond. Nguyen et al. (2015) also noted this, showing that the most
364
significant reduction was immediately after the inlet, however monthly variation showed that in certain months
365
further reduction could be gained through the remaining length of pond, and in other months no further reduction
366
was found. The effect of this observation was also noted by Blaustein et al. (2013) who identified that a piece-wise
367
linear trend with respect to time was most common amongst temperature dependence studies in dark conditions in
368
laboratory experiments. Temperature, however, is not a primary cause of inactivation when lower than 45 °C
369
(McGuigan et al., 1998; Craggs et al., 2004). Therefore, to explain the results of these studies, there may be another
370
mechanism occurring within WSPs immediately after entering that causes a large die-off that is not seen through the
371
remainder of the pond, such as predation or simply dilution by horizontal mixing.
372
Considering that the sunlight decay equations (e.g. Eq. 14) and many temperature-dependent relationships (e.g. de
373
Brauwere et al., 2011; Blaustein et al., 2013) are first order and if applied to the distribution model (Eq. 2), they will
374
produce log-linear die-off patterns similar to the calculated results shown in Fig. 6 for our plug flow model.
375
Therefore a more complete kinetic model should be based on a detailed understanding of the individual mechanisms
376
rather than regression of field data against one parameter, such as temperature or UV light exposure. Based on
377
laboratory and field data by Maïga et al. (2009a) and Nguyen et al. (2015), the present model has demonstrated that
378
the interaction of sunlight damage and stratification plays a significant role. The diurnal pattern with lower near-
379
surface E. coli concentrations measured during daylight hours supports this. Extending the present model to include
380
other mechanisms, once the details and appropriate rate-kinetics have been established, is straight forward.
381
5. Conclusion
382
A one-dimensional model for temperature distribution, coupled with a concentration distribution is proposed. It has
383
been shown to perform well against measured data and provides further elucidation of the physics behind
384
observations made of pathogen concentrations in a maturation pond. It is evident that daytime stratification causes
385
higher reductions within the surface region of the pond from sunlight disinfection. Vertical mixing from natural
386
convection at night-time brings the non-disinfected, daytime concentration, to the surface and the subsequent
387
outflow concentration is much higher at night. The measured outflow E. coli concentration from the maturation pond
388
had far higher fluctuations indicating that other disinfection mechanisms are most likely occurring, or a higher
AC C
EP
TE D
M AN U
SC
RI PT
362
15
ACCEPTED MANUSCRIPT
importance to sunlight disinfection is required to reproduce this effect. Incorporation of altered equations for
390
turbulent diffusive transport and kinetic models is easily achieved, making the models’ application versatile. The
391
proposed model offers the simplest solution to solving pathogen die-off by including transient boundary conditions
392
and accounting for vertical mixing and inactivation, thus offering clear advantages to current depth-integrated
393
models.
394
Acknowledgements
395
We would like to acknowledge the funding provided from the Australian Water Recycling Centre of Excellence, the
396
Queensland Government Science Fund Support and Queensland Urban Utilities.
397
References
398
Alvarado, A., Sanchez, E., Durazno, G., Vesvikar, M., Nopens, I., 2012. CFD analysis of sludge accumulation and
402 403 404 405
SC
M AN U
401
APHA, 1992. Standard Methods for the Examination of Water and Wastewater, 18th ed. American Public Health Assosiation, Washington, DC, USA.
Ashworth, J., Skinner, M., 2011. Waste Stabilisation Pond Design Manual, Power and Water Corporation, Northern Territory, Australia.
TE D
400
hydraulic performance of a waste stabilization pond. Water Science and Technology 66 (11), 2370-2377.
Badrot-Nico, F., Guinot, V., Brissaud, F., 2009. Fluid flow pattern and water residence time in waste stabilisation ponds. Water Science and Technology 59 (6), 1061–1068.
EP
399
RI PT
389
Bednarz, T.P., Lei, C., Patterson, J.C., 2009a. An experimental study of unsteady natural convection in a reservoir
407
model subject to periodic thermal forcing using combined PIV and PIT techniques. Experiments in Fluids 47 (1),
408
107–117.
409 410 411 412
AC C
406
Bednarz, T.P., Lei, C., Patterson, J.C., 2009b. A numerical study of unsteady natural convection induced by iso-flux surface cooling in a reservoir model. International Journal of Heat and Mass Transfer 52 (1-2), 56–66. Bichai, F., Smeets, P.W., 2013. Using QMRA-based regulation as a water quality management tool in the water security challenge: experience from the Netherlands and Australia. Water Research 47 (20), 7315-7326.
16
ACCEPTED MANUSCRIPT
414 415
Blaustein, R.A., Pachepsky, Y., Hill, R.L., Shelton, D.R., Whelan, G., 2013. Escherichia coli survival in waters: temperature dependence. Water Research 47 (2), 569-578. Bohrerova, Z., Rosenblum, J., Linden, K.G., 2015. Importance of recovery of E. coli in water following ultraviolet
416
light
417
http://dx.doi.org/10.1061/(ASCE)EE.1943-7870.0000922.
disinfection.
Journal
of
Environmental
Engineering
141
(6),
04014094.
RI PT
413
Bolton, N.F., Cromar, N.J., Hallsworth, P., Fallowfield, H.J., 2010. A review of the factors affecting sunlight
419
inactivation of micro-organisms in waste stabilisation ponds: preliminary results for enterococci. Water Science
420
and Technology 61 (4), 885–890.
423 424 425 426 427 428
in a stabilisation pond. Water Science and Technology 48 (2), 75-80.
M AN U
422
Brissaud, F., Tournoud, M.G., Drakides, C., Lazarova, V., 2003. Mixing and its impact on faecal coliform removal
Chang, K.A., Cowen, E.A., 2002. Turbulent Prandtl number in neutrally buoyant turbulent round jet. Journal of Engineering Mechanics 128 (10), 1082-1087.
Craggs, R.J., Zwart, A., Nagels, J.W., Davies-Colley, R.J., 2004. Modelling sunlight disinfection in a high rate pond. Ecological Engineering 22 (2), 113-122.
TE D
421
SC
418
Curtis, T.P., Mara, D.D., Dixo, N.G.H., Silva, S.A., 1994. Light penetration in waste stabilization ponds. Water Research 28 (5), 1031-1038.
Curtis, T.P, Mara, D.D, Silva, S.A., 1992a. Influence of ph, oxygen, and humic substances on ability of sunlight to
430
damage fecal coliforms in waste stabilization pond water. Applied and Environmental Microbiology 58 (4), 1335–
431
1343.
433 434 435 436 437
AC C
432
EP
429
Curtis, T.P., Mara, D.D., Silva, S.A., 1992b. The effect of sunlight on faecal coliforms in ponds: implications for research and design. Water Science and Technology 26 (7-8), 1729-1738. Dake, J., Harlem, D., 1969. Thermal stratification in lakes: analytical and laboratory studies. Water Resources Research 5 (2), 484–495. Davies-Colley, R.J., Craggs, R., Park, J., Nagels, J., 2005. Optical characteristics of waste stabilization ponds: recommendations for monitoring. Water Science and Technology 51 (12), 153–161.
17
ACCEPTED MANUSCRIPT
438 439
Davies-Colley, R.J., Donnison, A.M., Speed, D.J., 2000. Towards a mechanistic understanding of pond disinfection. Water Science and Technology 42 (10-11), 149–158. Davies-Colley, R.J., Donnison, A.M., Speed, D.J., Ross, C.M., Nagels, J.W., 1999. Inactivation of faecal indicator
441
microorganisms in waste stabilisation ponds: interactions of environmental factors with sunlight. Water Research
442
33 (5), 1220-1230.
445 446 447 448 449 450
concentrations in the tidal Scheldt river and estuary. Water Research 45 (9), 2724-2738.
SC
444
de Brauwere, A., de Brye, B., Servais, P., Passerat, J., Deleersnijder, E., 2011. Modelling Escherichia coli
Elyasi, S., Taghipour, F., 2006. Simulation of UV photoreactor for water disinfection in Eulerian framework. Chemical Engineering Science 61 (14), 4741-4749.
M AN U
443
RI PT
440
Gliah, O., Kruczek, B., Etemad, S.G., Thibault, J., 2011. The effective sky temperature: an enigmatic concept. Heat and Mass Transfer 47 (9), 1171-1180.
Gu, R., Stefan, H.G., 1995. Stratification dynamics in wastewater stabilization ponds. Water Research 29 (8), 1909– 1923.
Henderson-Sellers, B., 1984. Engineering Limnology. Pitman Advanced Publishing Program, Boston, Massachusetts.
452
Henderson-Sellers, B., 1985. New formulation of eddy diffusion thermocline models. Applied Mathematical
455 456 457 458 459 460 461
Herb, W.R., Stefan, H.G., 2005. Dynamics of vertical mixing in a shallow lake with submersed macrophytes. Water
EP
454
Modelling 9 (6), 441-446.
Resources Research 41, http://dx.doi.org/10.1029/2003WR002613. Incropera, F., DeWitt, D., 1996. Fundamentals of heat and mass transfer. fourth ed. John Wiley & Sons, Inc., United
AC C
453
TE D
451
States.
Jacobs, A.F.G., Heusinkveld, B.G., Kraai, A., Paaijmans, K.P., 2008. Diurnal temperature fluctuations in an artificial small shallow water body. International Journal of Biometeorology 52 (4), 271–280. Jemli, M., Alouini, Z., Sabbahi, S., Gueddari, M., 2002. Destruction of fecal bacteria in wastewater by three photosensitizers. Journal of Environmental Monitoring 4 (4), 511-516.
18
ACCEPTED MANUSCRIPT
465 466 467 468 469 470 471
Kimball, B.A., 1983. Conduction transfer functions for predicting heat fluxes into various soils. Transactions of the ASAE 26 (1), 211-218.
RI PT
464
coli in clear water versus waste stabilization pond water. Water Research 50, 307–317.
Kirk, J.T.O., 2010. Light and Photosynthesis in Aquatic Ecosystems, third ed. Cambridge University Press, Cambridge, United Kingdom.
Kohn, T., Nelson, K.L., 2007. Sunlight-mediated inactivation of ms2 coliphage via exogenous singlet oxygen
SC
463
Kadir, K., Nelson, K.L., 2014. Sunlight mediated inactivation mechanisms of Enterococcus faecalis and Escherichia
produced by sensitizers in natural waters. Environmental Science and Technology 41 (1), 192–197. Kollu, K., Örmeci, B., 2012. Effect of particles and bioflocculation on ultraviolet disinfection of Escherichia coli. Water Research 46 (3), 750-760.
M AN U
462
472
Kollu, K., Örmeci, B., 2015. Regrowth potential of bacteria after ultraviolet disinfection in the absence of light and
473
dark repair. Journal of Environmental Engineering 141 (3), 04014069. http://dx.doi.org/10.1061/(ASCE)EE.1943-
474
7870.0000905.
Lamoureux, J., Tiersch, T.R., Hall, S.G., 2006. Pond heat and temperature regulation (PHATR): Modeling
476
temperature and energy balances in earthen outdoor aquaculture ponds. Aquacultural Engineering 34 (2), 103–116.
477
Losordo, T.M., Piedrahita, R.H., 1991. Modelling temperature variation and thermal stratification in shallow aquacultural ponds. Ecological Modelling 54 (3-4), 189–226.
EP
478
TE D
475
Lumley, J., Panofsky, H., 1964. The structure of atmospheric turbulence. Intersience Publishers, New York.
480
Maïga, Y., Denyigba, K., Wethe, J., Ouattara, A.S., 2009a. Sunlight inactivation of Escherichia coli in waste
481
stabilization microcosms in a Sahelian region (Ouagadougou, Burkina Faso). Journal of Photochemistry and
482
Photobiology B: Biology 94 (2), 113–119.
AC C
479
483
Maïga, Y., Wethe, J., Denyigba, K., Ouattara, A.S., 2009b. The impact of pond depth and environmental conditions
484
on sunlight inactivation of Escherichia coli and enterococci in wastewater in a warm climate. Canadian Journal of
485
Microbiology 55 (12), 1364-1374.
19
ACCEPTED MANUSCRIPT
486
Maraccini, P.A., Mattioli, M C., Sassoubre, L.M., Cao, Y., Griffith, J.F., Ervin, J.S., Van De Werfhorst, L.C., 2016a.
487
Solar Inactivation of Enterococci and Escherichia coli in Natural Waters: Effects of Water Absorbance and Depth.
488
Environmental Science and Technology 50 (10), 5068-5076. Maraccini, P.A., Wenk, J., Boehm, A.B., 2016b. Photoinactivation of eight health-relevant bacterial species:
490
Determining the importance of the exogenous indirect mechanism. Environmental Science and Technology 50
491
(10), 5050-5059.
RI PT
489
McGuigan, K.G., Joyce, T.M., Conroy, R.M., Gillespie, J.B., Elmore-Meegan, M., 1998. Solar disinfection of
493
drinking water contained in transparent plastic bottles: characterizing the bacterial inactivation process. Journal of
494
Applied Microbiology 84 (6), 1138-1148.
496
Nelson, K., 2000. Ultraviolet light disinfection of wastewater stabilization pond effluents. Water Science and Technology 42 (10-11), 165–170.
M AN U
495
SC
492
Nguyen, M.T., Jasper, J.T., Boehm, A.B., Nelson, K.L., 2015. Sunlight inactivation of fecal indicator bacteria in
498
open-water unit process treatment wetlands: Modeling endogenous and exogenous inactivation rates. Water
499
Research 83, 282–292.
TE D
497
500
Paaijmans, K.P., Jacobs, A.F.G., Takken, W., Heusinkveld, B.G., Githeko, A.K., Dicke, M., Holtslag, A.A.M., 2008.
501
Observations and model estimates of diurnal water temperature dynamics in mosquito breeding sites in western
502
Kenya. Hydrological Processes 22 (24), 4789-4801.
505 506 507 508 509 510
EP
504
Pandey, D., Lee, R., Paden, J., 1995. Effects of atmospheric emissivity on clear-sky temperatures. Atmospheric Environment 29 (16), 2201-2204.
AC C
503
Passos, R.G., von Sperling, M., Ribeiro, T.B., 2014. Hydrodynamic evaluation of a full-scale facultative pond by computational fluid dynamics (CFD) and field measurements. Water Science and Technology 70 (3), 569-575. Patankar, S.V., 1980. Numerical heat transfer and fluid flow. Series in computational methods in mechanics and thermal sciences, Hemisphere Pub. Corp, New York. ISBN 0-891-16522-3. Pérez-Garcı́a, M., 2004. Simplified modelling of the nocturnal clear sky atmospheric radiation for environmental applications. Ecological Modelling 180 (2-3), 395-406.
20
ACCEPTED MANUSCRIPT
513 514 515 516 517
epilimnion of north-temperate Lake Opeongo, Canada. Aquatic Sciences 76 (2), 187-201. Reynolds, A.J., 1975. The prediction of turbulent Prandtl and Schmidt numbers. Internation Journal of Heat and Mass Transfer 18 (9), 1055-1069.
RI PT
512
Pernica, P., Wells, M.G., MacIntyre, S., 2013. Persistent weak thermal stratification inhibits mixing in the
Scheierling, S.M., Bartone, C.R., Mara, D.D., Drechsel, P., 2011. Towards an agenda for improving wastewater use in agriculture. Water International 36 (4), 420-440.
Schuch, A.P., Menck, C.F., 2010. The genotoxic effects of DNA lesions induced by artificial UV-radiation and
518
sunlight.
519
http://dx.doi.org/10.1016/j.jphotobiol.2010.03.004.
of
Photochemistry
and
Photobiology
B:
M AN U
Journal
SC
511
Biology
99
(3),
111-116.
520
Sheludchenko, M., Padovan, A., Katouli, M., Stratton, H., 2016. Removal of fecal indicators, pathogenic bacteria,
521
adenovirus, cryptosporidium and giardia (oo)cysts in waste stabilization ponds in northern and eastern Australia.
522
International
523
http://dx.doi.org/10.3390/ijerph13010096.
525
of
Environmental
Research
and
Public
Health
13
(1),
96.
Shilton, A., Harrison, J., 2003. Integration of coliform decay within a CFD (computational fluid dynamic) model of a
TE D
524
Journal
waste stabilisation pond. Water Science and Technology 48 (2), 205-210. Sinton, L.W., Hall, C.H., Lynch, P.A., Davies-Colley, R.J., 2002. Sunlight inactivation of fecal indicator bacteria
527
and bacteriophages from waste stabilization pond effluent in fresh and saline waters. Applied and Environmental
528
Microbiology 68 (3), 1122–1131.
530
Smith, I.R., 1979. Hydraulic conditions in isothermal lakes. Freshwater Biology 9 (2), 119–145.
AC C
529
EP
526
http://dx.doi.org/10.1111/j.1365-2427.1979.tb01496.x.
531
Spencer, J.W., 1971. Fourier series representation of the position of the sun. Search 2, 172.
532
Sundaram, T.R., Rehm, R.G., 1973. The seasonal thermal structure of deep temperate lakes. Tellus 25 (2), 157–168.
533 534
http://dx.doi.org/10.1111/j.2153-3490.1973.tb01602.x. Tennekes, H., Lumley, J.L., 1972. A first course in turbulence. MIT Press, USA pp 143-144.
21
ACCEPTED MANUSCRIPT
537 538 539 540 541 542
Atmospheric Environment 41 (37), 8091-8099. Weigand, B., Ferguson, J.R., Crawford, M.E., 1997. An extended Kays and Crawford turbulent Prandtl number model. Internation Journal of Heat and Mass Transfer 40 (17), 4191-4196.
RI PT
536
Tominaga, Y., Stathopoulos, T., 2007. Turbulent Schmidt numbers for CFD analysis with various types of flowfield.
Wensink, H.H., Dunkel, J., Heidenreich, S., Drescher, K., Goldstein, R.E., Lowen, H., Yeomans, J.M., 2012. Mesoscale turbulence in living fluids. PNAS 109 (36), 14308-14313.
Yang, P., Xing, Z., Fong, D.A., Monismith, S.G., Tan, K.M., Lo, E.Y.M., 2015. Observations of vertical eddy
SC
535
diffusivities in a shallow tropical reservoir. Journal of Hydro-environment Research 9 (3), 441-451.
AC C
EP
TE D
M AN U
543
22
ACCEPTED MANUSCRIPT
Figure Captions:
545
Fig. 1 - Maturation pond layout showing: (a) the location of two thermistor chains, baffles, north direction and inlet
546
(thermistor chain 1 is located at the outlet) and (b) a thermistor chain schematic showing dimensions and sensor
547
numbers.
RI PT
544
548
Fig. 2 - Model concept emphasising (a) example control volumes (one control volume for each calculation location),
550
calculation locations (filled circles) and material location limits (bounded by solid lines) with example eddies in the
551
water column. (b) An unstable temperature gradient is approximated in the simulation as an equivalent temperature
552
where the two hatched areas equate to the same energy.
M AN U
SC
549
553 554
Fig. 3 - Experimental observations of (a) air temperature and wind speed, (b) solar radiation and atmospheric
555
pressure, and water column temperatures recorded by thermistor chains 1 and 2 for (c) and (d), respectively.
556
Fig. 4 – Comparison of (a-b) experimental temperatures to modelled results of (c) temperature and (d) E. coli. The
558
period starts from 28 February 2015 6 am for two days. Simulation results illustrate periods of stratification and
559
mixing. Legend labels align with thermistor positions given by Fig. 1b.
TE D
557
EP
560
Fig. 5 – Experimental data showing the (a) average E. coli concentrations of duplicate samples measured from the
562
inlet and outlet and (b) E. coli log reductions highlighting that daytime (10 am – 2 pm) sampled concentrations are
563
mostly lower than night-time concentrations.
564
AC C
561
565
Fig. 6 - Simulated log reductions from the surface element and experimental points located between a theoretical
566
residence time of 12 to 20 days. The effect of complete mixing and night-time convection are shown as additional
567
simulations.
568
23
ACCEPTED MANUSCRIPT
Tables:
570
Table 1 – Observed averages of radiation parameters with 1 standard deviation shown in parenthesis. Number of
571
samples was 7 throughout a single day. Spectra
R+\
UV Fraction
r
-
m-1
UVB
5.61 (0.2)
0.54 (0.1)
39.2 (3.24)
UVA
94.4 (0.2)
0.48 (0.1)
44.6 (5.3)
PAR
-
0.39 (0.09)
18.5 (3.2)
M AN U
SC
%
RI PT
569
AC C
EP
TE D
572
24
ACCEPTED MANUSCRIPT
35 m
Buoy S1
5m
Outlet
M AN U
60 m
S2
S4
AC C
EP
TE D
N
S3 0.9 m
1
RI PT
5m
SC
2 3m
Inlet S5
a)
Weight b)
RI PT
T
SC
z(+ve)
M AN U
WL
Control Volumes
ACCEPTED MANUSCRIPT
SL
AC C
EP
TE D
Water
Soil
a) Model Concept
b) Unsteady Thermal Gradient
a)
7.5
20
5
10
2.5
SC
30
0 1500
−2
b)
104
1000
103 102
500
101
0
0.6
0
30
TE D
0.4
35
0.8 d) Chain 2 0.6
EP
0.4
25
0.2
AC C
0 22
23
24 25 26 27 Time (day of February 2015)
28
01
20
Temperature (°C)
100
0.2
Radiation (W m )
0 105
0.8 c) Chain 1
Water Column Height (m)
RI PT −1
10
Wind Speed (m s )
40
M AN U
Atmospheric Pres. (kPa)
Air Temperature (°C)
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
30
b) Experimental Chain 2
35
SC
25
30 25 40
c) Modelled temperature 35 30 25
surface mixed stratified bottom
TE D
Temperature (°C)
RI PT
S1 S2 S3 S4 S5
35
40 Temperature (°C)
a) Experimental Chain 1
M AN U
Temperature (°C)
40
stratified
0.4
completely mixed
d) Modelled E. coli
natural convection progessive mixing
C/C 0
EP
0.3
night time
AC C
0.2
0.1 06
surface die−off continuously mixed layer
12
18 00 06 12 18 00 Time (day hour of 28 Feb / 01 March 2015)
06
6
a) Experimental raw data
10
5
10
4
M AN U
Enumeration (CFU 100 mL−1)
7
10
10
3
10
Inlet Outlet
2
10
1
10
0
b) Calculated E. coli log
10
−2 −3 −4
reduction
TE D
log10 C/C 0
−1
SC
RI PT
ACCEPTED MANUSCRIPT
Night−time
Daytime
AC C
EP
−5 05 F 08 F 11 F 14 F 17 F 20 F 23 F 26 F 01 M 04 M 07 M 10 M Time (day of February/March 2015)
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
0
log 10 C/C 0
−1
−2
−3
TE D
−4
daytime peak Eq. 13 Eq. 14 Eq. 15 Experiment Eq. 13 Complete vertical mixing Eq. 13 Natural convection only
AC C
EP
−5 22 F 24 F 26 F 28 F 02 M 04 M 06 M 08 M 10 M 12 M 14 M Time (day of February/March 2015)
ACCEPTED MANUSCRIPT
A practical model for sunlight disinfection of a subtropical maturation pond Highlights:
EP
TE D
M AN U
SC
RI PT
A practical model for the prediction of E. coli die-off is proposed Vertical sunlight inactivation and mixing is accounted for Stratification isolates bottom pathogens from sunlight Night-time mixing redistributes pathogens for following day Transient boundary conditions reproduce effects of observed weather conditions
AC C
• • • • •