Team:Cornell/Model

Team:Cornell - 2019.igem.org

Modeling
Overview

Our bioreactor is similar to a traditional packed-bed reactor. In packed bed reactors, porous pellets support a small catalyst pellet at the center. Likewise, in our system, inert alginate beads support E. coli - effectively a catalyst for microcystin breakdown. For a fluid carried through the bioreactor, there are three steps for mass transport and the reaction.

1. Mass transfer through the boundary layer
2. Molecular diffusion down the length of the pore
3. Reaction

Mass Transfer

We modeled mass transfer through the boundary layer using a simple mass transfer model. The model is based on correlated data for forced convection through boundary layers.


Where đšœ is the mass flux and k is the mass transfer coefficient. The mass transfer coefficient was determined by correlation.


Here, D is the diffusion constant for microcystin-LR, d is the pellet diameter, Re is the dimensionless Reynolds number, and Sc is the dimensionless Schmidt number. The Reynolds number is related to the fluid flow regime - a ratio of the inertial to viscous forces in the fluid. The Schmidt number relates the resistance of momentum diffusion to mass diffusion. We found that the diffusion constant D = 1.4 × 10-6 cm2/s according to Zastepa et. al (1).

We anticipated that the boundary layer diffusion would likely not be the limiting factor in our system due to the relatively inviscid behavior of water. Regardless, we aimed to introduce mixing and eddy flow in the reactor to induce turbulent flow, since the mass transfer coefficient is proportional to the Reynolds number (turbulent flows are characterized by high Reynolds numbers). Figure 1 illustrates the dependence of the mass transfer coefficient on the degree of turbulence, or mixing, in the system.


Figure 1. Influence of turbulence on the mass transfer coefficient. Reynolds numbers above roughly 2000 indicate turbulent flow, or a high degree of mixing.

Molecular Diffusion and Reaction

We coupled our modeling for molecular diffusion and the actual reaction - the classical treatment by chemical engineers. The E. coli are encapsulated in porous alginate pellets. The microcystins, once they diffuse through the boundary layer, must diffuse down the length of a pore and then react in the bacteria. We approximated the reaction as a first-order reaction. Michaelis-Menten kinetics, commonly used in biological systems, is approximately a first-order reaction at low substrate (microcystin) concentrations.


To relate the reaction rate to the timescale on which pore diffusion takes place we introduce a new dimensionless group, the Thiele modulus (2).


R is the pellet radius, D is the diffusion constant for microcystin, and k’ is the rate constant divided by the total pellet volume. In a randomly packed reactor, the total pellet volume is approximately 64% of the total reactor volume.

We can relate the Thiele modulus to another group called the effectiveness factor, the ratio of the actual reaction rate to the ideal reaction rate. The effectiveness factor tells us how much mass transfer resistance (on a molecular level) harms the overall reaction rate. Analytically, the effectiveness factor is


Where coth is the hyperbolic cotangent function. Figures 2 and 3 predict how the effectiveness factor varies with the Thiele modulus and pellet diameter, respectively.


Figure 2. Effectiveness factor vs. Thiele Modulus, for a rate constant of k = 0.1.


Figure 3. Effectiveness factor vs. Pellet Diameter.

Our modeling has informed our decision to minimize the size of the pellet to decrease the characteristic time for molecular diffusion down the length of the pore. We made the pellets as small as we reasonably could (3 mm diameter). We lacked the ability with our available tools to make smaller pellets; however, decreasing the pellet size further should be a trivial matter with more advanced manufacturing techniques and greater control over the process.

Fluidics Model

We modeled the fluid flow of our bioreactor with our nozzle using COMSOL to see if there was computational support for mixing inside the bioreactor. The negative space of the inlet, which is the volume that the fluid would flow through, of our bioreactor was used as the geometrical model in Comsol. We first assumed that flow through the nozzle was turbulent and followed a Îș-Δ model of flow, which utilizes a wall function to bridge the region between the wall and the fully developed flow. A wall function significantly reduces the computation time though the solution near the walls is less accurate. We further assumed that mixing results near the beginning of the bioreactor would be indicative of mixing throughout the bioreactor. When alginate beads were incorporated we assumed that the alginate beads operated under ideal conditions, are static and deflect fluid similar to a wall. All tests were conducted with the material properties of water flowing through the inlet at 10 m/s and exiting the outlet at similar pressure to the inlet.

Initially, a COMSOL model was generated, composing of only the nozzle and the beginning of the bioreactor. Significant mixing was shown to occur within the nozzle head with mixing occurring within the bioreactor, though only minimal mixing was observed in the center of the bioreactor. As you can see below, the streamline exhibits eddy flow and turbulent mixing, an indication that our bioreactor will be well mixed.



Next we incorporated the alginate beads into the model. The beads were assumed to be 3mm in diameter or approximately 0.1 inches and were symmetrically distributed within the bioreactor with some spacing between beads.



Results showed significant mixing shortly after entering the bioreactor with more mixing occurring in the center of the bioreactor than in the initial model. The results from this model provided analytical support for the occurrence of mixing within the bioreactor. The resulting arrow lines of our model is displayed below.



1. Zastepa, A., Pick, F. R., & Blais, J. M. (2017). Distribution and flux of microcystin congeners in lake sediments. Lake and Reservoir Management, 33(4), 444-451.
2. Levenspiel, O. (1999). Chemical Reaction Engineering (3rd ed.). John Wiley and Sons.