Mastering Agitator Design and Impeller Selection

Whether you are designing a continuous stirred-tank reactor (CSTR), scaling up a fermentation vessel, or establishing a bulk liquid blending operation, specifying the correct agitator is paramount. A misaligned impeller selection can lead to unmixed "dead zones", damaged product through excessive shear, or mechanically overloaded drive motors.

1. Principles of Industrial Mixing

Agitation is used to achieve phase dispersion, heat transfer, solid suspension, or miscible blending. Three timescales characterise mixing performance:

Flow Patterns

Every impeller generates one of three bulk flow patterns — axial (parallel to shaft, high circulation, suits blending/suspension), radial (outward to walls, high local shear, suits gas dispersion/emulsification), or tangential (circumferential rotation). Tangential flow is undesirable in low-viscosity service because it produces solid-body rotation and a surface vortex with negligible top-to-bottom turnover; it is only useful for very-high-viscosity fluids where close-clearance impellers deliberately use wall proximity to shear and fold the fluid. See the impeller table for flow-pattern assignments by type.

Impeller Selection Flowchart

Selecting an impeller involves matching the mechanical characteristics to the primary fluid property (viscosity) and the unit operation's main objective. Use this decision tree as a starting guide:

2. Common Impeller Types

Note: "Close-clearance" impellers cover a range of designs. Helical ribbons typically become advantageous starting around 20,000–50,000 cP, while anchors are best suited for the upper end of that range and beyond (>50,000 cP). The flowchart threshold reflects the point where close-clearance designs generally start outperforming standard turbine impellers.

Impeller-to-Tank Diameter Ratio (D/T)

$D/T$ is one of the most consequential sizing decisions: it influences power draw, pumping capacity, and the Zwietering $S$ factor (Section 5). Standard turbines run at $D/T \approx 0.3$–$0.5$ (smaller = more shear, larger = more flow for the same power); close-clearance impellers run at $D/T \approx 0.9$–$0.98$ where the narrow wall gap is essential to their function.

Multiple Impellers on a Single Shaft

When $H/T$ exceeds roughly $1.2$–$1.5$, engineers stack two or more impellers on the same shaft to maintain circulation throughout the full liquid depth. Total power does not add linearly — impellers spaced closer than $\sim 1$–$1.5D$ interact hydraulically and each draws less than its standalone value. Keep spacing above this minimum and verify combined power against correlation data.

3. Crucial Mixing Parameters: Power and Reynolds Numbers

The Impeller Reynolds Number ($Re$) sets the flow regime and therefore how $N_p$ is evaluated:

• $N_{Re} > 10,000$: Turbulent — $N_p$ constant.
• $10 < N_{Re} < 10,000$: Transitional — $N_p$ varies; read from correlation charts.
• $N_{Re} < 10$: Laminar — viscous forces dominate.

Rather than reading $N_p$ from a chart for each regime separately, a convenient single expression that spans laminar through turbulent flow is the two-asymptote model (Nagata, 1975):

Here $K_p$ is a laminar power constant specific to the impeller type (typical literature values: Rushton turbine $\approx 71.5$, 45° PBT $\approx 36.5$, hydrofoil $\approx 33.0$, anchor $\approx 220$–$400$, highly geometry-dependent on $D/T$ and wall clearance), and $N_{p,turb}$ is the constant turbulent power number from a table such as the one in Section 2. At low $Re$, the $K_p/Re$ term dominates and $N_p$ rises steeply as viscous effects take over; at high $Re$, $N_p$ flattens to the turbulent plateau $N_{p,turb}$. Taking the maximum of the two terms gives a smooth, continuous estimate of $N_p$ across the full Reynolds number range without needing to manually identify which regime applies — this is the basis used by automated calculators (including the CheCalc Agitated Reactor Calculator) to evaluate Eq. 2 directly from $Re$.

The constant-$N_p$ turbulent assumption holds only for properly baffled vessels. In an unbaffled tank, solid-body rotation and a surface vortex develop at high $Re$; the Froude Number $Fr = N^2 D/g$ characterises this effect, and $N_p$ becomes a function of $Fr$ as well as $Re$, so Eq. 2 underestimates mechanical loading. Baffles (see Common Mistakes) eliminate this dependency and are the standard assumption throughout this guide.

Calculating Required Power

Once the flow regime is established and $N_p$ is determined, the un-geared power drawn by the impeller is calculated:

Gassed vs. Ungassed Power

Equation 2 gives ungassed power. When gas is sparged below the impeller, bubbles cavitate behind the blades and the effective fluid density drops — gassed power $P_g$ is typically $40$–$60\%$ of $P$ for a Rushton at normal gas rates. Despite this, the drive motor must be sized for the ungassed startup condition; sizing only for $P_g$ is a common error that causes overload trips whenever gas flow is interrupted. The ratio $P_g/P$ depends on the gas flow number ($Fl = Q_g / N D^3$) and impeller type, and is read from published correlation charts.

Effect of Baffle Configuration on Power Number

The $N_{p,turb}$ values in the impeller table assume a fully baffled vessel (4 full-height baffles, 90° spacing). Apply the following corrections for other configurations:

• Standard (4 full baffles): No correction — use table values directly.
• Partial baffling (beaver-tail, glass-lined reactors): $N_p \times 0.75$.
• No baffles: $N_p \times 0.40$ (Rushton) or $\times 0.50$ (axial impellers). Lower power draw comes at the cost of poor circulation and vortexing — rarely a sound trade-off outside high-viscosity laminar service.

4. Case Studies: Power Calculation in Practice

5. Solid Suspension: The Zwietering Equation

One of the most frequent uses for an axial flow impeller (like a PBT or hydrofoil) is suspending solid particles in a liquid—such as catalyst beads or dissolving crystals. If the agitator runs too slowly, solids pile up on the tank floor forming a "fillet," meaning they are isolated from the reaction.

The critical speed required to just lift all particles off the bottom (so that no particle rests for more than 1-2 seconds) is called the Just Suspended Speed ($N_{js}$). The industry standard correlation for this is the Zwietering Equation:

Suspension Quality Above $N_{js}$

For processes where suspension itself is the goal (preventing settling), operating near $N_{js}$ with a modest margin is sufficient. For mass-transfer-limited reactions (dissolution, catalysis, leaching), uniform concentration throughout the vessel height matters — homogeneous suspension — which can require $1.5$–$2 \times N_{js}$, with a large power penalty since $P \propto N^3$. Confirm whether your requirement is "off the floor" or "uniformly distributed" before setting the design speed.

6. Mechanical Design: Shaft Sizing

Agitator shafts are subjected to both torsion (twisting from drag) and bending (from overhung impeller weight and hydraulic side loads). The torque ($T_q$) generated at the shaft is:

Sizing the shaft for this torque alone (pure torsion) would give an optimistic, undersized result, because it ignores the bending moment from the overhung impeller weight and hydraulic side loads discussed below. Rather than calculating the bending moment explicitly — which requires shaft length, bearing span, and impeller hydraulic force data that are often not available at the sizing stage — the standard industry approach (ASME B106.1M, also used by Paul et al., 2004) is the equivalent torque method: the actual torque is multiplied by a combined loading factor $K$ that acts as a conservative proxy for the unmeasured bending contribution.

The combined loading factor $K$ is selected based on the mechanical configuration, with commonly recommended values:

• $K = 1.5$: short shaft, single impeller close to the bottom bearing — minimal overhang and bending.
• $K = 2.0$: standard top-entry agitator configuration — the generally recommended minimum for most designs.
• $K = 3.0$: deep vessels or multiple impellers on a single shaft, where overhang and hydraulic side loads are larger.
• $K = 4.0$: long cantilevered shafts or configurations with high bending relative to torsion.

$K$ can be related back to the bending moment $M_b$ via the maximum shear stress theory, $K = T_e/T_q = \sqrt{1 + (M_b/T_q)^2}$ — so larger $K$ values correspond to bending moments that are large relative to the torque. Selecting $K$ is therefore a judgment call based on the expected geometry, and the CheCalc Agitated Reactor Calculator lets you choose $K$ directly (with $K=2.0$ pre-selected as the recommended minimum) rather than requiring a separate bending moment calculation.

Where the Bending Moment Comes From

The $K$ factor in Eq. 5 exists because real agitator shafts are simultaneously bent by two effects: the dead weight of the shaft and impeller(s) acting as an overhung cantilever from the gearbox bearing, and unsteady hydraulic side loads — particularly during startup, in unbaffled or partially-filled vessels, and for impellers mounted off-center. These loads are difficult to predict precisely, which is why they are captured indirectly through $K$ rather than calculated from first principles at the sizing stage.

A separate and equally important check is the shaft's critical speed — the rotational speed at which the shaft's natural frequency coincides with the operating speed, causing resonant lateral vibration ("shaft whip"). Long, slender, unsupported shafts (common in tall vessels or when a lower steady/stabilizer bearing is omitted) are especially susceptible. Good practice is to keep the operating speed below roughly $50\%$ to $60\%$ of the calculated first critical speed, which often becomes the limiting factor for shaft length and diameter in tall agitated vessels — even when the diameter from Eq. 5 satisfies the stress check. Critical speed depends on shaft length, support span, and diameter, and must be verified separately; it is not addressed by the $K$ factor.

Finally, the shaft diameter selected also affects the mechanical seal or packing at the point where the shaft penetrates the vessel. Mechanical seals in particular have limited tolerance for shaft runout and lateral deflection; a shaft that is technically strong enough for torsion and bending may still cause premature seal failure if deflection at the seal faces exceeds the seal manufacturer's allowable limits. Seal selection and shaft stiffness should therefore be considered together, not as independent decisions.

7. Scale-Up Principles

Taking a successful reaction from a 10-gallon lab pilot to a 1,000-gallon production tank is notoriously difficult. You cannot keep all geometric and dynamic parameters constant simultaneously. The primary scaling factor ($R$) is the ratio of the tank diameters, or roughly the cube root of the volume ratio: $R = (V_2 / V_1)^{1/3}$.

Engineers must choose a specific scaling criterion based on the most critical aspect of their process:

• Constant Power per Volume ($P/V$): Maintains identical mass transfer rates and gas dispersion. $N_2 = N_1 R^{-2/3}$.
• Constant Tip Speed ($\pi N D$): Maintains maximum shear rate, critical for sensitive crystallization or cell cultures. $N_2 = N_1 R^{-1}$.
• Equal Blend Time: Per the Ruszkowski (1994) correlation for turbulent baffled vessels, $\theta_{95} \propto 1/N$ for geometrically similar tanks — so maintaining the same absolute blend time requires the same speed ($N_2 = N_1$, i.e. $R^0$ exponent). This is typically impractical since the larger impeller at the same speed draws far more power ($P \propto D^5$). Equal blend time is therefore usually treated as a trade-off target and is most closely approximated in practice by the $P/V$ criterion.
• Equal Froude Number ($Fr = N^2 D/g$): Maintains the same surface vortex behavior and gravity-to-inertia force ratio, relevant mainly for unbaffled or partially baffled systems where free-surface effects matter. $N_2 = N_1 R^{-1/2}$.
• Constant Reynolds Number: $N_2 = N_1 R^{-2}$. Rarely practical, as it requires massive, slow-moving impellers that are not economically viable — the speed reduction is far steeper than any other criterion.
• Heat Transfer Scale-Up: For jacketed vessels, the impeller-side film heat transfer coefficient typically follows a relationship of the form $h \propto N^{0.67} D^{0.67}$ (the exact exponents depend on the specific Nusselt-number correlation used). This scales differently from both $P/V$ and tip-speed criteria, so a scale-up that preserves heat transfer performance will generally require a different speed than one that preserves mass transfer or shear — the two cannot both be held exactly constant.

In practice, a single process often has competing requirements — for example, a fermentation process needs adequate oxygen transfer (favoring constant $P/V$), gentle treatment of cells (favoring constant tip speed), and good jacket heat removal (favoring the heat transfer criterion above), and these three criteria rarely agree on the same production-scale speed. Real-world scale-ups are therefore usually a compromise: engineers calculate the speed implied by each relevant criterion, identify which one represents the actual limiting constraint for the process (often determined by pilot-scale experiments or prior plant experience), and may also consider changing the impeller type, diameter ratio, or number of impellers at the larger scale rather than relying on speed alone to satisfy every criterion simultaneously.

8. Common Mistakes to Avoid

Conclusion

Selecting the correct agitated impeller is a balance of understanding fluid rheology, identifying the correct macro-flow patterns, and running the hydraulic and mechanical sizing calculations. By systematically applying the Reynolds number to identify flow regimes, sizing the shaft using the equivalent torque method with an appropriate combined loading factor, and selecting the proper scale-up criteria, process engineers can prevent catastrophic mixing failures and design highly efficient tanks.