Continuous manufacturing is usually presented as a whole plant and adopted one step at a time, which is the right way round. The gain is not speed, it is that a small piece of equipment running for a long time does the work of a large one running briefly, so the capital and the facility shrink and the material spends less time waiting. The cost is control: a process with no batch boundary needs a definition of a lot, analytics fast enough to act on, and a plan for what happens to the material made while something was drifting. Deciding step by step is what makes those costs affordable.
Where continuous chromatography earns its complexity
Running several small columns in sequence so that one loads while another elutes uses the resin far harder than a single large column does, which is the whole economic argument: the same throughput on a fraction of the capital tied up in resin, and a capture step that is no longer sized by the biggest batch. What it costs is valves, a controller and a method that has to be developed as a cycle rather than a run. It pays first on an expensive resin and a feed that is always available, and it pays least on a cheap polishing step that is already fast.
What continuous bioprocessing changes upstream and down
Upstream the change is perfusion: cells held at high density with medium exchanged continuously, which gives a steadier product quality profile than a fed batch peak and asks for a cell retention device that works for weeks. Downstream the change is that the purification sees a constant, dilute stream rather than a concentrated harvest, which suits chromatography and suits filtration less. The two halves have to be designed together, because a perfusion upstream feeding a batch downstream just moves the surge tank around.
When a continuous bioreactor is the wrong choice
A vessel run for weeks is exposed for weeks: one contamination event costs the whole campaign rather than one batch, and a line that drifts genetically has time to do it. That is why this is chosen for products whose quality benefits from a steady state, and avoided where the cell line is unstable, where the product is unstable in the medium, or where the facility cannot hold sterility for that long. The honest comparison is not yield per litre, it is grams per campaign against the risk of losing the campaign.
Continuous flow centrifugation, and the harvest nobody plans
Harvest is where a continuous train most often reverts to batch, because a disc stack or a tubular bowl running continuously is a real machine with real cleaning requirements and it is easy to specify one too small. Sizing follows the solids loading rather than the flow, and the number to ask for is the clarification achieved at the flow you will actually run rather than the maximum on the datasheet. Where the feed is a high density perfusion culture, filtration often beats centrifugation on the same duty.
The control and the analytics that make any of it legal
Two things have to exist before a regulator will look at it: a definition of a lot in terms of time and material, so that a result can be traced to what it was measured on, and measurements fast enough to act on rather than fast enough to report. That is what process analytical technology means in practice, and it is the part that takes the longest to build. A continuous step with batch analytics is a batch step that happens to run smoothly.
Deciding one step at a time
The sequence that works is to find the step that is the bottleneck or the largest piece of capital, convert that one, and keep a surge tank on either side of it so the rest of the train does not have to change. That keeps the validation local and gives a number to argue the next conversion with. A program that starts by designing a whole continuous plant tends to stop at the point where the analytics bill arrives.
A rocking bioreactor for the seed train
A rocking bioreactor agitates a bag on a platform rather than with an impeller, which gives low shear mixing and no cleaning, and it is used mostly for seed expansion and for shear sensitive cultures. Gas transfer comes from surface exchange, which sets the ceiling on cell density. Rock angle, rate and fill volume are the three parameters that have to be recorded for a transfer.
A batch bioreactor and what the mode decides
A batch bioreactor runs to exhaustion with no feed, which makes it the simplest mode and the one where nutrient depletion and by-product accumulation define the end point. Fed batch extends it by adding concentrated feed, and perfusion replaces medium continuously. The mode is chosen by the product's stability in the vessel rather than by the titre alone.
A lab scale bioreactor and what it has to predict
A lab scale bioreactor exists to predict a larger one, so its value is in the parameters that scale: power per volume, gas transfer coefficient, tip speed and mixing time, rather than its litres. Geometric similarity to the intended large vessel is what makes the data transferable. Instrumentation and control matter more at this scale than volume, since the point is the data.
A multi parallel bioreactor and the experiments it enables
A multi parallel bioreactor runs many small vessels under independent control, which turns process development from a series into a designed experiment, and that is what compresses a development timeline. The specification to compare is how much of the large vessel's control is genuinely reproduced per position, especially gassing and pH control, since a system that only controls temperature is a shaker.
Questions people ask about continuous manufacturing
What counts as a lot when there is no batch?
Whatever you define and can defend: a period of time, a quantity of material, or the output between two events such as a column cycle or a medium change. The definition has to let a result be traced to the material it describes, and it has to be written before the process runs rather than after a deviation.
Does it reduce cost of goods?
Usually through capital and facility rather than through materials: smaller equipment, less space, less time in process. Where the resin or the medium dominates the cost, the saving comes from using them harder, which is the same argument. It rarely reduces the analytical bill, which often rises.
Can an existing batch process be converted?
Step by step, and the earlier steps are the easier ones. Conversion means redeveloping that step as a cycle or a steady state, showing comparability of the material either side of the change, and adding the measurements to control it. A conversion presented as a straight swap has not been costed.