Вы находитесь на странице: 1из 1

Use of a Rotating Mesh in Computational Fluid

Dynamics (CFD) Modeling of a Cell Culture Bioreactor


Equipped with a Packed-Bed Impeller
Joseph McInerney1*, Zach Greenleaf2, Ma Sha2, Michael Thomas1
1
Eppendorf Manufacturing Corporation, Enfield, CT, USA; 2Eppendorf, Inc., Enfield, CT, USA; *Presenting author: mcinerney.j@eppendorf.com

Abstract
Anchorage dependent cells, such as Vero cells, are widely used as a platform for viral vaccine production. We have previously >>It is most economical to develop new biological and Single-use vessel development and characterisation
published a high density Vero cell culture method using the BioBLU® 5p Single-Ues Vessel. The Eppendorf BioBLU 5p Single-Use pharmaceutical processes at the smallest scales >>Mixingt time
possible; ideally with equipment that can fit on a >>Flow patterns
Vessel uses a packed-bed for cell attachment and a patented packed-bed impeller to gently circulate the nutrients and oxygen the
laboratory bench. >>Power number
cells need to grow. The packed-bed impeller moves liquid by using the centrifugal force generated as liquid is pulled up through
>>Only when a bioprocess produces desirable results at >>Laboratory testing with cells
its exit nozzles. The centrifugal force of the flow rules out the use of the Multiple Reference Frame (MRF) method to model the the bench and pilot scales should it be scaled up to
flow by CFD, so a rotating mesh method must be used. In the rotating mesh method, the impeller and mesh directly surrounding the manufacturing scale.
the impeller moves with the impeller inside a stationary global mesh. >>Use of single-use equipment saves time, because the
bioreactor does not need to be sterilized before and
cleaned after the run.

  
BioBLU 5p Single-Use Vessel Power number Torque measurement
Vessel specifications Impeller specifications The bioreactor characteristic used as the Experimental torque measurement
benchmark to compare the CFD model The moment on the impeller shaft is measured with a custom
BioBLU 5p BioBLU 5p
to the experimental result is the power torque measuring system that is installed between the
Maximum gas flow (SLPM) 57.5 Impeller style Packed-bed
impeller number. electric motor and the impeller drive shaft coupling on top of
Maximum gas flow (VVM) 2.0 Impeller material Polycarbonate the BioBLU 5p vessel.
Vessel type Single-Use Impeller quantity 1 2 π (M-Md)
Working volume (L) 3.75 Power number Np =
Impeller diameter (D) (mm) 80.25
Vmax height* (mm) 192 ρ N2d5
Impeller diameter: vessel ID 0.48
Vessel inner diameter (ID) (mm) 166 Max. tip speed (m/s) 0.8 M: torque (N•m) A drive motor and custom torque measuring
Vmax height: vessel ID 1.15 Md: dry vessel torque (N•m) assembly on top of a BioBLU 5p.
Max. agitation (rpm) 200 ρ: fluid density (kg/m3)
*: Vmax height = height from bottom of the vessel to liquid top N: agitation speed (rps)
BioBLU 5p Single-Use Vessel d: impeller outer diameter (m)
surface at maximum vessel working volume


CFD model Model geometry Rotating mesh method >>ANSYS Fluent® software solves
>>Original CAD drawing is >>Bioreactor geometry is broken RANS equations (Reynolds Aver-
Multiple reference frame (MRF) method is the industry standard. In MRF, the impeller blades are held stationary, and the flow is set in motion inside a imported into ANSYS® up into a mix of tetrahedral and aged Navier-Stokes) for each ele-
zone surrounding the impeller. The force from the liquid on the blades is the same as if the blades were moving and the liquid was initially stationary.
For the packed-bed impeller (6), the primary driver of the flow is the centrifugal force imparted to the water as it exits the impeller nozzles. As stationary
geometry modeler; hexagonal mesh elements (2.9 ment during each iterations
Stationary
nozzles would not cause centrifugal force, the MRF method cannot be used to CFD model the packed-bed impeller. In this case, a rotating mesh method SpaceClaims million) mesh >>10 iterations per time step. The
is used to model the packed-bed impeller. In the rotating mesh method, the impeller and mesh directly surrounding the impeller moves with the impeller >>Model axis and rotational >>Mesh is separated into an outer time step size is fixed at 4 de-
inside a stationary global mesh. The rotating mesh moves and is regenerated between each iteration. This regeneration is time consuming during the CFD
run, and requires careful mesh planning and care in building the case prior to the run. This care in building cannot be stressed enough. The end result of axis must be the same stationary and an inner rotating grees of rotation between time
The geometry model shows packed-bed Rotating
the rotating mesh is a more realistic simulation. The impeller revolves naturally, the results can be more accurate than MRF, and accurate mixing times impeller inside the cell-culture basket mesh, that contains the impeller. steps which allows a stable, high-
can be obtained as well as the power number. mesh
filled with cell attachment matrix. Contact faces are mesh interfaces. resolution rotating solution.

 
CFD base equations Results
For the CFD modeling, ANSYS Fluent® is used with Fluent’s RANS solver (Reynolds Averaged Navier-Stokes). The CFD software is able to plot the flow paths, velocity magnitudes, and velocity vectors of the water around the inside of the bioreactor.
RANS solves the continuity (2) and conservation of momentum (3) equations for fluid through the control volumes
delineated by the meshed geometry.

 
 ·      (2)
t


 · (    p     g F (3)
t

k-ε turbulence models solve the two transport equations for kinetic energy (k) and eddy energy dissipation (ε). In the
Realizable k-ε model, the kinetic energy transport equation is: View of flow pathlines View of flow pathlines The velocity magnitude of the flow in a cutting plane
through the impeller exit nozzles.
 
t
           
     G  Gb      Y  S

(4)

and the eddy energy dissipation transport equation is:


 
 
  
              C G  C Gb)  C   S (5)
t 

  C 

: density, : turbulent viscosity

: velocity vector;
p: pressure C C C C: constants derived from
experiments conducted on
: viscous shear stress tensor a wide range of flows The velocity flow vector in a cutting plane through Velocity vectors a plane through the center of the vessel.
g: gravity the impeller exit nozzles. The velocity scale is meant to emphasize the slower flow
F: external force in the vessel at the expense of a distorted view of the
,: Prandtl numbers faster flow through and around the impeller.
G: kinetic energy generation

Gb: buoyancy generation
Y:fluctuating effects BioBLU 5p Np laboratory power number values are
S: external energy source term obtained from a torque meter mounted on top of the
BioBLU 5p vessel. The CFD simulations use the BioBLU
Assumptions 5p geometry with a rotating mesh.
There are over 100 assumptions that are made in the computer simulation. Making these assumptions is not a trivial Observations:
matter, as some assumptions might radically change the results. A few of the more important assumptions are:
>>CFD results for this impeller type are 1.4 times
>>The degree of rotation allows sufficient resolution for accurate results
greater than laboratory results
>>Treatment of the water surface as a flat surface rather than a concave air-water interface
>>The CFD results could be optimized to better
>>Geometry simplifications such as eliminating or smoothing smaller geometry details do not hurt accuracy
match the average values of the laboratory results
>>Mesh density (2.9 million elements) is assumed to give mesh independent results Power number vs. tip speed experimental and default
CFD curves for a 3.75 L Eppendorf BioBLU 5p packed- by adjusting the turbulence model as was done in
>>Turbulence models are assumed to give good approximation of turbulence bed cell culture perfusion bioreactor. Power number vs. tip speed for various impeller
reference 5. types.

 References Acknowlegments
Conclusion [1] Bin Li and Ma Sha. Scale-Up of Escherichia coli Fermentation from Small
Scale to Pilot Scale Using Eppendorf Fermentation Systems. Eppendorf We thank
>>The use of the rotating mesh method is acceptable in CFD modeling of packed-bed impeller bioreactors. Application Note No. 306, 2016.
>> Using the default turbulence constants, the CFD power number results are 1.4 times greater than the power numbers derived from lab experiments on a BioBLU 5p Single-Use
[2] ANSYS, Inc. ANSYS Fluent Theory Guide. Release 17.0. January 2016. ANSYS support:
Canonsburg, PA.
Vessel. This is a decent first-pass result, given that the full range of possible power numbers for different types of vessels and impellers is on the order of 30 times (0.5 to 15). [3] Han XF., Willard S., and Sha M., Cell Culture Scale-Up in BioBLU® Single- Jeremy McCaslin,
>> The CFD model can be optimized from a few lab experiments; and will be optimized in further studies. CFD works hand in hand with experiment. Use Vessels. Eppendorf Scalability Presentation, 2016.
[4] Z. Greenleaf, XF Han, J. McInerney, and M. Sha. P/V Scale-UP of Packed-Bed
Surya Deb,
>>Now that the degree of CFD error is known from lab comparison, most of engineering of new bioreactors can be done by CFD, and then only checked in the laboratory in the final Perfusion Vessels. Eppendorf Internal white paper. Robert Payne
phase of development. An improved design in CFD modeling will translate to an improved design in reality. [5] J. McInerney, M. Aguilar, M. Thomas, M. Sha. Optimization of Computational
Fluid Dynamics (CFD) Modeling using Experimentally Obtained
>>The ability of CFD models to predict a commercial scale process from bench scale results can save hundreds of hours of design time and prevent costly mistakes in the construction
Characteristics of Stirred-Tank Fermentors . Poster. BPI 2017. Boston.
of large commercial scale bioreactors. [6] Freedman, D., Zhenbin, Z. Agitator for a fermentation and tissue culturing
vessel. United States Patent 4,634,675. Issued Jan. 6, 1987.


Your local distributor: www.eppendorf.com/contact · Eppendorf AG · Barkhausenweg 1 · 22339 Hamburg · Germany · eppendorf@eppendorf.com

ANSYS® and Fluent® are registered trademarks of ANSYS, Inc., USA. Eppendorf®, the Eppendorf Brand Design, and BioBLU® are registered trademarks of Eppendorf AG, Germany.
www.eppendorf.com
All rights reserved, including graphics and images. Copyright © 2018 by Eppendorf AG.

Вам также может понравиться