Chemical Engineering Journal 222 (2013) 8593

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal

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

A combined respirometertitrimeter for the determination of microalgae kinetics: Experimental data collection and modelling
Bjorge Decostere a,b, Natascha Janssens a, Andres Alvarado a,d, Thomas Maere a, Peter Goethals b,c, Stijn W.H. Van Hulle a,b, Ingmar Nopens a,
a

BIOMATH, Department of Mathematical Modelling, Statistics and Bioinformatics, Ghent University, Coupure Links 653, B-9000 Gent, Belgium EnBiChem Research Group, University College West Flanders, Ghent University Association, Graaf Karel de Goedelaan 5, 8500 Kortrijk, Belgium Aquatic Ecology Research Unit, Ghent University, Plateaustraat 22, B-9000 Gent, Belgium d DIUC, Universidad de Cuenca, Av. 12 de Abril s/n Cuenca, Ecuador, Belgium
b c

h i g h l i g h t s
" A microalgae respirometertitrimeter and protocol was developed. " Signicant stripping of CO2 takes place even without aeration. " Recovery of HCO3 was higher than 90% both from DO and proton production data. " Different respirograms were successfully modelled and kinetics determined.

a r t i c l e

i n f o

a b s t r a c t
The potential of microalgae for wastewater treatment has recently led to signicant surge in research towards economically more viable and technologically optimised systems. In this context, mathematical modelling has not been used to its full capacity. In this work, a combined respirometrictitrimetric unit for determination of microalgae kinetics and an experimental protocol is proposed. It was found that overall oxygen production was lower than stoichiometrically expected, which could be attributed to CO2-transfer to the gas phase. A basic model for microalgae growth on inorganic carbon and oxygen production is proposed and was successfully calibrated using several respirometrictitrimetric datasets. The model structure was based on the activated sludge models (ASMs) and can now be extended with impact of additional degrees of freedom. To our knowledge it is the rst time that a respirometrictitrimetric approach is applied to microalgae and used to calibrate an ASM based kinetic model. Crown Copyright 2013 Published by Elsevier B.V. All rights reserved.

Article history: Received 29 June 2012 Received in revised form 19 January 2013 Accepted 31 January 2013 Available online 9 February 2013 Keywords: Microalgae Kinetics Respirometry Titrimetry Modelling

1. Introduction Microalgae are present in a variety of environmental systems. Sometimes they are undesired, e.g. in river systems where they can cause eutrophication when becoming abundant. In treatment technologies like waste stabilisation ponds on the other hand, they are wanted and cultivated to serve as oxygen producers and for additional nutrient removal. Recently, also a lot of attention has gone to the application for carbon sequestration to reduce greenhouse gas emissions (e.g. from industrial exhaust gases) [1] and as a renewable source for products currently made from crude oil [2]. The efciency of these algae processes is, however, rather low (especially in view of upscaling) and optimisation is required to make them cost-effective.
Corresponding author. Tel.: +32 92645939; fax: +32 92646220.
E-mail address: Ingmar.Nopens@UGent.be (I. Nopens).

When biotechnological processes are applied for environmental technologies (e.g. activated sludge, biolters, fermentors, etc.), biomass characterisation in terms of kinetics is of crucial importance for system design and optimisation. A method that is often used to accomplish this in the context of activated sludge waste water treatment is respirometry [3]. This method measures the consumption rate of O2 and translates this into an oxygen uptake rate (OUR) which is then coupled to the kinetics of the organisms. The metabolism of algal biomass is somewhat different, i.e., algae produce oxygen through photosynthesis, hereby using an inorganic carbon source (CO2 or HCO 3 ) and the energy of light. With abundant light, a respirometric batch setup will then result in a negative oxygen uptake rate, or in other words an oxygen production rate (OPR). In the absence of light, endogeneous algae respiration will lead to a positive oxygen uptake rate (OUR). Given the similarities, the determination of microalgae kinetics from OPR

1385-8947/$ - see front matter Crown Copyright 2013 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2013.01.103

86

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

Nomenclature Symbols bmax DCO2 DO2 k1 k2 KCO2 KHCO3 KLa XALG Y1 Y2 Y3 algae biomass concentration (mg DW l1) 1 algal growth yield on HCO HCO 3 (g DW g 3) algal growth yield on CO2 (g DW g1 CO2) oxygen production yield (g O2 g1 DW)

l lmax
PH PO2 PO2,max SCO2 Ssat CO2 SO2 Ssat O2

maximum algal decay rate (d1) diffusion coefcient of CO2 in water (m3 d1) diffusion coefcient of O2 in water (m3 d1) reaction rate for hydratation of CO2 (d1) 1 reaction rate for dissociation of HCO 3 (d ) half saturation constant for growth on CO2 (g CO2 m3) half saturation constant for growth on HCO 3 3 (g HCO 3 m ) oxygen transfer rate (d1) specic growth rate (d1) maximum specic growth rate (d1) specic proton addition rate (mg H+ mg1 DW d1) photosynthetic activity (mg O2 mg1 DW) maximum photosynthetic activity (mg O2 mg1 DW) dissolved carbon dioxide concentration (mg CO2 l1) dissolved carbon dioxide concentration at saturation (mg CO2 l1) dissolved oxygen concentration (mg O2 l1) dissolved oxygen concentration at saturation (mg O2 l1)

Abbreviations ASM activated sludge models BSAR background signal addition rate (mg H+ l1 d1) CO2TR carbon dioxide transfer rate (mg CO2 l1 d1) DO dissolved oxygen (mg O2 l1) DW dry weight (mg DW l1) PAR proton addition rate (mg H+ l1 d1) OPR oxygen production rate (mg O2 l1 d1) OTR oxygen transfer rate (mg O2 l1 d1) OUR oxygen uptake rate (mg O2 l1 d1) PAR photosynthetic active radiation TIC total inorganic carbon (mg l1) TPAR total proton addition rate (mg H+ d1)

curves, in analogy with bacterial respirometry experiments, is an elegant method to measure microalgae kinetics. In literature, respirometry is in some cases accompanied by titrimetry for activated sludge, providing an independent measure of biological activity, which is helpful when calibrating models [4]. This titrimetric approach exploits a pH-effect that is governed by the organisms metabolism [5]. Since microalgae use carbon in its inorganic form, the carbonaceous equilibrium and, hence, the pH will be inuenced. Whether the rate of the latter is directly related to microalgae kinetics will be tested. Although estimation of microalgal photosynthetic activity by measuring O2 evolution are ambiguously described in literature [6] the methods described are very diverse [7,8] and the data collected from different works present important discrepancies, and in some cases calculations are lacking. By combining titrimetry with respirometry it is our goal to dene and take into account the different aspects that occur during the microalgal photosynthetic activity. As such the use of respirometry and titrimetry for determination of microalgae kinetics is investigated. A respirometer setup including a titrimetric approach is proposed along with a protocol to successfully perform respirometrictitrimetric experiments that provide a maximum of information. Furthermore, a kinetic model taking into account inorganic carbon limitation is proposed and calibrated. This research is an initial required step to further promote models for microalgae system optimisation.

pH control algorithm implemented in LabView (http://www. ni.com). At the same time this provided carbon source to the system to obtain high growth yields. Mixing through air sparging prevented the microalgae to settle or attach to the reactor wall. The separate respirometric tests were seeded with microalgae from the cultivation reactor after centrifugation at 534g for 20 min (Sorvall RC-5B, Dupont instruments). Subsequently, the concentrated algae were diluted in 1000 ml of freshly prepared growth medium. To prevent reaction vessel overow, only 700 ml of the algal suspension was transferred to the 1 l respirometer reactor vessel. The reactor which was operated batch wise and was spiked with a 50 ml NaHCO3 ultra pure water solution (VWR International, Belgium). The total suspended solids concentration of the algal solutions was determined according to standard methods [11]. The concentrations of total nitrogen and total phosphorus were determined before and after each experiment using Hach Lange test kits LCK340 and LCK 350 (Hach-Lange, Belgium), respectively. 2.2. The algal respirometer A schematic of the algal respirometer is shown in Fig. 1. It is inspired by a respirometrictitrimetric setup used for activated sludge [5]. The 1 l reactor vessel was heat-jacketed to allow temperature control (Alpha R8, http://www.lauda.de) enabling the exploration of system behaviour at different temperatures (default at 20 C). The light cage enclosing the reactor entirely consisted of eight uorescent lamps (Grolux T8 18W, Sylvana). Light intensity was measured using a photosynthetic active radiation (PAR) light sensor (PAR mini, PP-systems). Dissolved oxygen (DO) and pH were measured online with an oxygen (Inpro 6100, Mettler Toledo) and pH electrode (Inpro 4250, Mettler Toledo) and the data logged using a PCI-MIO16XE-50 data acquisition card using LabView (http://www.ni. com). The DO sensor delay (determined to be 0.53 s) was taken into account according to Vanrolleghem and Spanjers [3]. The observed dissolved oxygen dynamics can be represented by:

2. Materials and methods 2.1. Cultivation of microalgae for kinetic experiments The strain of microalgae used for the respirometric experiments was Chlorella vulgaris. This strain was cultured in a 10 l breeding reactor. The growth medium used was a variant of the BG-11 medium [9]. In order to prevent phosphorus limitation, the medium was slightly modied, i.e. the phosphorus concentration was increased for the N:P ratio to comply to the Redeld ratio, dened as 106C:16 N:1P [10]. The pH of the culture was controlled by adding pulses of CO2 into the culture and was carried out by a

dSO OPR OTR dt

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

87

Fig. 1. A schematic illustration of the combined algal respirometrictitrimetric unit.

with SO the dissolved oxygen concentration (mg l1) and where the oxygen production rate (OPR) represents the rate of oxygen produced by the microalgae through consumption of an inorganic carbon source and the oxygen transfer rate (OTR) represents the diffusion of oxygen to the atmosphere. The pH was controlled online at a user dened setpoint using a banded (0.05 pH) onoff feedback control algorithm implemented in LabView by dosing HCl or NaOH through two 3-way pinch valves (Z530A, SIRAI, Italy). The rate and amount of 0.5 M HCl and 0.5 M NaOH dosed into the reactor vessel constitutes the titrimetric data. 2.3. Mass transfer of oxygen and carbon dioxide to the atmosphere The mass transfer of oxygen to the atmosphere (or stripping) can be quantied using Eq. (2). It consists of the oxygen transfer coefcient from liquid to gas phase (KLa) and the difference between the dissolved oxygen concentration in the reaction vessel (usually higher than the saturated value due to the microalgal oxygen production) and the dissolved saturation concentration at a given temperature. Consequently, the oxygen transfer rate (OTR) can be calculated as:

lated from the KLa for O2 by multiplying with a reduction factor [13,14]. The CO2 transfer rate can be calculated as follows:

CO2TR K L a

s DCO2 sat S SCO2 DO2 CO2

with CO2TR the carbon dioxide transfer rate (mg l1 d1), DO2 the diffusion coefcient of oxygen in water and DO2 the diffusion coefcient of carbon dioxide in water, respectively 1.65 104 m3 d1 and 1.73 104 m3 d1 [15]. Ssat CO2 is the saturation concentration (mg l1) and SCO2 the concentration of carbon dioxide in the solution (mg l1). The saturation concentration is calculated from the atmospheric concentration (assumed to be constant and equal to 390 ppmv or 8.86 lmol l1) and the Henry constant (assumed to 1 be constant and equal to 0.83 Mgas M liquid [16]). As such the satura1 tion concentration equals 7.37 lmol l or 0.32 mg l1. 2.4. Processing of the titrimetric data As can be deducted from Eqs. (5) and (6) [17], consumption of dissolved CO2 or HCO 3 by microalgae results in a pH increase in the system as protons are consumed in the process. Therefore, acid will need to be dosed to maintain a constant pH.
2 106 CO2 122 H2 O 16 NO 3 18 H HPO4

OTR K L aSO Ssat O

1

with KLa (d ) the oxygen transfer coefcient (from liquid to gas phase) and Ssat O the dissolved oxygen concentration at saturation (mg l1). The dissolved oxygen concentration at saturation is temperature dependent and can be calculated according to ASCE [12], with T in C:
3 2 5 3 Ssat O 14:65 0:41T 7:99 10 T 7:78 10 T

! C106 H263 O110 N16 P 138 O2

2 106 HCO 3 16 H2 O 16 NO3 124 H HPO4

! C106 H263 O110 N16 P 138 O2

The KLa can be directly estimated from the dissolved oxygen prole. Indeed, when all inorganic carbon source is depleted (i.e. OPR = 0) the dissolved oxygen decrease can solely be attributed to transfer to the atmosphere. This point can easily be detected in the prole (see further). Next to O2, also CO2 is stripped from the liquid phase. The rate at which this happens, depends on the saturated CO2 concentration and the mass transfer coefcient (KLa) for CO2, which was calcu-

To interpret the titrimetric data, several processes impacting pH should be accounted for: CO2 consumption for algal growth, atmospheric CO2 diffusion and algal respiration. Consumption of CO2 by algae results in a decreasing CO2 partial pressure. As such, CO2 diffusion from the atmosphere to the reactor liquid will occur, causing the production of protons in the system (Eq. (7)) [18,19]. However, the CO2 that diffuses into the liquid system will also be consumed by the algae. Moreover, the consumption of dissolved oxygen due to respiration and simultaneous CO2 production also shifts the H2 CO3 =HCO 3 equilibrium to the right (Eq. (8)).

88

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

CO2 H2 O $ H2 CO3 H and K 1 HCO 3

H 10pK a1 CO2

2 HCO 3 $ CO3 H and K 2

CO2 3 H 10pK a2 HCO 3

Fig. 2. Visual representation of the calculation of the amount of protons added due to the consumption of carbon source during a microalgae respirometric experiment.

The abovementioned effects result in a titrimetric background signal, i.e. the background signal addition rate (BSAR) [20]. The amount of protons consumed by this BSAR needs to be corrected for when calculating the net proton addition rate due to CO2 consumption by microalgae (i.e., proton addition rate or HAR). This can be clearly observed in Fig. 2. The rst part of the curve (before the knee) corresponds to the period where CO2 is consumed by microalgae, whereas the second part is only due to the BSAR. The slope obtained from the rst part of the curve represents the total rate of proton addition (PTAR), including the BSAR. The latter can be determined from the second part of the curve. Subtracting this from the PTAR yields the HAR.

1 Fig. 3. Example of a respirometric experiment (top) and the resulting OPR and OTR (bottom) in case of 100 mg l1 HCO and 3 and an algae concentration of 478 mg DW l with a light intensity of 65 lE m2 s1. pH is controlled at 7.5. Temperature is set at 15 C.

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

89

2.5. Modelling software To model the respirometrictitrimetric data the modelling and simulation platform WEST (mikebydhi.com, Denmark) was used. Parameter estimations were performed using the Simplexalgorithm [21].

3. Results and discussion 3.1. Data collection and derived information Fig. 3 (top) shows a typical result of a respirometrictitrimetric experiment. The results are repeatable and similar proles were obtained when certain experimental conditions were modied (e.g. different spiked quantities of bicarbonate, different algal biomass concentration). The goal of this paper is, however, to discuss the specicities of these experiments rather than investigating the impacts of different experimental conditions, which is regarded as future work. The dissolved oxygen prole in Fig. 3 (top) is the result of the balance between (1) oxygen produced by the algae during the consumption of the pulse of HCO 3 which was added to the system and (2) oxygen removed from the system through transport to the

atmosphere (Eq. (1)). Upon addition of inorganic carbon source (in this experiment 100 mg l1 HCO 3 ), the initial dynamic equilibrium is disturbed as more dissolved oxygen is produced than removed. This results in a rapid increase of the dissolved oxygen concentration. However, this increase is limited by the maximum growth rate of the algae (i.e. metabolic limitation) which leads to a new steady state (plateau in time interval 0.10.15 d). At some point the inorganic carbon source is depleted (approx. 0.16 d) and limits the DO production. This leads to a decrease in DO, eventually returning to the state the system was in prior to the addition of inorganic carbon source. Along with the consumption of bicarbonate, protons are removed from the system. Due to the fact that pH is controlled at a xed setpoint (here 7.5), proton addition is needed. According to the dashed line in Fig. 3 (top), this happens at a constant rate (PTAR = 11.848 mg l1 d1) during the time interval between spiking and depletion of bicarbonate. After depletion, the proton addition reduces to the BSAR level, in this case 0.141 mg l1 d1, or about 1% of the PTAR. This results in a HAR 1 1 for the consumption of the pulse of HCO d . 3 of 11.707 mg l The specic HAR (expressed per unit biomass) at the beginning of the experiment (PH) is determined to be 0.024 mg H+ mg1 DW. The total amount of protons added for the consumption of the added pulse of HCO 3 can be determined by integrating the titrimetric prole yielding 1.859 g H+ l1. Eq. (6) allows an exact

1 Fig. 4. Visualisation of the CO2 transfer rate (top) and HCO HCO3 and 478 mg DW l1 at pH 7.5 and 15 C . 3 and CO2 concentration (bottom) with 100 mg l

90 Table 1 Gujer matrix for the algal kinetic model. Process 1. Growth on HCO 3 XALG (g DW m3) 1 1 1

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

3 S HCO3 (g HCO3 m )

SCO2 (g CO2 m3)

3 SCO3 (g CO2 3 m )

SO2 (g O2 m3) Y3

Process rate

1/Y1 1/Y2

lmax l

 

SHCO K CO2 SHCO

3 3

 

K CO2 K CO2 SCO2

X ALG

2. Growth on CO2 3. Microalgae decay 4. Transfer rate O2 5. Transfer rate CO2 6. CO2 hydration 7. HCO 3 dissociation

Y3 1

SCO2 max K CO SCO 2 2

X ALG

bmax X ALG 1 61 61 44 60 K L aSsat O2 SO2 q sat 2 K L a DCO DO2 SCO2 SCO2   SHCO SCO2 k1 44 10pH 61:K31   SHCO SCO3 k2 613 10pH 60:K 2

Table 2 Parameters used in the model with their corresponding value range obtained from literature and the chosen range for the optimisation of the model. Parameter Literature range 0.111 0.0030.1 0.549 0.761 1.24 0.0616.1 0.0444.4 0.2414.4 2221105 1041012 6.36 10.33 Assigned value 0.01 0.549 0.761 1.24 3 0.2 10000 10000 6.36 10.33 Assigned range 0.111 Unit d 1 d 1 g DW g1 HCO 3 g DW g1 CO2 1 g O2 g DW 3 g HCO 3 m g CO2 m3 d 1 d 1 d 1

lmax
bmax Y1 Y2 Y3 K HCO3 KCO2 KLa k1 k2 pK1 pK2

527

calculation of the stoichiometrically required amount of acid (given the equations hold): a concentration of 100 mg l1 HCO3 yields 1.918 mg H+ l1 that needs to be added to maintain a xed pH. Hence, the titrimetric method had in this case a recovery rate of 97%, proving to be accurate. The calculated OPR and OTR from the dissolved oxygen prole are shown in Fig. 3 (bottom). The maximum OPR of the algae after spiking with the inorganic carbon has an average of about 250 mg O2 l1 d1. Hence, the maximum rate of oxygen production per unit of DW of algae (PO2,max) equals 0.523 mg O2 mg DW1 d1. The total amount of oxygen produced is determined by integrating the OPR curve and equals 39.94 mg O2. This is, according to Eq. (6), equivalent to 58.57 mg l1 HCO 3 , i.e. signicantly lower than the theoretical amount that can be produced from the amount of 1 HCO ) added to the system, being 68.3 mg l1 O2. 3 (100 mg l From these results the recovery rate only amounts to about 58.48%. This low recovery can be explained as follows. First, respiration is not taken into account when interpreting the data. Indeed, micro-algae use oxygen for their maintenance metabolism, thus lowering the total amount of oxygen produced. This type of respiration is called dark respiration [15]. In addition, photorespiration can occur at high oxygen to carbon dioxide ratio in the solution, and as such inhibition of the photosynthesis occurs [22]. As Birmingham et al. [23] stressed, that photorespiration is only inhibited at the CO2 saturation level in the water and this was probably not the case during the experiments, it should be taken in account for interpreting the data. The calculation of the theoretical amount of oxygen produced from Eq. (4) is therefore most probably an overestimation of the net amount of oxygen produced as the amount lost by (photo)respiration need to be subtracted. Dark respiration, photorespiration and light limitation are therefore processes that need to be considered when interpreting the

calculated data. However, this level of detail was not pursued here and, hence, further research is recommended to quantify the contributing effect of these processes on the oxygen production and carbon consumption. A possible approach to quantify dark respiration would be to perform a respirometric test in the absence of light. Photorespiration could be measured by 14CO2 and CO2 depletion after addition of NaH14CO2 to the algae solution [23]. At this stage of the research these experiments were not yet performed. Another possible explanation is that not the entire amount of inorganic carbon source is available for the algae to be consumed and as such for oxygen production. As mentioned before, next to oxygen stripping to the atmosphere, also stripping of carbon dioxide occurs. An initial concentration of 100 mg l1 HCO3 or 0.001639 mol l1 total inorganic carbon (TIC) corresponds, based on Eq. (7), to 0.000149 mol l1 H2CO3, 0.001489 mol l1 HCO 3 and 9.39 107 mol l1 CO2 at pH 7.5. Consumption of 1 mol 3 TIC leads to addition of 1 mol H+. As such the concentration of CO2 in the respirometer can be calculated based on the addition of protons. Further, according to Eq. (4), the CO2 transfer rate can be calculated from this CO2 concentration and is depicted in Fig. 4, proving to be signicant. Integrating this curve results in 22.82 mg l1 HCO 3 TIC transfer to the atmosphere. Accounting for this loss of TIC, a total recovery of 92.98% is obtained. As such it can be concluded that there is a signicant amount of TIC that is not available for the microalgae due to stripping resulting in rather low recovery when expressed in the amount of oxygen produced and that CO2 transfer should be incorporated in both data interpretation and modelling. 3.2. Modelling of the respirometric data To describe the respiration behaviour of microalgae, a rst basic kinetic model was set up based on the experimental observations. It contains ve state variables: microalgae biomass concentration, concentrations of the different carbon species in the aqueous sys2 tem (HCO 3 , CO2 and CO3 ) and dissolved oxygen concentration. The model was inspired by Gehring et al. [24] and Alex et al. [13], which are similar to the River Water Quality Model by Reichert et al. [25]. The next sections describe the model in more detail through the seven processes that it accounts for. The nal model presentation is for the rst time based on the activated sludge type models (ASM) [26], allowing (1) straightforward interchange with existing waste water treatment models and (2) extension of the presented model. Further the kinetic model presents a trade-off between detailed metabolic models (e.g. Kliphuis et al. [27]) and oversimplied kinetic models (e.g. Nedbal et al. [28]). 3.2.1. Algal growth and decay kinetics The basic principle of modelling the microalgae biomass concentration (XALG) is that the maximum growth rate of algae is

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

91

Fig. 5. Comparison of experimental data and model predictions for respirometric data. (a) 75 mg NaHCO3 added to 267 mg l1 DW, (b) 75 mg NaHCO3 added to 252 mg l1 DW and (c) 150 mg NaHCO3 added to 459 mg l1 DW.

limited by different factors. The specic growth rate (l) is generally modelled by multiplying the maximum growth rate (lmax) with these limiting factors such as substrate limitation or light

availability. However, for now, these limiting factors are not taken into account and only a maximum growth rate is used. The maximum algal decay rate in Eq. (9) is represented by bmax:

92

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

l lmax bmax

Since the microalgae used in the respirometric tests are suspended in the growth medium with a sufcient amount of macro-and micronutrients, the nutrients are assumed not to be limiting for their growth in the current experimental setting. Because temperature and light intensity is kept constant in the different experiments studied here, no factor for the temperature dependency not light intensity dependency for the growth rate of microalgae is included in the model at this stage. The inorganic carbon source (C-substrate), however, is consumed in the respirometer and it becomes limiting for the algae growth. The availability of CO2 and HCO 3 is therefore modelled by a Monod function. CO2 is able to cross cell membranes and enters directly into the cell by diffusion. Contrarily, the uptake of HCO 3 requires a transporter system or its prior conversion to CO2 [29]. Therefore, CO2 will be preferentially taken up by the microalgae. Given this, an inhibition term in the HCO 3 uptake has been incorporated in the model (Table 1). The growth rate of the microalgae can now be determined as the specic growth rate multiplied by the biomass concentration (q = lXALG). It should be stressed that at this stage of the investigation no microalgae respiration is included in the model, due the fact that not enough information was present in the experimental data to estimate the contribution of this process (only 10% oxygen was not accounted for). 3.2.2. Inorganic carbon species As CO2 and HCO 3 can both be used by algae as carbon source for growth [18,19,29,30], these inorganic carbon sources, together
with CO2 3 are incorporated in the model (note that the model will be used in future for cases where different inorganic carbon sources are added). The concentration of these carbon species are related to each other by the governing chemical equilibra (Eqs. 2 (7) and (8), [17] HCO 3 is dosed to the system and CO2 and CO3 are formed in the aqueous environment by dehydration and dissociation of HCO 3 , respectively. To calculate the rate at which chemical conversion between the three carbon species takes place, the concentrations of the carbon species need to be converted into pK a1 H mol l1, as the dissociation constants (K 1 HCO 3 CO2 10 pK a2 H and K 2 CO2 ) are expressed in this unit and 3 HCO 10
3

proton addition rate can be modelled from the consumption of HCO 3 and CO2. Moreover the BSAR due to the chemical equilibria of the different carbon species in the water and the diffusion of CO2 between the atmosphere and the liquid phase [30] inuences the proton balance. 3.2.4. Summary of the model An overview of the modelled processes and the reaction rates is given in the Gujer matrix of the model presented in Table 1. Each modelled process is presented with their corresponding process rates and stoichiometric coefcients. 3.2.5. Parameter values Default values for the parameters are summarized in Table 2 and were obtained from literature [13,31,32,24,33,34,25,15]. In case parameters were used in an estimation (see further) the range that was used is provided. The dissociation constants (pK) for the hydration of CO2 and the dissociation of HCO 3 were taken as 6.36 and 10.33, respectively [17]. The rate constants for these reactions were chosen to be 10,000 for k1 and k2 [24], respectively indicating very fast reactions. The values for the yields for the production of biomass from HCO 3 or dissolved CO2 and the yield of oxygen production were determined stoichiometrically from Eqs. (5) and (6). The maximum decay rate bmax was set to a value of 0.01 d1 because tests were short and the decay rate was considered not to play a signicant role. At this stage of the research, and according to good modelling practice, only two parameters were estimated, KLa and lmax. 3.3. Model calibration The model was optimised by tting its output to three different data sets of the respirometric experiments with 75 mg NaHCO3 (or 1 72.6 mg l1 HCO DW of algae (Fig. 5a), 3 ) added to 267 mg l 75 mg NaHCO3 added to 252 mg l1 (Fig. 5b) and 150 mg NaHCO3 1 (or 145.2 mg l1 HCO DW (Fig. 5c). The 3 ) added to 459 mg l model is able to describe the DO-prole acceptably well. The values of the optimised parameters are presented in Table 3. As can been seen the maximum growth rates of the three different experiments are very similar and are comparable with values found in literature. For these three experiments the KLa spans the range from 15.83 to 26.79 d1. This might be explained by the fact that non-controllable conditions, such as the atmospheric pressure and the salinity of the algal solution, were different during the experiments. Further research should be conducted to investigate the correlation between the two tted parameters. The consumption rate of inorganic carbon is also plotted and is very similar in the tested cases. Good model performance, i.e. t to the experimental data, and similar values for the calibrated parameters for the different datasets were obtained. However, a calibrated model is only able to predict respirometer behaviour provided that the experimental conditions are similar. Hence, model predictions should be interpreted with care when predicting effects outside the range of

the concentrations of the different species in the model are expressed in g m3. From the chemical equilibria the equilibrium concentrations of the three species are calculated and subtracted from the actual concentration of the inorganic carbon source. Consequently, the value that is obtained is proportional to a driving force determined by the difference between the equilibrium and the actual concentration. This value is then multiplied by a rate constant k1 or k2 to obtain a process rate to express the change
2 in the concentrations of HCO 3 , CO2 and CO3 because the system strives for a chemical equilibrium [15].

3.2.3. Oxygen production and proton consumption The dynamic dissolved oxygen concentration is determined from a balance between the oxygen production rate (OPR) and oxygen transfer rate (OTR) as was discussed before (Eq. (1)). Stoichiometrically, 1.24 g of oxygen is produced for the production of 1 g of biomass (Eqs. (5) and (6)). This value represents the oxygen production yield (Y3) and is used to calculate the OPR from the biomass concentration present in the system. The rate of proton addition is determined by the removal rate of carbon source spiked to the algal respirometer. As can be deducted from Eqs. (5) and (6), consumption of one 1 g of HCO 3 leads to a removal of 19.18 103 g of protons, whereas consumption of 3 one gram of HCO g of protons. Hence, the 3 removes 3.86 10

Table 3 Values of the optimised parameters of the model tted to the measured data presented in Fig. 5. Biomass (mg l1) 458 252 235 Bicarbonate pulse (mg l1) 150 75 75 KL a (d1) 26.79 19.44 15.84

lmax
(d1) 0.52 0.48 0.52

Cost 1000 1731 5787

B. Decostere et al. / Chemical Engineering Journal 222 (2013) 8593

93

values for which it was validated (e.g. larger spiked amounts of carbon source) or for degrees of freedom that were not yet validated (e.g. light intensity, temperature). However, the model results are promising and form a solid base for future research in exploring microalgae system behaviour and its optimisation. 4. Conclusion The main ndings can be summarized as:  A respirometrictitrimetric setup and protocol for microalgae was developed that allows determination of microalgae kinetic parameters, inspired by classic respirometry in the activated sludge process.  The recovery of the respirometric data appeared to be lower than stoichiometrically expected. This is shown to be attributed to CO2 stripping during the experiment.  A basic model was constructed to describe the observed behaviour of respirograms and titrimetric results, yielding a good prediction of the observed system behaviour. However, the model needs to be extended to be validated for wider experimental ranges.

Acknowledgements This work has been supported by a grant from the VLIR-UOS (Flemish Inter-University Council University Development Cooperation) Program and by ETAPA (Empresa Pblica Municipal de Telecomunicaciones, Agua Potable, Alcantarillado y Saneamiento de Cuenca, Ecuador). References
[1] S. Van den Hende, H. Vervaeren, S. Desmet, N. Boon, Bioocculation of microalgae and bacteria combined with ue gas to improve sewage treatment, New Biotechnology 29 (1) (2011) 2331. [2] Ganapathy Sivakumar, Jianfeng Xu, Robert W. Thompson, Ying Yang, Paula Randol-Smith, Pamela J. Weathers, Integrated green algal technology for bioremediation and biofuel, Bioresource Technology 107 (2012) 19. [3] P.A. Vanrolleghem, H. Spanjers, A hybrid respirometric method for more reliable assessment of activated sludge model parameter, Water Science and Technology 37 (1998) 237246. [4] B. Petersen, K. Gernaey, P.A. Vanrolleghem, Practical identiability of model parameters by combined respirometrictitrimetric measurements, Water Science and Technology 43 (7) (2001) 347356. [5] K. Gernaey, B. Petersen, D. Dochain, P.A. Vanrolleghem, Modelling aerobic carbon source degradation processes using titrimetric data and combined respirometrictitrimetric data: structural and practical identiability, Biotechnology and Bioengineering 79 (2002) 754769. [6] T.B. Hancke, K. Hancke, G. Johnsen, E. Sakshaug, Rate of O2 production derived from pulse-amplitude-modulated uorescence. Testing three biooptical approaches against measured O2 production rate, Journal of Phycology 44 (2008) 803813. [7] Z. Dubinsky, P.G. Falkowski, A.F. Post, U.M. Van Hes, A system for measuring phytoplankton photosynthesis in dened light eld with an oxygen electrode, Journal for Plankton Research 9 (1987) 607612. [8] M. Janssen, M. De Winter, J. Tramper, L.R. Mur, J. Snel, R.H. Wijffels, Efciency of light utilization of Chlamydomonas reinhardtii under medium-duration light/ dark cycles, Journal of Biotechnology 78 (2000) 123137.

[9] R.Y. Stanier, R. Kunisawa, M. Mandel, G. CohenBazire, Purication and properties of unicellular blue green algae (Order Chroococcales), Bacteriological Reviews 35 (1971) 171205. [10] J.U. Grobbelaar, Algal nutrition, in: A. Richmond (Ed.), Handbook of Microalgal Culture: Cornwall, Blackwell Publishing, 2004, pp. 97115 (Chapter 6). [11] APHA, Standard Methods for the Examination of Water and Wastewater, 18th ed., American Public Health Association, Inc. (APHA), New York, USA, 2005. [12] American Society of Civil Engineers (ASCE), Standard Guidelines for In-Process Oxygen Transfer Testing, New York, USA, 1996. [13] J. Alex, H. Pastagiya, N.C. Holm, First results of the development of a combined high rate biomass-algal model for wastewater treatment applications. In: Proceedings of 2nd IWA/WEF Wastewater Treatment Modelling Seminar, Mont-Sainte-Anne, Canada, March 2830, 2010. [14] G. Sin, Systematic Calibration of Activated Sludge Models, PhD Thesis, Ghent University, Belgium, 2004. [15] G. Wolf, C. Picioreanu, M.C.M. van Loosdrecht, Kinetic modeling of phototrophic biolms: the PHOBIA model, Biotechnology and Bioengineering 97 (2007) 10641079. [16] R. Sander, Compilation of Henrys Law Constants for Inorganic and Organic Species of Potential Importance in Environmental Chemistry (Version 3), 1999. <http://www.henrys-law.org>. [17] W. Stumm, J.J Morgan, Aquatic Chemistry, John Wiley & Sons, Inc., New York, 1996. [18] L. Binaghi, A. Del Borghi, A. Lodi, A. Converti, M. Del Borghi, Batch and fedbatch uptake of carbon dioxide by Spirulina platensis, Process Biochemistry 38 (2003) 13411346. [19] J.C. Goldman, Inorganic carbon availability and the growth of large marine diatoms, Marine Ecology-Progress Series 180 (1999) 8191. [20] G. Sin, R. Govoreanu, N. Boon, G. Schelstraete, P.A. Vanrolleghem, Evaluation of the impacts of model-based operation of SBRs on activated sludge microbial community, Water Science and Technology 54 (1) (2006) 157166. [21] J. Nelder, R. Mead, A simplex method for function minimization, Computer Journal 7 (1965) 308313. [22] Nigel D.H. Lloyd, David T. Canvin, David A. Culver, Photosynthesis and photorespiration in algae, Plant physiology 59 (1977) 936940. [23] B.C. Birmingham, J.R. Colman, Brian Colman, Measurement of photorespiration in algae, Plant Physiology 69 (1981) 259262. [24] T. Gehring, J.D. Silva, O. Kehl, A.B. Castilhos, R.H.R. Costa, F. Uhlenhut, J. Alex, H. Horn, M. Wichern, Modelling waste stabilisation ponds with an extended version of ASM3, Water Science and Technology 61 (2010) 713720. [25] P. Reichert, D. Borchardt, M. Henze, W. Rauch, P. Shanahan, L. Somlyody, P. Vanrolleghem, River water quality model no. 1 (RWQM1): II. Biochemical process equations, Water Science and Technology 43 (2001) 1130. [26] M. Henze, W. Gujer, T. Matsuo, M. Van Loosdrecht, Activated Sludge Models ASM1, ASM2, ASM2d and ASM3, Scientic and Technical Reports, IWA Publishing, London, UK, 2000. [27] A. Kliphuis, D. Martens, M. Janssen, R.H. Wijffels, Effect of O2:CO2 ratio on the primary metabolism of chlamydomonas reinhardtii, Biotechnology and Bioengineering 108 (10) (2010) 23902402. [28] L. Nedbal, J. Cerveny, N. Keren, A. Kaplan, Experimental validation of a nonequilibrium model of CO2 uxes between gas, liquid medium and algae in a at-panel photobioreactor, Journal of industrial Microbiology and Biotechnology 37 (2010) 13191326. [29] S. Van den Hende, H. Vervaeren, N. Boon, Flue gas compounds and microalgae: (bio-)chemical interactions leading to biotechnological opportunities, Biotechnology Advances 30 (2012) 14051424. [30] G. Ifrim, M. Titica, M. Barbu, L. Boillereaux, G. Cogne, S. Caraman, J. Legrand, Multivariable feedback linearizing control of chlamydomonas reinhardtii photoautotrophic growth process in a torus photobioreactor, Chemical Engineering Journal 218 (2013) 191203. [31] T. Asaeda, T. Van Bon, Modelling the effects of macrophytes on algal blooming in eutrophic shallow lakes, Ecological Modelling 104 (1997) 261287. [32] D. Dochain, S. Gregoire, A. Pauss, M. Schaegger, Dynamical modelling of a waste stabilization pond, Bioprocess and Biosystems Engineering 26 (2003) 1926. [33] S. Kayombo, T.S.A. Mbwette, A.W. Mayo, J.H.Y. Katima, S.E. Jorgensen, Modelling diurnal variation of dissolved oxygen in waste stabilization ponds, Ecological Modelling 127 (2000) 2131. [34] M. Omlin, P. Reichert, R. Forster, Biogeochemical model of Lake Zurich: model equations and results, Ecological Modelling 141 (2001) 77103.