Bench scale bioreactor selection and scale up planning

Building a scale down model in a bench top bioreactor

1. Decide what will be held constant across scales. Power per volume, tip speed, oxygen transfer coefficient and mixing time cannot all be matched simultaneously. Choosing which one governs your process is the whole of scale up strategy.
2. Instrument the small vessel like the large one. Dissolved oxygen, pH, temperature and off gas at minimum, with the same control strategy. A vessel without the instrumentation of its target scale generates data that does not transfer.
3. Run enough vessels to see variation. A single small vessel gives one trajectory. Parallel vessels give the variability, which is what tells you whether a difference between conditions is real.
4. Include the harvest in the model. Clarification by depth filtration and concentration by tangential flow are where a good culture becomes an unusable harvest. Model them at small scale rather than discovering the problem at pilot.
5. Record everything the large scale will need. Feed strategy, timings, in process controls and their acceptance ranges, written as a process description rather than as a set of habits. That document is what transfers, not the vessel.

You cannot match everything

Geometric similarity, power input, tip speed, mixing time and oxygen transfer scale differently with volume, and holding one constant means letting the others change. That is why scale up is a choice rather than a calculation.

Identify the parameter your process is actually sensitive to, by experiment where possible, and design the scale down model around it.

The deliverable is a description

What leaves a small scale programme is not the vessel but a process description with critical parameters, ranges and justification. A receiving site can run that; it cannot run a set of habits.

Write it as the work proceeds. Reconstructing the reasoning behind a parameter range from old notebooks is slow and sometimes impossible.

A bioreactor system, and what comes with the vessel

A bioreactor system is the vessel plus everything that makes it controllable: the drive and impeller, gas supply and sparger, the pH, dissolved oxygen and temperature probes with their transmitters, pumps for feed and base, and the controller and software that log the run. Vessels are quoted alone often enough that a comparison falls apart on the probes, which are consumable and need calibration. Ask what the price includes, how many pumps and gas channels the controller drives, and whether the software exports the batch record you will need later.

A lab fermenter, and how it differs from a cell culture vessel

A lab fermenter and a cell culture bioreactor are the same apparatus tuned for different organisms. Microbial work is fast, hungry and hot: high agitation from a Rushton impeller, oxygen transfer measured in vessel volumes per minute, a heavy heat load to remove and a vessel that takes steam sterilisation in place. Mammalian work is the opposite, a gentle marine impeller, sparging that will not shear cells, and carbon dioxide in the gas mix for pH. A vessel sold for one will run the other badly, so match the impeller, the gas train and the cooling to the organism before the volume.

A small scale bioreactor, and what the scale is for

A small scale bioreactor exists to answer questions cheaply: media and feed screening, a design of experiments run across parallel vessels, and the characterisation a process needs before anyone books a suite. What it does not do is predict the large scale on its own, because mixing time, oxygen transfer and shear do not scale with volume. Keep the parameters that transfer, the volumetric power input, the oxygen transfer coefficient and the tip speed, and record them per run, so a result from a two litre vessel arrives at two hundred litres with the arithmetic already done.

An optical dissolved oxygen probe against a Clark cell

A Clark type dissolved oxygen probe consumes oxygen at a cathode behind a membrane, so it drifts as the electrolyte ages, needs the membrane changed and reads slowly at low tension. An optical probe reads the lifetime of a luminescent dye instead: nothing is consumed, there is no membrane to change, it tolerates repeated steam sterilisation and it holds calibration far longer, which is what makes it standard on single use vessels where nobody can service a probe. The cost is the transmitter and the spot, which photobleaches and is replaced on a schedule rather than when it fails.

A polarographic do probe, and what the membrane asks of you

A polarographic or Clark type oxygen probe holds an electrolyte and a cathode behind a gas-permeable membrane, and oxygen is consumed at the cathode to make the signal. That design is why it needs the things an optical probe does not: a membrane change on a schedule, electrolyte topped up or replaced, a polarisation period of several hours before calibration, and a stirred sample so the consumption does not deplete the film at the membrane. It is cheap, it is well understood, and it is the wrong probe for a vessel nobody can open mid-campaign.

Common questions

Does a small vessel predict a large one?

Only if it was designed to. Matching the governing parameter, instrumenting equivalently and running the same control strategy is what makes it a scale down model rather than a small experiment.

How many parallel benchtop bioreactors are useful?

Enough to estimate vessel to vessel variation, which usually means at least three per condition. A single vessel per condition cannot separate a treatment effect from run variability.

Why include harvest at small scale?

Because clarification and concentration frequently fail on cultures that looked excellent. Modelling them early prevents a pilot campaign discovering that the harvest cannot be processed.

Browse by service class

• Compare Gene synthesis on advertised prices