Reverse Operating Pumps Part 5

Welcome back to Pumping Corner for what I hope will be the final part of this series. It was originally intended to be a trilogy, but in keeping with my limited organizational and writing skills the subject (and you the readers), had other plans.

With that out of the way, let's conclude the work we started in Part 4 where we estimated the performance of a PAT (Pump As Turbine) when only the standard pump HQ curve was available. Here we'll talk about what other things need to be considered in order to ensure the machine will operate as intended.

Ok so I computed my PAT characteristic, now what ?

In Part 4 we went through a 5 step process to compute the expected characteristic of the PAT. Note that this only tells us what the expected performance will be without consideration of any other limiting factors. There are however a number of design constraints we need to check before we can consider the overall performance limits. The main ones are:

1. Power limits of the PAT due allowable stress
2. Radial and axial thrust
3. Rotor deflection
4. L10 bearing life
5. Bearing lubrication and cooling
6. Mechanical seal configuration
7. Maximum running speed - Overspeed

Power Limits and allowable stress

If I compare the power characteristics for the PAT selected in Part 4, I get the following

As can be seen from overlaying the power characteristics of the pump and turbine modes, operating in the turbine mode can result in significantly more power being transmitted through the rotor.

The maximum available head that the PAT will handle needs to be defined as this in turn defines the maximum power.

It is very important that the rotor be checked for torsional stress at the normal locations (typically the points of minimum diameter). The keys and keyways on the rotor require checking for shear and compressive stresses.

It is probable that some upgrading of the materials of construction will be necessary. Alternatively the rotor design can be modified although this is a more invasive, costly and time consuming approach.

Radial Thrust

From the studies I've seen, it is generally safe to assume that the maximum radial thrust coefficient in pump mode will not be exceeded in turbine mode. The one exception would be operation at flows very much higher than the pump BEP flow (typically >350%).

That said the characteristic of the radial thrust is very different and that does have implications for the flowrate at which bearing life is lowest and rotor deflection is highest. I've plotted the typical characteristic you would expect to see in a non dimensional form below (note this is for a typical impeller in the range of nq 10 to 50). While in pump mode, the highest radial thrust occurs at zero flow, in turbine mode the radial thrust is lowest at the flowrate corresponding to zero power output and then rises continually from there.

As can be seen it is only when the turbine flows exceed 3 times the pump BEP flow that we would start to need to consider a radial thrust coefficient higher than that normally seen in the pump.

However the resulting radial thrust is a function both of the radial thrust coefficient Kr and the differential head of the pump.

Because our differential head in turbine mode is higher, it is possible that the radial thrust will also be higher and this needs to be checked using the Kr curve characteristic shown above or from some other source.

Axial Thrust

Similar to the situation with radial thrust, the axial thrust maximum magnitude** in a PAT can be assumed to similar to that experienced in pumping mode.

(**for identical differential heads, so if your turbine has a higher differential head, your axial thrust will also correspondingly increase).

There is one notable exception to this and that is for radial or mixed flow pumps. This is because the momentum component is reversed in a PAT and for those machines this is a component of the resulting axial thrust.

Taking the standard calculation method from HI 14.3 (14.3.5.4.3.3.8 to be exact), we can compute the momentum component using the formula shown below.

I'm not going to do that here because in most cases the axial thrust due to momentum change is not significant. However if you have single stage machine with a small impeller eye diameter that results in fluid velocity through the eye >30 ft/s (10m/s), I recommend that it should be checked.

Where:

Fm = axial thrust in N or lbf (due to momentum change through the impeller)

s = fluid specific gravity

Q = fluid flow rate in m3/h or USGPM

D1 = impeller eye diameter in mm or inches

Rotor Deflection

As noted above in the section on radial thrust, the magnitude of this force in turbine mode can potentially exceed the values seen in pump mode.

In cases where the computed turbine radial thrust exceeds that of the pump, the rotor deflection must be checked. There isn't one single calculation method for this as it depends greatly on the type of pump being evaluated.

However in general

• Classically stiff rotors need to demonstrate that their deflection will not exceed the minimum radial clearances present in the pump
• Multistage rotors which rely on the Lomakin effect need to be evaluated in the running condition to confirm there is sufficient fluid support at the wear rings and bushings to prevent rotor to stator contact.

L10 Bearing Life

As noted above in the section on radial and axial thrust, the expected magnitude of these forces in turbine mode can in some cases exceed the values seen in pump mode.

Additionally, because the characteristics of force vs. flowrate are different, we need to evaluate whether the rolling element bearings have adequate life over the expected operating range. What I mean by this is the PAT designer should be aiming for an acceptable L10 life (typically >25K hours) based on the user's expected head and flow ranges. Outside of that range a lower L10 life may be acceptable depending on the frequency of excursion.

Computation methodology of the bearing L10 life is the same for turbine mode as for pump mode and hence I'm not going to cover it in detail.

The bearing selections should also be checked for maximum allowable operating speed based on the expected runaway speed of the turbine (see also the section below).

Bearing Lubrication/Cooling

The most obvious concern here is that in turbine mode the PAT rotor is rotating in the opposite direction to that seen in pump mode. Depending on the type of bearings fitted this may (or may not) be a problem. A few of the things to check are listed here:

1. If your PAT has a constant level oiler, it may need to moved to the opposite side of the bearing housing. Constant level oilers are sensitive to the direction of rotation as shown in the diagram below.
2. If you PAT utilizes oil ring lubrication, a check needs to be made that the oil ring will function correctly under reverse rotation. This will depend on how the internal features in the bearing are arranged to channel the oil thrown from the oil ring.
3. For pumps with forced lubrication a check needs to be made to confirm that the bearings can function with reverse rotation. The oil feed into journal bearings may be on the incorrect side. If you have optimized tilting pad radial or axial bearings these may have pivot points that only function correctly with a specific sense of rotation. Designs with directed oil feed such as LEG (Leading Edge Grooves) etc. will need to be configured for the reverse rotation.
4. Many optimized shaft driven cooling fans will only function with specific sense of rotation. If your pump has one of these it will be to be swapped out for a bi-directional fan or a fan made to suite the reverse rotation

Mechanical Seal configuration

Most simple (multi-spring) single mechanical seals can operate in either rotation. However when a gas seal or a seal with a pumping ring is utilized, these are often manufactured for one sense of rotation.

It normally is simply a matter of specifying the reverse rotation when purchasing the seals.

Overspeed

For those of you who have been following this reverse running pumps series, hopefully you will be aware that a PAT is susceptible to an overspeed condition in the case where there is a loss of load. This can happen due to generator malfunction, loss of grid connection or even failure of the coupling between the pump and generator.

Part 2 of the series describes the scenario where the pump will reach a runaway speed (overspeed) where the torque developed falls to zero.

Depending on the specific speed of the pump this runaway speed can be significantly higher than the normal operating speed (see the graph below taken from Part 1).

There are two ways this can be managed:

1. A separate safety system can be installed to detect overspeed and shut off the fluid source and/or apply a brake
2. The pump rotor system can be designed to manage the overspeed without failure (this includes looking at the lateral rotordynamics, bearing selection +lubrication and mechanical seals).

Both options have tradeoffs.

• A separate safety system will add significantly to the First cost and will require regular testing. It can also be subject to false trips.
• Designing the pump rotor system for the overspeed condition might require significant pump redesign (which would negate some of the advantage of a PAT over a bespoke turbine).

Ultimately each system is different and needs to be examined on a case by case basis during specification of the system.

My general guidance is that pumps with a specific speed lower then 30 metric (1500 in US units), which have a relatively low runaway speed can generally tolerate the overspeed without modifications, making option 2 the better choice.

Ok, that's (hopefully) the final time I need to talk about reverse operating pumps for a while. I hope you found the series useful. Links to the other parts of the series are shown below.

As always - all comments, questions and criticisms are gratefully received.

Until next time Beatus Centrifuga

• Reverse operating pumps Part 1
• Reverse operating pumps Part 2
• Reverse operating pumps Part 3
• Reverse operating pumps Part 4