In spite of advancements in mechanical sealing technology, seal failures are still the most common cause of “pump repairs”. In fact, seal-related repairs represent approximately 60 to 70% of all centrifugal pump maintenance work. Therefore, seal-related repairs are an excellent area in which to concentrate pump reliability improvement programs.

Although a mechanical seal may be small enough to hold in your hand and simple in concept, it is actually a very complex device. The successful operation of mechanical seals in centrifugal pumps calls for careful attention to detail in several areas. These are:

1. Seal hardware (including sleeve, gland and gaskets)
2. Seal installation
3. Pump hardware (including piping)
4. Pump repair and installation
5. Pump operation (including flush to seal).

Of course, these areas are not independent, but often seal reliability can be improved by concentrating on individual areas to solve particular problems. Naturally, the pertinent area must be selected or the effort may be entirely wasted. Therefore, the first step in reducing seal failures is to establish a seal failure analysis program.

Failure Analysis

Since the purpose of a mechanical seal is to prevent, or at least limit, leakage, most users consider the seal to have “failed” when excessive leakage occurs. Therefore, all seals will fail – it is simply a matter of time. The machinery engineer must therefore be prepared to learn from each seal failure and use that information to improve future designs and applications so that seal life is improved. In order to do this, detailed records of each failure should be maintained.

Failure analysis is simply allowing the seal to tell why it did not perform as expected. This means acting the part of a detective. The following simple guidelines will make failure analysis easier and more consistent.

1. Know how a new seal looks
2. Know how a successfully applied seal looks
3. Examine the failed seal very carefully
4. Write down the differences between the new, successful and failed seals
5. Make up a consistent explanation about the differences

Although manufacturers and mechanics are very familiar with new seals, sometimes a machinery engineer may not see the unused item. He only sees the failures. Similarly, he may not see many successfully applied seals because they are not called to his attention. He is too busy looking at failures. On the other hand, some failure modes are so common that a mechanic may come to accept them as “normal wear”. For failure analysis to be useful, all differences in the new and failed seal should be noted.

These differences must be written down in a consistent format.   There is a great temptation to take mental notes and write them later. This temptation must be overcome and written notes made while examining the seal. If these notes become soiled, they can be re-written later.   A simple checklist can be a good reminder.  Another good idea is to use the seal layout drawing.

For many refineries and chemical plants, the following statistics have been developed from analysis of many pump seal failures:

The primary causes of seal failures may be grouped as

• Operating problems 30 to 40%
• Improper installation 20 to 30%
• Pump design/repair 10 to 20%
• Misapplied seal 10 to 20%
• Worn out 1 to 5%
• Miscellaneous 15 to 20%

These statistics show that most seals fail prematurely. They also show that these early failures can be prevented by simply selecting the proper seal, installing it correctly and operating the pump carefully.

Seal Hardware Failures

Another way to look at seal reliability is to insist that the seal must survive in spite of any problems with the pump or its operation. In this case, failures are examined from a hardware or design point of view. In other words, if a pump requires repair because the seal is leaking, the corrective measures are directed towards the seal hardware. This may not always be correct; however, the following statistics seem to apply to hardware failures in refinery and chemical plant pump seals:

• Rotating or stationary face 40 to 70%
• Loss of axial movement (hangup) 10 to 20%
• Fretting 10 to 20%
• Corrosion 10 to 20%
• Other 5 to 20%

Naturally, seals or pumps which were not properly installed or repaired are omitted from the above groups.

Seal faces.  In pump repairs caused by hardware failures, the seal faces are the apparent problem 40 to 70% of the time. These faces were severely worn or damaged. In some instances, an immediate improvement was made by simply changing face materials. For example, a stellite or ceramic face might be replaced with a solid tungsten carbide face. In other instances, the face materials could not be changed or improved and efforts were focused on other areas.

Because mechanical seal failures can be expensive, there is no excuse for using second rate materials. In many refinery and chemical plant services, a premium grade carbon and tungsten carbide or silicon carbide faces have proven reliable. Springs are frequently Alloy 20. Glands, sleeve, shells, adapters, etc are made from 316SS. In order to reduce inventories and avoid costly mix-ups, these same materials are also used in easy services — even cold water.

Hangup.  Loss of axial movement, or seal hangup, is caused when leakage accumulates and hardens beneath the “floating” element. In most seals, this is the rotating element. Hangup is a particularly troublesome problem in high temperature seals because any leakage tends to form “coke”. There are four approaches to solving hangup problems:

1. Change the liquid (external flush which will not solidify)
2. Reduce leakage (higher balance ratio and/or narrow face seal)
3. Remove leakage (steam quench)
4. Remove opportunity to hangup (change to a stationary seal).

Some stationary seals are designed so that there is a relative motion between the “floating” element and the rotating shaft. This relative motion makes it difficult for solids to accumulate and cause hangup. Because stationary metal bellows seals include this feature as well as narrow faces and are rated for high temperatures, they are often used in services which tend to cause hangup of conventional seals.

Fretting.   Fretting is caused by the constant axial rubbing of the “floating” element or its gasket against the sleeve or drive lugs. The movement may be less than a thousandth of an inch, but eventually the sleeve or drive lug is damaged. The damage may cause leakage directly or may restrict axial movement so that leakage occurs between the faces.

The solution to fretting problems usually takes one of the following approaches:

1. Improve seal alignment during installation
2. Reduce shaft axial float (adjust bearings)
3. Use O-rings for gaskets
4. Change to a bellows seal

Miscellaneous.  Sometimes a seal is observed to be leaking, but when removed and examined there is no obvious damage. This often happens when sealing light hydrocarbons. For these products, the liquid may flash to a vapor between the seal faces. Since most seals are designed for liquids, the resulting force imbalance causes the faces to “pop open”. The solution may involve changing the balance ratio, seal type or installing an external flush.

Seal Failures Caused by Pump Design

Sometimes seal failures are so strongly related to the pump and its performance that reliability can be improved only by modifying the pump. The pump may have a true design problem or may simply have been misapplied. Some of the more common pump related seal failures are:

• Operation at low flow
• Excessive shaft deflection
• Cavitation

Low flow. In spite of the performance curve, which shows operation from no flow to maximum, centrifugal pumps may not operate smoothly at less than about 50% of Best Efficiency Point (BEP) flow. Vibration and noise may increase markedly at less than this minimum stable flow. The result is a decrease in seal life. Resolution of this problem many require hydraulic modification by the pump manufacturer or a by-pass line to artifically increase flowrate.

Excessive shaft deflection. If the shaft deflects, the seal must move axially each revolution to compensate. API Standard 610 for centrifugal pumps specifies a maximum of 0.002″ shaft deflection at the location of the seal faces. Some older pumps and non-API pumps may not meet this specification. In particular, older pumps designed for packing may have excessive shaft deflection. Shaft deflection is reduced by increasing the shaft diameter and/or reducing the bearing span or shaft overhang.

Cavitation.  Cavitation has been rightly and wrongly blamed for many ills in both pumps and seals. Certainly cavitation increases pump vibration and vibration reduces seal life. Studies have shown that the simple 3% head loss rule which is used to define NPSHR may not adequately define the onset of cavitation problems. Also, operation at low flow sometimes produces symptoms similar to cavitation.

Cavitation problems are sometimes difficult to solve. Frequently, increasing the available NPSH is prohibitively expensive. Reducing the required NPSH is sometimes possible with special impellers or inducers. For these reasons, cavitation problems must be carefully addressed during the initial specification of the pumps.

 

Miscellaneous Comments on Failures

Coking.  In my experience, coking is almost never a problem for hydrocarbons at less than 300 F. Also, if the seal does not leak, coking cannot occur. Between 300 F and 500 F I’ve used stationary seals with narrow faces and a high balance ratio successfully. This combination reduces coking failures by allowing less leakage and isolating the axially flexible member from the rotating member thereby preventing hangup. Above 500 F, the tendency to coke is so strong that steam is required.

Sheared mating ring pins. Of course, this indicates a high torque as compared to the resistance of the pin and gasket fit.  This is often a problem with two hard faces are used and both faces are highly polished.

Spinning seat ring. This also indicates a high torque and light resistance.  Is there an anti-rotation pin? Have the fits been gradually opened up over the years of repair?

Worn drive notches in primary ring.  Again, a high torque. Also axial movement caused by misalignment.

Wedge extrusion and wear.  Excessive axial movement and wear.  Try to use an O-ring design instead of a wedge, U-cup, etc.

Worn springs and pockets. More indications of axial movement.

Retainer wear in drive dents.  Still more indication of excessive torque and axial movement.  Often a problem when the primary ring is harder than the retainer, for example, a tungsten carbide primary ring inside a stainless steel retainer.

Sleeve fretting.  Axial movement of dynamic secondary sealing element (wedge or even O-ring).