Why Seals Fail
Seals fail for a number of reasons. Your job is to pinpoint the
reason and fix it.
Here you are in a situation in which the seal has run for a
period well beyond the installation period. Its leaking and now you have to make a
decision. Has the seal failed or simply worn out? What you decide now will
determine whether you fit a replacement seal or seek out an alternative type. The
basics are simple.
A worn out seal will leak when the seal face has worn away
If we extend this criteria to all leaking seals it becomes sadly obvious that the majority
of seals, perhaps 85% of process seals, fail long before they are worn out.
This section is devoted to the three main reasons why seals
fail. Only three you say? Three main reasons and lots of routes to them.
Seals fail because ...
- The seal faces open.
- Heat causes a problem.
- The chemical environment causes a material failure.
OK so there is another category ... the installation failure,
but that's covered in the
Seal Faces Open
The shaft moves for many reasons, those that
affect the seal operation are:
- End play
- Thrust movement
- Temperature growth
- Impeller adjustment
- Bearing wear
- Bent shaft
- Shaft whip
- Shaft deflection (discharge closed)
- System NPSH incorrect causing cavitation
- Harmonic vibration, check the coupling, does it "hum"
or "buzz". Rubber couplings can operate with high degrees of misalignment
without total failure but cause problems for the seal.
- Impeller imbalance
- Slip-stick. Not surprisingly not much is known about what happens
between seal faces in service. There are theories. The faces acquire a film of liquid that
lubricates the seal surfaces, the carbon face wears slightly depositing a layer of carbon
on the stationary face so that the carbon face runs on carbon , but there is a condition
that causes the faces to vibrate open when pumping non-lubricating fluids. Fluids near
their vapour point, very hot water, can cause these conditions. The seal faces
"chatter " against each other in a slip-stick motion slipping when the drive lug
hits the seal head, bouncing round and momentarily stopping before being hit by the drive
lug again. To be a sealman you have to believe.
- Poor pump performance. This statement covers a host of sins.
Consider running two or three pumps into one discharge line, the odds are that the pump
performances will not be perfectly matched. Does it matter? Not really, unless you are
concerned about your seal life, because what is happening here? One or other of the pumps,
because of poor performance now combined with poor system design, will be experiencing
discharge throttling, tending to over load the impeller at the throat, causing turbulent
flow and shaft bending. Look into other causes of poor pump performance.
The seal runs against a stationary component. The
stationary is usually fitted into the seal plate which is bolted to the pump and sealed
with a gasket. Now, I do not want to sound too pedantic here but you have to realize
that the seal stationary has to be fitted square to the axis of the shaft and in proper
alignment with the axis of the pump shaft. The stationary has to be fitted into the
seal-plate square. None of this is easy to achieve and each error compounds the next. The
rotating head has to follow any misalignment from square that the stationary carries.
Every rotation of the shaft causes the rotating seal head to move back-and-forth twice.
Interfere with that movement and the faces are open.
Difficult as it is to get the stationary fitted correctly,
should you achieve it then other factors come into play to limit the excellence of your
work. Stress imposed by pipe strain, coupling misalignment, or plain thermal growth put
the pump casing out of shape just enough to cause the seal to work harder.
All of the items described mean that the shaft and seal are in
constant relative movement. If anything interferes with the free movement of the seal, the
When the faces open, dirt in the liquid penetrates the lapped
surfaces, embeds in the soft face which gradually changes to a grinding surface to score
and wear away the hard face of the stationary ring. Have you noticed this effect? Do you
look at your failed seals? You should, because on those faces lie clues to help you find
the faults opposing long seal life. Well when we have gotten through this section and onto
the tell tale signs I bet you will take a bit more notice of your failed seal bodies.
The main reasons why seal faces open are:
- The elastomer sticks to the shaft. Spring loaded elastomers will
stick to the shaft, O-rings will flex by 0.005" (0.13mm) and then roll. O-rings will
fret a shaft but spring loaded elastomers (teflon wedges, chevrons, etc.) can cause
serious surface damage to your shaft or sleeve leading to early seal failure. A leak under
the seal head looks very much like a face leak.
- The shaft is out on machining tolerance. Correct tolerance is
+0.000" to -0.002" from nominal. A packing sleeve is not machined to any close
tolerance, after all it is going to wear against the packing so its external dimension is
not too important. An oversize sleeve or shaft will cause the seal to hang-up, an under
size shaft or sleeve will prejudice the ability of the elastomers to seal the head to the
- The surface finish on the shaft/sleeve is too rough. A lathe
finish is not good enough. The finish should be at least 32 RMS and for that a ground
finish is required.
- Have you got a hardened shaft on your pump unit? The seal set
screws will not "bite" into the shaft and could slip causing the setting
dimension of the seal to alter.
- The pumped fluid changes state. Sea water, brine pumps, sugary
solutions, cause crystallizing when the salts come out of solution
or the sugars become caramelized. Other coking substances, heat transfer oil, tar, cause
similar problems. You will see the build up of material around the leak site.
- Solids can cause the seal head to stick to the shaft or restrict
the o-ring flexibility. Take a look at the double seal arrangement, back to back version.
Used on some services the O-ring could very quickly become clogged preventing the seal
head from moving to accommodate wear of the faces.
- Incorrect setting length at installation. You may never figure
this one out. Just make sure that the fitting dimension is correct when installing the
seal. Otherwise sometime in the future the seal will let go, usually after the pump is
stopped, and the faces will look good but only partly worn. What has happened is that the
spring pressure has reduced to the point where the seal leaks during idle periods. This
can be difficult to spot, unless you know what to look for ... and when.
- Fretting. Very small movements between components causes a
polishing action. The polishing action removes the surface molecules. On pump shafts made
of stainless materials the surface of the metal consists of chromium oxide. Elastomers
moving very slightly against this surface wipe away the oxide which immediately reforms.
The oxide is carried into the wiping surface changing its character completely. A rubber
ring coated with chromium oxide becomes more efficient as a polishing, grinding surface
and removes material at a faster rate. A "fret" ring is characterized by a
polish mark on the shaft surface at the point where the seal elastomer seals against the
shaft. If worn badly enough the fret ring can cause a new seal to fail on installation
because the elastomer cannot seal effectively due to the damage on the surface.
- Distortion of the stationary face. This is not common but the
stationary could be badly fitted leading to over tightening, especially the silicon
carbide grades which are designed with a lip to be clamped in the seal plate. Failure
under these circumstances may be confused with cracking due to heat checking of the
component. S.C grades of 99.9% only heat check if they are tightened un-evenly, so check
out your grade and suspect poor fitting if its a high grade material failing by cracking.
With other materials such as tungsten carbide, or plated surfaces, such as stellite,
consider the distorting effect of poor clamping if no other solution presents itself.
- Face Mis-centering or run-off. This is not common and is easy to
diagnose. The faces are not concentric and the rotating head comes off the stationary
track and picks up dirt. Scoring of the stationary and an off center running track gives
you all you need to know.
- Incorrect grade of O-ring material. Lots of things happen to
elastomers so check out the ones on your seal, are they swollen, hard, squashed, shiny,
- The seal hits something, it is prevented from moving to
- Lots of possibilities here, so I list a few.
- The shaft is bent and hitting the stationary face. You will
notice this pretty quick, but bear in mind that the running clearance of the seal
components and the shaft may be quite tight, so a small shaft displacement may not be
obvious, the seal will show you what is happening.
- Solids in the seal chamber hitting the seal.
- Incorrectly fitted gasket extending into the seal chamber. Split
casing pumps can suffer this problem.
- The shaft is not concentric with the seal chamber.
- Insufficient clearance in the seal chamber. Check this out if you
are changing seal type or intend using different materials to cope with other problems.
- A seal box recirc line is directed at the seal faces. Most seal
chambers have a radial flow insert when most seal manufacturer's will tell you that a
tangential flow insert is safer and causes less disturbance to the seal faces.
Heat Causes Seal Failures.
- Heat affects the elastomer. This the part most sensitive to
extremes of temperature.
- Heat can change the state of the fluid being pumped.
- Raising the temperature of corrosive liquids increases their
potency. A 16 deg F rise doubles the corrosion rate of most acids.
- Differential expansion rates can destroy plated seal surfaces.
Low grade silicon carbide will crack with sudden changes in temperature.
- Differential expansion of shaft and pump casing can change the
face loading by altering the fitting dimension.
We now have the over-view of heat related failures so let us
look in more detail at what is happening.
A wide range of elastomers are in use and many of
them are rubber compounds. Teflon materials have a predetermined heat range of
to 226 deg C beyond which Teflon breaks down and burns making small amounts of phosgene
gas. Teflon should not be used in temperatures close to its ultimate limit because it is a
heat insulator and local heat production may cause it to reach its ultimate temperature.
Rubber compounds are made by baking the material until it
is cured to a predetermined hardness or durometer. The various materials formed in this
way, nitrile, viton, buna-n, and others, are commonly found in sealing applications. Less
common is Kalrez a specialized compound with a high resistance to
chemical attack. Formed in a heat setting process, these materials continue to be affected
by the heat applied during the life of the seal. At temperatures beyond the range of the
rubber seal the material continues to harden. As it hardens the shape of the seal takes on
the shape of the groove if an O-ring or splits appear in rubber bellows as flexibility is
lost. O-rings take on a "compression" set and appear oval and feel hard to the
touch. O-rings are manufactured with a 10% tolerance oversize to allow for some
thermo-setting in service. At higher temperatures the elastomer life to full compression
set will depend upon the temperature and time at this temperature. The point for you is
that exceeding the range of the rubber parts of your seal will shorten the working life of
the seal and you need to bear this in mind.
An odd case, in Saudi Arabia I was called to a refinery that
had been under construction for several years and pumps had been installed, but not run,
for varying periods. Pumps under going test runs were leaking along the shafts.
Investigation showed that over time in ambient temperatures of 55 deg C the seal
elastomers had baked hard and vulcanized to the metal parts. All seals had to be changed.
Heat is generated from the friction running at the seal faces.
Depending upon the type of face material and the seal box environment a rise of around 25
deg C above the seal fluid temperature can occur. Look at your seal types, where is the
elastomer in relation to the seal faces. The nearer the elastomer is placed to the running
faces the greater the additional heat it will experience. The use of low friction seal
face combinations will reduce this effect. The carbon / ceramic combination has the lowest
friction rating with hard faces such as tungsten / tungsten faces the highest.
Unbalanced seals, because the face weight is varying with the
system pressure, can experience greater rises in face generated heat creating damage to
Excessive heat producing a temperature rise of 55 Deg C on a
Viton O-ring will reduce its useful life to less than 1000 hours running time. For a seal
that is expected to run for one year that is an 88% reduction in useful life. An 82 deg C
rise will reduce the life of the seal by 97%.
Loss of water to a cooling water jacket, loss of any cooling
arrangements puts your seals at risk.
I was called to a split-casing boiler feed
pump that was experiencing out-board seal failure. Normally I would expect more problems
with the in-board (coupling end ) seal due to less opportunity to dissipate the heat soak
along the shaft. Examination of regular temperature recordings made of the cooling water
system and seal box temperatures revealed that the out-board seal was being starved of
cooling water flow. Dismantling the orifice plate controlling the flow to the in-board
seal showed excessive wear enlarging the orifice and allowing through a larger proportion
of the flow. Replacing the orifice plate solved the problem. All can seem well with your
equipment but the seals will always let you know first when problems are arising.
Changing state of the fluid
Liquid gases and other volatile fluids can vaporize and freeze
water out of the air on the outside of the seal restricting movement. Shortly before I
took up my post in Saudi Arabia a liquid propane pump blew its seal open due to a build up
of ice around the seal faces. Liquid released into the atmosphere created a vast cloud of
highly flammable gas. Fortunately no one was hurt and no explosion occurred but it was a
close thing. It was thought appropriate to fit a double seal with a barrier fluid for
Liquids changing state to a gas experience enormous volume
increases. Water increases in volume by 1700 times, so a small drop vaporizing across a
seal face will explosively blow apart the faces. Boiler feed pumps and other hot water
pumps can be heard "popping" or "puffing" if the seals are not working
correctly. As the water droplets expand and open the seal faces more water rushes in to
cool the area, collapsing the steam bubble and causing the faces to snap shut. Another
small droplet penetrating the faces vaporizes and causes the faces to open again. Water
treatment crystals, entrained oxides, other dirt particles are trapped between the faces
as they close. Your seal is on its way to the scrap yard.
Some fluids crystallize with additional heat. Sea-water, brine,
and similar fluids leaking past your seal and drying out around the seal plate can build
up to affect the seal head and prevent it from moving. Crystals can also score the running
surfaces of the seal causing damage leading to failure.
Hydrocarbons form coke as they partially burn or vaporize.
Coking causes a hard solid to form around the seal effectively stopping it from moving
freely. A similar effect is seen in food plants handling product containing sugar. Sugar
escaping across a seal face can crystallize, or simply burn and coke. The signs are
un-mistakable on the seal face.
Heat can cause impurities to come out of solution and plate onto
seal surfaces, building up hard films or lacquers.
Heat can destroy seal faces.
I have mentioned some of these effects but I think a defined
list will help you.
Plated materials can experience differential expansion. Often
materials such as stellite are plated over stainless steel. The expansion rates are poorly
matched so operating outside of the design limits of the materials will cause strains to
appear in the plating interface, causing cracks to appear. The cracks will cause the
carbon face to wear dramatically fast.
The less expensive ceramic material (85%) will crack if cold
shocked. Sudden changes in temperature of 38 deg C or more will destroy the seal face. The
higher quality ceramic (99.9%) will cold shock if it is under distorting stress, properly
fitted and evenly clamped it will survive sudden changes in temperature. Get to know which
materials are being fitted into your seal installations.
Carbon rings using fillers and fitted into high temperature
pumps can have the filler material melt out of the carbon causing them to become porous
Poor carbons with voids can blister and pit as the trapped air
or gases expand and blows pieces off the carbon surface.
Lapped seal faces can distort, going out of flat. The effect of
touching the lapped surface with a finger is to coat the surface with dirt and skin oils
but also to distort the surface away from flat by the application of heat from your hand.
Distorted seal faces leak.
Heat increases the corrosiveness of most corrosive
- The carbon part of the seal will show signs of being attacked.
- O-ring grooves can be damaged limiting their ability to seal
- O-rings can become hard or start to crack, or become swollen and
- Metal surfaces can be attacked and appear pitted which will
prejudice the seals ability to work properly.
- Springs and other highly stressed parts can fail due to increased
Expansion due to heating effects.
All metals expand when heated. A stainless steel shaft 48"
long by 4" dia will grow 0.138" in length when heated through 300 deg F. The
working limit of most carbon seal faces is 0.125" . Seal compression is set at about
0.064" to produce the spring face weight. A seal mounted on a shaft moving by
0.138" with other expansion effects happening to the pump casing is in danger of
opening. Apart from ensuring the accurate placing of the seal on the pump shaft there is
little to be done to compensate for such movement. Tell-tale signs of inaccurate setting
of the seal will be where you need to be looking.
The shaft diameter will expand too, by about 0.010". The
seal material will expand also but under extreme circumstances this expansion can cause
the seal to hang-up on the shaft. Over-compression of the elastomers will limit their
effectiveness, as well as the other effects mentioned earlier.
Failure of materials is usually a sign of a mis-match of
material to environment. The substantial construction of seals excludes major failure of
some main component, so we concentrate on the effects of environmental attack on sensitive
- Chemical attack on the elastomer will cause it to swell.
- The carbon will appear pitted. Acid attack on carbon is directed
against the impurities. The reaction of the impurities to the acid solution cause holes
and pits to form, weakening the structure and producing a porous carbon. A higher grade of
carbon is required.
- The springs can break. Stainless steel is known to fail due to
chloride stress corrosion. Many single coil spring driven seals fail because the spring
breaks. They are usually in-expensive and over-engineered, but they still fail.
- Metals corrode. In seals where metal parts are designed to be
thin due to flexibility requirements, metal bellows seals, welding techniques used in
construction and material compatibility with mating components and pumped fluids are
factors that affect the life of a seal.
- Set screws clamping onto a hardened shaft material will not grip
properly, allowing the seal body to slip, leading to a range of other effects, but
ultimately to a seal failure.
- Plated seal faces are not corrosion resistant, so the plating
material can be removed from the surface.
This list is not exhaustive however comprehensive it may appear.
You will find some new problem and when you do I want to hear all about it. So do all the
other guys visiting this site. Look forward to hearing from you, I just know I will in