Vertical pumps suit a diverse range of industrial applications for services ranging from low to high pressure, say, from 0.3 barg to 700 barg, and cryogenic to high temperature, e.g., -190°C to +500°C. Moreover, they cover a wide range of flow rates, roughly from 6 to 60,000 m3/h. The pumps contain from 1 to 50 stages and run at speeds from 180 to 20,000 rpm. They come in numerous varieties.
A vertical wet-pit pump can handle applications involving clean or lightly contaminated liquids. The drive motor is mounted directly above the pumping head, which is immersed in the liquid. Such a pump has a common discharge and support column. It has a vertical line shaft with the slide bearings typically lubricated by the pumped liquid.
A vertical sump pump is used primarily in services with lightly contaminated liquids, slurries and liquids containing solids. It also has a motor mounted directly on top of the pumping head, which is suspended in the pumped liquid. Such a pump has a separate discharge pipe and support pipe column. It has a vertical line shaft with slide bearings or cantilever design without slide bearings.
Submerged pumps are the other major category of vertical pumps but there are other types and categories as well.
Bearings and the lubrication system significantly impact the pump’s reliability, maintenance and long-term operation. Indeed, bearings and their lubrication usually account for most problems. So, specify all bearings according to API-610 whether the pump is an API type or not. Likewise, insist upon a sophisticated lubrication oil system as per API-610 or via a robust specification (preferably with two lubrication pumps, dual filters, etc.) for any large vertical pump.
Thrust bearings are critical components. They should be designed to carry the weight of all rotating parts plus the hydraulic thrust (both steady state and dynamic forces) and should be readily accessible for inspection. The hydraulic design of the pump should ensure that residual loads on the thrust bearings are always downwards.
Because vertical pumps commonly are direct driven, couplings should have provisions to secure the hub to the shaft with minimal dismantling.
Vertical Pump Dynamics
A vertical pump, particularly a suspended unit, usually is a flexible system that runs at a speed located between or around natural frequencies. This makes it susceptible to resonant vibration if its separation margins and dynamic situation aren’t verified during design and installation. Preferably maintain a 10–15% margin of separation between any natural frequency of the support structure and operating speed (and its first few harmonics).
In case of high vibration, it may be possible to rebalance the rotor assembly on site. However, in the case of resonance, this usually won’t markedly reduce the vibration level. Furthermore, the balancing planes probably aren’t easily accessible on site and rebalancing has some limits. The unbalance condition somewhat depends upon temperature and other operating conditions. So, expect increased vibration levels during transient conditions, due to fouling of a pump impeller or for other reasons. Also, any changes to the pump, such as maintenance work on the electric motor, coupling, pump rotor, etc., can affect vibration level.
Basic structural elements typically include the foundation, pump structure and electric motor frame. Strive to use accurate foundation data in a dynamic analysis. However, the deflection of the foundation typically represents less than 5–8% of the total deflection of the structural elements. Therefore, if accurate foundation data aren’t available, use estimated or rough data.
Because of its flexible structure, a large vertical pump can get excited by random broadband turbulences. However, vibration levels due to turbulent excitation usually pose no major concerns.
Large vertical pumps located side by side can be excited by the foundation structure to some extent. Some cross-couplings of vibrations between pumps can occur via the foundation or soil; measurements can detect these. If manufacturing and construction tolerances are controlled, identical pumps will exhibit only minor difference in vibration characteristics. In other words, natural frequencies usually differ by approximately 0.5–1.5 Hz. Slightly different speeds and natural frequencies of identical pumps occasionally result in moderate beat phenomena.
Operation may change the natural frequency of a vertical pump system. For instance, liquid level can alter the natural frequency slightly. However, such a variation usually is small. As a very rough indication, you can expect 0.05–0.4-Hz higher bending frequency at low liquid level relative to high liquid level. Also, the natural frequency often increases when a vertical pump is stopped (because, for example, the pump column isn’t full of liquid). Such a change in the bending natural frequency might be on the order of 0.1–0.6 Hz.
A vertical pump, particularly one with a long shaft, has relatively large inertia in the driver and pump stages and is susceptible to torsional excitations. Therefore, it requires a complete torsional study.
Any critical vertical pump package demands careful condition monitoring. So, specify electric motor winding temperature sensors, an X-Y vibration probe for each bearing as well as an axial probe and key phasor for each shaft. In addition, insist upon an accelerometer at the pump casing to monitor the casing vibration — particularly for a large vertical pump with flexible structure. Ensure the number and location of vibration sensors will suffice for machinery diagnostics.
Plants often operate vertical pumps in parallel. For such operation, the “head versus flow” curves of the pumps ideally should match within 2–3% between the minimum continuous flow and the end of curve flow. Always check that all combinations of parallel running are efficient for long-term operation without any performance or reliability issues.
Parallel operation of pumps can be risky; indeed, it often causes problems. Use of dissimilar pumps poses a particularly challenging situation. In such cases, you should match the head at the rated point of each pump (not at the same volumetric-flow-rate points). The surrounding area of rated points should agree within certain limits (say, 3–4%). Check with the pump manufacturers about whether the expected combinations of parallel operation raise any issues. In addition, plot the operating points for all scenarios and compare them between the minimum continuous flow and the end of curve.
When two pumps are running in parallel and providing a total flow greater than either pump could produce individually, if one pump trips off for any reason it’s extremely important that the other pump doesn’t trip off on overload. Indeed, such a requirement is common and well established for many critical applications. This mandates the sizing of the electric motor drivers for the end of pump curve. If the electric motors weren’t sized for the end of curve, then the remaining pump would overload and trip off, ending all liquid flow. In the case of critical pump services, this would result in a disastrous and costly shutdown.
There’s also another economic driver for sizing the electric motors for the end of the curve — maintenance When a pump requires service, it’s desirable to maximize the flow from the remaining pump even if this puts that pump at a flow rate outside its preferred operating range. The plant may need to run the pump at that rate for a relatively short time to capture a high unit incentive or get through an emergency situation.
Not all services require sizing the electric motor driver for the end of curve. So, assess this case by case. Many process pumping services and some critical utility applications do demand sizing the driver for end-of-curve operation (with ample margins), though. A design team may decide to deviate slightly from this requirement based on the results of thorough studies and simulations of many different operating cases. However, I strongly recommend sizing the electric motor based on the end of curve. Because such sizing usually must be done in the bidding stage, you either should follow that guideline or conduct a full study then. Finally, it’s better to slightly oversize the pump driver than slightly undersize it.
Some pump manufacturers now don’t show the pump performance curve beyond 120% or 125% of the rated flow; I’ve recently seen such truncated curves from four vendors in different services. The main purpose is to avoid sizing drivers based on the above rule. In such cases, a major question is “where is the end of curve?” The curve usually should end at the point where the efficiency drops. The end of curve generally isn’t a point where the pump can operate continuously; therefore, for many pumps, the point of “maximum continuous flow” can’t be the real end of the curve.
Usually, a pump’s head-versus-capacity curve ends on the far-right side of its best efficiency point (BEP), say, roughly at 135%, 140% or 150% of BEP flow, where the pump shows relatively low efficiency or starts to cavitate or net positive suction head becomes a limiting factor. Obviously, points with efficiencies close to that at BEP or slightly lower can’t be by any measure the end of the curve.
Unfortunately, the consequence is that many vertical pumps have undersized electric motor drivers. Some vendors resort to any trick possible — cutting the curves, misusing and misinterpreting API guidelines or project specifications, etc. — to employ the smallest possible (and definitely undersized) driver.
Based on my experience, you can address this issue and get proper electric motor sizing in one of two ways:
1. Forcefully ask the vendor to extend the curve to show the real end and then insist upon sizing the driver based on the end of curve.
2. If a vendor won’t agree to option 1, define the point of 120% (or 125%) BEP flow as one of rated points. This automatically results in motor sizing for the power required in this newly defined rated point using all project and code margins (for example, 10% API margin).
Let’s now look at two cases — both involving multiple wet-pit vertically suspended diffuser-type pumps.
Seawater pumps. In our first case, two vertical pumps were operated in parallel to provide the 11,000-m3/h seawater required by a chemical complex. Each had a pump capacity of 5,500 m3/h and a discharge pressure of 3.5 barg, and used a direct-drive vertical electrical motor to provide a speed of 750 rpm. Each is a single-casing diffuser pump with discharge through the column. The discharge nozzle of each pump is 36 in. with an ASME B16.5 150-lb flange. The overall vertical pump train exceeds 12 m.
The BEP flow of each pump is around 6,300 m3/h; so, the rated-point flow is about 87% of the BEP flow. The end of the curve occurs at a flow of about 8,200 m3/h and a discharge pressure of 2.4 barg; these values represent around 130% of BEP flow (or 149% of rated flow) and 69% of the generated pressure, respectively.
The estimate for head rise to shutoff is around 28%, which is suitable for this service. The rated power for the rated point is around 760 kW. The electric motor is sized for the end of curve; therefore, it’s rated for 920 kW. The pump efficiency at the rated point is estimated to be around 84% which is relatively high for a vertical pump.
The selected impeller diameter is 720 mm; the maximum impeller diameter possible for the selected pump model is around 760 mm, enabling a 5.5% increase in diameter if required in the future (e.g., for a system revamp, plant expansion, malfunction, seawater consumption rise, etc.).
Three parallel pumps. Here, the design team didn’t perform an accurate dynamic analysis; later it was found that the pumps have bending natural frequencies just below the rotating speed. Each pump is expected to have two distinct natural modes at each natural frequency because of slight asymmetry. The pair of bending natural modes for these vertical pumps occurred at about 88% and 98% of the normal speed — the 2% separation margin was far too small. Therefore, the unavoidable unbalance of the pump rotor combined with the nearby natural frequency resulted in increased vibration levels. The sad reality is this can happen to many vertical pumps.
When the pumped liquid level was low, natural frequencies increased somewhat and shifted closer to the pump speed (making the margin less than 2%), leading to a slightly higher vibration response. The pumps were rebalanced on site to reduce the excitation. However, this didn’t improve the separation margin of the natural frequencies and, so, vibration levels remained unsatisfactory. Solving the issue required some pump modifications to increase the separation margin.
AMIN ALMASI is a rotating equipment consultant based in Sydney, Australia. Email him at firstname.lastname@example.org.