Peer-Reviewed PublicationBiotechnology & Bioengineering 2022 · with Sandoz, TU Graz, BASF and CureVac

Reactor Optimization – Sandoz Kundl
Impeller Redesign at 160 m³

Sandoz wanted a 160 m³ production bioreactor to draw less power by alternating radial and axial impellers. Tracer tests and CFD (computational fluid dynamics) showed which configuration would work before any impeller was changed in steel.

Read the Paper Contact Our Team
30% Shorter mixing time in a 160 m³ production reactor
20% Lower energy demand in the same reactor
3,740 MWh Energy saved per reactor each year

Challenge

Stirring a 160 m³ fermenter is a major cost factor.

In a large fermenter, a good part of the production cost goes into stirring and aerating the broth. The classic setup puts a Rushton turbine (radial-flow disc impeller) on every level. That breaks up the gas well, but it draws far more power than axial impellers do. Axial impellers save power, yet near the bottom of the tank they handle gas poorly.

Alternating the two types looks like a good compromise. Before touching the vessel, though, Sandoz had to know whether it would still disperse the gas and mix as fast while drawing less power in a real 160 m³ reactor (Ø 4.25 m, 100 rpm).

Technical drawing of the 160 m³ reactor with four impeller levels and probe positions
Production reactor with four impeller levels. Tracer goes in at the top, and probes in two planes record it. Figure from Bernauer et al., Biotechnology & Bioengineering 2022

Approach

Full-Scale Validation Before Virtual Optimization

Before anyone trusted the simulation, it had to match the real 160 m³ production reactor, and validation at that scale is almost never possible. A tracer was injected and followed with probes until the vessel was evenly mixed. SimVantage's GPU-accelerated CFD software reproduced the measured mixing curves. From then on, Sandoz engineers could try stirrer systems on the computer and compare them until they found the best one. Only that configuration went into the real reactor.

Tracer signal over time: two measurements and the CFD simulation lie on top of each other
Measured and simulated tracer signal in the 160 m³ reactor
Side-by-side CFD comparison of two impeller configurations and tracer mixing snapshots from 10 to 130 seconds
Flow field and spreading tracer over time in configurations A and B

Outcome

Implemented in Production and Publicly Reported

The new configuration saves roughly €350,000 and 3,740 MWh per reactor each year. With alternating radial and axial impellers, the reactor holds about as much gas as before (gas holdup, the gas volume fraction in the liquid) and mixes much better at the same operating conditions. In production, mixing time dropped by 30% and energy demand by 20%. Sandoz has reported the savings at Kundl publicly, along with the roll-out of the upgrade to further large fermenters.

Publication

Analyzing the effect of using axial impellers in large-scale bioreactors

Bernauer et al. · Biotechnology & Bioengineering 2022
Joint work with Sandoz, Graz University of Technology, BASF and CureVac

Companion paper: Characterization of the gas dispersion behavior of multiple impeller stages by flow regime analysis and CFD simulations, Bernauer et al., Biotechnology & Bioengineering 2021

Read the Paper All Publications

Facing a similar question? Let's talk.

Tell us about your reactor and your bottleneck. We will show you what simulation can quantify before you change anything in steel.

Book a Demo Contact Our Team All Case Studies