Latest / Mechanical Engineering Made Simple / (#106) Shaft Deflection Kills Mechanical Seals

Transcript

(#106) Shaft Deflection Kills Mechanical Seals · Mechanical Engineering Made Simple

• 0:00
  Welcome back to mechanical engineering made simple.
• 0:02
  It's it's great to be back here. Today we're doing a deep dive
• 0:05
  into a scenario that is probably going to trigger some PTSD for a
• 0:09
  lot of you listening. Oh.
• 0:10
  Yeah, I know exactly which scenario you're talking about.
• 0:13
  The phantom failure. You know the one.
• 0:16
  You have a critical pump or a mixer in a plant.
• 0:19
  It's midnight, the machine goes down, maintenance tears it
• 0:23
  apart, finds a blown mechanical seal, replaces it, and, you
• 0:27
  know, high fives everyone. Case closed.
• 0:30
  Until 2 weeks later when the exact same seal blows again.
• 0:33
  Exactly. And that is where the finger
• 0:36
  pointing starts. The seal guy blames the
• 0:38
  installation, the installation guy blames the bearing, and the
• 0:40
  bearing guy blames the lubricant.
• 0:42
  Always the lubricant. Right.
• 0:43
  But today we're going to look at the evidence provided in our
• 0:46
  source material and build a case against the suspect that usually
• 0:50
  walks away Scott free. The shaft.
• 0:53
  It's the literal backbone of the equipment, yet it's often the
• 0:56
  last thing checked. We're going to walk you through
• 0:58
  the physics of why this happens, moving past the basic torque
• 1:01
  calculations and getting into deflection, surface finish, and
• 1:05
  the human element of installation.
• 1:07
  It's a classic case of looking at the symptom rather than the
• 1:10
  disease. You mentioned torque
• 1:11
  calculations, and that is usually where the trouble
• 1:13
  begins. We call this the trap of just
• 1:16
  enough. Right, let's set the scene with
• 1:17
  the design face. Let's say we have an engineer,
• 1:19
  we'll call him Bob. Bob is designing a mixer shaft.
• 1:22
  He looks at the horsepower, looks at the RPM, calculates the
• 1:26
  torque, and picks the diameter. That guarantees the steel won't
• 1:29
  snap. And on paper, Bob is a hero.
• 1:32
  The safety factor says the shaft allows for, say, 20,000 PSI of
• 1:37
  shear stress, and he's only at 5000.
• 1:40
  It's never going to break. Then the machine is built and
• 1:42
  the seal leaks immediately. Bob is completely confused
• 1:46
  because his math says the shaft is strong.
• 1:48
  Enough. And he's right.
• 1:49
  It is strong enough, but he fell for the classic blunder he
• 1:53
  assumed torque chooses the diameter.
• 1:55
  It doesn't. Constraints choose the diameter.
• 1:58
  Specifically stiffness constraints.
• 2:01
  This is the difference between will it break and will it bend.
• 2:04
  Exactly. In the industry we lovingly call
• 2:07
  it noodling. A shaft can be perfectly capable
• 2:10
  of transmitting the torque without snapping, but still act
• 2:13
  like a wet noodle under radial loads.
• 2:16
  Right hydraulic. Forces or the weight of a
• 2:19
  massive impeller at the end of it.
• 2:21
  It's intuitive when you think about it.
• 2:23
  Steel is strong, but it is also elastic.
• 2:27
  If you have a long skinny shaft, it acts like a diving board.
• 2:30
  That's the perfect analogy. And if that diving board bounces
• 2:33
  too much right where the mechanical seal is trying to do
• 2:36
  its job, you have a massive problem.
• 2:38
  And this is where the source material gets very specific
• 2:41
  about limits. If that shaft deflects or bends
• 2:44
  even a tiny fraction of an inch right where it passes through
• 2:47
  the mechanical seal, you're in trouble.
• 2:49
  Deep. Trouble.
• 2:49
  The seal faces are lapped flat to within light bands.
• 2:52
  We are talking microscopic flatness.
• 2:55
  If the shaft bends, it literally forces those faces to open up on
• 2:58
  one side once they. Open, the fluid gets out, or
• 3:01
  worse, abrasive particles get in.
• 3:03
  The sources we looked at from Mixer analysis are pretty strict
• 3:05
  here. For a general shaft, you might
• 3:08
  get away with a total center line deflection of 125
• 3:12
  thousandths of an inch. That's 0.125.
• 3:15
  That sounds small. But for a mechanical seal.
• 3:18
  It's an order of magnitude tighter.
• 3:20
  You generally cannot exceed 0.0015 inches at the seal face.
• 3:26
  That's 1 1/2 thousandths. To put that in perspective for
• 3:29
  you, a human hair is about 3 thousandths, so we are talking
• 3:32
  about half the width of a human hair.
• 3:34
  If Bob's shaft bends the width of a single hair, his design has
• 3:38
  failed. And the worst part?
• 3:40
  When the machine stops, the shaft springs back to perfectly
• 3:43
  straight. The mechanic pulls it out, puts
• 3:46
  a straight edge on it and says the shaft is fine, but it was
• 3:49
  the murderer all along. So Bob needs to stop calculating
• 3:53
  for stress and start calculating for deflection.
• 3:55
  He needs to beef up the diameter not because the torque requires
• 3:58
  it, but because the seal requires it.
• 4:00
  Precisely, stiffness is the governing factor in 90% of
• 4:04
  rotating equipment design. OK, so let's say Bob fixes that.
• 4:07
  He makes the shaft thicker, it's stiff, doesn't noodle.
• 4:10
  But six months later, snap a clean break and it almost never
• 4:15
  breaks in the middle of a long smooth section does.
• 4:17
  It rarely. It almost always breaks at a
• 4:19
  transition, a shoulder, a keyway, a thread relief.
• 4:22
  This brings us to the concept of stress risers.
• 4:25
  I always like the analogy of water flowing through a pipe for
• 4:28
  this. It's the best way to visualize
• 4:30
  it, yeah. If you have a smooth straight
• 4:32
  pipe, the water flows logically. But if you have a sudden 90°
• 4:36
  turn or a sharp restriction, you get turbulence.
• 4:39
  You get pressure spikes right at that corner.
• 4:41
  Stress acts the exact same way. It flows through the metal.
• 4:44
  If you force that stress to navigate a sharp 90° corner
• 4:48
  where a bearing sits, the stress piles up at that corner.
• 4:52
  The technical term is stress concentration factors or KT.
• 4:55
  Right. And that factor can be huge.
• 4:57
  A sharp corner can multiply the local stress by a factor of 3 or
• 5:00
  4. So Bob calculated 5000PSI of
• 5:04
  stress, but at that tiny corner it's actually 20,000 PSI.
• 5:07
  And that's exactly where the crack initiates.
• 5:10
  It's insidious because the rest of the shaft is barely working,
• 5:14
  but that one corner is screaming for help.
• 5:16
  The fix is usually geometry. You need a fillet radius.
• 5:20
  You need a curve to guide that stress around the corner, just
• 5:23
  like a smooth bend and a pipe helps water flow.
• 5:26
  Now there's another geometry crime we need to discuss, and
• 5:29
  the Machinery's Handbook is pretty ruthless about this one.
• 5:32
  Key ways ah. The necessary evil We need key
• 5:36
  ways to transmit torque to the coupling or the impeller.
• 5:39
  But man, do they hurt the shaft. The source material threw out a
• 5:43
  number that actually surprised me.
• 5:44
  It suggested that if you cut a standard keyway into a shaft,
• 5:48
  you should consider that shaft's strength effectiveness to be
• 5:50
  reduced to roughly 75% of a solid shaft.
• 5:54
  It sounds high, but it makes total sense.
• 5:56
  You aren't just removing material, you are adding two
• 5:58
  sharp internal corners at the bottom of that groove.
• 6:01
  It's a double whammy of cross section reduction and stress
• 6:04
  concentration. So.
• 6:05
  If you were designing right to the limit and you forget to
• 6:07
  account for that 25% loss in capacity.
• 6:10
  You are operating in the danger zone and that's not even
• 6:13
  touching on the type of keyway. If you use a sled render keyway
• 6:17
  versus an end Milky Way, the stress profiles change
• 6:19
  completely. But the main take away for you
• 6:22
  is never assume a keyed shaft has the same rating as the round
• 6:25
  bar it came from. It effectively makes a 2 inch
• 6:28
  shaft behave like a 1.5 inch shaft in terms of fatigue limit.
• 6:31
  That is a fair approximation, yes.
• 6:34
  OK, moving on. Let's assume the geometry is
• 6:37
  good. The shaft is stiff, the corners
• 6:39
  are radius. The key way is accounted for.
• 6:42
  We move to the next layer of the onion, the mystery leak.
• 6:46
  This is the one that drives maintenance guys completely
• 6:48
  crazy. You pull the shaft and it looks
• 6:50
  perfect. It's shiny, it's smooth, but oil
• 6:53
  keeps pouring past the lip seal. And naturally they replace the
• 6:56
  seal thinking they just got a dead.
• 6:58
  Right, but this is where we have to look at the fishbone of
• 7:01
  failure, specifically the surface finish branch you
• 7:04
  mentioned shiny. In the world of ceiling, shiny
• 7:07
  can actually be a huge problem. Because if it's too smooth, it
• 7:10
  can't hold lubrication. Exactly, it's a Goldilocks
• 7:13
  situation. If the shaft is too rough, it
• 7:15
  acts like a file and grinds the seal lip down.
• 7:18
  If it's mirror polished, the seal runs dry, gets hot and
• 7:22
  blisters. You need a specific roughness
• 7:24
  average, usually between 10 and 20 raw to hold just enough oil
• 7:28
  to keep the seal happy. So it needs to be smooth, but
• 7:31
  not too smooth. OK, but what if the roughness is
• 7:34
  perfect? Then we look for the invisible
• 7:36
  killer machine lead. This is a fascinating concept.
• 7:40
  I remember reading about this years ago and thinking it
• 7:43
  sounded like magic. Explain how a shaft can have a
• 7:46
  lead. It comes down to how the shaft
• 7:48
  was ground at the factory. Imagine the grinding wheel
• 7:51
  touching the shaft. If that wheel moves laterally
• 7:53
  across the shaft while it's spinning, it leaves microscopic
• 7:56
  grooves. Like the threads on a screw, but
• 7:58
  tiny. Invisible to the naked eye, but
• 8:01
  physics doesn't care if you can see them.
• 8:03
  If those microscopic threads are oriented in the wrong direction
• 8:06
  relative to the shaft rotation, they act like an Archimedes
• 8:09
  screw. So the shaft is essentially a
• 8:11
  pump, a micro pump. It actively pumps the oil under
• 8:15
  the seal lip and out into the world.
• 8:17
  You can have the tightest seal in the world, but if the shaft
• 8:19
  is pumping fluid past it, it's going to leak.
• 8:22
  That is a nightmare to diagnose. How do you even prevent that?
• 8:25
  You have to specify plunge grinding or ensure the polishing
• 8:28
  removes that directional lead. It's a detail that doesn't show
• 8:32
  up on a standard print, but it literally makes or breaks the
• 8:35
  machine. It's wild to think that a
• 8:37
  microscopic groove can defeat a massive industrial pump.
• 8:41
  It's the. Butterfly effect of engineering.
• 8:43
  All right, let's talk timeline. We've covered immediate leaks,
• 8:46
  but what about the failure that happens three months down the
• 8:48
  road? The machine runs fine for a
• 8:50
  month, maybe 2, and then suddenly snap.
• 8:54
  This is fatigue, and This is why rotating equipment is so
• 8:57
  unforgiving compared to structural steel.
• 8:59
  If you build a bridge, the beam just sits there.
• 9:02
  It holds the load. Right, but a shaft is rotating.
• 9:05
  Think about a paper clip. You bend it once, it's fine.
• 9:08
  You bend it back, it's fine. You do that 50 times and it
• 9:11
  snaps. That is fatigue failure.
• 9:14
  And a rotating shaft is basically being bent back and
• 9:17
  forth thousands of times a minute.
• 9:18
  Every single rotation, if there is a bending load, like from a
• 9:21
  heavy belt or a mixer impeller, the top of the shaft is in
• 9:24
  tension and the bottom is in compression.
• 9:26
  Half a turn later, that reverses.
• 9:29
  It's called fully reversed bending stress.
• 9:31
  So even if the stress is well below the yield point, the point
• 9:35
  where it would permanently bend, it can still crack over time,
• 9:39
  correct? It's not a matter of if it
• 9:41
  fails, but when, unless you design for infinite life.
• 9:44
  And that brings us to the sleep at night numbers.
• 9:46
  I love that phrase, the sleep at night numbers.
• 9:48
  Yeah. Our source material, the
• 9:50
  Analysis MW document gives us some specific safety factors for
• 9:54
  this. Right, it suggests that for the
• 9:55
  drive end, the end connected to the motor, you want a safety
• 9:59
  factor greater than 1.5. 1.5 That seems reasonable.
• 10:02
  But for the stub end, the end inside the tank or the process,
• 10:05
  you want a safety factor greater than two point.
• 10:08
  Zero. Why the difference?
• 10:09
  Why does the stub end need to be stronger?
• 10:11
  Because the stub end is the Wild West, it's subjected to impact
• 10:15
  loads from the process. Corrosion, weird hydraulics,
• 10:19
  maybe chunks of solid material hitting the impeller.
• 10:22
  It lives a much harder life. If you are designing under these
• 10:25
  numbers, you are rolling the dice.
• 10:28
  So if Bob wants to sleep at night, he double s the required
• 10:31
  strength on the wet end of the shaft.
• 10:33
  Absolutely. It's cheap insurance compared to
• 10:36
  a catastrophic failure in the field.
• 10:38
  We've covered the design, the geometry, the finish and the
• 10:41
  fatigue, but we have to talk about the wild card, the
• 10:45
  variable that ruins even the best engineering, the human
• 10:49
  element. Installation.
• 10:51
  This is where hearts are broken. You can do everything perfectly
• 10:53
  right in CAD, but if the person installing the bearing is having
• 10:56
  a bad day, it's game over. The sources list a few common
• 11:00
  murder weapons here. The first one is the cocked
• 11:02
  bearing. This is incredibly common.
• 11:04
  You're pressing a bearing onto a shaft.
• 11:06
  Maybe it goes on a little crooked.
• 11:08
  Or, and this is very common, you're tighten the set screws on
• 11:11
  a lock nut. And you just crank them down in
• 11:12
  the circle, right, 1234? And that is exactly how you kill
• 11:16
  a bearing. If you tighten them in a circle,
• 11:19
  you tilt the lock net. You have to use a star pattern,
• 11:22
  just like changing a tire on a car.
• 11:24
  If you don't, the inner ring of the bearing tilts relative to
• 11:28
  the shaft center line. So it's not sitting flat.
• 11:30
  Right, and as it spins, it creates a wobble.
• 11:34
  This leads to fretting corrosion, essentially
• 11:36
  microscopic rubbing between the shaft and the bearing.
• 11:39
  That's where you see that reddish brown dust when you take
• 11:41
  it apart. Exactly.
• 11:43
  That's fretting corrosion. It's the blood spatter of a
• 11:45
  cocked bearing. Eventually it loosens up, the
• 11:48
  shaft starts banging around and you have a total failure.
• 11:51
  Your bearing didn't fail, your installation did.
• 11:53
  Exactly. Another one is cleanliness.
• 11:55
  If you mount a bearing over a Burr or a tiny piece of dirt,
• 11:59
  you've created a stress riser and a misalignment before you
• 12:02
  even turn the switch on. And there's this concept of
• 12:05
  creep. From the SKF sources, this one
• 12:08
  sounds sinister. Creep.
• 12:09
  Is what happens when you're too gentle.
• 12:11
  If you don't have a tight enough interference fit, a press fit on
• 12:14
  the ring that rotates the friction isn't enough to hold
• 12:17
  it. So the shaft turns, but the
• 12:19
  bearing ring slips. It slips slowly, it creeps, and
• 12:24
  as it slips it acts like a lathe, turning down the shaft
• 12:28
  diameter. I have seen shafts that were
• 12:30
  worn down by half an inch because the bearing was spinning
• 12:33
  on the shaft instead of with the shaft.
• 12:35
  Brutal. So the fit tolerance, getting
• 12:38
  that FN one light drive fit or whatever this standard calls for
• 12:41
  is absolutely critical. It's not a suggestion, it's a
• 12:44
  requirement. If the ring rotates relative to
• 12:47
  the load, it must be an interference fit.
• 12:49
  So look, we've dissected the crime scene.
• 12:52
  If we want to solve this, if you want to be the engineer that
• 12:55
  stops the cycle of failure, what are the golden rules?
• 12:58
  It boils down to a philosophy change.
• 13:00
  First, stiffness over strength designed for deflection.
• 13:03
  First, if it's stiff enough for the seal, it's usually strong
• 13:06
  enough for the load. Second, soften the steps.
• 13:09
  Use generous fillets, don't choke the stress flow.
• 13:12
  Third, respect the surface, no machine lead.
• 13:15
  And finally, verify the fit. Don't let a bad install ruin a
• 13:19
  good design. Simple rules, but really hard to
• 13:21
  follow when you're under a tight deadline.
• 13:23
  The laws of physics don't care about your deadline.
• 13:25
  Well said. Now before we wrap this
• 13:27
  investigation, we have to hit the mailbag.
• 13:29
  We pulled the most common googled questions about shaft
• 13:32
  design to get some quick fire answers.
• 13:35
  These are the things people are searching for but might be
• 13:37
  afraid to ask their senior engineers.
• 13:39
  Let's hear them. Question 1.
• 13:41
  Is a hollow shaft weaker than a solid shaft?
• 13:45
  Say Bob wants to save some money on shipping weight, but he's
• 13:47
  worried about strength. The answer is suprisingly not
• 13:51
  much weaker. Think about where the stress is
• 13:54
  in torsion. It's on the outside skin of the
• 13:56
  shaft. The center of the shaft carries
• 13:59
  almost zero stress. So the center is just dead
• 14:02
  weight. Basically, it's just along for
• 14:04
  the ride. If you look at the math, torque
• 14:06
  capacity is a function of the outer diameter.
• 14:08
  Primarily. You can bore out the center,
• 14:10
  save significant mass and cost, and retain most of the torsional
• 14:13
  strength. Are there downsides?
• 14:15
  You have to be careful about wall thickness for buckling if
• 14:18
  the walls get too thin, but generally hollow shafts are a
• 14:21
  great engineering trick. It's why Dr. shafts and cars are
• 14:25
  hollow. Makes total sense.
• 14:27
  Question 2. Do I really need to balance a
• 14:30
  low speed shaft? I mean, if it's only turning 100
• 14:33
  RPM, does it even matter? Yes, this is the low speed myth.
• 14:37
  Even at 100 RPM, an unbalanced rotor creates centrifugal
• 14:41
  forces. Those forces hammer the bearings
• 14:44
  once per revolution. It's.
• 14:45
  Like a slow heartbeat of destruction.
• 14:47
  Exactly. It might not vibrate the floor
• 14:49
  like a high speed turbine, but it is eating the fatigue life of
• 14:52
  your bearings. A family that solves together
• 14:55
  stays together, and balance shafts keep the family of parts
• 14:58
  together. Let's see what she did there.
• 15:00
  Question three. My shaft is worn at the seal.
• 15:04
  There's a groove worn right into the metal.
• 15:06
  Do I have to scrap the whole thing and buy a new shaft?
• 15:08
  Please don't. This is a great maintenance
• 15:10
  hack. You can use a wear sleeve.
• 15:11
  Brands like SKF make these. They're often called speedy
• 15:14
  sleeves. How do they work?
• 15:15
  It's a super thin stainless steel sleeve.
• 15:18
  You press it right over the worn groove.
• 15:20
  It's so thin that you can usually use the exact same size
• 15:23
  seal over it. It gives you a brand new perfect
• 15:26
  surface without machining and without scrapping the shaft.
• 15:30
  That is a massive wallet saver right there.
• 15:32
  Yeah, especially if the shaft is customer huge.
• 15:35
  Absolutely. It's a standard repair procedure
• 15:37
  in many plants. Last question, what happens if I
• 15:40
  ignore the critical speed? Catastrophe.
• 15:42
  Every shaft has a natural frequency, a speed where it
• 15:45
  naturally wants to vibrate like a guitar string.
• 15:49
  If you operate near this critical speed, the vibration
• 15:52
  becomes self exciting. It feeds on itself it.
• 15:54
  Just shakes itself apart. Violently, you must ensure your
• 15:58
  operating speed is at least 20% away from that critical speed.
• 16:01
  You can be above it or below it, but do not hang out near it.
• 16:04
  It is a strict no go zone. So to recap our investigation
• 16:09
  today, the shaft is often the silent killer.
• 16:12
  It fails because of geometry, because it bends too much, or
• 16:16
  because we installed it wrong, rarely just because the torque
• 16:19
  was too high. That's the ultimate take away.
• 16:21
  The best engineers don't just calculate, they inspect the
• 16:24
  install. They look at the whole picture
• 16:26
  from the filet radius to the guy tightening the set screws.
• 16:29
  And here is the final provocative thought for you to
• 16:31
  chew on. With the rise of smart sensors
• 16:34
  and IoT in manufacturing, we're actually entering an era where
• 16:38
  you won't have to guess if a shaft is noodling or if a
• 16:40
  bearing is cocked. Imagine a future where the shaft
• 16:44
  itself tells you it's deflecting past that 0.0015 inch limit
• 16:48
  before the seal ever leaks. The days of the phantom failure
• 16:52
  might be numbered. That is a future I'm looking
• 16:54
  forward to. Me too.
• 16:55
  Well, if you haven't already, get over to our YouTube channel
• 16:58
  Mechanical Engineering Made Simple, where we post hands on
• 17:00
  how to videos, visual breakdowns, and practical
• 17:03
  engineering walkthroughs. You can actually use a family
• 17:05
  that solves together, stays together.