Pump Transients and Reverse Flow: 12 Engineer Questions Answered

Answered by Trey Walters, P.E. — Principal R&D Engineer, Datacor. Founder of Applied Flow Technology, author of AFT Impulse, ASME Fellow, member of the Hydraulic Institute water hammer committee. 40 years in thermal fluid systems engineering.

And Nelius Joubert — Principal Process Engineer, Sasol. Subject matter expert in process hydraulics, pumps and compressors, 33 years in heavy industry.

When a pump trips, flow reverses through it. Whether that reverse flow damages the pump, the check valve, or neither depends almost entirely on how fast the discharge valve closes. 

These questions came from engineers attending a webinar on a transient study at Sasol’s Secunda petrochemical plant, where the answer turned out to be a 6–12 second closure window.

The system these answers reference

Source: Joubert, N.J. & Walters, T.W. (2026), Pressure Surges 15, pp. 97–109.

How fast should a pump discharge check valve close?

Fast enough to stop reverse rotation before the pump restarts, slow enough not to exceed the valve’s design torque. At Sasol Secunda the safe window was 6–12 seconds. Closing in under 6 seconds exceeded the valve’s 35,500 Nm torque limit at 520 kPa differential pressure. Closing over 40 seconds let the pump reach 87% reverse speed before its programmed restart.

Is faster check valve closure always safer?

No. Faster closure protects the pump but loads the valve. On large valves the forces become the limiting factor. As Trey Walters put it during the session:

“Usually yes. But these valves are almost 2 meters in diameter. The forces are huge.”

Below a certain closure time, disc slamming and seat damage replace reverse rotation as the dominant failure mode.

Can a fast-closing check valve rupture the piping?

Yes in general, but not in every system — surge protection determines where the damage lands. Nelius Joubert:

“In other pump systems the spike can be much higher and it can be causing catastrophic rupture. In our case, although it could damage the valve, it wasn’t sufficient to damage the pipes.”

At Secunda a 50 m surge tower absorbed the pressure spike in all scenarios tested.

Does a check valve cause transient forces when flow is still moving forward?

Yes. Valve movement generates the forces, not flow direction. Trey Walters:

“If the valve position is changing, whether the flow is forward or backward, there will be forces on the piping assembly.”

How long does a tripped pump keep spinning in reverse?

Minutes, not seconds — and transient simulation predicts it accurately. At Secunda, measured against a stopwatch on the real pumps: reverse rotation began 4 seconds after the trip (AFT Impulse predicted 3.8 s), and the pump came to a complete stop after 4 minutes 30 seconds (predicted 4 min 31 s). The reverse-rotation behavior comes from four-quadrant pump data combined with rotating inertia.

Would adding a flywheel prevent pump reverse rotation?

Not on large pumps — they already carry enough inertia that adding more changes little. The Secunda rotating assembly carries 2,933 kg·m², and the reversing water column still stops it within 4 seconds. Nelius Joubert:

“I doubt it, because the size of these pumps already give it a large inertia.”

Note: rotating inertia is one of the largest uncertainty terms in a pump trip calculation. Sasol had it on the pump datasheet, which Walters called “highly unusual” — most engineers must estimate it.

Which quadrant does sustained reverse flow fall into?

Quadrant 3 — negative flow with negative rotational speed. A brief pump trip transits Q1 → Q2 → Q3 in seconds. A sustained drain-down, such as a slurry column returning from 140 m over 35 minutes, becomes a steady operating point in Q3 rather than a transient through it. Trey Walters:

“I am not an expert on this, but I understand the possibility of runaway conditions can occur in this kind of sustained reverse flow.”

Does AFT Impulse include pump data, or must you supply your own?

Both. For manufacturer performance data, Impulse connects to an online pump manufacturer database; for an installed system you enter datasheet values directly or import from Excel. For four-quadrant data — needed when a pump leaves its normal operating zone — Impulse ships with 26 published datasets drawn from decades of pump testing, selected by matching specific speed.

Can Impulse model column separation and vacuum during a pump trip?

Yes, with the caveat that column separation is inherently uncertain to model. Trey Walters:

“It’s a very uncertain area of modern science, but Impulse has the best models to do that.”

Column separation did not occur at Secunda because the 50 m surge tower kept the supply header above vapor pressure in every scenario.

Does pipe roughness affect water hammer results?

Not directly in large systems — but it shifts the starting conditions over time. Roughness is part of what eventually stops the flow and rotation, and changing it won’t meaningfully move the transient result at Secunda’s scale. The long-term effect matters more. Trey Walters:

“If the roughness changes enough — let’s say there’s some scaling, that’s very common in raw water intake systems — over time you get a degradation in flow. So you’re shifting your operating point on your pump curve. And so your transient actually starts at a different point.”

What if a cooling water system has no check valve by design?

There is usually still a valve closing somewhere, so reverse flow remains bounded. Trey Walters:

“This is very common in power plant condenser cooling water systems… In both cases, usually the motor-operated pump discharge valves do close — typically over 45–90 seconds. So the reverse flow stops at some point.”

Two cases apply: a single pump trip while others keep running, and a total trip where gravity pulls water back.

Is there a water hammer case study for a fire protection system?

Yes, for a nuclear power plant fire protection network — see Waterhammer Simulation and Mitigation for a Fire Protection Network at a Nuclear Power Plant. No petrochemical plant firefighting example is currently published.

Sources and further reading

• Sasol case study: Modeling Pump Transients With Reverse Flow

• Transient Evaluation of Pump Trips and Non-Return Valves in a Large Cooling Water System

• Practical Guidance on Modeling Pumps Under Transient Conditions

• Joubert, N.J. & Walters, T.W. (2026), “Transient evaluation of pump trips and non-return valves in a large cooling water system,” Pressure Surges 15, BHR Group, pp. 97–109.

Working through a transient problem of your own?

While most of these answers are general for all systems, some may be specific to the Sasol design. If you have specific questions regarding your design, request an Impulse demo or talk to an engineer to see what is right for your application.