Chemical Engineering Journal 268 (2015) 67–75

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Algae-dewatering using rotary drum vacuum filters: Process modeling,

simulation and techno-economics
Pinghai Shao a,⇑, Ken Darcovich a, Thom McCracken a, Guillermo Ordorica-Garcia a, Michael Reith b,
Stephen O’Leary b
a
Energy, Mining and Environment, National Research Council Canada, 1200 Montreal Road, Ottawa, Ontario K1A 0R6, Canada
b
Aquatic and Crop Resource Development, National Research Council Canada, 1411 Oxford Street, Halifax, N.S. B3H 3Z1, Canada

h i g h l i g h t s

 Mass transport was significantly affected by algal cake-layer compressibility.

 Cake-layer pore sizes were evaluated to determine the bursting pressure.
 Process capital investment and operating costs were assessed.
 The optimal operating conditions and minimum dewatering cost were identified.
 Two cost-sensitive domains were discovered to avoid.

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

Article history: Clean and energy-efficient rotary drum vacuum filtration was selected to conduct algae-dewatering. The
Received 13 November 2014 dynamic formation of an algal cake-layer on the filter surface was modeled by correlating the cake-layer
Received in revised form 22 December 2014 permeability to the physical parameters of algae and cake-layer. The compressibility of algal cake-layer
Accepted 9 January 2015
was taken into consideration in the modeling, and its effect on the algae-dewatering is discussed.
Available online 19 January 2015
The dewatering process was simulated to determine the process energy demand. Process economics
were assessed considering the dewatering cost, which includes capital investment and energy cost and
Keywords:
also labor, installation, maintenance and infrastructure. Optimal operating conditions and minimum
Algae-dewatering
Filtration
dewatering cost were achieved by process optimization, and two cost-sensitive zones in operating the
Process modeling and simulation filtration were identified. The techno-economics showed that the dewatering cost can be further reduced
Capital and energy by scaling up the process.
Process optimization Crown Copyright Ó 2015 Published by Elsevier B.V. All rights reserved.
Techno-economics

1. Introduction environment [13]. Cultivation of algae consumes large amounts

Filtration is a mature technology for the removal of particulate
matters from fluids. It has found wide industrial applications in
water treatment (e.g. turbidity removal) [1–3], food processing
(e.g. clarification of beer and liquors) [4–6], air cleaning (e.g. dust
and bacteria removal), etc. [7]. At present, various filter media
are available for the removal of these small particulates ranging
from submicron to 100 lm [8–11]. Filtration is based on size-
exclusion, so it does not require large energy input. A low pressure
(e.g. 20–30 kPa) can generate a high flux through these well-tuned
and highly porous filter media [12].
Microalgae, mostly in the range of 5–20 lm, are now playing a
key role in the ongoing endeavors to sustain energy and
⇑ Corresponding author. Tel.: +1 613 993 1700; fax: +1 613 991 2384.
E-mail address: pinghai.shao@nrc-cnrc.gc.ca (P. Shao).
of CO2, as a carbon source for algal biomass synthesis. Because
algae grow fast and possess high lipid contents, algae have been
mass cultivated to provide renewable energy and at the same time
mitigate CO2 emissions [14–16]. The concentration of algae in a
culture medium is usually 0.1 w/w%, and this concentration can
be enhanced up to 20 w/w% through centrifugal harvesting. The
traditional method to dewater this 20 w/w% harvested algae is
with thermal energy to evaporate most of the remaining 80% water
content [17]. Given the extremely high latent heat of water, algae-
dewatering by evaporation requires a very large energy input.
Alternative approaches have good potential to improve the energy
efficiency of algae-dewatering.
One major objective of our algal biofuel project is to maximize
the energy return to help better address energy sustainability. Each
operating unit of this process can benefit from use of energy-effi-
cient technology. Energy-efficient filtration with the rotary drum

http://dx.doi.org/10.1016/j.cej.2015.01.029
1385-8947/Crown Copyright Ó 2015 Published by Elsevier B.V. All rights reserved.
68 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75
vacuum filter (RDVF) was selected to conduct algae-dewatering.
Continuous operation of filtration is favored by industry for its flex-
ibility in process scale-up and for not needing extra storage space
for intermediate product. In addition, by using the vacuum opera-
tion, the amount of filtrate water contained in the porous algal
cake-layer can be minimized. Consequently, much of the heating
burden can be removed from the subsequent process for algae-dry-
ing, which has been identified as one of the bottlenecks limiting
the overall energy efficiency of algal biofuel technology [18,19].
To demonstrate the RDVF filtration process for algae-dewater-
ing, a mathematical model was developed to model the filtrate
water transport through the algal cake-layer as well as the under-
neath filter medium. The transport resistances of algal cake-layer
and filter medium were correlated to their physical properties.
Correlating the cake-layer resistance to its characteristic parame-
ters serves to make the model fundamental, straightforward and
readily applicable for simulation. Algae are a soft material, and
compression of the algal cake-layer in the filtration process must
be taken into consideration in this model, in order to properly sim-
ulate the observed behaviors of the algal cake-layer. A value was
determined for the equivalent pore size in the cake-layer, which
was a parameter required for modeling the algal cake-layer
dewatering.
Process economics is an important measure characterizing the
viability of a technology. Literature survey shows that little work
has been carried out on the cost analysis of algae dewatering
[20]. To determine costs of algae-dewatering with the filter, the
process capital and operational energy costs were evaluated. Pro-
cess optimization was performed to demonstrate the potential of
reducing this dewatering cost by identifying the best operating
conditions. Analysis of the global dewatering cost, which consisted
of energy, capital and installation, infrastructure, labor and mainte-
nance, suggests that the proportion of costs can be significantly
reduced through appropriate scale-up. Thus, scenarios for further
reducing this dewatering cost were also explored by considering
overall process scale.
2. An overview of algae-dewatering process
Algae-dewatering using a rotary drum vacuum filter is illus-
trated in Fig. 1. The rotary drum is partially submerged in the har-
vested algal feed. Water in the algae feed is drawn into the drum
Cake-dewatering compartment
Filtering compartment
Algae feed vessel
Fig. 1. Flow-chart of algae-dewatering using a rotary drum vacuum filter with three distinctive zones for algae-filtering, algae-dewatering, and algae-discharging.
through the filter medium to the low pressure zone through use
of a vacuum pump. Algae are thus held and accumulated on the
external surface of the filter to form a dynamic algal cake-layer.
The filtrate water passing through the filter and the drain piping
is first collected in the vacuum receiver and is finally moved out by
a water pump. The algal cake-layer keeps building up on the filter
surface until it is carried out of the feed tank as shown in Fig. 1. The
water contained in the algal cake-layer is removed in the subse-
quent algae-dewatering zone under a controlled vacuum pressure
during the portion of a drum revolution between exiting the liquid
region and prior to arriving at a scraping blade, which removes it
from the drum surface. This dewatered algal cake is peeled off
the drum filter surface and collected for drying.
3. Modeling of the constant pressure filtration of centrifuge-
harvested algae
3.1. Transport of filtrate water through the algal cake-layer
Modeling the dynamic formation of algal cake-layer on the filter
surface is a challenge. Some assumptions should first be made,
including (i) neglecting the effects of tangential shear induced by
filter rotation in the feed tank, on the formation of algal cake-layer
atop the filter surface, (ii) neglecting the effects of mixers in the
feed tank, and (iii) assuming all the algae moving toward the filter
surface with the water in the filtration process are retained, and
build up on the filter surface to form a uniform algal cake-layer
on the filter surface.
According to Darcy’s law, the flow-rate of the filtrate can be
written as [21–23],
dV A  Dp1
¼ ð1Þ
dt lR
where V is the cumulative volume of filtrate water, A is the effective
area for filtration, Dp1 is the pressure difference across the algal
cake-layer as shown in Fig. 2, l is viscosity of water, and R is the
hydraulic resistance of the cake-layer. Algal cake-layer resistance
is assumed to be proportional to the cake-layer thickness, ‘, by,
R ¼ bðr algae ; eÞ  ‘ ð2Þ
where b is the specific resistance of the cake-layer, related to ralgae
and porosity e of the algal cake-layer.
Cake-discharging compartment
Vacuum pump
Vacuum receiver
11
Water pump
 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75 69

Water

Algae

p1 Algae cake

p2 Filter medium

Fig. 2. Algal cake-layer containing network pathways for water filtrate and the structure of the underneath filter medium.

According to Kozeny and Carman [24–26], the specific resis- Inserting Eq. (11) into Eq. (7) yields,
tance is determined by,
1 V
Kð1  eÞ2 S2v ‘¼   ð12Þ
bðr algae ; eÞ ¼ ð3Þ Að1  eÞ 0:6qalgae  1
e3 C algae

where K is a constant equal to 5 for spherical packing [24,25], and Sv Substituting Eq. (12) to Eq. (6) yields
is the surface area contained in a unit solid volume of the packing.
Thus, dV Dp1
¼ ð13Þ
4p r 2alage dt  eÞlV
45ð1 
3 0:6qalgae
Sv ¼ 4 ¼ ð4Þ A 2 e3 r 2 C algae
1
algae
3
p r3algae ralgae
Eq. (13) is the fundamental equation governing water trans-
Inserting Eq. (4)into Eq. (3)yields port through the network channel as depicted in Fig. 2. Dp1 is
the driving force for water transport through the cake-layer,
45ð1  eÞ2
bðr algae ; eÞ ¼ ð5Þ and the denominator term represents the hydraulic resistance
e3 r2algae for the filtrate water, revealing that a larger filter area, a bigger
Inserting Eq. (5) into Eqs. (2) and (1) gives Eq. (6) algal size, a greater cake-layer porosity, and a lower algae concen-
tration all contribute to smaller hydraulic resistance for water
dV e3 r2algae A  Dp1 transport.
¼ ð6Þ
dt 45ð1  eÞ2 l  ‘

The thickness of algal cake-layer is determined by the volume of 3.2. Transport of filtrate water through the filter medium
the algae accumulated on the surface of filter medium by
ð1  eÞA‘ ¼ V algae ð7Þ The filter medium either in cloth, polyester mesh or nylon-
coated metal thread for the rotary drum vacuum filter is consid-
The volume fraction of algae in the algae feed is dependent
ered to have uniform permeability over its entire surface. To
upon the algae concentration, which is defined in Eq. (8). Note in facilitate the treatment of mass transport through these usually
wet algae, about 60% are dry algal biomass [17],
non-cylindrical pore structures, they are modeled as an array of
V algae qalgae  0:6 uniform cylindrical pores of a hydraulic radius. The equivalent pore
C algae ¼ ð8Þ size of these media is determined by [27] Eq. (14),
V feed
where V algae is the volume of the wet algae, and qalgae is the specific
2S
gravity of the wet algae. According to Eq. (8), the volume fraction of rmedium ¼ ð14Þ
L
algae in the algae feed can be written as:
V algae C algae where S is the intersectional area of the non-cylindrical pore, and L
/¼ ¼ ð9Þ
V feed 0:6qalgae is the circumference of the pore. The flow-rate of water through the
filter medium can thus be governed by the Hagen–Poiseuille Equa-
The volumes of algae and water filtrate are linked to the volume tion [28,29] as
fraction by Eq. (10)
V algae / dV Dp2
¼ ð10Þ ¼ 8la‘medium
ð15Þ
V 1/ dt
Aemedium r2
medium
Inserting Eq. (9) into Eq. (10) gives
For better filtration efficiency, a thin layer of pre-coat of fine
V particulates (e.g. diatomaceous earth) may need to be deposited
V algae ¼ 0:6q  ð11Þ
algae
1 on the surface of the filter medium. In such cases, an additional
C algae
resistance is included in Eq. (15) by using a correction factor, a.
70 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75

3.3. Transport of filtrate water through the algal cake-layer and the through direct image analysis of the algal cake-layer, the effect of
filter medium clogging, if any, on pore size can thus be assessed by a simplified
model as developed in this work. Thus-achieved pore size is an
The resistances of the algal cake-layer and the filter medium are equivalent pore size.
in-series, therefore these two resistances are additive and the flow- This equivalent pore size of the cake-layer is evaluated by
rate of the filtrate water can finally be formulated as, assuming that the porous algal cake-layer is comprised of an
ensemble of cylindrical pores (2r) that pose a same hydraulic resis-
dV Dp
¼ ð16Þ tance for water transport as the network pathway does. Therefore,
dt  eÞlV
45ð1 þ 8la‘medium
referring back to Eqs. (13) and (15), it can be stated,
0:6qalgae Aemedium r 2
A2 e3 r2 C algae
1 medium
algae
8la‘cake-layer 45ð1  eÞlV
¼   ð21Þ
where Dp ¼ Dp1 þ Dp2 Aer 2 0:6q
A2 e3 r 2algae C algae  1
In practice, the driving force Dp is generally kept as a constant algae

during filtration. Integrating of Eq. (16) under a constant pressure where a in this case stands for the tortuosity of the network path-
difference gives way. Inserting Eq. (7)into Eq. (21)yields the equivalent pore radius, r
45l 1e 8la‘medium rffiffiffiffiffiffi
   V2 þ V ¼ Dp  t ð17Þ 8a e
0:6q e3
Aemedium r 2medium r¼   r algae ð22Þ
2A2 r 2algae C algae 1 45 1  e
algae

Based on Eq. (17), for a given filtration time, t, the volume of the
filtrate water can be evaluated. The capacity of a filter for process-
ing the algae feed is thus determined by Eq. (18),
V 1
F filter ¼  ð18Þ
T 1/
where T is the filter rotation cycle time, which is three times as long
as the filtration time (i.e. t = T/3) since only one-third of the filter as
shown in Fig. 1 is working for the filtration, while the rest of the
time is for cake-layer-dewatering and cake-layer-discharging.
A requirement for algae-dewatering using vacuum pressure is
that the applied pressure difference must be large enough to over-
come resistance arising from the surface tension of water. There-
fore, the minimum pressure required for displacing the water out
of the algal cake-layer can be determined by Eq. (23) [12,33],
2r cos h
pdewatering ¼ ð23Þ
r
where r is the surface tension of water, and h is the contact angle of
water on the surface of algae.

4. Energy and capital costs, and process economics

3.4. Compressibility of the algal cake-layer and equivalent pore size of
the cake-layer
Algae deform under the applied pressure in filtration, giving rise
to compression of the algal cake-layer. Compression of the cake-
layer reduces the porosity, which can be expressed as,
k
e ¼ Bp
where k is the compression index, B is a constant, and p is the com-
pressive pressure. An extensive literature survey showed that there
had been no reported data on the compressibility of algal cake-
layer. Therefore, the compression properties of a bio-cake [30,31],
consisting mostly of yeast cells, and expected to have similar com-
pression properties to that of algae, were adopted to investigate the
impacts of compression on water transport through an algal
cake-layer. The obtained pressure-dependent mean porosity of the
bio-cake is shown in Eq. (20),
e ¼ 2:67p0:46
where p is the pressure in kPa. The critical pressure for compression
is around 17 kPa, under which the porosity of the algal cake-layer is
a constant. With the estimated mean porosity of the algal cake-
layer, the volumetric flow-rate of filtrate water and processing
capacity of the filter under various operating conditions can be eval-
uated using Eqs. (17) and (18).
Another important parameter of algal cake-layer is cake-layer
pore size. As shown in Fig. 2, the filtrate channel in the cake-layer
is actually a network structure. This situation may also be compli-
cated with the network clogging. Network clogging may be caused
by small-sized algae in the algal feed, and/or by the extracellular
organic matter (EOM) [32] released by algae particularly in the
case of high pressure filtration. Obviously, clogging of the network
would result in a reduced cake-layer porosity and pore size. Note
the probability of clogging in the cake-layer is impossibly difficult
to model. An alternative approach is taken in this modeling. By
linking the clogging to the alternation in the cake-layer porosity
Assuming the flow-rate of the harvested algae to be dewatered
is F feed , the number of filters required for processing the algae is
thus
F feed
n¼ ð24Þ
F filter
ð19Þ
In Fig. 1, supposing the filtration is performed at the vacuum
pressure of pfiltration , the total energy consumed by the water pump
is thus
1
W water pump ¼ F feed ð1  /Þðpambient  pvacuum receiver Þ ð25Þ
g
pvacuum receiver ¼ pfiltration  qwater gR  Dpresistance ð26Þ
Dpresistance ¼ 0:5qwater gR ð27Þ
where R is the radius of the filter drum, and g is the energy effi-
ð20Þ
ciency of the water pump. As depicted in Fig. 1, the pressure head
loss due to the piping resistance is not a constant since the flow-rate
of the filtrate water passing through the drum piping varies
throughout the filtration process. The piping in the filtration system
is relatively short and large piping diameter can be chosen to avoid
any significant head loss. For simplicity, according to the above
rational, this resistance is assumed to be a constant equivalent to
50% of the static pressure difference in the system. Strictly speaking,
the static pressure difference created by the water present in the
connecting piping in the drum is also not a constant, dependent
upon the location of the connecting piping within the filtration
zone. As shown in Eq. (26), the maximal value (R) was used to esti-
mate this static pressure. A simple geometric calculation based on
the drum in Fig. 1 shows that no significant error was introduced
with this treatment.
The vacuum pump in the system is intended to maintain a
steady vacuum pressure in the vacuum receiver to ensure a contin-
uous water flow in filtration. In the algae dewatering zone, the air
 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75 71

flow passing through the cake-layer will also be excavated with the industrial project. Supposing the labor requirement of a process
vacuum pump. Evaluation of energy consumption of the vacuum is proportional to its scale, and the salary of the worker is
pump needs to estimate this air flow-rate, which is determined 50 kUS$/yr, the labor cost allocated to this project can be deter-
by the Hagen–Poiseuille equation [12,29,34] as shown in Eq. (28), mined by

e  Adewatering r2 p20  p2dewatering C labor ¼ 3  50000=50=365=24 ð39Þ

Nair ¼  ð28Þ
16RT l a‘ The overall algae-dewatering cost should include all the cost
where l is the viscosity of air, p0 is the atmospheric pressure, and a items detailed in Eqs. (32), (38) and (39)
is the tortuosity of pores in the cake-layer. Therefore, the energy C dewatering ¼ C energy þ C depreciation þ C labor ð40Þ
consumption of the vacuum pump for the isentropic compression
of the air flow is evaluated by Eq. (29) [35], It is anticipated that this overall cost as defined in Eq. (36)can
2 3 be minimized by identifying the optimal operating conditions
!k1
kN air RT 4 p0
k
n including the filtration vacuum pressure, p, and the filter rotation
W vacuum ¼  15  ð29Þ cycle time, T.
k1 pdewatering g

where k is the ratio of heat capacities of air at constant pressure and
constant volume. The energy required to maintain the rotation of
the filter drum is assessed by Eq. (30) [28],
 
2Alx2 R 1 MgRx n
W rotation ¼  þ ð30Þ
3 1  k2 16 g of 0.6 m, and it provides a working area of 1.69 m2. The thickness
where x is the angular velocity of the rotating drum filter, k is the
ratio of radius of the drum filter to that of the feed tank, the first
term is the drag force experienced by the rotating filter in the feed
tank, and the second term is the torque resistance.
Therefore, the total energy of the RDVF dewatering system is
W total ¼ W water pump þ W vacuum þ W rotation
Assuming the price of industrial electricity is 0.05 US$/kWh, the
energy cost is thus
C energy ¼ 0:05  W total
Supposing the unit price of the rotary filter is Pc, the capital cost
for purchase of the filters is
C 1 ¼ n  Pc
As a rule of thumb, the installation of the dewatering system
can be considered to be 60% of the major capital cost [35,36] of
the process.
C 2 ¼ 0:6  n  Pc
The room temperature operation of the process also requires
the dewatering facility and the accompanying system to be
installed indoors. Infrastructure cost is assessed by assuming the
space required by the whole system and the open space needed
for process maintenance and operation, which is estimated at
50% of the capital cost.
5. Results and discussion
5.1. Algal cake-layer thickness
The sized filter has a drum diameter of 0.9 m and a drum length
of algal cake-layer as shown in Fig. 3 was investigated over a wide
range of operating conditions: vacuum pressure ranging from 5 to
60 kPa, and filter rotation cycle time varying from 10 to 120 s. Nat-
urally, for a specific operating pressure, a thicker algal cake-layer
corresponds to a longer cycle time. As shown in Fig. 3, the impact
of the working pressure on the cake-layer thickness is more com-
ð31Þ plicated. Roughly speaking, algae-dewatering benefited from using
a higher filtration pressure under 10 kPa, as indicated by the
increased algal cake-layer thickness. Negative impact was
observed instead when the filtration pressure was higher than
ð32Þ 20 kPa. It was seen that the buildup rate of algal cake-layer was
remarkably slowed by the increased working pressure. This
resulted from the compressibility of algae. For a compressible algal
cake-layer, the filtration pressure plays two roles, one is the driving
ð33Þ force for the filtration, and the other is the driver for cake-layer
compression, and the other occurs when the working pressure is
raised higher than the critical pressure (17 kPa) of algae compres-
sion. Any pressure surpassing this critical pressure densifies the
algal cake-layer, and causes a higher transport resistance for fil-
ð34Þ trate water, resulting in a reduced throughput for water and a cor-
responding lesser algal cake-layer thickness. It is shown in Table 1
that the porosity of the algal cake-layer is strongly dependent upon
the working pressure. There is almost 45% cake-layer porosity loss
70
10 kPa
20 kPa
60
Algae Cake Layer Thickness (mm)

C 3 ¼ 0:5  n  Pc ð35Þ
Empirically, it has been shown that the maintenance cost is 50
30 kPa
usually 10% [35,36] of the sum of the above three items,
40 5 kPa
C 4 ¼ 0:21  n  Pc ð36Þ
Therefore, the total capital investment is 30 40 kPa

C capital ¼ 2:31  n  Pc ð37Þ 60 kPa

20

Assuming the life span of the filter with maintenance is 20 years

[35], the hourly depreciation of the investment is estimated as 10

C depreciation ¼ 2:31  n  Pc=20=365=24 ð38Þ 0

0 20 40 60 80 100 120 140 160
In the present scenario, it is assumed that three workers are
Filter rotation cycle (sec)
required to work on a three-shift rotation for an industrial project
for dewatering 5000 m3 algae feed daily. The scale of this studied Fig. 3. Dependence of algal cake-layer upon the filter rotation cycle and the
demonstration project is 50 times smaller than this referred filtration pressure.
72 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75

Table 1 220
Effect of pressure on the porosity of compressible algae cake-layer.

Pressure Unit Porosity

Processing capacity of the filter (m3/d)

190
10 kPa 0.73
20 kPa 0.67
30 kPa 0.56
160
40 kPa 0.49
50 kPa 0.44
60 kPa 0.41
130

100
as the working pressure is enhanced from 10 to 60 kPa. Overall it
appears better to conduct the dewatering at a lower pressure so
20 kPa
as to avoid significant compression of the cake-layer and avoid a 70 10 kPa
level of filtration resistance which hinders efficiency. 30 kPa
Filter-based algae-dewatering is a continuous process, and the 40 kPa
60 kPa
cake-layer must be removed from the filter within the cake-dis- 40
charging zone. It is therefore required that algal cake-layer formed 0 20 40 60 80 100 120 140 160

on the filer surface be thick enough to easily remove the cake-layer Filter rotation cycle time (sec)
in the algae-discharging zone. In Fig. 3 it is shown that the algal
Fig. 5. Processing capacity of the filter under different operating conditions with
cake-layer thicknesses are mostly in the range of 10–50 mm, suit- the filtration pressure, and filter rotation cycle varying from 10 to 60 kPa, and 10 to
able for effective removal by the separator blade. 120 s, respectively.

5.2. Equivalent pore size in algal cake-layer and the minimum

dewatering pressure
The estimated equivalent pore size and the minimum pressure
for cake-layer dewatering are illustrated in Fig. 4. At a higher filter-
ing pressure, it is found that the cake-layer pore size is narrowed
with a denser algae packing, which is caused by compression, in
consistency with the pressure-dependent cake-layer porosity sum-
marized in Table 1.
As indicated in Fig. 4, the minimum pressure for algae-dewater-
ing is lower than the filtration pressure except in the case of
10 kPa. Thus, the filtration pressure is high enough to extract the
water out of the cake-layer, and there is no further need to increase
the vacuum level, indicating the feasibility of algae dewatering by
the RDVF filter. As shown in Eqs. (22) and (23), the equivalent pore
size is proportional to the algal size, and the minimum dewatering
pressure is inversely proportional to the size of algae. This implies
that a smaller algae size makes algal cake dewatering difficult: it
requires a higher vacuum level to displace the water out of the net-
work pores. In this case, the cake-layer dewatering conducted at a
100
Pore radius of algae cake-layer (micron) or
same vacuum level as in algae filtration obviates the complex vac-
uum system design for the rotary drum filter. In practice, a small
quantity of water will be retained in the cake-layer after dewater-
ing. This residual water is about 1–2 w/w% of the wet cake, which
is removed in the subsequent process of algae-drying.
5.3. Processing capacity of the RDVF filter
Under different operating conditions, the capacity of the
selected RDVF filter for processing 20 w/w% algae feed is depicted
in Fig. 5. It can be seen that filter rotation cycle time varies inver-
sely with algae filtration capacity.
This capacity reduction results from the formation of a thicker
algal cake-layer, which creates a larger hydraulic resistance when
the filter is allowed to have more time working in the filtration
zone. The impact of cake-layer compression on the processing
capacity is also observed. It is shown in Fig. 5 that operating the fil-
ter at 20 kPa gives the highest processing capacity, while in Fig. 3
the largest thickness of algal cake-layer occurs at 10 kPa of pres-
sure. This difference is attributed to the compression of algal
cake-layer: a looser and thus thicker cake-layer is built up at the
lower working pressure of 10 kPa as shown in Fig. 3, but with
greater cake compression at 20 kPa, a higher mass throughput is

still achieved. As indicated in Table 1, when moving from 10 to

minimum dewatering pressure (kPa)

Minimum dewatering pressure

Table 2
Number of the selected filter units required to accomplish the processing duty of
100 m3/d 20 w/w% algae feed coming from the algae-harvesting centrifuges.
10
Cycle Number of filters
Sec 10 kPa 20 kPa 30 kPa 40 kPa 60 kPa
10 0.60 0.47 0.54 0.60 0.68
20 0.78 0.63 0.75 0.83 0.95
Pore radius 30 0.92 0.76 0.91 1.01 1.16
40 1.04 0.86 1.04 1.17 1.34
1 50 1.15 0.96 1.16 1.30 1.50
0 10 20 30 40 50 60 70 60 1.24 1.04 1.26 1.42 1.64
70 1.33 1.12 1.36 1.53 1.77
Filtration pressure (kPa) 80 1.41 1.19 1.45 1.63 1.89
90 1.49 1.26 1.54 1.73 2.00
Fig. 4. Effects of filtration pressure on the pore radius of the algal cake-layer and 100 1.56 1.32 1.62 1.82 2.11
the minimum dewatering vacuum pressure. The assumed parameter values are: 110 1.63 1.38 1.70 1.91 2.21
algae radius equal to 5 lm; surface tension of water equal to72 lJ/m2; tortuosity of 120 1.70 1.44 1.77 1.99 2.31
the pore typically equal to 2 according to the geometry; contact angle equal to p/4.
 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75 73

160 1.4 120

60 kPa 110
1.2

Optimized hourly dewatering cost (US$/h)

140
100
40 kPa
1

Optimized cycle time (sec.)

120
Capital investment (kUS$)

30 kPa 90

0.8 80
100 10 kPa

20 kPa
80
60
40
20
0 5 10
Energy demand (kW)
Fig. 6. Plot of capital investment versus process energy demand under different
operating conditions with the cycle time varying from 10 to 120 s and the filtration
pressure varied from 10, 20, 40, to 60 kPa.
20 kPa of pressure, the porosity of the cake-layer was reduced from
0.73 to 0.67, and also the equivalent pore size went from 8 to 6 lm
as shown in Fig. 4. The trade-off between cake compressibility and
its increased resistance to water permeation produces no further
benefits at pressures above 20 kPa. At these higher pressures, algae
throughput becomes progressively less.
It is also shown in Fig. 5 that by operating the filter at a higher
rotation speed, more feed can be processed. For example, at 20 kPa,
the filter dewaters only 70 m3 algae feed per day at a rotation cycle
of 120 s, and the capacity is doubled with a cycle time of 30 s. High
rotary speeds offer a large potential for increasing algae product
rate albeit with greater energy consumption.
Table 2 summarizes the number of the filters needed for a pro-
ject with a planned capacity of processing 100 m3 20 w/w% algae
1.70
0.6 70
60
0.4
50
0.2
40
0 30
0 20 40 60 80
15 20 Pressure difference (kPa)
Fig. 8. Plot of the minimized hourly dewatering cost, and optimized cycle time
versus the filtration pressure.
feed per day as a function of operational parameters. The full range
of parameters in Table 2 produces a possible range of 0.6–2.31 for
the number of required drum filter units. This translates to a choice
of 1, 2 or 3 units, which in turn encompass only some correspond-
ing operational requires from Table 2. Given that the present pro-
cess in on the demonstration scale, it should be noted that a full
industrial installation with large daily throughputs by one or two
orders of magnitude, will regimes a much large number of filtra-
tion units and offers commensurately more configuration flexibil-
ity. For the present process, one filter would be suitable for
fulfilling the total daily dewatering duty, operated at a rotation
cycle time between 30 and 40 s.
Industrial projects are much larger in scale than a demonstra-
tion project. Larger filter units than the one considered here could
be deployed to treat a much larger daily feed. Larger filters have a
proportionally smaller footprint within a process. Savings in capi-
tal investment can be achieved with larger filters because they
have a net lower price per unit of filtering area.

1.60
100
1.50 Pressure (kPa) Dewatering Cost (US$/h)
90
Hourly dewatering cost (US$/h)

60
1.40 80

70
1.30
Pressure difference (kPa)

1.30 40
60
1.20
1.20 30
50
10
40
1.10 1.10
20 30
1.00 A
20
0.93
0.90 10
0 20 40 60 80 100 120 140 B
Filter rotation cycle (sec.) 0
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
Fig. 7. Hourly dewatering cost using the rotary filter under different operating Cycle time (sec.)
conditions. The processing duty is 100 m3/d, the price of electricity is 0.05 US$/kWh,
the unit price of the filter is 27.5 kUS$/pc and the life-span of the filter is 20 years, Fig. 9. Hourly dewatering cost contours on the domain of operating conditions. The
and the labor cost for the specified dewatering duty is 0.342 US$/h. processing duty is 100 m3/d algae feed containing 20 w/w% dry algal biomass.
74 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75

5.4. Process energy and capital investments Table 3

A report of process techno-economics.

The process capital investment was plotted in Fig. 6 against the Item Amount Unit
process energy demand, including the energy required by the fil- Feed 100 m3/d
trate water pump, vacuum pumps, and by rotating the drum filter Algae conc. 20 w/w%
in the feed tank. Across the range of operating conditions, differ- Capital investment 54.73 kUS$
ences greater than a factor of 3 arise in both the process capital Equipment 23.69 kUS$
Installation 14.21 kUS$
investment and energy demand. For example, operating the filter
Infrastructure 11.85 kUS$
at a higher rotation speed of 10 s per cycle, the resulting capital Project life-span 20 yr
investment can be more than 3 times less compared to the case Capital depreciation 2.74 kUS$/yr
with a rotation cycle of 120 s, while the process energy demand Power 5.59 kW
Energy consumption 176.3 GJ/yr
is nearly 4 times as much. For a given processing duty, a large vari-
Energy cost 2.45 kUS$/yr
ety of options are available through balancing the trade-off Labor 3.0 kUS$/yr
between the process capital and energy investment: the process Maintenance 0.25 kUS$/yr
capital investment can be saved by using more energy or vice Annual dewatering cost 8.23 kUS$/yr
versa. From the viewpoint of process economics, this interplay of
factors can be engineered to minimize the overall dewatering cost,
including the capital and the operating costs.
5.6. Techno-economics of the algae-dewatering
5.5. Dewatering cost and process optimization
The list of optimal parameters arrived at through techno-eco-
The dewatering cost (Eq. (36)) under various operating condi- nomic evaluation of the algae-dewatering process are summarized
tions is shown in Fig. 7. A conic curve with a minimum is obtained in Table 3. For processing 100 m3 of algae feed daily, the total cap-
for each of process. The higher dewatering cost, on the left side of ital investment for the project is 54.73 kUS$ with a process life-
the curve is due to the high energy consumption by the filter, span of 20 years, and the annual capital depreciation is
which is operated at a high rotational speed with a cycle time less 2.74 kUS$. To accomplish this processing duty, it needs 5.59 kW
than 20 s. It is also shown in Fig. 7 that operating the filter at too power input, corresponding to a yearly energy consumption of
slow a speed (with a cycle time greater than 40 s) also gives rise 176.3 GJ and a yearly energy cost of 2.4 kUS$. The energy-efficient
to a higher dewatering cost. In this case, capital costs are higher nature of this filtration dewatering process is highlighted by com-
with lower throughput rates. Fig. 7 shows that the hourly dewater- paring with an evaporation process which requires 66,000 GJ of
ing cost is strongly dependent upon the operating conditions, and energy per year. The annual labor cost of this project is 3 kUS$
can vary widely. Process simulation enables a comprehensive over- and the annual spending on process maintenance is 0.25 kUS$.
view of the impact of operating conditions on the dewatering cost, As a total, the aggregate cost for this project is 8.23 kUS$/yr. For
and a means to select process parameters for reliable and econom- simplicity, interest costs related to initial capital investment were
ical operation. not included in this study. This is not an unreasonable assumption,
For each of the 5 filtration pressures ranging from 10 to 60 kPa as there exist many subsidy programs which assist the develop-
shown in Fig. 7, an optimized cycle time and a minimized dewater- ment of clean energy processes [37].
ing cost can be determined. Plotting these minimized dewatering A breakdown of the annual dewatering cost is shown in Fig. 10.
costs and the optimized cycle times with respect to the corre- The biggest share, 37%, comes from labor, which is 7% higher than
sponding filtration pressure helps identify the optimum operating the second biggest share from energy. In this case, the abnormally
conditions and the minimum dewatering cost. As shown in Fig. 8, high proportion of labor costs results from the small scale of the
the minimum dewatering cost of 0.93 US$/h can be achieved by operation. Currently in algal biofuel industry, typical daily lipids
operating the filter at the filtration pressure of 20 kPa, and at the production capacity of plants are in the range of 90–300 ton/d.
rotation cycle of 40 s. [37], corresponding to a dewatering duty of 1500–5000 m3/d har-
A contours plot for dewatering cost levels covering a wide range vested algae feed (20 w/w%). This scale is not large enough to cut
of the operating conditions and the major range of the simulated
dewatering cost from 0.93 to 1.30 US$/h are shown in Fig. 9.
Two cost-sensitive areas A and B were found in the domain of
operating conditions. Operating the filter in area A comes with
great energy expense from the greater rotational resistance experi- capital
enced by the filter. While operating the filter in area B, with a low labor 23%
filtration driving force, much more capital investment for the pro- 37%
cessing is required, rendering the dewatering process less cost-
effective. In view of these considerations, the filter should operate
with at least 10 kPa of filtration pressure, and with a rotation cycle
upwards of 20 s.
This identified cost-sensitivity is also valuable for offering bet- energy
infrastructure 30%
ter control over the operating variables. For such dewatering pro-
7%
cesses, it is very difficult to maintain the operating conditions
such as the vacuum pressure completely constant. Normally, the
pressure would fluctuate in the process within control tolerance.
Fig. 9. suggests that a positive deviation relative to the optimum
maintenance
operating conditions (p > popt. in the pressure, and T > Topt. in the 3%
cycle time) would incur less increase in dewatering cost. A more
favorable control scheme can be formulated with this cost- Fig. 10. Breakdown of the dewatering cost among labor, energy, equipment,
sensitivity understanding. infrastructure and maintenance.
 P. Shao et al. / Chemical Engineering Journal 268 (2015) 67–75
down the labor cost. In this simulation, the assumed industrial
scale for estimating the labor allocated to this 100 m3/d demon-
stration project is 5000 m3/d. Under this scenario, there is a great
potential for scaling up the algae-dewatering process and thus
reducing the dewatering unit cost. Given this scale is, for example,
doubled. The labor cost allocated to this simulated project would
be cut in half.
6. Conclusions
A mathematical model was developed for the algae-dewatering
process by a rotary drum vacuum filter. Algal cake-layer is consid-
erably compressed as the filtration pressure is increased beyond a
critical pressure, and the compressed algal cake-layer becomes less
porous and develops smaller equivalent pore size. As a result, the
densified cake-layer presents a much higher resistance for the fil-
tration process.
Simulations showed that the pressure required to displace the
filtrate water out of the porous algal cake-layer is generally lower
than the filtration pressure. High pressure systems are not required
for dewatering the algal cake-layer. Consequently, the water con-
tent in the collected algae can be reduced to 1–2%, and the energy
required for the subsequent algal biomass drying which has a high
thermal demand can be significantly reduced.
For processing 100 m3/d 20 w/w% algae stream, the capital
investment for the project is 54.73 kUS$, and the power required
to fulfill this processing duty is 5.59 kW and the yearly energy cost
is 2.45 kUS$, is much less than needed by the evaporation
approach. Process optimization indicated that the vacuum pres-
sure in the filter should be controlled at 20 kPa, and the rotation
cycle of the filter should be around 40 s so that the dewatering cost
can be reduced to its minimum of 0.93 US$/h. Breakdown of this
cost reveals that 37% goes to labor, which represents the biggest
share. This results from the relatively small scale of the algal bio-
fuel project. Analysis indicates that scaling up the project will bring
significant unit cost benefits.
Acknowledgement
This research was funded by the HDE4-SIM, Algae Carbon
Conversion Flagship Program of National Research Council Canada
(NRCC).
References
[1] A. Motta, C. Borges, K. Esquerre, A. Kiperstok, Oil Produced Water treatment for
oil removal by an integration of coalescer bed and microfiltration membrane
processes, J. Membr. Sci. 469 (1) (2014) 371–378.
[2] K. Szymanska, A.I. Zouboulis, D. Zamboulis, Hybrid ozonation–microfiltration
system for the treatment of surface water using ceramic membrane, J. Membr.
Sci. 468 (150) (2014) 163–171.
[3] R. Fabris, E. Lee, C.W.K. Chow, V. Chen, M. Drikas, Pre-treatments to reduce
fouling of low pressure micro-filtration (MF) membranes, J. Membr. Sci. 289
(1–2) (2007) 231–240.
[4] A. Laval, Beer filtration: top end beer from cross-flow filters, Filtr. Sep. 44
(2007) 40–41.
[5] Société IndustrielledelaValléedel’Aygues (SIVA), Crossflow filtration: a new
approach to wine clarification, Filtr. Sep. 44 (2007) 36–39.
[6] L.M.J. de Carvalho, I.M. de Castro, C.A.B. da Silva, A study of retention of sugars
in the process of clarification of pine apple juice (Ananas comosus, L.Merril) by
micro- and ultra-filtration, J. Food Eng. 87 (2008) 447–454.
75
[7] C. Choo, C. Tien, Hydrosol deposition in fabrious beds, Sep. Technol. 1 (3)
(1991) 122–131.
[8] X. Wu, C. Tien, Polydispersed aerosol filtration in granular media, Sep. Technol.
5 (2) (1995) 63–75.
[9] R. Bhave, T. Kuritz, L. Powell, D. Adcock, Membrane-based energy efficient
dewatering of microalgae in biofuels production and recovery of value added
co-products, Environ. Sci. Technol. 46 (2012) 5599–5606.
[10] P.N. Patel, M.A. Mehaia, M. Cheryan, Cross-flow membrane filtration of yeast
suspensions, J. Biotechnol. 5 (1987) 1–16.
[11] J.A. Bobbitt, R.P. Betts, The removal of bacteria from solutions by membrane
filtration, J. Microbiol. Methods 16 (1992) 215–220.
[12] P. Shao, R.Y.M. Huang, X. Feng, W. Anderson, Gas–liquid displacement method
for estimating membrane pore-size distributions, AIChE J. 50 (2004) 557–565.
[13] M.L. Gerardo, D.L. Oatley-Radcliffe, R.W. Lovitt, Integration of membrane
technology in microalgae biorefineries, J. Membr. Sci. 464 (15) (2014) 86–99.
[14] E. Jacob-Lopes, T.T. Franco, From oil refinery to microalgae biorefinery, J. CO2
Util. (2013), http://dx.doi.org/10.1016/j.jcou.2013.06.001.
[15] G.G. Zaimes, V. Khanna, Microalgae biomass production pathways: evaluation
of life cycle environmental impacts, Biotechnol. Biofuels 6 (2013) 88.
[16] F. Ma, M.A. Hanna, Biodiesel production: a review, Bioresour. Technol. 70
(1999) 1–15.
[17] L. Lardon, A. Hellas, A. Sialve, J.P. Steyer, O. Bernard, Life-cycle assessment of
biodiesel production from microalgae, Environ. Sci. Technol. 43 (17) (2009)
6475–6481.
[18] O. Jorquera, A. Kipersok, E.A. Sales, M. Embirucu, M.L. Ghirardi, Comparative
energy life-cycle analyses of microalgae biomass production in open ponds
and photobioreactors, Bioresour. Technol. 101 (2010) 1406–1413.
[19] I. Rawat, R.R. Kumar, T.M.F. Bux, Biodiesel from microalgae: a critical
evaluation from laboratory to large scale production, Appl. Energy 103
(2013) 444–467.
[20] A.J. Dassey, C.S. Theegala, Harvesting economics and strategies using
centrifugation for cost effective separation of microalgae cells for biodiesel
applications, Bioresour. Technol. 128 (2013) 241–245.
[21] M. Reza, M.P. Krishna, Darcy’s law-based model for wicking in paper-like
swelling porous media, AIChE J. 56 (9) (2010) 2257–2267.
[22] J.P. Heller, The interpretation of model experiments for the displacement of
fluids through porous media, AIChE J. 9 (40) (1963) 452–459.
[23] M. Reza, T. Hua, M.P. Krishna, Numerical simulation of liquid absorption in
paper-like swelling porous media, AIChE J. 58 (8) (2012) 2536–2544.
[24] J. Shabanian, J. Chaouki, Hydrodynamics of a gas–solid fluidized bed with
thermally induced interparticle forces, Chem. Eng. J. 259 (2015) 135–152.
[25] G. Foley, A review of factors affecting filter cake properties in dead-end
microfiltration of microbial suspensions, J. Membr. Sci. 274 (2006) 28–46.
[26] K.C.E. Östergren, A.C. Trägårdh, G.G. Enstad, J. Mosby, Deformation of a
chromatographic bed during steady-state liquid flow, AIChE J. 44 (1) (1998) 2–
12.
[27] Y. Wang, M. Brannock, S. Cox, G. Leslie, CFD simulations of membrane filtration
zone in a submerged hollow fibre membrane bioreactor using a porous media
approach, J. Membr. Sci. 363 (2010) 57–66.
[28] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenmena, vol. I, John Wiley &
Sons Inc., New York, London, 1960.
[29] P. Shao, R.Y.M. Huang, An analytical approach to the gas pressure drop in
hollow fiber membranes, J. Membr. Sci. 271 (1–2) (2006) 69–76.
[30] W. Lee, W. Cheong, K. Yeon, B. Hwang, C. Lee, Correlation between local TMP
distribution and bio-cake porosity on the membrane in a submerged MBR, J.
Membr. Sci. 332 (2009) 50–55.
[31] S. Hong, W. Lee, H. Oh, K. Yeon, B. Wang, C. Lee, The effects of intermittent
aeration on the characteristics of bio-cake layers in a membrane bioreactor,
Environ. Sci. Technol. 41 (2007) 6270–6276.
[32] S. Babel, S. Takizawa, Microfiltration membrane fouling and cake behavior
during algae filtration, Desalination 261 (2010) 46–51.
[33] L.A. Segura, P.G. Toledo, Pore-level modeling of isothermal drying of pore
networks: effects of gravity and pore shape and size distributions on
saturation and transport parameters, Chem. Eng. J. 111 (2005) 237–252.
[34] A. Hernfindez, J.I. Calvo, P. Prfidanos, F. Tejerina, Pore size distributions in
microporous membranes, a critical analysis of the bubble point extended
method, J. Membr. Sci. 112 (1996) 1–12.
[35] P. Shao, M. Dal-Cin, M.D. Guiver, A. Kumar, Simulation of membrane-based
CO2 capture in a coal-fired power plant, J. Membr. Sci. 427 (2013) 450–459.
[36] P. Shao, M. Dal-Cin, A. Kumar, H. Li, D.P. Singh, Design and economics of a
hybrid membrane-temperature swing adsorption process for upgrading
biogas, J. Membr. Sci. 413–414 (2012) 17–28.
[37] The report of NAABB, National Alliance for Advanced Biofuels and Bio-
products, USA.