Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Tubular photobioreactor model for

light dark cycles control


Yulnderson Salguero-Rodrguez, Cesar Augusto Gmez-Perez, Jairo Jos Espinosa Oviedo
Departamento de Energa elctrica y automtica
Universidad Nacional de Colombia - UNAL
Facultad de minas, Medelln, Colombia
ysalgueror@unal.edu.co, cagomez@unal.edu.co, jespinov@unal.edu.co

AbstractThis paper shows a model of a photobioreactor reactors of identical dimensions that did not implement the LD
(PBR) combining different phenomena inside a reactor for cycles. In the last years the use of static mixers or flow obstacles
microalgae culture such as: fluid dynamics, photosynthesis has been studied. Static mixers are an effective method that
kinetics and mass balances. The proposed model has a new increases the degree of mixing in the microalgae culture [3],
characteristic; it is the inclusion of an empirical equation allowing the microalgae to move between illuminated and dark
which takes into account the Light and Dark cycles areas inside the reactor. Among some flow obstacles we can
behavior in a dynamic mathematical model for PBR mention studies of Perner [9] with static mixers, Wu [10] with
production prediction. Finally, the model of PBR was spirals, Zhang [11] with baffles and Cheng [5] with
simulated with a regulatory PI control strategy, showing crossmixers. The fluid dynamics was simulated using
adequate results of the system in closed loop. Computational Fluid Dynamics (CFD). Moreover,
experimental application showed that the productivity in
KeywordsPhotobioreactor, Photosynthesis, Light - Dark biomass for PBRs with mixers increased between 30% and 80%
cycles, Microalgae, PSU model. in relation to PBR without static mixers.
I. INTRODUCTION
Furthermore, photosynthesis models from literature [12], [13]
Microalgae are a diverse group of prokaryotic and eukaryotic have shown that LD cycles are beneficial for microalgae
photosynthetic organisms with a simple structure that allows growth. Even though the literature on PBR fluid dynamic
them to grow quickly [1]. Microalgae can be used for different design has considered the LD cycles, there is not a PBR
applications such as: biodiesel and food production, biofilters dynamic model that can predict the effect of LD cycles on the
in water treatment, pharmaceutical and cosmetic applications PBR production. The construction of this dynamic model can
among many others [2]. have application on PBR optimization and control.
The mixing has an important influence on the microalgae This paper presents the model of a tubular PBR which combines
growth [3]. Posten [4] considered that one main factor to obtain aspects such as: fluid dynamics, lighting attenuation models,
an efficient PBR is to improve the agitation and air supply microalgae kinetic growth models, and material balances, in
process. A well mixing prevents sediment formation, increases order to let control actions taking into account LD cycles. In the
mass transfer, and improves exchange between coexisting first part the introductory aspects of this model were presented.
phases [5]. Moreover, light gradients in PBRs make the The second part mentions the mathematical and simulation
microalgae receive illumination depending on their location. tools. In the third section the simulation results will be
An effective mixing produces microalgae movement from presented in a study case with PI control.
illuminated zones to dark zones [6]. This cycle of microalgae
moving between both zones is known as Light-Dark (LD) cycle II. MATHEMATICAL MODELS AND TOOLS
or flashing light effect. Frequency of LD cycles is an important
parameter of the lighting regime and has an influence on A. PBR model
productivity [5]. LD cycles reduce light inhibition, and The PBR can be divided into two main parts. The first section
therefore improve the performance of a PBR. This effect of LD is the solar receiver. In the second section, there is a bubble
cycles is considered important for the design of PBR [7]. column. In order to study photosynthesis rate we will focus on
the solar receiver. More details on the modeled system can be
Since 2001, different designs have been presented in which the found in Fernndez [14], in which they explained the operation
concept of LD cycles has been incorporated; for example, of the whole PBR. Fig. 1 shows a scheme of the PBR. For
Degen [8] presented a panel PBR in which cells were exposed modeling purposes, the solar receiver can be divided into n
to a flashing light, it improved efficiency using light with the sections where the output of each subsystem is the input of the
LD cycles. When performing batch assays, the biomass next one.
productivity for Chlorella vulgaris was 70% higher than
Where total mass of the analyzed element is , and the mass
of carbon dioxide, oxygen and biomass is represented by
, , . The variable kinetics for carbon dioxide,
oxygen and biomass are represented by , , .
It is possible to establish that there is not net accumulation of
mass in the system (5).

= (5)

It is assumed that mixing achieved has homogeneous


conditions, and due to it, the properties inside the element are
equivalent to the values of the properties at the output. Then,
carbon dioxide system mass can be evaluated in (6).
Fig. 1. PBR Scheme. Adapted from Fernndez [14].
= [ ] (6)
Fig. 2 shows the process systems diagram. In the initial element
which we denoted by 1, there is a mass flow , with a mass Here, is CO2 molecular weight, is the system volume
concentration of carbon dioxide , a mass oxygen and [ ] is molar carbon dioxide concentration. Global
concentration , and a mass concentration of biomass . equation for photosynthesis can be written in (7) [15]. Like the
The output of element 1 corresponds to the input of element 2. stoichiometric ratio for carbon dioxide and oxygen is 1:1, their
The subindex under each of the variables denote the element of reaction molar rates, are equivalent as shown in (8).
the solar receiver associated.
+ + 8 + (7)
= (8)

It is possible to establish a molar ratio of 1:8 for the amount of


carbon dioxide and oxygen in the reaction and the photons ( )
used, expressing them in grams we have (9) and (10).


= = 5,5 (9)


Fig. 2. Process system diagram of solar receiver divided into sections.
= =4 (10)

The PBR model has the following assumptions: An approximate composition of microalgae in mass percentage
The receiver is fed with saturated CO2 in culture medium. is 70% in carbon [16], then it is possible to make an equivalence
Uniform and constant temperature. between the fixed carbon and the amount of biomass in the
Oxygen concentration in the system at any point is less system, according to (11) and (12).
than its saturation value on the medium.
System has a homogeneous phase. = = 1,5 (11)


Operating regime of the pump is at its operation point 1,5 = 2,14 (12)
,

The variables of interest in PBR are the concentration of


Carbon dioxide rate can be considered independent of the
biomass, and chemical species involved in photosynthesis such
concentration of the same because the system has enough
as: CO2 and O2. The mass balance for each of the variables and carbon dioxide for the photosynthesis reaction shown in (7). On
the global mass balance in the initial element are presented from the other hand, the attenuation of the illumination due to the
(1) to (4). high biomass concentrations frequently used can generate dark
areas. Due to it, the illumination is considered as the limiting
= (1) factor of this reaction, which influences the specific growth rate
( ) by kinetic models. The stoichiometry of the reaction and the
= (2) specific growth rate achieved are related to the dioxide rate in
(13). is the biomass concentration in the study element.
= + (3)
Oxygen mass rate is given by (14). The biomass rate can be
= + (4) represented by (15).

,
= (13)
,

= (14) represents a model constant. The transport equation for is
,
shown in (27).
= (15)

The biomass can be rewritten as the product, between the = (26)
volume and the biomass concentration, as shown in (16).
+ = (( + ) )+ (27)
= (16)
The term is given by (28). The transport equation for is
Mass flow divided by density ( ) is equivalent to the volumetric shown in (29). In these equations represents the fluid density,
flow ( ), which can be expressed as the product between the average velocity, time, the dynamic viscosity, the
average flow velocity ( ) and the cross section area ( ). As the turbulence energy, the energy dissipation of turbulence. The
solar receiver is decomposed by several cylinder elements, the parameters used were taken from COMSOL manual [18].
volume of each element is given by the product between its area
and the length of the element ( ). By simplifying this = ( ( +( ) ) ( ) ) (28)
expression, (17) is obtained.
+ = (( + ) )+ (29)

= = = (17)
The model is simulated in COMSOL with boundary conditions
shown in (30). Where , represents the average velocity
Replacing previous equations in mass balances, the expressions normal to the flow area and is the atmospheric pressure;
(18), (19) and (20) are obtained. In this document, the analysis the last condition to accomplish the continuity equation.
for the first element has been shown, but it is enough to change
the subindex to analyze more than 1 element. Input =
[ ] ,
Output = (30)
= ([ ] [ ] ) (18) Wall =0
[ ] ,
= ([ ] [ ] )+ (19)
D. Particle tracking model
= ( )+ (20) The movement of particles in the fluid is simulated using a
model derived from Newtons laws, shown in (31) [19]. Where
represents the drag force (32) and the gravitational force
B. Kinetic model
(33).
Microalgae growth is going to be modeled taking into account
the photosynthesis process. Most of existing photosynthesis ( )
kinetic models have been constructed based on photosynthetic = + (31)
unit (PSU). The PSU model is fully described in [12]. The = ( ) (32)
mathematical expressions of the PSU model are shown in
( )
equations (21), (22), (23), (24) and (25) [12]. = (33)

= + + (21) In these equations, represents the velocity of the particle,


= (22) velocity of the fluid, the gravitational acceleration, is the
density of the particle, the density of the fluid, the mass
= (23)
of the particle, the particle diameter.
+ + =1 (24)
= (25) E. Illumination model
Illumination received by the system is calculated using a model
Where , , are the PSU fractions in the states resting, presented by Luo [20] showed in (34), it is a modification of
activated and inhibited. The other parameters are constants with the Beer-Lambert law.
a luminous intensity, the values reported for this model by Wu
and Merchuk [17] were used. This model can predict the effect = exp[( + ) ] (34)
of the light history (LD cycles) of cells on productivity [12].
C. Fluid dynamics model Where represents the external illumination, is the distance
of the cells to the illuminated surface, and are the
The turbulence model considers transport equations
extinction coefficients for the water and the cells respectively,
related to kinetic energy of turbulence, and energy dissipation
and finally is the cellular concentration of biomass. The
rate, . COMSOL models the viscosity with (26), in which
values of this model were reported by Bchet [21] and Luo [20].
F. Discret fourier transform where the presence of one or more dominant frequencies is
more evident. In the system several particles were simulated
The Fourier transformation is used to express periodic functions and a window function was applied to the illuminations data,
with a combination of sinusoidal basis functions. Data obtained then we proceeded to concatenate all illumination profiles and
from CFD simulation are defined at discrete points [9]. Due to they were analyzed with the Fourier transformation in the
the discrete data, it is then necessary to use the Discrete Fourier frequency domain. The illumination signal in the frequency
Transform (DFT) to evaluate the periodic behavior of simulated domain is shown in Fig. 4.
particles, this transformation is showed in (35).

()= ( ) exp( ) , = 0, 1, , 1 (35)

Where ( ) , = 0,1, , 1 is a uniformly sampled


sequence of data points with time sampling interval , the
total amount of samples and ( ), = 0, 1, , 1 the DFT
coefficients. It is important to say that DFT can be used to
evaluate periodicity in time, especially for LD cycles as showed
different author such as: Marshall [22] and Wu [23]. From the
DFT results, the power spectrum is computed and analyzed as
a graphic tool to evaluate the frequency characteristics from the
data.
Fig. 4. Frequencies in the power spectrum
III. RESULTS Regarding to the power spectrum in the frequency domain, it is
well known that the system shows several frequencies, in this
A. Illumination of the system case the frequencies associated with the highest peaks were
In order to achieve an adequate mixing in the system, a mixer taken in the spectrum. Then the illumination as a function of
was proposed which is shown in Fig. 3 a), it is formed by two time and velocity should be represented by a set of sinusoidal
crosses baffles. Inside the cylinder in the solar receiver first functions, with an amplitude ( ), a frequency ( ) and an offset
mixer of crossed baffles was located to a length of 0.25 m, so ( ) as shown in (36).
the flow profile is fully developed before arriving to the mixer,
the distance between the baffles was of 0.5 m. ( , ) = ( ) sin( ( ) ) + (36)

a) b) In this work, three main frequencies with their respective values


of the power spectrum were taken. Equations for the frequency
and the power are shown from (37) to (42), and then the
amplitude of every wave is calculated with (43). The particles
were simulated with velocities from 0.05 m/s to 0.55 m/s, in
Fig. 5 and 6 are showed the first frequency and the power
spectrum of the first frequency, where a linear relationship
between the variables and velocity was found.

Fig. 3. a) Geometry mixer proposed b) Illumination received for a microalga

After the simulation of the computational fluid dynamics (CFD)


and the particle tacking, both were simulated in COMSOL. The
trajectory of every microalga was indicative of a periodic
movement. Some interactions between microalgae particles
produce some disturbances; nevertheless, the proposed mixer
produced the switch from dark to illuminated zones. Fig. 3 b)
presents the illumination received by the microalga, it is Fig. 5. First frequency as a function of velocity
observed the periodic movement, and due to this movement, an
oscillatory illumination, the microalga moves between ( ) = 2.9798 0.1362 (37)
illuminated and dark areas during its trajectory. ( ) = 3.0205 0.1874 (38)
( ) = 3.0546 0.1671 (39)
The Fourier transformation allows us to analyze the data of the
illumination received by the particles in the frequency domain,
Fig. 8. Carbon dioxide concentration
Fig. 6. First power spectrum as function of velocity Oxygen and biomass concentration in the output of the system
are shown in Fig. 9 a) and b). Initially, both increase to their
( )= (40)
. maximum point, after that, they descend again because the
( )= (41) inverse reaction to (7). In others words, the reverse reaction in
. .
(7), in which photosynthesis is carried out, by means of the
( )= (42)
. fixation of carbon dioxide and release of oxygen. The system
( )
( )= 0.7018 (43) reaches the operation point of the system at 10000 s.
( )

The reconstructed profile for the particle average behavior is


shown in Fig. 7 using (36), it shows the average oscillatory
behavior of the particles alternating between illuminated areas
and dark areas, this illumination profile is used in the simulation
of the photosynthesis kinetics in order to obtain the specific
growth rate of the microalgae, and the behavior of the biomass
and the species involved in photosynthesis.
Fig. 9. Dynamics of concentration at the output of the PBR a) Oxygen b)
Biomass

C. PBR control
The control objective was to maintain the photosynthetic
efficiency at a desired level, and so the concentration of oxygen
at a fixed level. Therefore, the variable to be controlled is the
oxygen concentration at the output of the reactor, while the
manipulated variable is the velocity in the system, it has a direct
relationship with the frequencies of the LD cycles. The PBR
Fig.7. Illumination profile as time function system may be affected by two main disturbances; the first is
related to the initial concentration of biomass in the system and
B. PBR dynamics the second is related to the external illumination that the system
The illumination as function of the velocity and time was receives; external illumination varies with solar cycles and the
presented in (36). It is possible to see reflected the frequency of presence of clouds. For this system the performance of the
the LD cycles in the specific growth rate, with equations (21) controller will be analyzed to reach a fix value.
to (25) and (36). Then equations (18) to (20) extended with 10
elements are studied the dynamics of biomass and the chemical A regulatory control strategy based on PI controller is used,
species involved in the PBR. The PBR simulated had a total whose closed control loop is shown in Fig. 10, where a
length of 100 m, and a radius of 0.025 m. The simulation reference in the oxygen concentration ([ ] ) is fixed for the
parameters are shown in Table I. reactor output, in the last element corresponding to the 10
partitions analyzed ( [ ] ). Subsequently, the controller
TABLE I. Simulation parameters. calculates the velocity of the system ( ), this is used to generate
Parameter Unit Value an average illumination function according to (36). With the
[ ] mol/L 0.02 previously mentioned three-state kinetic model it is possible to
[ ] mol/L 0 calculate the specific growth rate ( ), it is used in the model of
g/L 180.151 the PBR to simulate the behavior of both biomass and oxygen
/ 0.3
IE / 450
in the system, which are products of the photosynthetic process.

The concentration of carbon dioxide in the output of the system The parameters of the controller were estimated using the
is shown in Fig. 8. Initially, the concentration decreases to its method of the reaction curve of Ziegers and Nichols [24] and
minimum point, then it rises again. were K = 43913.5 and = 1.5. Fig. 11 shows the
reference of the concentration of oxygen, whose level was set
at [ ] = 0.003 / , in this figure it is well known that microalgae, Eng. Life Sci., vol. 9, no. 3, pp. 165177, 2009.
the closed loop system is able to reach the reference level and [5] W. Cheng, J. Huang, and J. Chen, Computational fluid dynamics
simulation of mixing characteristics and light regime in tubular
establish it in 200 seconds, which is a relatively short time, photobioreactors with novel static mixers, J. Chem. Technol.
considering that the system in open loop takes about 14000 Biotechnol., vol. 91, no. 2, pp. 327335, 2014.
seconds to reach its operation point. [6] P. Hartmann, Q. Bchet, and O. Bernard, The effect of
photosynthesis time scales on microalgae productivity, Bioprocess
Biosyst. Eng., vol. 37, no. 1, pp. 1725, 2014.
[7] F. Rubio Camacho, F. Garca Camacho, J. M. Fernndez Sevilla, Y.
Chisti, and E. Molina Grima, A mechanistic model of
photosynthesis in microalgae, Biotechnol. Bioeng., vol. 81, no. 4,
pp. 459473, 2003.
[8] J. Degen, A. Uebele, A. Retze, U. Schmid-Staiger, and W. Trsch,
Fig. 10. Control scheme for PBR. A novel airlift photobioreactor with baffles for improved light
utilization through the flashing light effect, J. Biotechnol., vol. 92,
no. 2, pp. 8994, 2001.
[9] I. Perner-Nochta and C. Posten, Simulations of light intensity
variation in photobioreactors, J. Biotechnol., vol. 131, no. 3, pp.
276285, 2007.
[10] L. B. Wu, Z. Li, and Y. Z. Song, Hydrodynamic conditions in
designed spiral photobioreactors, Bioresour. Technol., vol. 101, no.
1, pp. 298303, 2010.
[11] T. Zhang, Dynamics of fluid and light intensity in mechanically
stirred photobioreactor, J. Biotechnol., vol. 168, no. 1, pp. 107116,
2013.
Fig. 11. Oxygen concentration in closed loop [12] X. Wu and J. C. Merchuk, A model integrating fluid dynamics in
photosynthesis and photoinhibition processes, Chem. Eng. Sci., vol.
IV. CONCLUSIONS 56, no. 11, pp. 35273538, 2001.
[13] A. Solimeno et al., New mechanistic model to simulate microalgae
The model presented in this paper combines in simple ways growth, Algal Res., vol. 12, pp. 350358, 2015.
different aspects of the PBR dynamics such as: fluid dynamics, [14] I. Fernndez, F. G. Acin, M. Berenguel, J. L. Guzmn, G. A.
illumination, photosynthesis and mass balances. Besides, it Andrade, and D. J. Pagano, A lumped parameter chemical-physical
takes into account the LD cycles effect on the behavior of the model for tubular photobioreactors, Chem. Eng. Sci., vol. 112, pp.
116129, 2014.
photosynthesis, and productivity achieved in the PBR. This
[15] G. Y. Jaramillo and E. M. Bustamante, En La Produccin
model can be used in different control strategy, due to its Agropecuaria, pp. 7390, 2005.
simplicity and that involves the effect of different variables in [16] J. F. Zijffers, K. J. Schippers, K. Zheng, M. Janssen, and J. Tramper,
the productivity of the PBR. In this study, a PI controller was Maximum Photosynthetic Yield of Green Microalgae in
applied, it showed satisfactory results for tracking the reference Photobioreactors, pp. 708718, 2010.
in less than 200 seconds. With the model presented in this [17] X. Wu and J. C. Merchuk, Simulation of algae growth in a bench-
paper, it is possible to establish optimal conditions of the PBR, scale bubble column reactor, Biotechnol. Bioeng., vol. 80, no. 2, pp.
156168, 2002.
with an optimization layer. Moreover, several control strategies
[18] COMSOL, Comsol Multiphysics Users Guide, Heat Transf.
could be tested to evaluate its performance. Branch, p. 709745, The Heat Transfer Branch, 2012.
[19] COMSOL, The Particle Tracing Module Users Guide, 2013.
V. AKNOWLEDGMENTS
[20] H. P. Luo and M. H. Al-Dahhan, Analyzing and Modeling of
The authors are grateful to the financial support of Photobioreactors by Combining First Principles of Physiology and
COLCIENCIAS and Universidad Nacional de Colombia Hydrodynamics, Biotechnol. Bioeng., vol. 85, no. 4, pp. 382393,
throught the program Jovenes investigadores e innovadores 2004.
convocatoria N 706 de 2015 [21] Q. Bchet, A. Shilton, and B. Guieysse, Modeling the effects of light
and temperature on algae growth: State of the art and critical
assessment for productivity prediction during outdoor cultivation,
VI. REFERENCES
Biotechnol. Adv., vol. 31, no. 8, pp. 16481663, 2013.
[1] J. P. Bitog et al., Application of computational fluid dynamics for [22] J. S. Marshall and Y. Huang, Simulation of light-limited algae
modeling and designing photobioreactors for microalgae production: growth in homogeneous turbulence, Chem. Eng. Sci., vol. 65, no.
A review, Comput. Electron. Agric., vol. 76, no. 2, pp. 131147, 12, pp. 38653875, 2010.
2011.
[23] B. Wu, Advances in the use of CFD to characterize, design and
[2] Q. Zhang, X. Wu, S. Xue, K. Liang, and W. Cong, Study of optimize bioenergy systems, Comput. Electron. Agric., vol. 93, pp.
hydrodynamic characteristics in tubular photobioreactors, 195208, 2013.
Bioprocess Biosyst. Eng., vol. 36, no. 2, pp. 143150, 2013.
[24] D. Seborg, T. Edgar, D. Mellichamp, and F. Doyle, Process
[3] J. Huang et al., Improving performance of flat-plate Dynamics and Control, Third edit. John Wiley & Sons, Inc., 2011.
photobioreactors by installation of novel internal mixers optimized
with computational fluid dynamics, Bioresour. Technol., vol. 182,
pp. 151159, 2015.
[4] C. Posten, Design principles of photo-bioreactors for cultivation of

You might also like