How Centrifugal Pump Curves Change With High-Viscosity Chemicals (And Why the Water Curve Lies to You)

The Water Curve Doesn't Survive Contact With Viscosity

With a centrifugal pump, high viscosity knocks down the water performance curve. The pump manufacturer's standard curve is normally based on water, and once you put a substantially more viscous chemical through that same pump, you should not expect it to make the same flow, head, or efficiency. (Source: KSB pump viscosity correction guidance)

Five Things Move, and Not in Your Favor

The easiest way to picture it is this. As viscosity goes up:

• Flow capacity goes down.
• Developed head goes down.
• Efficiency goes down — often substantially.
• The BEP (best efficiency point) moves down and to the left.
• Brake horsepower can increase, even though you're producing less useful hydraulic output, because you're spending more energy overcoming internal viscous drag.

That last point catches people. They'll take a pump curve and say "This pump only needs 7.5 horsepower on water, so we'll put a 10-horse motor on it." Then they put a thick polymer, glycol, oil, concentrated chemical, or slurry-like liquid through it and discover they've underestimated the motor load.

Shear and Friction Losses Eat the Useful Head

A centrifugal pump transfers energy by spinning the liquid through the impeller. With water, the liquid moves relatively freely through the impeller passages and casing. With a viscous fluid, you get greater shear and friction losses between:

• The liquid and impeller
• Liquid and casing
• Adjacent fluid layers
• Rotating and stationary surfaces

So more of the shaft energy becomes heat and internal loss instead of useful head. That's why efficiency is generally the first thing I expect to get hurt badly.

KSB, for example, shows an illustrative conversion at 500 cSt where a particular pump's correction factors were approximately:

Source: KSB.

Those are not universal correction factors — they depend on the pump, speed, specific speed, flow point, and viscosity — but they show you the magnitude of the problem. So a pump that looked excellent on a water curve might become a pretty lousy pump once the fluid gets thick.

You're Getting Squeezed From Both Directions

This is equally important. People sometimes correct the pump curve for viscosity but leave the system curve exactly where it was. Wrong. High viscosity also increases piping losses because Reynolds number drops and the friction factor changes. KSB specifically notes that the viscous fluid's kinematic viscosity must be used when calculating Reynolds number and pipe friction.

So you're getting squeezed from both directions:

• Pump curve comes down.
• System resistance goes up.
• Your actual operating point can move dramatically.

And when the flow becomes laminar, I wouldn't casually use the familiar assumption that friction head varies simply as Q-squared. The hydraulic behavior changes with Reynolds number.

There's a Viscosity Line Where I Start Paying Attention

KSB notes that noticeable centrifugal-pump curve effects generally begin around 20 cSt and above, after which empirical viscosity correction methods become appropriate.

That doesn't mean 19 cSt magically behaves like water. It means that around this region viscosity correction starts becoming significant enough that you shouldn't just read the manufacturer's water curve and call it done.

The standard engineering approach is to take the water performance and apply Hydraulic Institute viscosity correction factors — or a manufacturer-approved equivalent — to:

• Q (corrected flow)
• H (corrected head)
• Efficiency (corrected)

Then recalculate horsepower.

"600 cP" Means Nothing Without a Temperature Attached

This is huge with chemicals. If somebody tells me "The chemical viscosity is 600 cP," my next question is "At what temperature?" For many Newtonian liquids, viscosity can fall dramatically as temperature rises. (Source: KSB)

A product could be 1,000 cP at 40°F, 300 cP at 70°F, 100 cP at 100°F, depending on the chemistry. So I size the pump around the worst credible operating temperature, not somebody's room-temperature viscosity off a sales data sheet.

Polymers, Activated Polymer, and Sludge Deserve Another Warning

If we're talking about something like neat polymer, activated polymer, sludge, or another non-Newtonian fluid, then a simple viscosity correction may not adequately describe it. KSB notes that conventional viscosity correction methods apply to Newtonian fluids, while non-Newtonian behavior is more complicated.

That's where I want the actual rheology, shear rate, temperature, concentration, and preferably pump testing or manufacturer experience. At some viscosity level, I'll stop trying to force a centrifugal pump into the application entirely and start asking whether we should use a positive-displacement pump — progressive cavity, gear, lobe, peristaltic, diaphragm, etc. — because those machines generally handle viscosity fundamentally differently.

Never Select a Centrifugal Pump From the Water Curve

So my field rule is: Never select a centrifugal pump for a viscous chemical from the water curve. Get the chemical, viscosity at minimum/normal/maximum temperature, specific gravity, required GPM and TDH, pump speed and impeller diameter, then correct the pump curve and recalculate the system curve. That's when you find out what pump you've actually got.

Keep Going

• NPSH Available Less Than Required: Troubleshooting Guide →
• Browse Chemicals by Compatibility & Application →

Frequently Asked Questions