Why Centrifugation for Biologics Harvest
Centrifugation is the primary harvest clarification method for large-scale biologics manufacturing, processing over 80% of commercial monoclonal antibody harvests above 2,000 L. The method separates cells and debris from the product-containing supernatant using centrifugal force, typically achieving 95 to 99% cell removal in a single continuous pass.
For processes running at cell densities of 20 to 40 × 106 cells/mL in fed-batch bioreactors, centrifugation handles 2 to 10% volumetric solids at throughputs of 500 to 10,000 L/h. The alternative, tangential flow filtration (TFF), is preferred at smaller scales and for perfusion harvests, but disc-stack centrifugation dominates at production volumes because it processes the entire bioreactor contents in 1 to 4 hours without membrane fouling constraints.
The clarified centrate (supernatant after centrifugation) still contains submicron debris and soluble impurities. It is nearly always polished by depth filtration before entering the capture chromatography step. This two-stage harvest train (centrifugation followed by depth filtration) has been the industrial standard since the early 2000s.
Centrifuge Types for Bioprocess Applications
Five centrifuge types serve bioprocess harvest, each suited to different scales, solids loads, and shear tolerances. Disc-stack centrifuges with intermittent solids ejection are the dominant choice for GMP biologics production, but the right selection depends on feed characteristics and facility constraints.
| Type | g-Force | Throughput (L/h) | Solids Capacity | Best For |
|---|---|---|---|---|
| Disc-stack (intermittent ejecting) | 5,000–15,000 | 500–10,000 | 1–10% v/v, auto-eject | Production mAb/vaccine harvest |
| Disc-stack (continuous nozzle) | 5,000–12,000 | 1,000–20,000 | 5–25% v/v, continuous | Microbial fermentation (E. coli, yeast) |
| Tubular bowl | 15,000–60,000 | 20–500 | 2–5 kg/cycle, manual | Lab/pilot, fine particle recovery |
| Decanter | 2,000–5,000 | 1,000–50,000 | >15% v/v, continuous scroll | High-solids fermentation, cell paste |
| Single-use fluidized bed (kSep) | 1,000–3,000 | 10–600 | 1–5% v/v, enclosed | Single-use facilities, shear-sensitive |
Disc-stack centrifuge (intermittent solids ejecting) is the workhorse for mammalian cell culture harvest. The bowl contains 50 to 200 conical discs spaced 0.4 to 1.5 mm apart. Feed enters through a central distributor, flows outward between discs where centrifugal force sediments cells toward the underside of each disc, and the clarified centrate exits at the top. Accumulated solids are ejected at timed intervals (every 30 to 300 seconds) by hydraulically opening the bowl periphery. Alfa Laval BTPX and GEA Westfalia CSE are the most common platforms in GMP facilities.
Tubular bowl centrifuge generates the highest g-force (up to 60,000 g) but has limited solids capacity (2 to 5 kg per cycle) and requires manual disassembly for cleanout. It is used at lab and pilot scales for fine particle separations where high clarification efficiency is needed.
Decanter centrifuge uses a scroll conveyor inside a horizontal bowl to continuously discharge solids. It handles feeds with more than 15% solids and is the standard for microbial cell paste recovery (E. coli inclusion body harvest, yeast biomass). Decanters generate lower g-force (2,000 to 5,000 g) and produce coarser separation.
Single-use fluidized-bed centrifuge (kSep) suspends cells in a counterflow chamber where centrifugal force balances fluid drag, creating a fluidized bed of cells. The clarified supernatant overflows continuously. This design is fully enclosed and single-use, generating lower shear than disc-stack centrifuges. It is suited for gene therapy and vaccine harvest at scales up to 2,000 L.
The Sigma Factor: Operating Principle and Calculation
The Sigma factor (Σ) is the equivalent settling area of a centrifuge, expressed in m2. It represents the cross-sectional area of a gravity settler that would achieve the same separation performance as the centrifuge. This single parameter captures the combined effect of bowl geometry, disc count, and rotational speed, making it the standard metric for centrifuge comparison and scale-up.
For a disc-stack centrifuge, the Sigma factor is calculated as:
Sigma Factor Formula
Σ = 2π · n · ω² · (Ro³ − Ri³) / (3 · g · tanθ)
- n = number of discs (typically 50–200)
- ω = angular velocity (rad/s) = 2π · RPM / 60
- Ro = outer disc radius (m)
- Ri = inner disc radius (m)
- θ = half-angle of discs from the vertical axis (typically 35–50°)
- g = gravitational acceleration (9.81 m/s²)
For a tubular bowl centrifuge, the Sigma factor simplifies to:
Σ = π · ω² · L · (3Ro² + Ri²) / (4 · g)
where L is the bowl length. Because tubular bowls have no discs, their Sigma factor is driven primarily by bowl length and radius, giving typical values of 100 to 1,000 m².
The critical insight is that the feed flow rate Q at which a centrifuge achieves a target separation efficiency is directly proportional to its Sigma factor through Stokes settling velocity:
Q = 2 · vs · Σ · η
where vs is the Stokes settling velocity of the particle and η is the efficiency correction factor (0.4 to 0.7 for disc-stack centrifuges under real operating conditions).
How to Size a Centrifuge for Your Harvest
Centrifuge sizing starts with the Stokes settling velocity of the target particle and works backward to the required Sigma factor and feed rate. For CHO cell culture harvest, the target is removing intact and lysed cells (8 to 20 µm diameter) from the product-containing supernatant.
Stokes settling velocity describes how fast a spherical particle settles under gravity:
vs = d² · (ρp − ρf) / (18 · μ)
- d = particle diameter (m)
- ρp = particle density (kg/m³) — CHO cells: ~1,050–1,070 kg/m³
- ρf = fluid density (kg/m³) — cell culture media: ~1,010 kg/m³
- μ = dynamic viscosity (Pa·s) — media at 37°C: ~0.7 × 10−3 Pa·s
Worked Example: Sizing a Disc-Stack Centrifuge for CHO Harvest
Given: 2,000 L bioreactor, CHO cell culture at 25 × 106 cells/mL, mean cell diameter 15 µm, target harvest time 4 hours.
Step 1: Calculate Stokes settling velocity
vs = (15 × 10−6)² × (1,060 − 1,010) / (18 × 0.7 × 10−3)
vs = 2.25 × 10−10 × 50 / 0.0126
vs = 8.9 × 10−7 m/s = 3.2 mm/h
Step 2: Calculate required feed rate
Q = 2,000 L / 4 h = 500 L/h
Step 3: Calculate required Sigma factor
Using Q = 2 · vs · Σ · η, with η = 0.5:
Σ = Q / (2 · vs · η)
Σ = (500 / 3,600 × 10−3) / (2 × 8.9 × 10−7 × 0.5)
Σ = 1.39 × 10−4 / 8.9 × 10−7
Σ = 156 m²
Step 4: Select centrifuge
A Sigma of 156 m² is well within the range of a pilot disc-stack centrifuge (e.g., Alfa Laval BTPX 205 with Σ ≈ 3,000 m²). The BTPX 205 could therefore process this harvest at much higher feed rates. At 500 L/h, the actual Q/Σ = 0.17 L/(h·m²), giving comfortable margin for centrate clarity.
Note: Lower Q/Σ means longer residence time in the disc stack, producing clearer centrate but extending harvest time. The optimum is found by running a clarification curve at different feed rates.
Centrifugation Scale-Up Calculator
Calculate Sigma factors, Stokes settling velocity, Q/Σ ratios, and scale-up equivalence for disc-stack, tubular bowl, and decanter centrifuges.
Scale-Up Using Equivalent Settling Area
Centrifuge scale-up relies on maintaining the same Q/Σ ratio between lab and production equipment. If a lab tubular bowl centrifuge (Σ = 200 m²) clarifies a feed to acceptable turbidity at 20 L/h (Q/Σ = 0.1), a production disc-stack centrifuge (Σ = 30,000 m²) can process up to 3,000 L/h at the same Q/Σ of 0.1 L/(h·m²).
| Centrifuge Model | Type | Σ (m²) | Typical Q (L/h) | Scale |
|---|---|---|---|---|
| Beckman J2-21 | Tubular bowl | 100–500 | 5–50 | Lab |
| CARR P6 / Thermo DynaSpin | Single-use disc | 500–2,000 | 50–500 | Pilot |
| Alfa Laval BTPX 205 | Disc-stack | 2,000–5,000 | 200–2,000 | Pilot/Clinical |
| GEA Westfalia CSE 80 | Disc-stack | 5,000–12,000 | 500–4,000 | Clinical/Commercial |
| Alfa Laval BTPX 710 | Disc-stack | 20,000–60,000 | 2,000–10,000 | Commercial |
| kSep 6000S | Fluidized bed | 300–800 | 10–600 | Pilot/Clinical (SU) |
The efficiency correction factor η accounts for non-ideal flow between the discs. Stainless-steel disc-stack centrifuges in bioprocess service typically achieve η = 0.4 to 0.6, meaning the effective Sigma is 40 to 60% of the theoretical value. This correction arises from wake regions behind disc spacers, uneven feed distribution, and remixing during solids ejection.
When scaling up, generate a clarification curve at each scale by plotting centrate turbidity (NTU) versus Q/Σ. The curve should overlap between scales when the correction factor is properly calibrated. If the production centrifuge gives higher turbidity at the same Q/Σ, increase the correction factor or reduce feed rate.
A design-of-experiments approach can systematically optimize centrifuge operating parameters. Using a DOE generator, screen the effects of feed rate, bowl speed, ejection frequency, and back-pressure on centrate turbidity, product recovery, and cell lysis. A response surface design with 3 to 4 factors and 20 to 30 runs is typically sufficient to map the operating space and identify the Q/Σ sweet spot for your specific cell line.
Shear, Cell Lysis, and Product Quality Impact
Cell lysis during centrifugation is the primary product quality risk. When mammalian cells pass through the feed zone of a disc-stack centrifuge, they experience energy dissipation rates (EDR) of 105 to 106 W/kg, sufficient to rupture 5 to 15% of cells depending on their viability at harvest. Lysed cells release intracellular host cell proteins (HCP), DNA, and proteases into the centrate.
The consequences cascade through the downstream process. Additional HCP increases the chromatography burden and can cause Protein A ligand degradation. Released proteases may clip the target antibody during hold steps. DNA forms viscous aggregates that blind depth filters and increase turbidity. Joseph et al. demonstrated that a scale-down shear device generating EDR matching the feed zone of a pilot centrifuge (~2.4 × 105 W/kg) accurately predicted these quality effects, enabling development teams to characterize the harvest risk before committing production material.
Several strategies reduce centrifuge-induced lysis:
- Hermetic feed design: Eliminates the air-liquid interface in the feed zone, reducing EDR by 30 to 50%. Alfa Laval CultureOne and GEA pathfinder designs use this approach.
- Lower feed rate: Reducing Q/Σ below 0.10 L/(h·m²) decreases feed-zone shear at the cost of longer harvest time.
- Higher harvest viability: Cells at >90% viability are more resistant to shear than cells at 70 to 80% viability. Harvesting earlier sacrifices some titer but preserves product quality.
- Reduced bowl speed: Lowering RPM decreases EDR quadratically but also reduces Sigma, requiring lower feed rates. A DOE screening the interaction between RPM and feed rate identifies the optimum.
- Back-pressure adjustment: Higher back-pressure on the centrate outlet reduces turbulence in the feed zone.
| Parameter | No Centrifuge Shear | 5% Additional Lysis | 15% Additional Lysis |
|---|---|---|---|
| Centrate turbidity (NTU) | 30–80 | 80–200 | 200–500 |
| HCP in centrate (µg/mL) | 200–500 | 500–1,500 | 1,500–5,000 |
| DNA in centrate (µg/mL) | 1–5 | 5–20 | 20–100 |
| Depth filter capacity (L/m²) | 200–400 | 100–200 | 50–100 |
| Protein A DBC impact | Baseline | −5 to −10% | −10 to −25% |
Vendor Platform Comparison
Three vendor families dominate commercial biologics centrifugation: Alfa Laval BTPX series, GEA Westfalia CSE series, and single-use systems from kSep (Sartorius) and Thermo Fisher (DynaSpin). The choice depends on scale, facility type (stainless vs single-use), and CIP/SIP requirements.
| Platform | Vendor | Type | Max Flow (L/h) | CIP/SIP | Single-Use |
|---|---|---|---|---|---|
| BTPX 205 | Alfa Laval | Disc-stack, intermittent | 2,000 | Yes | No |
| BTPX 710 | Alfa Laval | Disc-stack, intermittent | 10,000 | Yes | No |
| CultureOne | Alfa Laval | Disc-stack, hermetic | 1,200 | No | Yes |
| CSE 80 | GEA Westfalia | Disc-stack, intermittent | 4,000 | Yes | No |
| CSE 170 | GEA Westfalia | Disc-stack, intermittent | 8,000 | Yes | No |
| kSep 6000S | Sartorius | Fluidized bed | 600 | No | Yes |
| DynaSpin | Thermo Fisher | Single-use disc | 500 | No | Yes |
Stainless-steel disc-stack centrifuges require CIP with 0.5 M NaOH and SIP at 121°C between campaigns, adding 4 to 8 hours per turnaround. Their capital cost ranges from $200,000 to $1,500,000 depending on bowl size, but they are amortized over 15 to 20 years of multi-product campaigns. Single-use systems eliminate cleaning validation but have higher per-batch consumable costs ($2,000 to $8,000 per run for disposable inserts). For facilities running fewer than 50 batches per year at volumes below 2,000 L, single-use centrifugation can be cost-competitive when cleaning validation savings are included.
Scale-Up Calculator
Calculate scale-up parameters including constant P/V, constant kLa, constant tip speed, and constant Reynolds number for bioreactors from lab to production scale.
Integration with Depth Filtration
Centrifugation alone rarely produces centrate clean enough for direct capture chromatography. Even well-optimized disc-stack centrifugation leaves 50 to 500 NTU of turbidity in the centrate, composed of cell debris, colloidal lipids, and precipitated DNA. Capture chromatography columns require feed below 10 NTU to avoid fouling and protect dynamic binding capacity.
The standard industrial approach is a two-stage harvest train:
- Primary clarification: disc-stack centrifugation removes 95 to 99% of intact cells and large debris, reducing turbidity from 2,000+ NTU to 50 to 500 NTU.
- Secondary clarification: depth filtration (e.g., Millistak+ HC, Clarisolve) removes remaining submicron particles, DNA, and lipids, reaching below 10 NTU. Typical depth filter loading is 100 to 400 L/m², depending on centrate quality.
The connection between centrifuge and depth filter performance is direct: every percentage point of additional cell lysis during centrifugation reduces depth filter capacity by 10 to 30% because lysed cell contents create fine, incompressible particles that blind the filter. Operating the centrifuge at lower Q/Σ (cleaner centrate) saves money on depth filters.
For processes at smaller scales (below 200 L) or in fully single-use facilities, some teams replace centrifugation with primary depth filtration alone or with flocculation followed by depth filtration. At larger scales, the throughput advantage of centrifugation is decisive: a disc-stack centrifuge processes a 2,000 L bioreactor in 2 to 4 hours, while equivalent depth filter area would require 10+ m² of filter media.
Filtration Calculator
Size depth filters, sterile filters, and TFF membranes. Calculate filter area, loading capacity, and Vmax for bioprocess clarification and formulation.
Frequently Asked Questions
What type of centrifuge is used for large-scale biologics harvest?
Disc-stack centrifuges with intermittent solids ejection are the standard for large-scale biologics harvest above 500 L. They process 500 to 10,000 L/h, handle 1 to 10% solids, and achieve 95 to 99% cell removal. Alfa Laval BTPX and GEA CSE series are the most widely installed platforms in commercial mAb manufacturing. For smaller scales or single-use workflows, fluidized-bed devices such as the kSep system offer gentler, enclosed separation.
How do you calculate the Sigma factor for a disc-stack centrifuge?
The Sigma factor for a disc-stack centrifuge is Σ = 2πnω²(Ro³ − Ri³) / (3g·tanθ), where n is the number of discs, ω is angular velocity in rad/s, Ro and Ri are the outer and inner disc radii, θ is the half-angle of the discs from vertical, and g is gravitational acceleration. This gives the area in m² of a gravity settler that would provide the same separation performance.
How do you scale up centrifugation for bioprocess harvest?
Maintain a constant ratio of feed flow rate Q to equivalent settling area Σ between lab and production centrifuges. If a lab centrifuge (Σ = 200 m²) clarifies at 20 L/h (Q/Σ = 0.1), a production centrifuge (Σ = 30,000 m²) can process 3,000 L/h at the same Q/Σ. Apply an efficiency correction of 0.4 to 0.7 for non-ideal flow in disc-stack centrifuges.
Does centrifugation damage cells and affect product quality?
Yes. The feed zone of a disc-stack centrifuge exposes cells to energy dissipation rates of 105 to 106 W/kg, lysing 5 to 15% of cells. This releases HCP, DNA, and proteases into the centrate, increasing downstream chromatography burden and potentially causing product clipping. Hermetic feed designs and lower feed rates reduce shear-induced lysis by 30 to 50%.
What is the difference between a disc-stack and tubular bowl centrifuge?
A tubular bowl generates higher g-force (up to 60,000 g) for fine separations but has limited solids capacity (2 to 5 kg/cycle, manual cleanout), suited for lab and pilot scales. A disc-stack has lower g-force (5,000 to 15,000 g) but much higher throughput (500 to 10,000 L/h) with automatic solids ejection, making it the production standard for biologics harvest.
Related Tools
- Centrifugation Scale-Up Calculator — Sigma factor, Stokes velocity, Q/Σ ratios for disc-stack, tubular bowl, and decanter centrifuges
- Scale-Up Calculator — Bioreactor scale-up by constant P/V, kLa, tip speed, or Reynolds number
- Filtration Calculator — Depth filter, sterile filter, and TFF membrane sizing
References
- Joseph A, Kenty B, Mollet M, et al. A scale-down mimic for mapping the process performance of centrifugation, depth and sterile filtration. Biotechnol Bioeng. 2016;113(9):1934–1941. doi:10.1002/bit.25967
- Joseph A, Goldrick S, Mollet M, et al. An automated laboratory-scale methodology for the generation of sheared mammalian cell culture samples. Biotechnol J. 2017;12(5):1600730. doi:10.1002/biot.201600730
- Shukla AA, Thömmes J. Recent advances in large-scale production of monoclonal antibodies and related proteins. Trends Biotechnol. 2010;28(5):253–261. doi:10.1016/j.tibtech.2010.02.001
- Ambler CM. The theory of scaling up laboratory data for the sedimentation type centrifuge. J Biochem Microbiol Technol Eng. 1959;1(2):185–205. doi:10.1002/jbmte.390010206