LES Simulation of Continuous Stirred Tank Reactors: Comparison to Theory and Experiment

John A. Thomas (M-Star)

Jason J. Giacomelli (PMSL)

Richard K. Grenville (PMSL)

AiChE 2015

Salt Lake City, UT

Wednesday, November 11, 2015

mstarcfd.com

Balancing throughput with mixing

mstarcfd.com Slide 2 of 14

P.N. Jones, N.G. Ozcan-Taskin, M. Yianneskis The use of momentum ratio to evaluate the performance of CSTRs Chem. Eng. Res. Des., 87 (2009), p. 485

CSTRs are everywhere, but most are not “ideal”

mstarcfd.com Slide 3 of 14

P.N. Jones, N.G. Ozcan-Taskin, M. Yianneskis The use of momentum ratio to evaluate the performance of CSTRs Chem. Eng. Res. Des., 87 (2009), p. 485

1 2

3 4

• The CSTR is an important concept in Chemical Engineering

• “The CSTR is normally run at steady state, and is usually operated so as to be quite well mixed”.

• What are consequences when this

assumption is invalid?

CSTRs are everywhere, but most are not “ideal”

mstarcfd.com Slide 4 of 14

P.N. Jones, N.G. Ozcan-Taskin, M. Yianneskis The use of momentum ratio to evaluate the performance of CSTRs Chem. Eng. Res. Des., 87 (2009), p. 485

1 2

3 4

1

2

3

4

Exit Age Dist. [-]

Time [-]

Traditional CFD and CSTRs

mstarcfd.com Slide 5 of 14

M. Liu, Age distribution and the degree of mixing in continuous flow stirred tank reactors. Chem. Eng. Sci., 69 (2012), p. 382

Speed Age

LBM is a proven transient CFD framework

mstarcfd.com Slide 6 of 14

Strong history in the community • Jack G. M. Eggels (Shell Oil) –Initial LBM-LES simulations of Rushton turbine (1996) •Harry E. A. Van den Akker (Delft University) –Additional detail into flow physics/dissipation •Jos J. Derksen (U. of Aberdeen) –Extensions to more exotic geometries/suspensions

M-Star DMT makes LBM universally accessible • Easy-to-use • HPC-friendly • Well-validated

Transient LES simulation using the DMT

mstarcfd.com Slide 7 of 14

DPBT

=0.1 m, Djet

=0.025 m

Q=0.3 L/s, N=1 s-1

~12 million lattice points. Runtime: 10 hours on 160 cores (POD)

https://youtu.be/EeK-9hgYXd4 https://youtu.be/2wuNnayjvYE

Measured ϴm

: 68.9 s

Predictions agree with experiment

mstarcfd.com Slide 8 of 14

Jones et al. Prediction from M-Star DMT

Predicted ϴm

: 68.5 s

P.N. Jones, N.G. Ozcan-Taskin, M. Yianneskis Chem. Eng. Res. Des., 87 (2009), p. 485

Exit Age Dist.

Tuning the system to minimize bypassing

mstarcfd.com

DPBT

=0.1 m, Djet

=0.025 m

Q=0.3 L/s, N=5 s-1

Slide 9 of 14

https://youtu.be/4gMlzx9rFaI https://youtu.be/7_ELcmc-7Rs

Jones, P.N. (2004). Doctoral Dissertation

Exit Age Dist.

The outlet data suggests minimal bypassing

Experimental data 0

0 0.5 1.0 1.5 0

0.1

1.0

10

Exit Age Dist.

DMT predictions

𝑡/𝜏 Time,

Slide 10 of 14

Testing various configurations in the DMT

mstarcfd.com Slide 11 of 14

Gives same exit age distribution as top injection! But, very high transient side forces

https://youtu.be/or_e3Ok_PN4 https://youtu.be/ye0Q0eZORRM

Easy to model time-accurate residence times

mstarcfd.com Slide 12 of 32

https://youtu.be/V_KBNuAsTpo

Easy to monitor homogenization times

mstarcfd.com Slide 13 of 14

Key points and summary

CSTRs are governed by inlet velocity and rotation rate No CSTR is ideal---need to consider back-mixing Experiments often reply on exit properties, which have limited insight LBM and the DMT offer mechanistic insights into reaction chemistry Boltzmann is a bridge between experiments and CFD

• DMT is time-accurate and requires no solid meshing • DMT conserves energy and runs very quickly

Slide 14 of 14

Thank you!

mstarcfd.com Slide 15 of 32

Thank you!