A bigger tank changes the clock and the power bill.
Scale a tank and its stirrer together. Hold blade speed, power per litre or blend time constant and compare what each choice costs.
Enlarge for bigger labels, then swipe sideways →
A 5.0× wider tank holds 125× as much liquid. Holding power per litre gives 2.92× blend time, 1.00× power per litre and 125.00× total power. Energy per litre through the blend is 2.92× because it includes the running time. Move elapsed time to compare equal-turn endpoints.
What this model assumes
Fully turbulent, low-viscosity, single-phase liquid blending in geometrically similar tanks with identical impeller type, relative impeller size, baffles and a fixed homogeneity criterion. With constant power number and density: P = Np ρ N³ D⁵; V ∝ D³; P/V ∝ N³ D²; blade-edge speed ∝ N D; blend time θ ∝ 1/N. All outputs are ratios to a small tank. Elapsed time is in small-tank blend times; 100 turns to blend and the linear dye-progress picture are invented to illustrate equal Nθ, not measured kinetics or a tracer simulation. Stirrer positions track those toy turns. Constant power per litre (W/L) is not constant energy per litre (J/L). Energy delivered through blending is (P/V)θ, shown separately. Real viscosity, gas, solids, geometry changes, multiple impellers, shear and heat/mass transfer can invalidate this model. No motor size or actual blend time is predicted.
Equal blend time can demand much more power per litre.
For the simplified turbulent case, the same tank geometry and impeller type need the same dimensionless number of turns to reach a fixed blend criterion. Slower stirrer speed therefore means a longer blend time. The dye picture uses an invented 100-turn endpoint to make that relationship visible.
Let the diameter ratio be λ. Geometric similarity gives a volume ratio λ³. With density and power number fixed, P ∝ N³D⁵, so power per volume scales as N³D². At equal blade-edge speed, N scales as 1/λ: blending takes λ times as long and power per litre falls to 1/λ.
At equal power per litre, N scales as λ^(−2/3), so blend time scales as λ^(2/3). At five times the diameter, volume is 125 times larger and blend time is about 2.92 times longer. At equal blend time, N stays fixed and power per litre rises as λ²: 25 times at five times the diameter.
Power per litre is a rate, in W/L. Energy per litre delivered through a blend also includes running time, in J/L. In the five-times case, equal power per litre gives about 2.92 times the energy per litre through blending. The two quantities cannot be used interchangeably.
These ratios assume fully turbulent, low-viscosity, single-phase blending and unchanged geometry and impeller type. They do not establish solids suspension, gas dispersion, cell damage, a motor rating or a real batch’s blend time.
A tank is five times wider and keeps the same blade-edge speed. What happens to blend time in this model?
Taking a real process
beyond the lab?
Jon and I are building RARE Labs toward an autonomous pilot plant builder using AI and robotics. Join the waitlist for updates, and tell us what you’re trying to scale.
Go a little deeper.
- David S. Dickey · Chemical Processing (2005)Don’t get mixed up by scale-up ↗
Author’s engineering equations for geometric similarity, tip speed, power and power per volume.
- Muhammad Arshad Chaudhry · BioProcess International (2024)Lessons in bioreactor scale-up: introductory principles ↗
Worked 125-fold-volume comparison: about 3× and 5× mixing time, and 25× power per volume.
- Grenville & Nienow · Handbook of Industrial Mixing (2004)Blending of miscible liquids ↗
Further reading on turbulent blending; chapter text was not accessible during this lesson’s verification.