![Relative viscosity μ r = μ/μ s of A. platensis suspensions as a function of their mass fractions, estimated by fitting the Krieger-Dougherty equation to the rheological data for shear rate γ̇ = 5 s −1 , with the maximum mass fraction set to 0.43. The resulting intrinsic viscosity value [μ] equals 39.24. The relative viscosity increased exponentially as a function of its mass fraction.](https://www.researchgate.net/profile/Alexander-Mathys/publication/325606950/figure/fig2/AS:638810744451080@1529315758725/Relative-viscosity-m-r-m-m-s-of-A-platensis-suspensions-as-a-function-of-their-mass_Q320.jpg)
Content uploaded by Alexander Mathys
Author content
All content in this area was uploaded by Alexander Mathys on Oct 15, 2018
Content may be subject to copyright.
Contents lists available at ScienceDirect
Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech
Comprehensive pulsed electric field (PEF) system analysis for microalgae
processing
Leandro Buchmann, Robin Bloch, Alexander Mathys
⁎
ETH Zurich, Institute of Food Nutrition and Health, Laboratory of Sustainable Food Processing, Schmelzbergstrasse 9, Zurich 8092, Switzerland
GRAPHICAL ABSTRACT
ARTICLE INFO
Keywords:
Ultrasonic Doppler velocity profiling
Microalgae
Arthrospira platensis
Rheological analysis
Multiphysics simulation
Pulsed electric field processing
ABSTRACT
Pulsed electric field (PEF) is an emerging nonthermal technique with promising applications in microalgae
biorefinery concepts. In this work, the flow field in continuous PEF processing and its influencing factors were
analyzed and energy input distributions in PEF treatment chambers were investigated. The results were obtained
using an interdisciplinary approach that combined multiphysics simulations with ultrasonic Doppler velocity
profiling (UVP) and rheological measurements of Arthrospira platensis suspensions as a case study for applications
in the biobased industry. UVP enabled non-invasive validation of multiphysics simulations. A. platensis sus-
pensions follow a non-Newtonian, shear-thinning behavior, and measurement data could be fitted with rheo-
logical functions, which were used as an input for fluid dynamics simulations. Within the present work, a
comprehensive system characterization was achieved that will facilitate research in the field of PEF processing.
1. Introduction
Nonthermal processes such as pulsed electric field (PEF) can be used
to effectively process biomass (Mahnič-Kalamiza et al., 2014; Rocha
et al., 2018; Vorobiev and Lebovka, 2008). For instance, PEF could be
used to gently pasteurize heat-sensitive liquids (Mathys et al., 2013;
Raso et al., 2006)orefficiently extract valuable compounds from mi-
croalgae (Goettel et al., 2013; ’t Lam et al., 2017; Kempkes et al., 2011;
Parniakov et al., 2015; Postma et al., 2016). Apart from these focus
areas, many other applications and advantages of PEF can be con-
sidered. Toepflet al. (2006) note the potential to use PEF to improve
environmental sustainability while saving energy and costs. Overall,
PEF has many promising applications for microalgae processing, such
as lipid extraction, stress inductions and contamination control within
microalgae cultures (Bensalem et al., 2018; Buchmann et al., 2018; Eing
et al., 2009; Rocha et al., 2018). Despite the long history of PEF re-
search and the apparent advantages of PEF, it is not yet widely used
commercially. Researchers mainly attribute this to the lack of treatment
https://doi.org/10.1016/j.biortech.2018.06.010
Received 3 May 2018; Received in revised form 4 June 2018; Accepted 5 June 2018
⁎
Corresponding author: ETH Zurich, Institute of Food, Nutrition and Health IFNH, Head of Sustainable Food Processing Laboratory, Schmelzbergstrasse 9, Zurich 8092, Switzerland.
E-mail address: Alexander.Mathys@hest.ethz.ch (A. Mathys).
Bioresource Technology 265 (2018) 268–274
Available online 07 June 2018
0960-8524/ © 2018 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/).
T
homogeneity, comparability, and reproducibility of PEF research re-
sults (Buckow et al., 2010; Jaeger et al., 2009; Raso et al., 2016;
Buchmann et al., 2018). In fact, limited comparability and reproduci-
bility of PEF research has been reported as the main underlying pro-
blem and is viewed by Raso et al. (2016) as a barrier for the develop-
ment and wide use of the technology. Buckow et al. (2010) discusses
the high monetary costs caused by the nonuniformity of PEF treatments
and the high energy use resulting thereof. Consequently, many re-
searchers have rightly focused on understanding PEF treatment in-
homogeneities and on developing measures to reduce such in-
homogeneities. However, experimental validation of simulated results
is a challenging task (Buckow et al., 2010; Gerlach et al., 2008).
The ultrasonic Doppler velocity profiling (UVP) method can be
utilized to noninvasively determine velocity profiles and therefore to
validate flow field simulations. The method was developed at the Paul
Scherrer Institute in Switzerland (Takeda, 1995). UVP uses ultrasonic
echography and the Doppler shift frequency to measure the in-
stantaneous velocity profile of liquids (Takeda, 1995). The time delay
between the initiation and reception of the ultrasound pulse gives in-
formation on the reflection position, allowing velocity determination by
incorporation of the Doppler shift frequency (Takeda, 2012). Compared
to other flow measuring methods, UVP can be used on opaque liquids,
and since it is noninvasive, it does not influence the velocity profile
(Wiklund et al., 2007).
To link UVP measurements to simulation results, the characteristic
suspension viscosities must be known. Analysis of characteristic sus-
pension viscosities as used for flow field simulations and experiments
can be conducted using rheological measurements (Ewoldt et al., 2015).
Thereby, the viscosity of fluids as a function of the applied force or
shear rate can be determined. Microalgae suspensions from Arthrospira
platensis serve as a promising model system for such analysis since they
are widely used in research and application.
When simulating Newtonian fluids, a constant viscosity value can be
assumed for the entire simulated process. On the other hand, simula-
tions of non-Newtonian fluids are more complex and require a viscosity
function as an input.
Noninvasive system analysis can be executed by integration of
suspension characteristics into numerical simulations. Simulations can
be used to improve PEF in an iterative process, for example, by con-
stantly simulating and optimizing the treatment chamber geometry
(Buckow et al., 2010). Furthermore, computational tools can aid in
understanding the different process factors involved in PEF and how
those influence the treatment (Fiala et al., 2001; Gerlach et al., 2008;
Meneses et al., 2011c). For those reasons, numerical simulations are
used in this work to gain a better understanding of flow fields in PEF
processing. Although simulation models might seem plausible, their
solutions might not always be accurate, especially for complex pro-
blems (Barbosa-Cánovas et al., 2011). Consequently, it is essential to
validate the simulations with experimental data whenever possible.
This principle is also accounted for in the experimental portion of this
work. Multiple papers in the past decade have described or reviewed
the basic physics laws, principles, equations and boundary conditions
underlying numerical simulations of PEF (Buckow et al., 2010; Gerlach
et al., 2008; Krauss et al., 2011; Meneses et al., 2011a; Wölken et al.,
2017). The focus of research on PEF simulations has been on improving
the electric field homogeneity (Álvarez et al., 2006; Fiala et al., 2001;
Meneses et al., 2011b; Zhang et al., 1995). Within this study, a com-
prehensive approach is taken to achieve a homogenous and comparable
energy input from PEF, considering electric and flow field in-
homogeneities. This novel approach combines multiphysics simulation
with noninvasive UVP measurement and rheological validation, en-
abling a comprehensive PEF system analysis and laying the foundation
for improved microalgae processing.
2. Materials and methods
2.1. Experimental setup
Experiments were conducted using a 10 mm and 1 mm diameter
polycarbonate treatment chamber (manufactured at ETH Zürich,
Switzerland). The treatment chamber was clamped into a retainer with
the flow aligned vertically. Polypropylene-based tubes with a 2.79 mm
internal diameter (SC0319A, Cole-Parmer GmbH, Wertheim, Germany)
were attached to the treatment chamber on both sides as an inlet and
outlet for the treated liquid. A peristaltic pump (MS-4/12-100
ISMATEC®, Cole-Parmer GmbH, Wertheim, Germany) was used at the
highest speed (99 rpm) to pump the liquid through the PEF treatment
cell from the bottom up, as described elsewhere (Goettel et al., 2013).
Prior to flow measurements, it was ensured that the treatment chamber
was thoroughly filled with liquid so that there were no remaining air
bubbles and the flow reached steady-state conditions.
2.2. Arthrospira platensis suspensions
Arthrospira platensis suspensions were used as a model system for
conducting flow behavior experiments. A. platensis suspensions of four
different concentrations in deionized water were prepared (5, 25, 60
and 100 g L
−1
); using A. platensis powder (PREMIUM II, origin China,
Institute for Food and Environmental Research ILU, Bergholz-
Rehbruecke, Germany). The suspensions were thoroughly stirred and
shaken in a flask to achieve a homogenous suspension. Flow experi-
ments were conducted within two hours of production to minimize any
decomposition of the suspension. In addition, a magnetic stirrer was
used during the entire flow profiling experiment to prevent sedi-
mentation.
2.3. Measuring mass flow rate
The mass flow rate through each parallel plate treatment chamber
(10 mm and 1 mm gap) was measured for the water and A. platensis
suspensions used in the flow field measurements. The liquids were
pumped through the treatment cell as described in Section 2.1 and
collected in a 50 mL falcon tube. The time for 20 mL to flow through the
system was recorded. From the mass flow rate
m
̇
, the cross-sectional
area of the inlet A, and the fluid density ρ, the average flow velocity uin
the inlet were calculated for different inlet diameters (Eq. (1)). These
measurements served as a guide for the selection range of the inlet
velocity value u
0
in the simulations. Using the actual experimental inlet
velocity as a boundary condition in the simulation allowed an experi-
mental validation.
=m uAρ
̇·· (1)
2.4. UVP measurements
A 10 mm lab-scale parallel plate continuous PEF system as described
in Section 2.1 was used for the UVP experiments. An ultrasonic profiler
(UB-Lab, UBERTONE, Strasbourg, France) was used for measurements,
together with two 4 MHz and one 8 MHz ultrasound transducer (UB-
Lab, UBERTONE, Strasbourg, France). Round holes approximately
6 mm deep with a diameter only marginally larger than the ultrasound
transducer were drilled into the outer polycarbonate shell of the
treatment cell at three incident angles θ
1
(0° or perpendicular to the
liquid flow, 45°, 60°) to hold the ultrasound transducers in place. Water
as a reference medium, as well as three different concentrations of A.
platensis suspensions, were used for the UVP measurements (see section
2.2). Inert copolyamide acoustic reflector beads (size 80–200 μm, den-
sity 1.07 g cm
−3
, MET-FLOW S.A., Lausanne, Switzerland) were added
to the suspensions and the water in order to facilitate the reflections of
the ultrasound waves and to improve the measurement signal. The
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
269
results were averaged over ten measurements for each investigated
suspension and water. For an in-depth method description and experi-
mental procedure, refer to Takeda (2012).
2.5. Rheological measurements
Rheological analyses were conducted on the A. platensis suspensions
and water as described in Section 2.2. The flow behavior was analyzed
using a stress-controlled rheometer (Physica MCR 501, Anton Paar,
Graz, Austria) with double gap geometry (DG 26.7, Anton Paar, Graz,
Austria). For each measurement, a liquid volume of 3.8 mL was loaded
into the outer cylinder. Samples were sheared at shear rates
γ̇
(s
−1
)of
500 s
−1
–0.5 s
−1
and vice versa. By reversing the shear rate, it was
ensured that a possible sedimentation of A. platensis during rheological
measurements was accounted for. The resulting measurement points
were visualized in a scatter plot showing the viscosity versus the shear
rate. The best-fitting rheological equation was either the Herschel-
Bulkley equation (Eq. (2)) or the non-Newtonian power law (Eq. (3))
(Mezger, 2015; Spagnolie, 2015).
=+ττ Kγ
̇n
0(2)
=τKγ
̇
n
(3)
with shear stress
τ
(Pa), yield stress
τ
0
(Pa), flow consistency index K(Pa
s
n
) and flow behavior index n(–). Incorporation of the generalized
Newtonian law =μγ τγ
(
(̇)/
̇), allows Eq. (2) to be rewritten. The re-
sulting equation gives the viscosity µ(Pa s) of the non-Newtonian fluid
as a function of the shear rate, given a shear rate equal to or greater
than the zero shear rate
γ̇
0
(s
−1
) (Eq. (4)).
=+ ≥
−−
μ
τγ Kγ γ γ
̇ ̇ ,̇ ̇
n
011
0
(4)
This function can now be applied onto rheological data to obtain
values for the yield stress, flow consistency index, and flow behavior
index. The resulting functions were used as viscosity functions in the
multiphysics simulations, as described in Section 2.6.
The rheological results were further used to derive an equation for
the A. platensis suspension’s viscosity as a function of its concentration,
using the Krieger-Dougherty relation (Eq. (5)). The Krieger-Dougherty
relation (Eq. (5)) can be used to express the suspension viscosity as a
function of the particle concentration (Krieger and Dougherty, 1959).
While this relation is strictly valid for the low shear Newtonian plateau,
it can be used at the lower end of the measured shear rate range (Zhang
et al., 2013).
==−
−
μ
μμ ϕ ϕ/(1/)
rs v μϕ
max []
max
(5)
where μ
r
(–) denotes the relative viscosity of the suspensions, μ(Pa s)
represents the effective viscosity of the suspensions, μ
s
(Pa s) represents
the viscosity of the suspending medium, ɸ
max
(–) is the maximum vo-
lume fraction, ɸ
v
(–) represents the volume fraction, and [μ](–) denotes
the intrinsic viscosity of the suspended particles. Ciferri (1983) describe
the morphology of A. platensis as helical filaments. Consequently, the
shape of A. platensis filaments can be approximated as a cylinder with
an aspect ratio of 1:10. This ratio was confirmed in microscopic images.
Metzner (1985) gives values for the maximum packing fraction of short
fibers with different aspect ratios. Pan (1993) provides a graph for the
maximum volume fraction as a function of the aspect ratio. Both
sources suggest that the value for maximum packing fraction of elon-
gated structures with aspect ratios close to 1:10 can be approximated at
ϕ
max
= 0.43.
In the absence of a value for the volume fraction of A. platensis
suspensions, the mass fraction was used instead. The mass fraction was
approximated from the mass concentration using a value for the free
water content in the A. platensis powder. The water content in the
utilized A. platensis powder was determined to be at 6.56% by using a
moisture analyzer (HR73/HA-P43, Mettler-Toledo International Inc.,
Columbus OH, USA). The water content of the powder was subtracted
from the powder mass and added to the mass of the suspending medium
water to calculate the mass fraction.
2.6. Computational fluid dynamics simulation
COMSOL multiphysics
®
software (version 5.3, Comsol Inc.,
Burlington MA, USA) was used to conduct multiphysics simulations of
the flow field in PEF treatment chambers. The geometries of the parallel
plate treatment chambers available at ETH Zurich, Sustainable Food
Processing Laboratory were recreated in COMSOL multiphysics
®
. The
simulations for parallel plate treatment chambers were conducted using
a 3D model. The geometry of a co-linear treatment cell was simulated
using a 2D axisymmetric approach.
The Reynolds numbers for the lab-scale geometries investigated in
this work were all well below the critical Reynolds value of 2300, even
at the highest velocities and lowest viscosities that were simulated.
Thus, Reynolds numbers were sufficiently small that a laminar physics
model was used to simulate the flow field. Boundary conditions for the
fluid properties, i.e., the fluid density and the dynamic viscosity, were
defined for every simulation. The density was set to equal the density of
water for all simulations (ρ= 1000 kg m
−3
). To simulate the flow of
water, the standard dynamic viscosity of water as deposited in the
software was used (μ=0.001 Pa s). The flow behavior of the A. platensis
suspensions used for the experimental part of this work was measured
in rheological trials as described in Section 2.5. The results of these
experiments were used to construct viscosity functions for every con-
centration of A. platensis solution measured. The viscosity functions
were then integrated into the simulation. The presented work only
considered stationary solutions of the simulation, as the system had
reached its equilibrium and was in a steady state. Therefore, in-
vestigating stationary solutions was sufficient to describe the situation
present in most PEF research applications. Further, effects of tem-
perature on media parameters were neglected due to the low energy
inputs studied.
2.7. Data analysis
Data analysis was conducted by an independent t-test. The con-
fidence interval was 95% for all experiments. Statistical results were
obtained using the software IBM SPSS Statistics (IBM Corp., Armonk
NY, USA).
3. Results and discussion
To set up the multiphysics simulations, the mass flow rate was de-
termined prior to all other experiments. It was found that the mass flow
rate was equal to 3.03·10
−7
kg s
−1
, 2.94·10
−7
kg s
−1
, 2.86·10
−7
kg s
−1
and 2.78·10
−7
kg s
−1
for water and for A. platensis suspension with
concentrations of 5 g L
−1
,25gL
−1
and 100 g L
−1
, respectively. Based
on the mass flow rate, the inlet velocities were calculated for different
inlet configurations using Eq. (1).
3.1. Flow behavior of Arthrospira platensis suspensions
Rheological measurements demonstrated that A. platensis suspen-
sions showed a non-Newtonian, shear-thinning behavior, and mea-
surement data could be fitted with rheological functions (Fig. 1).
Lower-concentration suspensions (5 g L
−1
,25gL
−1
) were best fitted
using the Herschel-Bulkley equation (Eq. (2)) whereas higher-con-
centration suspensions (60 g L
−1
, 100 g L
−1
) were best fitted with the
non-Newtonian power law (Eq. (3)).
The flow consistency index and the flow behavior index of A. pla-
tensis suspensions, which together define the viscosity functions at the
respective concentration, are shown in Table 1 in accordance with Eq.
(4). The rheological results for A. platensis suspensions correspond well
with findings by Bernaerts et al. (2017), who characterized rheological
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
270
behavior for suspensions of seven different strains of microalgae, in-
cluding A. platensis. The rheological behavior depends on the in-
vestigated strain and biopolymer composition; therefore, it must be
analyzed for every strain independently.
The investigated A. platensis suspensions showed an increasing
viscosity with increasing concentration (Fig. 1). Therefore, the max-
imum packing fraction was integrated with the rheological results to
obtain a Krieger-Dougherty plot illustrating the dependency of cell
concentration on relative viscosity (Fig. 2). The viscosity increased
exponentially with increasing concentration. Hence, the resulting plot
illustrates that the empirical equation formulated by Krieger and
Dougherty accurately approximates the viscosity of A. platensis sus-
pensions as a function of their concentration. The maximum mass
fraction was set to 0.43 in accordance with the literature. The data
shown in Fig. 2 were obtained for a shear rate of 5 s
−1
. In accordance
with Eq. (5), a value for the intrinsic viscosity of 39.24 was obtained.
This value is significantly higher than the intrinsic viscosity of 24.7
for Chlorella vulgaris reported by Zhang et al. (2013). This observation
means that the A. platensis cells contribute more to the viscosity of the
suspension than C. vulgaris cells do. In view of the morphological dif-
ferences between C. vulgaris and A. platensis, this difference is not sur-
prising. The long filamentous structure of A. platensis interacts more
strongly with the medium than the spherical cell of C. vulgaris, resulting
in a higher viscosity of the suspended cells.
3.2. Experimental validation of flow fields
Experimental validation of the fluid dynamic simulations was con-
ducted using a parallel plate treatment chamber of 10 mm diameter
applying UVP, as described in Section 2.4. The results of the mea-
surements are presented in Fig. 3 and are compared with the simulation
results for exactly the same treatment chamber geometry, using the
viscosity functions obtained from the rheological measurements. The
results of the experiments and simulations correlate. Based on an in-
dependent t-test, no significant difference between UVP and simulation
results was found for water and 25 g L
−1
and 100 g L
−1
A. platensis
solutions. Both in the simulation and in the flow profiling results, the
velocity profile was narrow and pronounced at low viscosity but flat-
tened to become more uniform at higher viscosity.
However, a statistically significant difference between the simula-
tion and UVP occurred at suspension concentrations of 5 g L
−1
(p < 0.05). The reason for this deviation might be the sensitivity of the
UVP method. As described in Section 2.4, the number of microalgae
cells in the 5 g L
−1
suspension might simply be too low for the ultra-
sound transducer to receive a signal significantly different from that of
water. Nevertheless, UVP validates the simulation results well and re-
inforces that simulations are a useful means to test the basic principles
in processes such as PEF.
While the validity and value of the simulations has already been
confirmed by other researchers who have used simulations for PEF
(Buckow et al., 2010; Fiala et al., 2001; Gerlach et al., 2008; Meneses
et al., 2011b), using the UVP method to validate the simulation results
Fig. 1. Rheological measurement results for water and different A. platensis
suspension concentrations, fitted with rheological equations. The viscosity in-
creased for increasing concentrations. The suspensions exhibited a non-
Newtonian, shear thinning behavior. In comparison, water showed Newtonian
behavior, characterized by the horizontal line at viscosity μ=0.001 Pa s.
Table 1
Values for yield stress τ
0
times shear rate
−
γ
|
̇|
1
,flow consistency index K, and
flow behavior index n of A. platensis suspensions obtained from the Herschel-
Bulkley and non-Newtonian power law fits on the rheological measurements.
−
τγ|̇|
0
1
Kn
(Pa s) (Pa s
n
)(–)
A. platensis 5 g·L
−1
0.0012 0.0063 −0.1091
A. platensis 25 g·L
−1
0.0016 0.0085 −0.0161
A. platensis 60 g·L
−1
–0.0093 −0.1952
A. platensis 100 g·L
−1
–0.0751 −0.3703
Fig. 2. Relative viscosity μ
r
=μ/μ
s
of A. platensis suspensions as a function of
their mass fractions, estimated by fitting the Krieger-Dougherty equation to the
rheological data for shear rate
γ
̇
=5s
−1
, with the maximum mass fraction set
to 0.43. The resulting intrinsic viscosity value [μ] equals 39.24. The relative
viscosity increased exponentially as a function of its mass fraction.
Fig. 3. a) Ultrasonic Doppler velocity profiling (UVP) measurement and b) si-
mulation results for water and suspensions with different A. platensis con-
centrations in the 10 mm diameter treatment chamber.
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
271
is a new approach. Some researchers have raised the concern that
measuring equipment might disturb the flow and thus make an ex-
perimental validation of flow experiments difficult (Buckow et al.,
2010; Gerlach et al., 2008). The UVP method addresses these concerns
well, since it can be used on a broad range of liquids and suspensions,
even opaque and non-Newtonian suspensions, without disturbing the
flow field. However, when testing lab-scale equipment, it is necessary to
increase the scale to a diameter in the range of 10 mm or more in order
to obtain sufficient data points during UVP measurements.
3.3. Simulation based comprehensive system analysis
3.3.1. Effect of inlet velocity
The effect of changing inlet velocity was investigated in both the
treatment chamber geometries of 10 mm and 1 mm diameter and for
fluids of different viscosity. Fig. 4 shows the velocity profiles in the
center of the treatment chambers for different inlet velocities (u
0
). In
both geometries, the velocity profiles had a parabolic shape, as ex-
pected from laminar flow profiles. The velocity was at its maximum in
the center of the treatment chamber and tended towards zero near the
walls. The maximum velocity increased with increasing inlet velocity.
Therefore, the differences in velocity across the treatment chamber
became larger with increasing inlet velocity, assuming laminar flow.
This basic principle was confirmed in simulations for all different
medium viscosities and treatment chamber geometries tested. The in-
fluence of inlet velocity on the flow profile correlated for all tested
viscosity and treatment chamber configurations. Therefore, reducing
the inlet velocity would increase the treatment homogeneity in parallel
plate treatment chambers, provided a laminar flow field.
3.3.2. Effect of viscosity
Liquid flow in continuous PEF was simulated for different viscos-
ities, using the viscosity functions determined in Section 3.1 for the
different concentrations of A. platensis. The inlet velocities and all other
parameters were kept constant to investigate the effect that viscosity
alone had on the velocity profile. The simulation results are displayed
in Fig. 5 for both parallel plate treatment chamber geometries. In the
10 mm diameter parallel plate chamber, the effect of viscosity (Fig. 1)
was clearly visible (Fig. 5a). At a low viscosity equal to that of water,
the maximum velocity in the treatment chamber was the highest. There
was a highly pronounced velocity profile and recirculation zones near
the chamber walls. These recirculation zones could be reduced with an
optimized treatment chamber inlet design. Nevertheless, with in-
creasing viscosity, the peak velocity decreased and the velocity profile
became flatter and more homogeneous overall. In the 1 mm diameter
chamber, the velocity profile appeared to be almost independent from
the viscosity (Fig. 5b). Within narrow treatment chambers, the friction
from the walls and the interacting forces between the fluid and the
walls dominated the fluid flow. The wall forces appear to have influ-
enced the flow field all the way to the center of the treatment chamber.
On the other hand, with larger geometries such as the 10 mm diameter
chamber, the wall appeared to exert its effect on liquid fractions that
were nearby, but the viscosity of the fluid governed the liquid flow in
the center.
Goettel et al. (2013) observed that the conductivity of PEF-treated
microalgae (A. protothecoides SAG 211-7a) suspensions increased with
increasing biomass concentration, making the PEF treatments at higher
concentrations more efficient. An increase in concentration led to an
increase in viscosity, as illustrated in section 3.1. Therefore, based on
the results presented in Fig. 5, an improvement in treatment homo-
geneity and energy efficiency accompanied the increased biomass
concentration. This finding further emphasizes the importance of flow
field considerations in continuous PEF systems.
Overall, the results showed that the medium viscosity does have an
impact on the flow profiles, although this effect seems to be dependent
on the size of the treatment chamber. Higher viscosities in general
improved the flow field uniformity and reduced the peak viscosity in
the center of the treatment chamber. When scaling up from a treatment
chamber with one or only a few millimeters in diameter, one should be
aware that the impact of viscosity might become more important with
increasing scale.
3.4. Energy input comparison between different treatment chambers
In PEF treatments, the specific energy input is crucial to assess
treatment effectivity and to compare results from different laboratories.
In parallel plate treatment chambers where the electric field is uniform,
variabilities in the flow fields account for the entire inhomogeneity in
energy input. Therefore, only the flow field must be simulated without
having to consider the electric field distribution. In contrast, with co-
linear treatment chambers, both the flow field and the electric field are
inhomogeneous. Therefore, there are two independent factors that both
contribute to potential treatment inhomogeneities. It may not be suf-
ficient to only investigate one of these two factors. To make a con-
clusive statement on the overall treatment homogeneity, specific energy
input distributions were calculated for both co-linear and parallel plate
treatment chambers, combining the results from flow field and electric
field simulations. To make the results as comparable as possible, a co-
linear cell with a 4 mm gap between the insulators (the geometry uti-
lized by Meneses et al. (2011c)) was compared to a parallel plate
treatment chamber with a 4 mm electrode gap (the geometry utilized by
Goettel et al. (2013)). All other parameters such as inlet velocity,
Fig. 4. Simulation results showing the effect of increasing inlet velocity u
0
on
the velocity profiles in the parallel plate treatment chambers for a) A. platensis
suspension of concentration 100 g L
−1
in the 10 mm diameter parallel plate
chamber and b) water in the 1 mm diameter parallel plate chamber.
Fig. 5. Simulation results showing the effect of increasing viscosity, according
to Fig. 1, on the velocity profiles in the treatment chambers for different con-
centrations of A. platensis suspensions for the a) 10 mm and b) 1mm diameter
parallel plate chambers.
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
272
viscosity, electric potential, pulse width, pulse repetition rate, and
medium conductivity were kept equal for both treatment chambers
(Table 2). The results for the energy input distribution are visualized in
Fig. 6. Since there was a much larger treatment zone in the parallel
plate chamber and the liquid was exposed to a higher pulse number, the
specific energy input was approximately ten times higher than that of
the co-linear chamber. However, by reducing the pulse repetition fre-
quency by the same factor of ten, the energy input for both treatment
chambers correlated. The shape of the energy input distributions was
almost the same for both geometries. Therefore, for the chosen para-
meters, both treatments were approximately equally nonuniform, with
equal amounts of deviation from the average treatment effect. How-
ever, as illustrated by the dotted line (· ·) in Fig. 6, there was a sig-
nificant difference between the actual energy input distribution in the
co-linear chamber considering electric and flow field, and the energy
distribution assuming a uniform flow field. This difference further un-
derlines the necessity to consider the flow field in continuous PEF.
The parameters of the energy input had to be examined more closely
to understand what made up the inhomogeneities in both treatment
chamber geometries and what was different between the two geome-
tries. The parallel plate chamber had a uniform electric field but a
nonuniform flow field. On the other hand, the co-linear chamber had a
more uniform flow field due to its pinched geometry. However, the
electric field of the co-linear chamber was nonuniform; it was stronger
near the walls and weaker towards the center of the chamber.
Therefore, the liquid near the walls that was already flowing more
slowly experienced a higher field, while the fast-moving liquid in the
center experienced a lower electric field. Consequently, the flow field
was more uniform in the co-linear chamber, but the electric field
nonuniformity exacerbated the effect of the inhomogeneous flow field.
Overall, it is possible that under certain conditions, either the co-linear
or the parallel plate chamber is more uniform concerning the energy
input distribution. In the case simulated in Fig. 6, the effects of electric
field and flow field described above seemed to be in balance, resulting
in correlating energy input profiles for both treatment chambers.
However, any of the medium or process parameters can be modified,
and the modifications might favor one of the two geometries more than
the other. While for different reasons, both the co-linear and parallel
plate treatment chambers exhibited a nonuniform energy input dis-
tribution. Therefore, it is important to calculate and compare the en-
ergy input distributions every time when changing between the two
geometries, optimizing the treatment chamber designs, modifying
process parameters, or scaling up the process.
The dashed line (- -) in Fig. 6 shows the specific energy input dis-
tribution of a batch parallel plate treatment chamber with the same
dimensions as the continuous chamber. All other parameters such as the
electric potential, conductivity, pulse width and frequency were kept
the same as for the continuous process (Table 2). It became evident that
the energy input in the continuous process deviated greatly from a
batch process, even when using the otherwise exact same process
parameters and geometry. To quantify the extent of the deviation, one
can disregard the fluid adjacent to the wall, and focus on the 80 percent
of fluid in the center of the chamber. In these middle 80 percent of the
parallel plate chamber volume, the average specific energy input was
approximately 165 J kg
−1
for the continuous system and three times
higher for the batch system at approximately 700 J kg
−1
. At the same
time, the approximately 10 percent of liquid nearest to the walls dis-
played an extremely high energy input that was significantly beyond
the energy input of 700 J kg
−1
expected from the batch treatment and
may have led to overtreated microalgae cells. This difference between
batch and continuous PEF processes is striking, and it is highly relevant
for improving reproducibility in PEF research. Considering the large
deviations between batch and continuous PEF, one must be extremely
cautious when transferring process knowledge or settings from batch to
continuous processes during scale-up.
4. Conclusions
Sustainable and economically viable microalgae-based biorefinery
concepts require cost-effective processing. PEF processing demon-
strated relevant applications in microalgae valorization; however, the
reproducibility of results was low. In continuous PEF processing electric
and flow field, characterization is crucial. The novel approach to
combine multiphysics simulation with noninvasive UVP measurements
and rheological validation enabled a comprehensive PEF system char-
acterization and control. The results presented in this work will allow
for better reproducibility of results, facilitating the research on PEF and
subsequent microalgae biorefinery approaches. Further research is re-
quired in the area of turbulent flow fields for scale-up considerations.
Acknowledgements
The authors gratefully acknowledge the ETH World Food System
Center (Project “NewAlgae”, grant number: 2-72235-17), Dr. Wolfgang
Frey and Dr. Christian Gusbeth from the Karlsruhe Institute of
Technology (KIT), the ETH Zürich Foundation, as well as Prof. Dr. Erich
J. Windhab, Dr. Damien Dufour, Pascal Bertsch, Daniel Kiechl and
Bruno Pfister from the ETH Zürich Food Process Engineering
Laboratory for their support.
Conflict of interest statement
The authors declare no conflict of interest.
Table 2
Parameters used for the calculation of the energy input distribution in the
parallel plate and co-linear treatment chamber geometries. The electric po-
tential U, the pulse width τ
p
, the medium conductivity σ, the pulse repetition
frequency fand the treatment chamber diameter D
in
were all defined the same
for both treatment chamber geometries, providing the same conditions for both
chambers in order to make the results comparable. Due to the differences in
geometry, the treatment chamber length Lwas larger for the parallel plate
chamber than for the co-linear chamber.
Uτ
p
σfD
in
L
(V) (µs) (mS cm
−1
)(s
−1
) (mm) (mm)
parallel plate 1000 1 4 9 4 28
co-linear 1000 1 4 9 4 5
Fig. 6. Simulated distribution of specific energy input across treatment cham-
bers for parallel plate and co-linear geometries, both with 4 mm diameter and
assuming the viscosity of water. All other parameters (inlet velocity, electric
potential, pulse width, pulse repetition rate, and medium conductivity) were
also kept equal for both treatment chambers (Table 2). The dotted (· ·) line re-
presents the energy input distribution in the co-linear chamber when assuming
an average or uniform flow field and only considering the nonuniform electric
field. The dashed (- -) line represents the energy input in a batch parallel plate
system with no flow and a uniform electric field.
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
273
References
’t Lam, G.P., Postma, P.R., Fernandes, D.A., Timmermans, R.A.H., Vermuë, M.H., Barbosa,
M.J., Eppink, M.H.M., Wijffels, R.H., Olivieri, G., 2017. Pulsed electric field for
protein release of the microalgae Chlorella vulgaris and Neochloris oleoabundans. Algal
Res. 24, 181–187.
Álvarez, I., Condón, S., Raso, J., 2006. Microbial inactivation by pulsed electric fields. In:
Raso, J., Heinz, V. (Eds.), Pulsed Electric Fields Technology for the Food Industry.
Food Engineering Series, Springer, Boston, MA, pp. 97–129.
Barbosa-Cánovas, G.V., Ghani Albaali, A., Juliano, P., Knoerzer, K., 2011. Introduction to
innovative food processing technologies: Background, advantages, issues, and need
for multiphysics modeling. In: Knoerzer, K., Juliano, P., Roupas, P., Versteeg, C.
(Eds.), Innovative Food Processing Technologies: Advances in Multiphysics
Simulation. John Wiley & Sons, Ltd., pp. 3–21.
Bensalem, S., Lopes, F., Bodénès, P., Pareau, D., Français, O., Le Pioufle, B., 2018.
Understanding the mechanisms of lipid extraction from microalga Chlamydomonas
reinhardtii after electrical field solicitations and mechanical stress within a micro-
fluidic device. Bioresour. Technol. 257, 129–136.
Bernaerts, T.M.M., Panozzo, A., Doumen, V., Foubert, I., Gheysen, L., Goiris, K.,
Moldenaers, P., Hendrickx, M.E., Van Loey, A.M., 2017. Microalgal biomass as a
(multi)functional ingredient in food products: rheological properties of microalgal
suspensions as affected by mechanical and thermal processing. Algal Res. 25,
452–463.
Buchmann, L., Böcker, L., Frey, W., Haberkorn, I., Nyffeler, M., Mathys, A., 2018. Energy
input assessment for nanosecond pulsed electric field processing and its application in
a case study with Chlorella vulgaris. Innov. Food Sci. Emerg. Technol. 47, 445–453.
Buckow, R., Schroeder, S., Berres, P., Baumann, P., Knoerzer, K., 2010. Simulation and
evaluation of pilot-scale pulsed electric field (PEF) processing. J. Food Eng. 101,
67–77.
Ciferri, O., 1983. Spirulina, the edible microorganism. Microbiol. Rev. 47, 551–578.
Eing, C.J., Bonnet, S., Pacher, M., Puchta, H., Frey, W., 2009. Effects of nanosecond
pulsed electric field exposure on Arabidopsis thaliana. IEEE Trans. Dielectr. Electr.
Insul. 16, 1322–1328.
Ewoldt, R.H., Johnston, M.T., Caretta, L.M., 2015. Experimental challenges of shear
rheology: how to avoid bad data. In: Spagnolie, S.E. (Ed.), Complex Fluids in
Biological Systems. Springer Science+Business Media, New York, pp. 207–241.
Fiala, A., Wouters, P.C., van den Bosch, E., Creyghton, Y.L., 2001. Coupled electrical-fluid
model of pulsed electric field treatment in a model food system. Innov. Food Sci.
Emerg. Technol. 2, 229–238.
Gerlach, D., Alleborn, N., Baars, A., Delgado, A., Moritz, J., Knorr, D., 2008. Numerical
simulations of pulsed electric fields for food preservation: a review. Innov. Food Sci.
Emerg. Technol. 9, 408–417.
Goettel, M., Eing, C., Gusbeth, C., Straessner, R., Frey, W., 2013. Pulsed electric field
assisted extraction of intracellular valuables from microalgae. Algal Res. 2, 401–408.
Jaeger, H., Meneses, N., Knorr, D., 2009. Impact of PEF treatment inhomogeneity such as
electric field distribution, flow characteristics and temperature effects on the in-
activation of E. coli and milk alkaline phosphatase. Innov. Food Sci. Emerg. Technol.
10, 470–480.
Kempkes, M.A., Roth, I., Gaudreau, M.P.J., 2011. Pulsed Electric Field (PEF) Method for
Continuous Enhanced Extraction of Oil and Lipids From Small Aquatic Plants. Patent
Application. US2011/0107655 A1.
Krauss, J., Rauh, C., Delgado, A., 2011. Chapter 11 Novel, multi –objective optimization
of pulsed electric field processing for liquid food treatment. In: Knoerzer, K., Juliano,
P., Roupas, P., Versteeg, C. (Eds.), Innovative Food Processing Technologies:
Advances in Multiphysics Simulation. John Wiley & Sons, Ltd., pp. 209–231.
Krieger, I.M., Dougherty, T.J., 1959. A mechanism for non-newtonian flow in suspensions
of rigid spheres. Trans. Soc. Rheol. 3, 137–152.
Mahnič-Kalamiza, S., Vorobiev, E., Miklavčič, D., 2014. Electroporation in food
processing and biorefinery. J. Membr. Biol. 247, 1279–1304.
Mathys, A., Toepfl, S., Siemer, C., Favre, L., Benyacoub, J., & Hansen, C. E., 2013. Pulsed
Electric Field Treatment Process and Dairy Product Comprising Bioactive Molecules
Obtainable by the Process. Patent Application. EP 2543254(A1). NESTEC S.A.
Meneses, N., Jaeger, H., Knorr, D., 2011a. Minimization of thermal impact by application
of electrode cooling in a co-linear PEF treatment chamber. J. Food Sci. 76, 536–543.
Meneses, N., Jaeger, H., Knorr, D., 2011b. Computational fluid dynamics applied in
pulsed electric field preservation of liquid foods. In: Knoerzer, K., Juliano, P., Roupas,
P., Versteeg, C. (Eds.), Innovative Food Processing Technologies: Advances in
Multiphysics Simulation. John Wiley & Sons, Ltd., pp. 193–207.
Meneses, N., Jaeger, H., Moritz, J., Knorr, D., 2011c. Impact of insulator shape, flow rate
and electrical parameters on inactivation of E. coli using a continuous co-linear PEF
system. Innov. Food Sci. Emerg. Technol. 12, 6–12.
Metzner, A.B., 1985. Rheology of suspensions in polymeric liquids. J. Rheol. 29, 739–775.
Mezger, T.G., 2015. Applied Rheology, first ed. Anton Paar GmbH, Graz, Austria.
Pan, N., 1993. Theoretical determination of the optimal fiber volume fraction and fiber-
matrix property compatibility of short fiber composites. Polym. Compos. 14, 85–93.
Parniakov, O., Barba, F.J., Grimi, N., Marchal, L., Jubeau, S., Lebovka, N., Vorobiev, E.,
2015. Pulsed electric field and pH assisted selective extraction of intracellular com-
ponents from microalgae Nannochloropsis. Algal Res. 8, 128–134.
Postma, P.R., Pataro, G., Capitoli, M., Barbosa, M.J., Wijffels, R.H., Eppink, M.H.M.M.,
Olivieri, G., Ferrari, G., 2016. Selective extraction of intracellular components from
the microalga Chlorella vulgaris by combined pulsed electric field-temperature treat-
ment. Bioresour. Technol. 203, 80–88.
Raso, J., Calderón, M.L., Góngora, M., Barbosa-Cánovas, G.V., Swanson, B.G., 2006.
Inactivation of Zygosaccharomyces bailii in fruit juices by heat, high hydrostatic
pressure and pulsed electric fields. J. Food Sci. 63, 1042–1044.
Raso, J., Frey, W., Ferrari, G., Pataro, G., Knorr, D., Teissie, J., Miklavčič, D., 2016.
Recommendations guidelines on the key information to be reported in studies of
application of PEF technology in food and biotechnological processes. Innov. Food
Sci. Emerg. Technol. 37, 312–321.
Rocha, C.M.R., Genisheva, Z., Ferreira-Santos, P., Rodrigues, R., Vicente, A.A., Teixeira,
J.A., Pereira, R.N., 2018. Electric field-based technologies for valorization of bior-
esources. Bioresour. Technol. 254, 325–339.
Spagnolie, S.E. (Ed.), 2015. Complex Fluids in Biological Systems. Springer Science
+Business Media, New York.
Takeda, Y. (Ed.), 2012. Ultrasonic Doppler Velocity Profiler for Fluid Flow. Springer,
Japan.
Takeda, Y., 1995. Velocity profile measurement by ultrasonic Doppler method. Exp.
Therm. Fluid Sci. 10, 444–453.
Toepfl, S., Mathys, A., Heinz, V., Knorr, D., 2006. Review: potential of high hydrostatic
pressure and pulsed electric fields for energy efficient and environmentally friendly
food processing. Food Rev. Int. 22, 405–423.
Vorobiev, E., Lebovka, N. (Eds.), 2008. Electrotechnologies for Extraction from Food
Plants and Biomaterials. Springer Science+Business Media, New York.
Wiklund, J., Shahram, I., Stading, M., 2007. Methodology for in-line rheology by ultra-
sound Doppler velocity profiling and pressure difference techniques. Chem. Eng. Sci.
62, 4277–4293.
Wölken, T., Sailer, J., Maldonado-Parra, F.D., Horneber, T., Rauh, C., 2017. Application of
numerical simulation techniques for modeling pulsed electric field processing. In:
Miklavcic, D. (Ed.), Handbook of Electroporation. Springer International Publishing
AG, pp. 1237–1267.
Zhang, X., Jiang, Z., Chen, L., Chou, A., Yan, H., Zuo, Y.Y., Zhang, X., 2013. Influence of
cell properties on rheological characterization of microalgae suspensions. Bioresour.
Technol. 139, 209–213.
Zhang, Q., Qin, B.L., Barbosa-Cánovas, G.V., Swanson, B.G., 1995. Inactivation of E. coli
for food pasteurization by high-strength pulsed electric fields. J. Food Proc. Pres. 19,
103–118.
L. Buchmann et al. Bioresource Technology 265 (2018) 268–274
274
