Process Development Scale-Up Steps for Biologics: From Shake Flask to 2,000 L

Process Development & Validation

Biologics scale-up moves a process through four steps: shake flask and 3–5 L bench bioreactors, 50 L single-use pilot runs, 100–500 L stainless-steel or single-use vessels, and 1,000–2,000 L commercial trains. At each step, oxygen transfer, mixing, shear and CO2 removal change non-linearly, and downstream columns and filters are resized to keep residence time and load constant.

Insights · 9 min read · Updated 2026

Key definition. Scale-up is the transfer of a bioprocess from a small vessel to a larger one while holding product quality and yield within the ranges defined during process development. Because power input, mixing time and gas transfer do not scale proportionally with volume, it is a sequence of engineering decisions, not a multiplication.

In this article

Why scale-up fails: parameters that do not scale linearly
Step 1: shake flask and 3–5 L bench-top bioreactors
Step 2: 50 L single-use pilot runs
Step 3: 100–500 L, stainless steel versus single-use
Step 4: 1,000–2,000 L commercial single-use trains
Scaling downstream: column diameter, UF area, buffer volumes
Scale-down models and why validation depends on them
Frequently asked questions

Why scale-up fails: parameters that do not scale linearly

Scale-up fails when a team holds the wrong parameter constant. Volume rises with the cube of vessel diameter but surface area with the square, so every surface-dependent process, including gas exchange, heat removal and base addition, weakens as the vessel grows. The table lists the parameters that drift most and the usual response.

Parameter	What happens at larger scale	Common criterion or mitigation
Oxygen transfer (kLa)	Falls per unit volume; sparger and agitation must compensate	Hold kLa or oxygen transfer rate constant; switch from microsparger to drilled-hole sparger to limit foam
Mixing time	Rises from seconds to minutes; pH and nutrient gradients appear near feed and base addition points	Hold power per volume (P/V, W/m³) constant; relocate feed lines to the impeller zone
Shear	Impeller tip speed rises if P/V is held; bubble burst at the surface is the main cell-damage mechanism	Cap tip speed; maintain shear-protectant surfactant concentration
Dissolved CO2 (pCO2)	Accumulates because stripping depends on gas flow; hydrostatic pressure raises solubility at the base of tall vessels	Increase overlay or sparge air flow; use larger bubbles for stripping

No single criterion works alone. Constant P/V preserves mixing but may push tip speed past the shear limit; constant tip speed starves mixing. Most mammalian process development scale up steps hold P/V within a tip-speed ceiling, then tune gassing for kLa and CO2 stripping.

Step 1: shake flask and 3–5 L bench-top bioreactors

The first of the process development scale up steps converts a shake-flask process into a controlled bioreactor process, and this is where the parameter set for every later stage is defined. Shake flasks have no pH or DO control; a 3 L or 5 L stirred bioreactor introduces both, with base addition, sparging and defined agitation. Differences in titer or growth between flask and bench reactor are usually the first signal that the clone or medium is sensitive to pCO2 or pH control strategy.

This step should produce a process description with ranges for temperature, pH, DO, agitation, feed schedule and harvest criteria, plus the oxygen uptake rate at peak cell density, which sets the kLa target for every later vessel. It is also where the seed train expansion ratio, typically 1:5 to 1:10 per passage, is fixed.

EirGenix in practice

EirGenix's process development group runs 3 L and 5 L bench-top bioreactors for mammalian work and 2 L bench-top fermentors for microbial processes, and lists media screening, feeding optimization and critical parameter control among its upstream activities.

Step 2: 50 L single-use pilot runs

A 50 L single-use bioreactor is the first vessel whose geometry resembles production equipment, and it is the step at which most scale-up problems become visible. The 10- to 17-fold jump from a 3–5 L vessel is large enough to expose mixing and CO2 issues, but small enough that a failed run costs days rather than weeks.

Three outputs justify the pilot step. It confirms that the seed train delivers the inoculation density at the planned cell age. It generates tens of grams of material for downstream development at representative column sizes. And it provides the first at-scale data set against which a scale-down model can later be qualified. Two or three 50 L batches at the center of the operating ranges are common practice before committing a 500 L or 2,000 L slot.

EirGenix in practice

EirGenix lists 50 L single-use bioreactors for mammalian process development, alongside a 20 L fermentor for microbial pilot work.

Step 3: 100–500 L, stainless steel versus single-use

At 100–500 L the choice between stainless steel and single-use becomes a program decision, because it affects batch cost, changeover time and the validation package. Stainless-steel vessels support higher gas flow rates and pressures and have decades of cleaning-validation precedent. Single-use vessels remove cleaning and steam-in-place from the critical path, shorten turnaround between products and limit cross-contamination risk in multi-product suites, at the cost of consumable spend and an extractables and leachables assessment per film.

For clinical supply, 200–500 L batches of a monoclonal antibody at 3–5 g/L deliver roughly 0.5–2 kg of drug substance, enough for most Phase 1 and Phase 2 programs. The scale-up step from 50 L to 500 L exposes any dependence on a specific sparger or impeller geometry, so kLa curves and mixing-time data for each vessel should be on file before the first GMP batch.

EirGenix in practice

EirGenix's Xizhi site runs 100 L and 500 L stainless-steel bioreactors alongside 50 L, 200 L and 1,000 L single-use bioreactors, with purification, media and buffer preparation, washer and autoclave rooms on site.

Step 4: 1,000–2,000 L commercial single-use trains

The final scale-up step into 1,000–2,000 L single-use trains is where supply economics are decided, and the technical risk is dominated by gas transfer and CO2 control rather than chemistry. A 2,000 L single-use bioreactor holds about two meters of liquid, so hydrostatic pressure and the long bubble-rise path change the balance between oxygen delivery and CO2 stripping. Processes that ran comfortably at 500 L can see pCO2 exceed 100 mmHg at 2,000 L unless the sparge strategy is redesigned.

Commercial single-use trains also change the seed train. Reaching a 2,000 L inoculation density typically requires a 500 L N-1 stage, sometimes run in perfusion to shorten the production phase. Batches per year, harvest volume per batch and the downstream equipment that must process it within validated hold times become the design constraints.

EirGenix in practice

EirGenix's Zhubei site, operational since 2019, runs 2,000 L single-use bioreactor trains with an annual antibody production capacity of up to 1,000 kg. Across both sites the company lists single-use trains from 50 L to 2,000 L and stainless-steel trains from 100 L to 500 L.

Scaling downstream: column diameter, UF area, buffer volumes

Downstream scale-up follows a different rule: keep the time and load each unit of product experiences constant, and scale the equipment cross-section to match the mass. For chromatography that means fixed bed height, linear velocity and residence time, with throughput raised by column diameter. For membranes it means fixed load per unit area, flux and transmembrane pressure, with area raised to match volume.

Unit operation	Hold constant	Scale by	Example path
Chromatography	Bed height (cm), linear velocity (cm/h), residence time (min), load (g protein per L resin)	Column diameter; column volume rises with the square of diameter	1 mL prepacked column → 14 cm → 60 cm diameter; cross-section grows about 18-fold from 14 cm to 60 cm
Depth filtration	Throughput (L/m²), flux (L/m²/h), differential pressure	Filter area	Lab capsule → pilot pod → production pod holder
Ultrafiltration/diafiltration	Load (g/m²), flux, TMP, diavolumes	Membrane area	88 cm² → 0.57 m² in development → 0.1–14 m² in GMP
Viral filtration	Throughput (L/m²), pressure, filtration time	Filter area	Small-area device → production cartridges in parallel
Buffers	Column volumes per step, concentration, hold time	Volume scales with column volume	Bench liters → 50–700 L preparation vessels

Buffer volume is the constraint most often underestimated. A 2,000 L harvest at 5 g/L yields about 10 kg of antibody; a 10 kg batch captured on Protein A loaded at 40 g/L needs about 250 L of resin-cycles — for example a 60 L column run over four cycles — and at roughly 20 column volumes per cycle, several thousand liters of buffer per batch.

EirGenix in practice

EirGenix's downstream development equipment spans MF/UF systems from 88 cm² to 0.57 m² and columns from 1 mL HiTrap to 14 cm BPG on ÄKTA Prime, Pilot and Avant systems; its GMP mammalian facility lists 0.1–14 m² UF units, 5–60 cm low-pressure chromatography columns and 50–700 L media and buffer preparation.

Scale-down models and why validation depends on them

Every scale-up step needs a qualified scale-down model, because process characterization and viral clearance studies run at bench scale and are only valid if the small vessel behaves like the large one. A scale-down model is a bench system, typically 2–5 L upstream or a small column downstream, whose outputs are statistically comparable to production data at the same set points.

Qualification compares growth, metabolite and product-quality profiles between scales using equivalence tests or offset analysis, with ICH Q8(R2) framing the design space work that follows. A consistent offset, such as lower peak cell density at bench scale, must be understood and documented rather than ignored. If a change in scale shifts product quality, ICH Q5E (2004) defines the comparability exercise, beginning with analytical characterization and extending to non-clinical or clinical data only if analytics cannot resolve the difference.

Frequently asked questions

What scale-up factor is reasonable per step?

A factor of 5 to 10 per step is conventional for mammalian culture, which is why a typical path runs 3–5 L, 50 L, 200–500 L and then 2,000 L. Larger jumps are possible when both vessels are engineering-characterized and a qualified scale-down model exists. The limit is usually CO2 stripping and mixing, not cell growth.

How do you choose between stainless steel and single-use?

Single-use suits multi-product facilities, clinical supply and processes up to 2,000 L, where fast changeover and no cleaning validation outweigh consumable cost. Stainless steel suits long single-product campaigns, high gas demand and scales above 2,000 L. Many sites run both.

What if product quality changes after scale-up?

ICH Q5E applies. The sponsor must demonstrate comparability between pre- and post-change material, starting with side-by-side analytical characterization of critical quality attributes such as glycosylation, charge variants and aggregates. If analytics cannot establish comparability, non-clinical or clinical bridging may be required, so the cheapest path is to find the scale-sensitive parameter and bring it back within range.

About EirGenix

EirGenix, Inc. (stock code 6589) is a Taiwan-based biologics CDMO and biosimilar developer founded in 2012, with cGMP facilities in Xizhi (New Taipei City) and Zhubei (Hsinchu County) and a European subsidiary in Munich. Its mammalian manufacturing site holds Taiwan FDA PIC/S GMP certification (2014), Japan PMDA approval (2020) and US FDA approval (2023), and the company's trastuzumab biosimilar EIRGASUN®/HERWENDA® received EU marketing authorization in November 2023 with Sandoz as marketing authorization holder. EirGenix provides cell line development, process development, analytical and quality control services, cGMP manufacturing in mammalian and microbial systems, and ADC and bioconjugate development and manufacturing.

Scale your process with EirGenix

Discuss your process development scale up steps from bench to 2,000 L with the EirGenix team.

Scale your process with EirGenix

References