Why Are There Holes in My New Impeller? Part 3

 

Common Pumping Mistakes

Open Impellers

True open impellers are not commonly used in industrial pumps because the demands of service require sufficient vane support (webbing or shrouds) to buttress the required torque loads and to maintain the relative geometric position of the vanes.

Be aware that sometimes the vanes on open impellers will move over time and service due to material stress relief issues and other operational force factors. This tendency to move (aka, spring) normally makes them unusable for robust applications. On the positive side, because there is no shroud or webbing either front or back on a true open impeller, there is little to no surface area for the pressure to act on, so minimal axial force is generated.

Semi-Open Impellers

American National Standards Institute (ANSI)-style (B73.1 / B73.2) and International Organization for Standardization (ISO)-style (5199/2858) pumps are single-stage end suction types that use semi-open impellers 98.5 percent of the time (author’s unsubstantiated estimate). Note that there are scores of end suction pump models using semi-open impellers that are not designed to ANSI or ISO specifications such as industrial stock and process pumps.

Semi-open impellers have a shroud on one side only. The simple geometry of this construction makes these impellers less expensive to manufacture. Semi–open impellers also possess the ability to pass through more and larger solids including stringy and fibrous materials than closed impellers.

Typically, an enclosed impeller is more efficient than a semi-open impeller of similar geometry, but it is possible for a semi-open impeller, held to a very tight casing clearance setting (aka, leakage) to be as efficient as a closed impeller, at least for that length of time before the effects of wear open the clearance.

High vs. Low Specific Speed Impellers

Axial flow pumps with impellers of high specific speed will not incorporate shrouds or web supports between the vanes and the impeller will appear more like a boat propeller than an impeller. Do not confuse these with open-style impellers in the low specific speed ranges.

The consequence of the semi-open design is an impeller with higher unbalanced axial forces when compared to a similar size in the fully closed designs. Under normal operating conditions, there will be a much higher force on the back of the impeller than on the front. The resultant force exerts in a direction toward the suction and the pump’s thrust bearing counteracts that force. Please note that even in normal operation, there can be momentary forces acting in either direction, so the thrust bearing should be designed/selected to handle forces in both directions.

A new medium-sized ANSI pump will develop upwards of 850 pounds of axial force and as the pump wears that force increases. The axial force will be reduced proportionally as the suction pressure increases. These factors and more must be considered when calculating the required bearing size and design life span. A typical ANSI pump thrust bearing is nominally designed for a minimum of 17,500 hours.

A pump designer could simply install bigger thrust bearings and not otherwise address the axial force, but the bigger bearings require bigger shafts and that requires bigger frames and housings—all of those things result in a pump that has a bigger initial cost and a higher maintenance cost.

The more practical method used to reduce the axial thrust on end suction pumps and semi-open impellers is to reduce the axial force on the rear shroud by reducing the pressure. This is why the incorporation of balance holes, pump out vanes or both comes into play as axial force reduction methods. Pump out vanes are a design compromise and so there is a small, but acceptable, trade-off with efficiency and power consumption when using this approach.

IMAGE 1: Pump out vanes are small vanes on the backside of the impeller shroud. (Image courtesy of the author)

Pump Out Vanes

Pump out vanes are sometimes simply referred to as POVs, expellers, back vanes or back ribs and are just what the name implies: they are small vanes on the backside of the impeller shroud (Image 1). Pump out vanes are probably the most cost-effective method of axial force reduction.

As previously stated, the main purpose for pump out vanes is to reduce the pressure behind the impeller. Reducing the pressure in this area reduces the axial force pushing the impeller toward suction. The function of pump out vanes is similar to standard impeller vanes in that they impart a velocity to the fluid contained in the annulus area behind the impeller.

As the impeller rotates, the liquid that is adjacent to the impeller shroud will also gain velocity due to the simple phenomena of disc friction. Adding the pump out vanes to the rear shroud simply enhances the desired effect by increasing the angular velocity of the liquid. Increasing the liquid velocity reduces the pressure on the shroud and, therefore, the magnitude of the axial force.

A.J. Stepanoff conducted research and developed formulas based on his experiments with pump out vanes in the 1950s and ’60s. He determined that the angular velocity of the liquid in the annulus area will be about 50 percent of the impeller shroud speed (and pump out vanes). Liquid near the shroud is moving at almost the same speed, but it slows down fairly quickly as you increase the distance from the shroud. Subsequent research has demonstrated that the accurate determination of the true angular velocity is more complicated than my statement, but the basic premise is sound.

One area of contention is the distance from the pump out vane to the stuffing box/seal chamber (think nearest obstruction). The distance defines the axial gap of the annulus chamber behind the impeller. As the impeller moves away from the stuffing box, whether through wear or operator settings, the resultant gap will at some point become too great for the pump out vanes to exert their full effect. Most pump experts agree a performance drop will start to occur when the gap begins to exceed 0.060 inches (1.5 mm).

The pressure-velocity relationship, aka, Bernoulli’s principle: The first law of thermodynamics deals with the conservation of energy and states that energy is always conserved and that it can neither be created nor destroyed, but (and this is the important part) it can be altered in form. A liquid moving in a pipe or a pump casing is in a simple pressure-velocity relationship (energy conservation). If the velocity goes up, the pressure will be reduced and vice versa.

For the general purposes of this column, you can consider the description of pump out vane performance as a simple example of Daniel Bernoulli’s equation and the pressure velocity relationship (with sincere apologies to my Applied Fluid Mechanics 301 professor, Dr. R.G.).

The efficiency and effectiveness of pump out vanes is a direct function of at least six design factors. The first four factors are listed in order of importance:

  1. The radius (length) of the vane
  2. The height of the vane
  3. The number of vanes
  4. The clearance between the pump out vane and the nearest obstruction (stuffing box/seal chamber)

How these factors affect the efficacy of the pump out vanes to reduce thrust is sometimes a subject for debate among pump designers, CFD software developers and engineering researchers. Overlooking some questionable portions of the debate, there remains many aspects of the concept that are agreed on, as follows.

The longer the vane (radius length), the more effective it is to increase the liquid velocity. The caveat is that the longer the vane, the more power is required and it is a function at the fifth power. The bonus is that the manufacturing process of casting the impeller with pump out vanes remains both easy and economical. The height of the vane as it stands out in profile from the shroud (aka, proud) will make it more effective, but at some point you must trade off pressure reduction with the incremental loss of pump efficiency and the inherent increase in power consumption.

The number of pump out vanes is a direct factor in pressure reduction. The more vanes added to the shroud, the more effective the axial force reduction, up to an optimum number of approximately six. Adding more than six vanes will, in most all cases, have no positive effect and so adding a seventh vane will usually be a wasted effort of neutral results. Adding an eighth pump out vane to the design will start to reduce the benefits.

Up for argument: Other pump out vane design factors are the width of the vane and the shape or geometry—think angle, placement and curvature with relation to shaft centerline, i.e., is the vane curved or straight? At this time, the definitive effect of these factors remains both untested and confusing for me. Lastly, often the pump out vanes are where they are simply for economy of production and/or sweeping debris from the space, versus an axial force reduction method.

An additional benefit of using pump out vanes is that the pressure in the seal chamber/stuffing box will also be reduced. This pressure reduction feature works to prolong the life and reduce related costs for the packing or mechanical seal. The reduced pressure effect can sometimes create unintended problems if the product has a high temperature or if the pump is in a lift situation (due to vapor pressure issues).

Either way, there is a higher probability of flashing the liquid to vapor and caution should be exercised. These issues are easily solved with the addition of a flush/seal support system, like a plan 11 or 32, and adding a throttle or restrictive bushing in the bottom of the seal chamber/stuffing box. The portion of the pump world that is governed by API 610 does not endorse pump out vanes as an acceptable method of axial thrust force reduction (API 610 11th section 6.7.1).

Pump out vanes do not increase leakage (bypass) as a balance drum or back ringed impeller would, and they also do not affect the resultant thrust as impeller wear ring clearances inevitably open with wear.

Caution: It is possible to axially adjust the pump shaft such that the pump out vanes are too close to the stuffing box/seal chamber and unintentionally create a vacuum in the annulus and stuffing box area. The vacuum will cause product flashing issues and dramatically shorten the packing or seal life (see remedy above).

Balance Holes

Some impellers will have both balance holes and pump out vanes, others just balance holes or just pump out vanes. Regardless, the reason for the balance holes remains the same, and it is another acceptable method to reduce the axial force and resultant thrust.

Similar to pump out vanes, the balance holes will reduce the pressure in the stuffing box and so, again, be vigilant of the product vapor pressure. It is recommended that you do not use impellers with balance holes in lift situations as the potential for liquid product flashing increases. A lift situation is defined as when the source of liquid being pumped is below the centerline of the pump impeller—that is, the impeller is not flooded. Should the pumped liquid have solids, suspended or entrained, there is risk of clogging the balance holes/passages. Flow from the balance holes will create some level of imbalance and disturbance at the impeller eye, which adds inefficiency.

An impeller with balance holes will make the pump slightly less efficient and require more horsepower to operate. Consequently, the pump is more expensive when calculating the total cost of ownership (TCO). The antithesis is that without the balance holes, the bearing life is shortened.

Operating Tip

The magnitude of axial thrust in a centrifugal pump is a direct function of the differential head. Thus, where the pump operates on its curve will have a direct effect. The further left you operate from the best efficiency point (BEP), the more axial thrust will be generated.

Maintenance Tip

In all of these examples of balancing axial force, it is always important to ensure the pump rotors and impellers are centrally located (both mechanically and hydraulically) in their respective casings/volutes and diffusors. Due to casting variances, machining practices and machine “stack tolerances” (cumulative effect), there may be instances where an impeller will require the addition or removal of a shim and/or some axial position adjustment like a shaft sleeve nut. The purpose of these features is to ensure the rotating mechanical/hydraulic center runs true to the static center of the volute or diffusor. On a new pump, these centering procedure steps were completed during the factory assembly by a knowledgeable technician, but this is also why many better pump designs offer axial adjustment features or suggest and/or support the possible use of shims as required on rebuilds and rotor replacements in the field.

Compromises: Grandma Said ‘You Can’t Get Something for Nothing’

The most efficient and direct method for axial thrust compensation is to simply transfer it to the thrust bearing, which will work in smaller pumps but becomes proportionally expensive as the pump size increases.

Balance drums have a wasteful leakage rate. Semi-open impellers have efficiency robbing pump out vanes, balance holes or both. Cross-over castings are expensive and add another level of complication to the pump. Closed impellers with back rings and balance holes becomes less efficient and more expensive, additionally as the ring clearance increases, so does the axial thrust. Back rings are an added cost, both initially and in subsequent maintenance evolutions, and though seemingly a minor point, they make the pump physically longer.

If you do not purposely design for or add features to balance the axial forces, the thrust bearing will need to be really big, expensive and probably unreliable.

Remember These Two Things

In the end, two things are for sure in the pump world: the laws of physics are strictly enforced, and you can’t get something for nothing.

-Jim Elsey

 

An Easy Open and Shut Case…

Open or Closed Tanks and the Effect on NPSHA Calculations


When working through NPSHA calculations for pump applications we need to know if the suction supply tank is open to the atmosphere or not. If it is an open tank the calculation is easy; as we just convert the ambient pressure to head for the first factor in the NPSHA calculation. Don’t forget to convert to absolute values and compensate atmospheric pressure for the local altitude above sea level. If the tank is closed, then we need to do a little more work converting the pressure or vacuum to absolute head for the calculation.

The Issue:

Often the customer will tell us the tank is closed to atmosphere, but it really isn’t; consequently the NPSHA calculations will be incorrect. The resulting NPSHA error will lead to a noncompetitive pump selection.

Sometimes the suction supply tank appears to be closed to atmospheric pressure, but if you look closer at the tank you will see it has breather valves installed. If there is a breather valve installed the tank pressure will always be very close to atmospheric pressure. It is very common, especially in the oil and gas world and also in the chemical and petro-chem arenas to use breather valves on the big bulk tanks. You may actually witness these breather valves on any size tank because the owner needs to protect the investment. Please realize these scenarios may also include rail tank cars, but do not confuse these examples with tank cars that are specifically designed to be pressurized or placed under vacuum for unloading purposes.

The Breather Valves Protect the Supply Tanks from:

  • Overpressure (rupture) and or vacuum (implosion) issues
  • Fire protection
  • Evaporation; loss of product
  • Corrosion protection

Another purpose is to prevent excess air and or water (plus other bad stuff like general pollution, Oand NOx) from destroying the product integrity while it is in the tank. The purpose is to protect the product from outside influence and or to protect the outside environment from the product.

These protection/breather valves are normally required by EPA and or OSHA …they are not just a good idea, they are often a legal requirement in many product applications. Most tank owners apply the same rules to all of their tanks regardless of the product, tank size or location. Note that both the EPA and OSHA will defer to API 2000 for the selection and sizing criteria for the breather valves.

So…I Just Want to do the NPSHA Calculation, What Now?:
If the tank has a breather valve, the answer is to simply use the local atmospheric pressure for the NPSHA calculations, because the actual tank pressure is going to be very close.

What if There is no Breather Valve and the Tank is Really Closed Off to Atmosphere?:

When you have a closed tank; I recommend you read my two Pumps and Systems articles on this subject from October and November of 2018, where the basics are covered.
https://www.pumpsandsystems.com/how-calculate-npsha-systems-under-vacuum
https://www.pumpsandsystems.com/calculate-npsha-closed-pressurized-system
If after reading you are still in doubt, call your RSM or engineering for assistance.

Last Comment:
Given a pump system with a supply tank open to atmosphere: note that on the suction line to an operating pump it is not uncommon to have a pressure lower than ambient. You may only expect this situation on a pump that is involved in a suction lift, but even for a flooded suction condition the suction pressure at the pump inlet can be at a vacuum. You can accurately calculate the actual pressure (vacuum) anywhere along the line by using Bernoulli’s Equation. Open and shut case… Easy peasy – lemon squeezy.

References:
OSHA 1910.106 July 1985
API 2000 Venting for Tanks 7th Edition 2014
API 12 (49 CFR 195.264)(b)(1) Specification for tanks
API 650 (CFR 132(b)(3) Specifications for large welded tanks
API Bulletin 2521 (Evaporation Reduction)
API Bulletin 2513 (Evaporation Losses)
EPA 40 CFR 112. Note: This regulation does not actually use the terms “aboveground storage tank.” Instead the term “bulk storage container”.
DOT (various/numerous with respect to rail cars)

-Jim Elsey

 

Why Are There Holes in My New Impeller? Part 1

 

radial or axial
Common Pumping Mistakes

Ever wonder why some impellers, all sparkly fresh and new from the factory, have holes in them, or why some impellers have those funny mini-vanes on the back side? The short answer is to reduce axial thrust, reduce the pressure in the stuffing box (seal chamber) and to preclude the collection of foreign debris in the annulus (chamber) behind the impeller. You may think that internal pump forces are not important, that the thrust bearing in the pump is really just window dressing and pump operating/running clearances have little to do with any of this. Unless your pumps run trouble free for more than five to eight years between maintenance, perhaps there is something here.

Regardless of size or design, whenever you operate a centrifugal pump, there are always dynamic forces to be managed and collectively known as the “total dynamic load.” The total dynamic load is the summation of all the forces that manifest throughout the pump as either radial and/or axial loads. The pump designer will address all of the forces by incorporating features to manage, reduce and/or eliminate the effects. And of course, as with everything in machinery, computer software and real life, there are always compromises to be made. Grandma always said, “You can’t get something for nothing.”

Radial or axialIMAGE 1: Radial or axial, the resultant forces are mostly due to the hydraulic pressure acting on the impeller. F=PA. Force is equal to the pressure (P) working on the effective surface area (A) presented by the impeller shroud(s). (Images courtesy of the author)

Axial Force

There are several dynamic factors that must be addressed in the pump, but the two important ones are the axial and radial forces. This column will address the management of axial forces in the pump. We will look at radial forces in a future column. The actual magnitude of these forces is calculable, but we will skip over those formulas this time and just address the cause and effect.

Axial forces are those that act on the rotor in a direction parallel to the shaft. Axial force, if not properly addressed, will try really hard to push the rotor out one end of the pump. In reality, there are forces acting in all directions on the rotor, but it is the vector summation of the axial components that is our concern for now.

Force (F) as defined for this application, is the total summation of the pressure applied per unit area. Therefore, the higher the pressure and/or the larger the area, the greater the force. The fundamental equation is F = PA; where P = pressure and A = area. This force is also a vector and so it has properties of both magnitude and direction.

The resultant axial forces in the pump exist due to higher pressures acting on one side of a surface and/or acting on a larger surface area when compared to the opposite side. The major thrust contributor in the pump is the pressure working on the large surface areas presented by the impeller shroud(s). But realize that shaft shoulders and impeller hubs also present surface areas and even the small diameter at the end of the shaft (protruding out of the pump at the coupling end) has atmospheric pressure pushing on it.

Pump out vanes are one method to manage axial force.

IMAGE 2: Pump out vanes are one method to manage axial force.

In every case involving axial thrust, the magnitude will be proportionally reduced by an increase in the suction pressure when compared to atmospheric, but of course at some point the suction pressure can be too high for the pump casing or related components such as the packing or mechanical seal. After balancing or reducing most of the axial forces in the pump, the residual axial forces will still need to be carried by the thrust bearing. The thrust bearing transmits the axial force to the pump frame housing and eventually to the foundation.

As hypothetical design questions, ask yourself what type and size of thrust bearing would you put in the pump, and then ask how long do you want the bearing to last? Finally, if the thrust force doesn’t have a pump foundation to transfer to, where does it go?

Methods to Manage Axial Forces in the Pump

There are several methods used in modern pump design to both mitigate and address the effects of axial force. Each approach has positive and negative attributes. Not all methods are covered in this column.

The pure mechanical approach simply designs the thrust bearing to absorb all of the thrust.

Next, there is the hydraulic configuration approach that attempts to reduce the thrust as much as possible via the basic pump design. For example, a back-to-back impeller pump where thrust is balanced due to symmetry. Other methods employ design features and balancing devices that address the forces by using balance drums, balance passages (holes), back ringed impellers, pump out vanes or some combination of these methods. Regardless of the approach in the real world, the pump will still require a thrust bearing of some type and size because no pump (or system) is perfectly balanced over the full range of operations.

Vertical Pumps

Vertical pumps, like the turbine style, will typically use a thrust bearing in the motor that is designed to carry all of the axial thrust from the pump. This is a unique design feature compared to all other types of pumps. There are exceptions were the pump thrust bearing is incorporated into the pump support housing.

Most vertical pumps will have all of the impellers arranged in the same direction. There are industrial grade multistage process pumps that will incorporate a crossover component so that half (or some design mandated percentage) of the impellers are facing one way and the remaining face the opposite direction. A few designs will also incorporate a double suction first-stage impeller.

This column is not covering all types of vertical pumps, and I will not address vertical submersible or cantilever pumps.

Within the normal range of operating conditions, a vertical pump will experience a resultant downthrust, which places the pump shaft in tension. The rotor is designed to be in tension and will experience issues when it is not. Consequently, if there is an upset in pump operations where an upthrust condition overcomes the downthrust for any length of time, there can be serious problems including bearing damage and shaft breakage. Most vertical pump rotors and the motor thrust bearing are designed to accommodate momentary thrust upsets that occur during startup and shutdown.

During normal pump operation, the magnitude of the thrust factor varies with the impeller design, the number of stages, the length of the pump, where the pump is operating on its performance curve and the head generated per stage. Some impellers are hydraulically balanced and some are not, and on some pumps it is a simple option for the application engineer to select on the build sheet to choose either. Consequently, the feature of balanced impellers needs to be declared to all the cooks in the kitchen, especially when choosing the driver and the thrust bearing.

It is both important and prudent to discuss all of these details with the manufacturer and/or an expert in the matter prior to changing where and how the pump will operate. Something as simple and unexpected as changing the type of bowl bearing, radial line shaft bearing or a change in the axial clearance setting (lift) could affect the success or failure of the pump application.

Vertical pumps: locating the pump thrust bearing in the motor makes for a more unique (less interchangeable) and expensive motor. On the pro side of the pros and cons list is that the motor is specifically designed for the application.

Vertical pump impellers typically do not use pump out vanes (for thrust reduction) in the majority of designs because of axial clearance uniformity issues when setting the pump rotor lift (short version is variance and stack tolerances), but many designs have the option to incorporate balance holes in the impeller(s) to reduce the thrust.

-Jim Elsey

 

What’s the speed limit?

How fast was I going officer?


Speed is a critical limit for any pump, but even more so for Positive Displacement (PD) pumps. The maximum allowable speed of a PD pump is determined by several factors including the viscosity and temperature of the pumpage. Other important factors are the level of abrasives in the product, acceleration head, and the Net Positive Suction Head Required (NPSHR).

Commercially available and cost effective electric induction motors nominally operate at speeds well above the optimum PD pump speeds, consequently some method must be used to reduce the drive output speed. Direct drive is just not all that common in most Internal Gear Pump (IGP) and Progressive Cavity (PC) applications for this reason.

The boundary for PD pump speed will typically be managed with either a gear reducer or a Variable Frequency Drive (VFD) and/or a combination of the two. For even more precise flow modifications a servo motor can be used in conjunction with both a gear reducer and a VFD.

Speed Kills

As the product temperature and/or abrasive concentrations increase, the pump should be operated at even slower speeds to reduce the inevitable wear and increase reliability. This may also mean a bigger and slower pump is required. Pump wear is exponentially proportional to speed. Even for relatively small increases in speed the wear rate can increase by a factor of eight.

Prior to purchase, the allowable speed range for the pump should be reviewed so that the correct choice of materials and speed control are made to achieve the lowest Total Cost of Ownership (TCO) and Mean Time between Failures and Repairs (MTBF/R).

Controlling Speed

Gear reducers (aka “gear sets” or “gear boxes”) are both essential and common components in the drive train of many PD pumps. Unfortunately the fixed output of a gear box will lock the end user into one operating speed. Therefore, the pump’s hydraulic duty point (at some speed) and the maximum allowable speed must both be considered when making the selection.

One additional benefit of a gear reducer is the increase in the amount of torque delivered to the pump shaft.  Gear sets are frequently referred to as “torque multipliers” for this reason. Adding a gear set may potentially reduce the required motor size when compared to direct drive.

VFDs are often used in applications where speed dependent flow requirements will/can vary over a range. A VFD in conjunction with a gear reducer will allow the pump to operate across an acceptable range of speeds, while simultaneously providing the required proportional flow rate.

Note: Pump speed limits must be applied when initially programming the VFD.  The VFD operational parameters must be set within the pump’s speed limits to avoid the critical and common over speed mistake. 

Don’t get Pulled Over by the Pump Police for Speeding: 

Operating correctly saves time and money.

Teaching owners and end users about pump boundaries allows them to choose smart solutions for their application and ensure the equipment is effective, efficient and reliable… reducing the TCO and MTBF/R.

If you have any questions please contact your Regional Manager or Engineering in Green Bay.

 

Why Are There Holes in My New Impeller? Part 2

The second installment of Jim Elsey discussing how the impeller affects axial thrust and stuffing box pressure.

 

IMAGE 1: Typical multistage pump with impellers all facing in the same direction (Images courtesy of the author)

-Jim Elsey

Balance Drums

In larger and higher horsepower multistage horizontal pumps, all of the impellers face in the same direction (toward suction) and the resultant axial force is almost negated by the use of a balance drum (or simply “drum”). The drum may be of either a straight or stepped design (aka, “two diameter drum”). Some pumps will use balance rings and/or balance discs; these will not be covered. Of interest, know that the axial force can exceed several tons in these pumps.

The drum is keyed and fixed to the shaft after the last stage and, therefore, it rotates at the same speed as and at a fixed position on the shaft. The drum rotates inside of a close clearance stationary throttle bushing. This rotor arrangement and drum geometry permits a small and fixed amount of product leakage back to a chamber that is at or near suction pressure. On a two-diameter drum, the variable axial position of the shaft allows either more or less leakage at any instant to change the chamber pressure on the low pressure side and counteract the axial force. This action occurs instantaneously and automatically as a direct function of the simple design. The amount of total axial movement of the rotor in this evolution is very small—perhaps a few thousandths of an inch.

It is not the intent of this column to describe the balance drum operation in more detail, but a basic description would explain that by using discharge pressure on one side of the drum and suction pressure on the opposite side, the majority of the axial forces will be balanced. Not all balance drum designs will balance the full magnitude of axial thrust as the pump operates over the entire range of its curve, so it may become a case of “caveat emptor,” or in English, “let the buyer beware.”

When you choose a balance drum in the pump design, there is the added cost and complexity of the component itself and its subsequent incorporation into the pump along with the added maintenance. Setting the balance drum clearance(s) in the field is a tedious and complex task that is frequently done incorrectly, all of which leads to expensive damage and downtime. Balance drum designs may also experience significant difficulty with system pressure transients, so always design (size) to have some residual thrust (versus total negation at ideal conditions) with the calculated leakage rates adjusted for a high wear condition. The drum leakage rate does reduce the pump’s efficiency, but in my opinion, it remains an acceptable tradeoff compared with alternative designs in this power range of pumps.

Opposed or Back-to-Back Impellers

A common example of a back-to-back arrangement would be a pump where two single-stage closed impellers are placed back-to-back, one facing each direction so as to balance the axial forces. A partition and/or pressure breakdown bushing between the stages keeps the stage pressures separated. This design balances the thrust almost perfectly and there is little to no reduction in efficiency.

In other large, multistage pump designs, the axial force is managed by using an opposed impellers method, just on a different scale from the back-to-back illustration. For example, in an 8-stage pump, four impellers face one direction and the other four face the opposite way. In this manner, most of the axial force can be reduced. Note that the number of stages facing in one direction is not always an even number when compared to the number facing the opposite direction. On pumps of this style, and especially with four or more stages, you must be vigilant to ensure the rotor is correctly positioned (axially) or there will be issues (damage) with the thrust bearings regardless of type.

The potential remains for high pressure differentials (pressure breakdown) between stages in these pumps and as the wear and subsequent leakage increases with time the thrust will also increase. Years ago, A.J. Stepanoff conducted lab tests on a two-stage (back-to-back) pump. He measured an increase in axial thrust from an initial 200 pounds with the interstage bushing clearance at 0.014 inches to over 1,050 pounds when the clearance increased to 0.060 inches. On many multistage pump models, the downside of the opposing impeller concept is if your system requires an odd number of stages to meet a specific hydraulic condition point. Opposing impeller pumps require a “crossover piece” somewhere in the pump to redirect the flow direction. The crossover casting increases the cost and complexity of the pump.

One last comment on multistage opposing pumps. For this design, there are X number of impellers facing different ways, consequently there are both right-handed and left-handed impellers, and often the first stage is totally different from all the rest (different inlet eye to address net positive suction head [NPSH] issues). To most people and untrained mechanics, all of the impellers will look the same. What could possibly go wrong?

Whack a Mole

The issue of addressing and designing for the pressure breakdown between stages becomes, in my opinion, a game of “whack a mole.” The hypothetical question is: Do you design to contain the pressure per stage, or for some portion of the pump, or for all the stages? For example, on a two-stage pump set up with back-to-back impellers, there is a bushing requirement to contain/control the pressure between the two stages and again at the high side stuffing box. To carry that idea forward, on an 8-stage pump, do you design to break down the pressure after each stage, after four or after all eight?

Dual Suction Impeller

A common example and a slight variation to opposed impeller designs is used on single-stage horizontal split case pumps where one impeller has two symmetrical inlet “eyes” in opposition at 180 degrees to each other. These dual suction impellers are also referred to as double entry impellers. In principle, the net effect is a fully balanced axial force. In fact, the dual suction design is the most successful (but not necessarily the most cost effective) of all the methods. The thrust bearing for these pumps is only required for upset conditions such as startup and shutdown and other off-design operations.

IMAGE 2: Single-stage end suction closed impeller. The resultant axial force will be in a direction toward suction.

As is the case with all pumps, the opposing forces are never fully balanced due to minor differences in the casings and impellers. Understand that these parts are typically cast and carry the potential for imperfections and/or asymmetry inherent to castings. The dual suction design is also limited in the number of stages that can be practically applied, usually three at maximum. At more than three stages, the pump simply becomes too long.

Dual suction impellers may also be applied on the first stage of high end vertical pumps and are frequently used for condensate applications because the net positive suction head available (NPSHa) for these systems is so low and the dual eye impeller requirements for net positive suction head required (NPSHr) will typically be 50 percent of a standard impeller.

Closed Impellers & Use of Back Rings

Closed impellers have both front and back shrouds, and so the surrounding pressure works from both sides of the impeller. There will be a resultant mismatch in the axial force due to the asymmetry of the impeller eye area. The result is less force acting on the eye area and results in a higher force acting on the rear shroud pushing the impeller toward suction.

To counter the thrust mismatch, many single-stage end suction pumps with closed impellers will incorporate a back ring on the rear shroud to reduce the axial thrust. (Some technical references refer to this back ring as an annular seal.) Adding the back ring to the impeller and drilling balance passages (holes) in the impeller shroud creates an area of lower pressure in the chamber behind the impeller. The lower pressure yields a lower axial force. An alternate and, in my opinion, a better design choice is to create an internal passage in the casing back to suction in lieu of drilling balance passages (holes) in the impeller shroud.

The back ring (annular seal) has a similar geometry and appearance of a front wear ring, however, the main purpose is different. The rear ring is there to simply reduce (pressure breakdown) the product leakage into the chamber. Again, by creating the low pressure chamber on the back side of the impeller, the force is reduced.

IMAGE 3: Dual suction Impeller. Axial force is balanced.

This back ring feature is common on pumps in heavy industrial and other critical services including many American Petroleum Institute (API) 610 applications. One advantage of the design is that thrust mitigation is not a direct function of the rotor’s axial position. A caveat with this design is that as the back ring clearance wears the effect is diminished. As a result, the magnitude of axial thrust will vary with the ring wear. Some designs will handle this issue better than others due to a better understanding of all of the principles involved, consequently a more accurate calculation of the total area for the balance passages as compared to the annular clearance. Proper sizing and placement of the annular ring as a ratio and proportion of the balance bleed passages is both paramount and critical.

Note that some closed impeller designs will use pump out vanes on the rear shroud, a feature that will be covered in more detail in another section. Some closed impellers will also incorporate pump out vanes on the front shroud of the impeller for the sole purpose of keeping debris away and do not exist for axial thrust mitigation. Other closed impeller designs use pump out vanes on the rear shroud more to rid the area of debris than to reduce thrust.

Not often used in industrial pumps, but some commercial pump casings will incorporate stationary (static) vanes or fins as an integral part of the casing at the front of the impeller to reduce axial thrust.