Scale up at constant residence time and bed height

You have a method that works on a small column. Now you need more product per run. The goal is a larger column that behaves like the small one: same binding, same separation, same step lengths in column volumes (CV). This tutorial shows the arithmetic, the assumptions behind it, and the checks you still need to run.

1. Scale by column volume and residence time

Copying the flow rate in mL/min to a bigger column does not keep the method. The same mL/min through a wider column means a lower linear velocity and a much longer residence time. Loads, washes and gradients written in mL would also cover a smaller fraction of the larger bed.

Instead, write the method per column volume and keep the residence time τ = CV / Q:

• Flow: Q₂ = Q₁ × CV₂ / CV₁.
• Load: keep the load per mL of bed. Load₂ = Load₁ × CV₂ / CV₁, at the same feed concentration.
• Gradients, washes and equilibration: keep the same number of CV. The volume in mL scales with CV.

The usual route keeps bed height h and increases the inner diameter. Then CV₂ / CV₁ = A₂ / A₁ = (d₂ / d₁)², and keeping τ also keeps the linear velocity u = Q / A × 60. Vendor handbooks list exactly this: keep bed height, flow velocity, sample concentration and gradient volume in CV; increase column diameter, flow rate and load.

Every step then takes the same time at both scales, because a step of n CV at flow Q lasts n × CV / Q = n × τ. Process time per cycle does not grow with scale, apart from the effects in section 4.

Worked example

Small column: 1.0 cm inner diameter, 10 cm bed, 2.6 mL/min. CV₁ = π × 0.5² × 10 = 7.85 mL, u = 199 cm/h and τ = 3.02 min. Large column: 2.6 cm inner diameter, same 10 cm bed. CV₂ = 53.1 mL, a CV ratio of 6.76. The large-scale flow is 2.6 × 6.76 = 17.6 mL/min, at the same 199 cm/h. A 50 mL load becomes 338 mL. A 20 CV gradient grows from 157 mL to 1062 mL and takes 60.4 min at both scales. These are illustrative numbers and the calculator defaults; check your own resin limits.

2. What changes with a different bed height

Bed height is not always fixed. Development is often done on short beds, and a process column may be packed taller to fit a given resin volume into an available diameter. If you keep τ but change h, the linear velocity changes by the same factor: u₂ = u₁ × h₂ / h₁. A bed twice as tall runs at twice the velocity.

One vendor data file states that, for its resins, binding capacity stays the same as long as residence time is constant, even when development used shorter beds. Treat this as something to confirm for your resin and feed, not as a general law.

Pressure is the cost. For laminar flow through a packed bed, the Kozeny–Carman equation (a form of Darcy’s law) gives a pressure drop proportional to velocity, viscosity and bed height:

Δp = k · u · µ · h · (1 − ε)² / (d_p² · ε³),

with particle diameter d_p, bed void fraction ε and a constant k. So Δp ∝ h · u. At constant τ, u is proportional to h, so a rigid bed would see Δp grow with h². Doubling the bed height roughly quadruples the bed pressure drop. This is our arithmetic from the equation. It assumes the same resin, packing and liquid, and a bed that does not compress. Real agarose-based beds compress under flow, which lowers ε and raises Δp further.

3. Wider columns lose wall support

In a narrow column, friction against the wall carries part of the bed’s weight and drag. In a wide column, most of the bed is far from the wall. The resin there compresses more at the same velocity, and pressure drop rises more than a lab column predicted. Pressure-flow studies on compressible media show that the column aspect ratio (bed height over diameter) changes pressure drop, and that unexpectedly high pressures can result in manufacturing.

So a velocity that runs fine on a 1 cm column is not proven safe on a 30 cm column, even at the same bed height. Use the resin’s pressure-flow data for large columns, or measure a pressure-flow curve in the packed column. This is one reason process beds are often kept to moderate heights.

4. Extra-column volume does not scale

The system adds volume outside the bed: tubing, valves, mixer, detector cell and fraction-collector delay. That volume belongs to the instrument, not the column. Its effect shrinks relative to a larger column. A system volume that is 10% of a 1 mL column is 0.1% of a 100 mL column. That is our arithmetic, but the consequences are practical:

• System mixing and dispersion act on a fixed volume, which is a larger share of a small column’s volume. From that ratio we expect that gradients and peaks can look different at the two scales even when the column behaves the same. This is an inference, not a measured result; overlay both runs on a CV axis to see whether it matters for your systems.
• Delay volumes for gradients and fraction collection change when you move to a different system. Measure them again, as in the dynamic binding capacity tutorial.
• Time to fill and flush lines adds to process time, and it does not scale with CV either.

5. Verify at the new scale

1. Confirm the operating window. Check the large-scale velocity and expected pressure against the resin and column limits, with your buffer at its run temperature. Use the flow and residence-time converter for the numbers.
2. Check capacity at the residence time you will run. If τ changed, or the bed height changed, measure dynamic binding capacity at the new τ before fixing the load. Follow the DBC tutorial.
3. Check the packed column. A larger column must be packed well; test its efficiency and peak shape before loading product. See fundamentals for plate count.
4. Compare a run with the small scale. Overlay the chromatograms in Chromatobench; divide volumes by CV to compare scales. Compare recovery and pool purity with assays.

Common pitfalls

• Copied mL/min: the residence time goes up, the run gets slower, and the separation is not the one you developed.
• Copied mL volumes: a 20 mL gradient that was 20 CV on a 1 mL column is under 1 CV on a 25 mL column.
• Taller bed, same velocity: keeping u while raising h lengthens τ. Every step gets slower, and capacity measured at the old τ no longer applies; see the DBC tutorial on residence time.
• Taller bed, same τ: u rises with h and pressure faster still. Check the limits.
• Different resin particle size at scale: pressure and resolution change. The calculator assumes the same resin.

Sources and scope

This is an editorial workflow. The worked example, calculator defaults and figure are original teaching material, not data for a named resin. The h² pressure estimate and the dead-volume arithmetic are our derivations from the stated equations.

• Cytiva™, Ion Exchange Chromatography: Principles and Methods (handbook) — Table 2.3 and the scale-up steps: keep bed height, flow velocity, sample concentration and gradient volume in CV; increase column diameter, flow rate and sample load. Appendix 5 gives the cm/h to mL/min conversion.
• Cytiva™, Capto™ SP ImpRes and Capto™ Q ImpRes data file — vendor statement that scale-up usually keeps bed height and flow velocity, and that binding capacity for those resins is kept when residence time is constant, even if development used shorter beds.
• Erickson (2019), Chromatography Scale-up: Don’t Get Tied Down By Bed Height, Pharmaceutical Online, by a Cytiva™ system engineer — argues for scaling at constant CV/h (residence time) rather than fixed bed height to fit the column to the needed volume.
• Cytiva™, Develop your column packing methods based on pressure-flow measurements (procedure) — the Kozeny–Carman equation as written above, loss of wall support in wider columns causing more compression and pressure drop, and the need to measure pressure-flow behaviour for compressible resins.
• Stickel and Fotopoulos (2001), Pressure-Flow Relationships for Packed Beds of Compressible Chromatography Media at Laboratory and Production Scale, Biotechnology Progress 17, 744–751 — aspect ratio and wall effects change pressure drop in agarose beds and can give unexpectedly high pressures at manufacturing scale. Cited from the abstract.
• Fancy et al. (2025), Unlocking efficiency in column chromatography with packed bed supporting inserts, Frontiers in Bioengineering and Biotechnology 13, 1613174 — lack of wall support compresses resin near the column centre; bed height is typically limited to about 30 cm to avoid compression. Open access, CC BY.

For a textbook treatment of process development and scale-up, see Carta and Jungbauer, Protein Chromatography: Process Development and Scale-Up (Wiley-VCH, 2020). Listed as further reading; no statement on this page depends on it. Use the instructions for your resin, column and system for all operating limits.