SmartScaler.
Transfer your process operating conditions between reactors in minutes.

SmartScaler is scale-up software that gives bioprocess teams CFD-grade answers in seconds, and nobody on the team needs to know CFD. It runs pretrained surrogate models, which are machine-learning models trained on validated CFD simulations.

Demo

From Source Reactor to Scaling Result in Under Two Minutes

You enter how your source reactor runs today and pick the target scale. SmartScaler returns matched operating parameters and shows how well each scaling criterion fits.

Challenge

Scale-up remains one of the highest-risk phases in bioprocess development.

Traditional scale-up still runs on correlations based on heuristics such as constant P/V. These scale-up rules fail on most non-standard production equipment and real-world fermentation broths.

What Changes Between Lab, Pilot and Production Scale

  • 01Mixing takes longer as the vessel gets bigger
  • 02Oxygen transfer shifts with vessel geometry and gas flow
  • 03Power input does not grow in proportion to volume
  • 04Production-scale runs are too expensive for many experiments

Solution

Smart Scale-Up Without CFD Complexity

SmartScaler packs validated CFD into pretrained surrogate models, so process engineers can use the results without setting up a simulation.

What Teams Do with SmartScaler

  • Compare scale-up strategies side by side before choosing one
  • Check whether mixing or mass transfer will limit the larger reactor
  • Spot risks that only appear at the larger scale
  • Back up engineering decisions before anything is built or changed

You work with the parameters your team already uses, such as stirrer speed, gas flow and fill volume.

SmartScaler workflow: source reactor operating parameters feed a pretrained surrogate (rapid-response) model of the cell microclimate, which produces the target reactor operating parameters
The surrogate model describes the conditions the cells experience, so SmartScaler can reproduce them in the target reactor.

Method

Pretrained Surrogate Models Built from Validated CFD

We simulate each reactor system once across its operating range and train machine-learning surrogate models on the results. Because the models come pretrained, a scaling question takes seconds and needs no new CFD run.

  1. Simulate the Operating RangeWe run mechanistic Lattice Boltzmann CFD of the reactor at many combinations of stirrer speed, fill volume and gas flow.
  2. Train Surrogate ModelsNeural networks learn single values such as kLa (oxygen transfer) and mixing time, and also how shear and energy dissipation are distributed through the tank.
  3. Validate and ProvideWe test the predictions against CFD runs the networks never saw in training. Only then does a model go into SmartScaler.
  4. Match Source and TargetAn optimizer searches for the operating point at the target scale where the cells see the same hydrodynamic environment as in your source reactor.

Today and Next

Today we build and validate the surrogate models and provide them in SmartScaler. If your reactor is not in the library yet, we create a model for it. Coming next, you will be able to train your own surrogates directly from SimVantage Bolt simulations.

See How the Surrogates Are Validated

Benefits

Built on Science. Validated in Industry.

In practice, that means five things for your team.

  • Know what to expect before the first run at the new scale
  • Catch problems before they turn into costly changes late in the project
  • Get more out of the target reactor
  • Give regulatory documentation a quantified basis for each scale-up choice
  • Move from pilot to production faster

Free Demo

Try SmartScaler yourself for free.

A free demo account lets you try SmartScaler on example reactors at your own pace. It runs in the cloud, so there is nothing to install, and you are not committing to anything.

Platform

SmartScaler in the SimVantage Platform

SmartScaler and SimVantage Bolt draw on the same CFD. Bolt is the tool for modeling your reactor and process in full detail. SmartScaler uses surrogate models built from that CFD to answer scale-up questions quickly, and our consulting team steps in when a question needs more than either tool.

Next, we are connecting the two more closely, so you can train your own surrogate models from SimVantage Bolt simulations.

See the Full Platform

Are you facing critical process design questions? Discuss your challenge with our team.

Whether you want more out of an existing reactor or are planning an industrial scale-up, we can help. Our mechanistic models have been validated in production reactors.