Improving microalgal growth by strengthening the flashing light effect simulated with computational fluid dynamics in a panel bioreactor with horizontal baffles

Biological CO2 elimination by photosynthetic microalgae is a sustainable way to mitigate CO2 from flue gas and other sources. Computational fluid dynamics was used to simulate algal cell movement with an enhanced flashing light effect in a novel panel bioreactor with horizontal baffles. Calculation results showed that the light/dark (L/D) cycle period decreased by 17.5% from 17.1 s to 14.1 s and that the horizontal fluid velocity increased by 95% while horizontal baffles were used under a 0.02 vvm air aeration rate and a microalgal concentration of 0.85 g L−1. The probability of the L/D cycle period within 5–10 s increased from 27.9% to 43.6%, indicating a 56% increase when horizontal baffles existed. It was proved by experiments that the mass-transfer coefficient increased by 31% and the mixing time decreased by 13% under a 0.06 vvm air aeration rate when horizontal baffles were used, and the algal biomass yield increased by ∼51% along with the decrease in the L/D cycle period when horizontal baffles were used.


Introduction
CO 2 is a greenhouse gas that mainly causes global warming and contributes to the formation of hostile environments. CO 2 biological elimination by photosynthetic microalgae is a crucial way to mitigate CO 2 from different sources, including the atmosphere and industrial exhaust gases, especially ue gas of coal-red power plants. 1 Lipid production from microalgae was also been optimized for new energy developments. 2,3 Microalgae cultivation is vital in the utilization of microalgal biomass. Various types of bioreactor have been extensively utilized for algal culture, such as a raceway pond, at panel bioreactors, and tubular reactors. Particularly, at panel bioreactors have many advantages, such as a large illuminated surface, suitability for outdoor cultivation, good productivity, and being easy to clean. 4 An appropriate mixed multiphase ow state in bioreactors is pivotal to supply CO 2 efficiently, eliminate produced oxygen, provide alternate periods of light/dark (L/D), equably distribute nutrients, and avoid cell sedimentation. 5 The inuence of L/D cycles with different light intensities on the growth of microalgae has been researched. [6][7][8][9] The vivid characterization of ow eld in photobioreactors via experiment is difficult and costly to achieve. 10 Developments of computational uid dynamics (CFD) and the availability of more powerful computers have paved the way for the modeling and designing of a bioreactor. 11,12 Soman et al. (2015) designed a bioreactor that combined at plate bioreactors and airli, and then further studied the superior liquid circulation properties of the bioreactor using CFD. 13 It was proved that the design had a better surface to volume ratio and hydrodynamic properties. Kommareddy et al. (2017) further simulated the algal growth and hydrodynamic properties in the same bioreactors. 14 Massart et al. (2014) established and validated a CFD hydrodynamic model for a at-panel airli bioreactor. 15 In this respect, experimental water ow rates and the liquid circulation in the riser of the reactor were compared with the CFD solution results. However, these cases were simple combinations of airli and at plate bioreactors. So the problems of airli bioreactors still existed in these design, such as low turbulent kinetic energy and low horizontal velocity in the downcomer. 16 Moreover, microalgal cells movement and L/D cycle in the bioreactor were not investigated. New structure that can overcome the drawbacks of usual airli at panel photobioreactors should be designed and investigated.
The CO 2 mass transfer performance of an airli at-plate bioreactor with at baffle and waved baffle was studied by Chen et al. (2016) 17 through the numerical and experimental methods. The results showed that the downcomer-to-riser cross-sectional area ratio played a major role on the mass transfer behavior of at-plate airli bioreactor. However, this research only studied the gas mass transfer in the reactor. The algal movement and L/D cycle of microalgal cells were not investigated. Moreover, algal cultivation validation was not performed. The at-plate PBRs equipped with internal mixers was developed and further optimized its structure parameters using CFD by . 18 The maximum cell concentration and biomass productivity were 11.3% and 22.2% higher than those in the archetype. However, microalgal cells movement and L/D cycle in the bioreactor were not investigated. Novel baffles named HTTP baffles that produce vortices to improve the uid velocity between light and dark areas in a atpanel bioreactor were developed by . 19 Fluid velocity between light and dark areas increased from $0.9 cm s À1 to $3.5 cm s À1 . Biomass yield increased by 70% with the enhanced ashing light effect. However, microalgal cells movement and L/D cycle in the new design remained unexplored due to the limitations of experimental measurement method.
In the present study, the movement of algal cells in the vortex ow eld produced by horizontal baffles was analyzed through CFD. The cell L/D cycle period, fraction of time that the microalgal cell was exposed to light zone (light time fraction), and the horizontal uid velocity were investigated at different gas aeration rates and microalgal concentrations. The results demonstrated that horizontal baffles can shorten the L/D cycle period of the microalgal cells and thereby improving the microalgal growth rate in at panel bioreactor.

Geometries of the at-panel reactor and horizontal baffles
Panel bioreactor (PBR) schematic with horizontal baffles was showed in Fig. 1. It is 20 cm long, 16 cm wide, and 90 cm high. The diameters of the horizontal baffles are 7 cm with the length of 20 cm. The axis of the lowest horizontal tube is placed 8 cm from the le wall and 9 cm above the bottom of the PBR. Five more horizontal baffles have the same x-axis coordinates and the same axis distance of 12 cm on the y-axis direction.

Flow simulation in the at-panel reactor
The PBR was 3D meshed using ANSYS ICEM CFD 15.0 (64 bit), and the simulation was conducted with ANSYS FLUENT 15.0 (64 bit). The Eulerian two-phase model was applied because using the multiphase model is unavoidable while bubbles occur in the photobioreactors. A standard k-3 model was chosen with rst-order exactness to describe the turbulent ow behavior inside the PBRs. In this model, turbulent dispersion force and gas-liquid interphase drag force were considered. The outer walls and the internal structures of the PBRs were set as no-slip boundary conditions to water. The outlet was set as degassing boundary, representing that only the gas in the dispersed phase could escape from the surface and the continuous phase could not go through the top surface. 18 The time step for the transient ow eld computation was set as 0.004 s. To conrm grid independency, three scale grids (532 525; 904 944; and 1 174 084) were used. Small difference was found between the computed values of and 1 174 084 cells. So, the mesh with 904 944 cells was adopted for all the cases.

Fluid velocity and L/D cycle period calculation
Fluid velocity and L/D cycle period were calculated according to the result of simulation. The vertical uid velocity (V z ) was calculated with the velocity of two lines on the z-axis direction (V z 1 and V z 2 ). Line 1 was from point ( The bottom of the PBR was set as (0, 0, 0), as described in Fig. 1a. A total of 1000 simulated particles were injected from two entrance ports; the two coordinates of the ports were (60 mm, 0 mm, 10 mm) and (À60 mm, 0 mm, 700 mm). The particle diameter used for the algal cells was 5 mm with a density of 1000 kg m À3 . The maximum particle tracking time was set to 60 s. Discrete random walk model, 20 were considered during the simulation, where d is the particle diameter, r is its density, andñ is the velocity vector, with the subscripts p stands for particle and f for the uid (the continuous phase).
The spatial position of each spherical particle was recorded every 0.1 s. The L/D cycle period (T av i ) of a microalgae cell was dened and calculated following the method described by Huang et al. 22 The average L/D cycle period 10 of each cell was used to calculate the average L/D cycle period of the entire respectively. All the parameters above were processed by MAT-LAB R2012b (64 bit) and Microso Office Excel.

Determination of critical depth between light and dark zone under various microalgal concentrations
A glass cylinder (diameter ¼ 10 cm) was lled to different depths with the microalgal solution. A photosensitive electrode (GLZ-C, Zhejiang Top Instrument Co., Ltd. China) was installed on the bottom of the cylinder to record light intensity, and illumination intensity (I) detected by photosensitive electrode was recorded as uid depths increasing in 1 cm increments. I 0 , which was the illumination intensity without microalgal solution, was recorded as 530 mmol m À2 s À1 . In this study, the critical I of Chlorella was set as 96.84 mmol m À2 s À1 according to the report, 23 and the depth corresponding to this critical I was dened as critical depth. In the PBR, the region where I was lower than the critical I or depth was bigger than the critical depth was dened as dark zone, and the remaining region was dened as light zone. Five concentrations of microalgal solution (C i ¼ 0.28, 0.56, 0.85, 1.1, 1.7 g L À1 ) were tested. Microalgal biomass were centrifugation at 8000 rpm for 5 min and then dried under 90 C for 24 h to test microalgal concentrations by gram per liter.

Experimental measurement
Velocity magnitude was tested by using the miniature ultrasonic doppler velocimeter (Boyida Technology Co., Ltd., China) to validate the correctness of the simulation results for velocity. Solution-phase mixing time were calculated according to the method of Mendoza et al. (2013). 24 Water was employed as test uid during the measurement. Initially, the water pH was adjusted to 4.0 AE 1 by adding chlorhydric acid (35%, w/v). Then, 0.20 ml NaOH solution (12 mol L À1 ) per liter of water was added as alkalinity tracer. The time was recorded when the alkalinity tracer was added. The mixing time is the required time for pH variations reaching to lower than 5% of the nal stable value. The pH probes were used to measure the response to the pH pulse at two positions in this PBR. The overall volumetric masstransfer coefficient K L a L was measured by the method of Sierra et al. (2008). 25 Water in the PBR was alternately aerated with air and N 2 . N 2 and air aeration rates were controlled by mass ow meter (SevenstarCS200, China). Then, mass-transfer coefficient was calculated according to the formula: dC L /dt ¼ k L a L (C* À C L ), where C* was the saturation concentration of dissolved oxygen. Dissolved oxygen probes (Mettler Toledo, InPro6850i/ 12/120) and pH probes (Mettler Toledo, InPro3253i/SG/120) were connected to transmitters and data acquisition soware (i-7017fc, ICP DAS, Taiwan). Every 0.1 s, measurements were automatically recorded.

Microalgal cultivation
Microalgae mutant Chlorella PY-ZU1 was cultivated with Bristol's solution (also called soil extract, SE) and measured by the same method described by Cheng et al. (2013). 26 Chlorella PY-ZU1 was cultivated in the at-plate PBR under 23 C with continuous illumination of 530 mmol m À2 s À1 . 15% CO 2 was continuously aerated into the culture medium with 0.02 vvm ow rate.  as periodic exposure of microalgae to light or rapid travel between dark and light zones. Flashing-light effect can be characterized by L/D cycle period. When microalgal concentration was 0.85 g L À1 , the critical depth from light zone to dark zone was test to be 2 cm. Fig. 2 showed the effects of gas aeration rate on light/dark cycle period in a PBR with or without horizontal baffles under a microalgal concentration of 0.85 g L À1 , which was analyzed according to the simulation result. While the air ow rate increased from 0.02 vvm to 0.04 vvm, the L/D cycle period decreased from 14.1 s to 10.7 s with the horizontal baffles, and decreased from 17.1 s to 12.1 s without the horizontal baffles. The addition of horizontal baffles decreased the L/D cycle period by 17.4% under 0.02 vvm air aeration rate. The ashing-light effect of the microalgae from dark area to light area was improved. Microalgal growth rate can be obviously improved with the enhanced ashing light effect. 27 Fig. 3b showed the vertical velocity and horizontal velocity of uid in a PBR with or without horizontal baffles, which was obtained from the simulation result. As the aeration rate increased from 0.02 vvm to 0.04 vvm, horizontal uid velocity increased from 1.0 cm s À1 to 1.3 cm s À1 with the horizontal baffles, and increased from 0.52 cm s À1 to 0.62 cm s À1 without the horizontal baffles [ Fig. 3a]. The velocity magnitudes in the PBRs with and without horizontal baffles were measured using a miniature ultrasonic Doppler velocimeter. When the aeration rate was 0.02 vvm, the test average velocity magnitudes were 0.95 cm s À1 with horizontal baffles and 0.48 cm s À1 without horizontal baffles, and the differences from the simulation results were 5% and 7.7%, respectively. This result demonstrated that the simulation result was acceptable.

Results and discussion
With the horizontal baffles, small scale vortex ow was developed in the PBR, around one horizontal tube baffle, or between two baffles. In the presence of horizontal baffles under air aeration rate of 0.02 vvm, the horizontal uid velocity signicantly increased by 1.8 times especially in the middle of the PBR. Thus, culture uid can be quickly moved from dark area to light area. Only a large vortex ow was generated within PBR without horizontal baffles. Then, uid ow direction mainly changed at the bottom and top of the PBR. 28 The major part of the microalgal culture uid cannot move quickly to the other side in the central section of the PBR. Several studies reported that ashing light effect of the microalgae from dark area to light area can enhance photosynthesis and improve the quality and quantity of microalgal biomass. For that reason, it is signicant to consider the integration of ashing light effect into microalgal cultivation systems. 17 The vertical uid velocity increased from 9.6 cm s À1 to 10.4 cm s À1 and from 10.2 cm s À1 to 13.9 cm s À1 as the gas aeration rate increased from 0.02 vvm to 0.04 vvm with and without the horizontal baffles. Vertical uid velocity decreased by 25% when the horizontal baffles were used under 0.04 vvm air aeration. Longer bubble residence time can be achieved with a slower vertical uid velocity. The CO 2 utilization efficiency was improved as more CO 2 was dissolved into the culture uid.
Fluid movement was obviously affected by horizontal baffles, especially in the upper section of the PBR [ Fig. 3b] as parts of air bubbles were blocked by the horizontal baffles. The vertical uid velocity was not increased further as aeration rate increased from 0.06 vvm to 0.1 vvm with the horizontal baffles. So the L/D cycle period was not decreased further. Vertical and horizontal uid velocities increased from 16.5 cm s À1 to 20.1 cm s À1 and from 0.7 cm s À1 to 0.9 cm s À1 , respectively, while gas aeration rate increased from 0.06 vvm to 0.1 vvm without horizontal baffles. A high intensity turbulent region was developed within the reactor. Hence, the L/D cycle period decreased from 9.6 s to 8.2 s while gas ow rate increased from 0.06 vvm to 0.1 vvm.
Light zone was only 2 cm depth from the light direction with a microalgal concentration of 0.85 g L À1 . So light time of algal cell was decreased with the increased vertical uid velocity This journal is © The Royal Society of Chemistry 2018 while gas aeration rate was increased from 0.02 vvm to 0.1 vvm. Thus, the light time fraction slightly decreased from $21% to $17% (Fig. 2). Flue-gas (especially from a coal-red power plant) aeration rate is restricted because it contains NO x and SO x , 29 high ashing light frequency can be obtained with horizontal baffles at a low gas aeration rate.

L/D cycle periods under different microalgal concentration
The L/D cycle period time was calculated based on different microalgal concentrations under 0.02 vvm gas aeration rate. The critical depth was decreased from 5 cm to 1 cm when the microalgal concentration was increased from 0.28 g L À1 to 1.7 g L À1 . L/D cycle period decreased from 15.6 s to 11.4 s and the light time fraction decreased from 39% to 17% when the microalgal concentration was increased from 0.28 g L À1 to 1.7 g L À1 in the presence of horizontal baffles (Fig. 4). The crosssectional ow area was periodically changed when horizontal baffles were used. Air was aerated close to right side of the wall (Fig. 1b). Small scale vortex ow was developed in the plate reactor, around one horizontal tube baffle, or between two of the horizontal baffles. Fluid uctuations increased when the critical depth line moved to the right side of the wall. Thus, the L/D cycle period decreased by 27% as the critical depth was decreased from 5 cm to 1 cm.
The L/D cycle period was $17 s and light time fraction decreased from 35% to 22% when the microalgal concentration was increased from 0.28 g L À1 to 1.7 g L À1 in the absence of horizontal baffles (Fig. 4). Flow direction of uid mostly changed at the bottom and top of the PBR. Most of the microalgae uid can not move quickly from dark area to light area in center section of the PBR. Movement of particles inside the PBR was relatively uniform; thus, the L/D cycle period time was not shortened with the increase of microalgal concentration.
The L/D cycle period, which was based on different microalgal concentrations, was the same at different uid depths. The dark area comprises roughly 90% of the cycle when the microalgal concentration was 1.1 g L À1 . Microalgal cells cannot timely move from one side to the other side in the PBR without horizontal baffles (Fig. 5a). Small light dark cycle period was good for cells growth. Most L/D cycle periods were longer than 5 s, and the probability of the L/D cycle period with 5-10 s was 27.9% (Fig. 5b). The uid ow direction was changed quickly, vortex ow elds were produced, and the vertical uid velocity increased from 0.52 cm s À1 to 1.0 cm s À1 when the horizontal baffles were used. Consequently, the cell was quickly moved from dark area to light area in the PBR (Fig. 5a). The probability of the L/D cycle period with 0-5 s increased from 1% to 5.9%, and the probability of the L/D cycle period with 5-10 s increased from 27.9% to 43.6% (Fig. 5b), indicating an increase of 56% when the horizontal baffles were used. The result showed that the cells has a bigger probability to gain a small light dark cycle period in PBR with horizontal baffles than that in PBR with baffles.

Solution mixing and mass transfer in the PBR with horizontal baffles
Effects of gas aeration rate on mass transfer coefficient and mixing time in PBR with or without the horizontal baffles were tested by experiment [ Fig. 6a]. When the gas aeration rate was increased from 0.02 vvm to 0.1 vvm, mass-transfer coefficient decreased from 48.1 s to 27.6 s without the horizontal baffles, and decreased from 39.6 s to 22.5 s with the horizontal baffles. When the horizontal baffles existed, mixing time decreased by 13% under 0.06 vvm gas aeration rate. The PBR mixing efficiency improved (bigger horizontal velocity) while horizontal baffles were used. Appropriate mixing condition allowed even nutrient distribution in the culture medium and accelerated microalgal growth. 30 When air aeration rate was increased from 0.02 vvm to 0.1 vvm, mass-transfer coefficient increased from 0.053 h À1 to 0.142 h À1 without the horizontal baffles, and increased from 0.081 h À1 to 0.208 h À1 with the horizontal baffles. The residence time of rising bubbles was higher in the culture solution because that solution vertical velocity decreased by 18% under 0.06 vvm gas aeration rate. Therefore, mass transfer coefficient increased by 31% under 0.06 vvm gas aeration rate, thus leading to higher mass transfer and accelerated CO 2 dissolution. 31,32

Increased microalgal biomass yield in the presence of horizontal baffles
Effects of horizontal baffles on microalgal growth rate and pH values versus time with 15% CO 2 under a aeration rate of 0.02 vvm were illustrated in Fig. 6b. The pH of SE culture decreased quickly from 6.8 to 5.8 during the rst 4 h because CO 2 quickly dissolved into the medium. The culture pH increased slowly aer 4 h, this increasing tendency of pH owing to the CO 2 uptake by microalgae. The uid ow direction was changed quickly, vortex ow elds were produced, and the vertical uid velocity increased from 0.52 cm s À1 to 1.0 cm s À1 when the horizontal baffles were used.
Consequently, the cell was quickly moved from dark area to light area in the PBR (Fig. 5a). The probability of the L/D cycle period with 5-10 s increased by 56% from 27.9% to 43.6% when the horizontal baffles were used. So the L/D cycle period decreased by 17.4% under an air aeration rate of 0.02 vvm when horizontal baffles were added. It is proposed that mixing has two separate but synergistic effects, for example, it not only moves the microalgal cells through a L/D cycle, but also decreases the boundary layer, which increases the rate of exchange through the cell wall of nutrients and metabolites. Thus, more nutrients can be uptake and light can be utilized more efficiently, so the biomass yield is increased. 33 The biomass yield in the h day increased by roughly 51% in comparison with the condition without baffles at a same aeration rate.

Conclusion
Hydrodynamic and ashing effect of a panel bioreactor with horizontal baffles were investigated by CFD simulation. The L/D cycle period decreased by 17.5% and the probability of the L/D cycle period within 5-10 s increased by 56% when horizontal baffles were used under 0.02 vvm gas aeration rate. Experiments conrmed the enhanced ow eld in the bioreactor. Mixing time decreased by 13% and mass-transfer coefficient increased by 31% under 0.06 vvm gas aeration rate. The microalgal biomass yield increased by 51% with the same light intensity. The optimized width of the PBR under different light intensities can be further investigated to increase the microalgal biomass yield per unit area.

Conflicts of interest
There are no conicts to declare.