Latest / Mechanical Engineering Made Simple / (#41) From Perfect Shape to Loose Bolt: Mastering the Full Spectrum of Precision Mechanical Engineering

Transcript

(#41) From Perfect Shape to Loose Bolt: Mastering the Full Spectrum of Precision Mechanical Engineering · Mechanical Engineering Made Simple

• 0:00
  Welcome back design junkies from Mechanical Engineering Made
• 0:02
  Simple. Today we're taking a really
• 0:04
  interesting deep dive. We're spanning pretty much the
• 0:08
  whole mechanical life cycle. We'll start with a really
• 0:10
  abstract forward-looking stuff like design optimization, asking
• 0:14
  you know, what is the perfect shape.
• 0:16
  And then we'll bring it right down to the the nuts and bolts
• 0:19
  literally, of making sure power gets transferred reliably.
• 0:23
  Our mission here is basically to give you a shortcut.
• 0:26
  We want you to be really well informed about how that high
• 0:28
  level theory like optimization connects with the the
• 0:32
  fundamental principles you need for precision right there on the
• 0:35
  shop floor. We're connecting the big design
• 0:37
  challenges with the tiny mechanical details.
• 0:39
  Yeah, exactly. And the material we've looked
• 0:41
  at, it really gives a good snapshot of the modern
• 0:43
  engineering landscape we're covering, You know, modern
• 0:45
  computational design ideas, the physics behind ultra precise
• 0:49
  things like flexors. Also industry standards for
• 0:53
  fasteners, tolerances and the practical side of drive systems,
• 0:56
  belts and chains. The goal is quick, solid insight
• 1:00
  and showing how you know a failure in just one little area,
• 1:02
  maybe a bad algorithm or even just a loose bolt.
• 1:05
  It can bring down the whole system.
• 1:06
  OK, let's kick off with efficiency.
• 1:08
  It's a huge topic. We talk about maximizing
• 1:11
  performance but using way less material.
• 1:13
  We're talking to policy optimization, and this is where
• 1:16
  it gets really cool, I think. It's not just tweaking a shape,
• 1:19
  it's almost like growing a structure with an algorithm.
• 1:22
  That's a great way to put it. We're essentially asking the
• 1:24
  software to come up with designs without our usual human bias
• 1:28
  towards, you know, straight lines and simple curves.
• 1:31
  And the results often look, well, organic, kind of like
• 1:34
  lattices. The impact is way more than just
• 1:37
  saving money on materials. By boosting performance and
• 1:40
  cutting waste, you get a real environmental benefit that's
• 1:43
  critical in fields like aerospace or automotive.
• 1:46
  It sounds amazing, like perfect designs automatically, but
• 1:50
  getting there the the model the computer uses must be incredibly
• 1:54
  complex, right? Oh, absolutely.
• 1:56
  It has to be what we call an integrated model.
• 1:58
  It needs to juggle 3 key things all at once.
• 2:01
  You've got complex load conditions, how the specific
• 2:03
  material actually behaves, often nonlinearly, and really
• 2:07
  importantly, the real world limits of manufacturing.
• 2:09
  Can you actually make this shape?
• 2:11
  This puts a lot of the hard weight upfront in the design
• 2:13
  phase. It basically gets rid of that
• 2:15
  expensive time wasting trial and error stuff later on.
• 2:17
  So how does the software find these shapes?
• 2:19
  It can look pretty wild. Is it just standard finite
• 2:21
  element analysis? Yeah.
• 2:23
  It's a it's kind of the next level.
• 2:25
  Beyond basic FEA often involves methods borrowed from AI.
• 2:29
  Think sophisticated stuff like genetic algorithms, particle
• 2:33
  swarm optimization, things like that.
• 2:35
  These let the system explore just millions of potential
• 2:38
  structural layouts, finding the one that hits the performance
• 2:41
  targets while being super efficient with material OK.
• 2:43
  If it's that powerful, why isn't everything designed this way
• 2:47
  now? Why hasn't it completely taken
• 2:49
  over from human designers? That's a really good practical
• 2:52
  point. The big hurdle is
• 2:54
  standardization. Different fields have totally
• 2:57
  different needs, right? A part optimized for stiffness
• 2:59
  in a building is worlds apart from one designed to minimize
• 3:02
  thermal changes in, say, a semiconductor machine.
• 3:05
  So because there's no single set of rules, the best optimization
• 3:08
  method changes depending on the job.
• 3:10
  Trying to get one standard approach to work everywhere,
• 3:12
  it's still really tricky. The hysics are just too
• 3:14
  different all. Right, let's shift gears.
• 3:16
  We've otimized the structure. Now how about controlling
• 3:19
  movement with extreme recision? This takes us into the realm of
• 3:23
  flexure systems. These are really elegant, often
• 3:26
  single piece parts that act like joints or bearings, but without
• 3:30
  the sliding parts. Exactly.
• 3:31
  The big win with flexures is their pure elastic behavior Yeah
• 3:36
  no friction, no stiction, stick slip problem and basically
• 3:39
  nowhere. Those are the big enemies if you
• 3:41
  need repeatability. But to get to the ultimate goal
• 3:44
  like Angstrom level repeatability, well, the design
• 3:47
  rules are incredibly. Straight angstrom.
• 3:48
  OK, that's that's mind bogglingly small. 1/10 of a
• 3:51
  nanometer, like positioning something within the width of a
• 3:54
  few atoms. What's the absolute must have
• 3:56
  physical requirement to even get close to that?
• 3:59
  It really comes down to the materials physics.
• 4:01
  First, you need incredibly low hysteresis in the material.
• 4:05
  It has to spring back perfectly forgetting it was ever bent.
• 4:09
  And the absolute critical part, you have to make sure there's
• 4:11
  zero dislocation motion inside the materials crystal structure.
• 4:16
  That means designing so the stress Sigma is kept much, much
• 4:20
  less than the yield stress Sigma.
• 4:22
  Yield. Normal engineering might push
• 4:24
  closer to yield. For this kind of precision, you
• 4:26
  need a huge safety margin to guarantee it stays purely
• 4:29
  elastic. OK, that explains why material
• 4:33
  choice is so key. Things like single crystal
• 4:35
  materials become really attractive.
• 4:37
  So once you've got the material right, how do you make sure the
• 4:39
  part only moves the way you want it to?
• 4:41
  Just along one specific axis, for example.
• 4:44
  That's where kinematics comes in.
• 4:46
  There's a fundamental principle called Blanding's Rule of
• 4:48
  Complementary patterns. It's basically a geometric law
• 4:51
  for designing constraints. It says every allowed line of
• 4:54
  rotation has to intersect every constraint surface.
• 4:57
  It's the rule book that dictates precisely how motion is allowed
• 5:00
  or starped, making sure you get just that single predictable
• 5:03
  degree of freedom you want. And once the geometry is
• 5:05
  defined, we can predict how stiff or compliant that motion
• 5:09
  will be. The classic example is the
• 5:11
  cantilever beam deflection equation, right?
• 5:13
  It links the shape and material properties.
• 5:15
  Does. If you look at that formula,
• 5:17
  deflection delta equals the force F times the length l ^3 /
• 5:21
  3 times Young's modulus E times the moment of inertia I.
• 5:25
  The key thing to notice there is how sensitive deflection is to
• 5:28
  geometry. Length is cubed, but the
• 5:31
  material property E is just a linear factor.
• 5:33
  It really highlights that in precision mechanics, getting the
• 5:36
  geometry right often has a bigger impact on stiffness than
• 5:39
  just picking a different material right.
• 5:41
  Let's pivot now. Maybe it's something less
• 5:43
  exotic, but just as vital. Bolted joints.
• 5:45
  When you tighten a bolt, you're clamping 2 flanges.
• 5:48
  Now think about the flange material and the bolt itself.
• 5:50
  They both stretch or compress a bit.
• 5:52
  They act like springs, responding to that preload
• 5:55
  force. A common question is, are those
• 5:57
  springs acting in parallel or are they in series?
• 6:00
  OK, they both experience the same preload force, but the bolt
• 6:04
  stretches and the flange compresses, so they have
• 6:07
  different displacements. Yeah, that means they have to be
• 6:09
  in series. Right, exactly.
• 6:10
  They're in series, and understanding that is
• 6:12
  fundamental if you need to calculate the overall stiffness
• 6:15
  of the joint. Now think practically.
• 6:18
  What about shear loads? An external force tries to slide
• 6:22
  the two clamped parts sideways. What actually stops them from
• 6:25
  moving? Well, it's the friction between
• 6:28
  the surfaces isn't it? Generated by that high clamping
• 6:31
  force from the preload? It's usually not the bolt Shank
• 6:34
  taking the shear load directly. Precisely.
• 6:36
  Friction does the heavy lifting for shear.
• 6:38
  If you have really high shear loads, you can't rely on
• 6:41
  friction alone. That's when you need to add
• 6:43
  things like separate dowel pins, or maybe use special shoulder
• 6:46
  bolts designed specifically to take that sideways load.
• 6:49
  But none of it works if that initial preload isn't correct.
• 6:53
  And getting that preload right? That's tricky.
• 6:56
  Using a torque wrench is common. It's easy, but the sources we
• 7:00
  looked at confirmed it's definitely not the most ultra
• 7:02
  repeatable. Way too many variables like
• 7:05
  friction under the bolt head and then the threads.
• 7:07
  It's really just an estimation based on torque, not a direct
• 7:10
  measurement of the tension in the bolt.
• 7:12
  For real precision you'd want to measure the actual stretch of
• 7:15
  the bolt, or maybe use strain gauges, but that gets
• 7:18
  impractical or expensive pretty quickly.
• 7:20
  If you can't measure that preload accurately, you're
• 7:23
  basically guessing about the joints.
• 7:25
  Reliability verification is key. And Speaking of verification,
• 7:29
  moving from forces to geometry, we need ways to check that our
• 7:33
  parts are made and assembled precisely.
• 7:35
  Which brings us to a real classic tool for checking
• 7:37
  angles, the 10 inch sign bar. For anyone listening who hasn't
• 7:40
  used one, it's a precision ground bar with two cylinders
• 7:43
  exactly 10 inches apart used on a surface plate with gauge
• 7:46
  blocks. It's such an elegant tool
• 7:48
  because it turns measuring an angle into a simple linear
• 7:51
  measurement. You put gauge blocks under one
• 7:54
  cylinder, creating a right angle triangle.
• 7:57
  The 10 inch bar is the hypotenuse and the stack of
• 7:59
  gauge blocks, let's call it height.
• 8:00
  H is the side opposite the angle you're setting or measuring.
• 8:04
  And here's the mental shortcut every machinist knows because
• 8:08
  the bar is 10 inches long. To find the height H needed for
• 8:11
  a certain angle, you just take the sine of that angle and
• 8:14
  multiply by 10. Or even simpler, just find the
• 8:16
  sign and shift the decimal .1 place to the right.
• 8:19
  No complex calculations needed. Right there on the inspection
• 8:22
  table. It's brilliant for quick,
• 8:24
  accurate checks. Yeah, that ability to translate
• 8:27
  a mathematical idea into a practical, repeatable
• 8:29
  measurement is really the heart of precision engineering.
• 8:32
  And underpinning all this verification are tolerance
• 8:35
  fundamentals. You have standards like unified
• 8:37
  threads, metric and profile threads, and those basic
• 8:40
  classifications of fit. You know, clearance, transition,
• 8:42
  interference fits. These define the essential
• 8:45
  boundaries for how parts mate together.
• 8:47
  Controlling those tolerances is what makes mass production
• 8:50
  actually work reliably. OK, we've got structurally sound
• 8:54
  parts precisely made and joined. Now let's talk about actually
• 8:58
  transmitting power reliably. Let's look at synchronous belt
• 9:01
  drives first. Things like timing belts.
• 9:04
  They're crucial. You need precise speed ratios or
• 9:07
  timing. Right when you're sizing one of
• 9:10
  these systems, a key calculation is the operating torque Q.
• 9:13
  You figure out torque Q in pound inches by taking the #63,025,
• 9:19
  multiplying that by the shaft horsepower, and then dividing
• 9:22
  the whole thing by the shaft RPM.
• 9:24
  And importantly, you generally base these calculations on the
• 9:26
  smaller sprocket, the one that spins faster, because that's
• 9:29
  usually the limiting factor for torque capacity.
• 9:32
  And belts don't always see a perfectly smooth load, right?
• 9:35
  So you need to build in a safety margin, a service factor.
• 9:37
  What's typical? Yeah, the recommended service
• 9:39
  factor is usually somewhere between say 1.5 and two point O.
• 9:43
  If you've got really harsh conditions or lots of shock
• 9:45
  loads, maybe constant starting and stopping, you definitely
• 9:49
  want to lean towards the higher end, towards 2.0, maybe even
• 9:52
  more, just to prevent early failure.
• 9:54
  What usually causes problems with these drives in the real
• 9:56
  world? What makes them fail or perform
• 9:58
  poorly? By far the biggest issue is Dr.
• 10:01
  misalignment. It comes in two main flavors.
• 10:04
  There's angular misalignment, that's when the shafts aren't
• 10:06
  perfectly parallel. And then there's parallel
• 10:08
  misalignment, where the shafts are parallel but the pulleys or
• 10:12
  sheaves aren't lined up in the same plane.
• 10:15
  Both of these create really big unnecessary forces trying to
• 10:19
  push the belt sideways. That leads to overheating,
• 10:22
  uneven wear, and definitely more noise.
• 10:25
  Speaking of noise, are synchronous belts just naturally
• 10:29
  louder? Historically, yeah, they tend to
• 10:31
  be noisier than, say, V belts. That's because you have rigid
• 10:34
  teeth meshing together, but manufacturers have been working
• 10:37
  on that. Some newer designs, like the
• 10:39
  Power Grip GT3 drives mentioned in our sources, they're
• 10:42
  significantly quieter. They have refined tooth shapes
• 10:45
  that mesh more smoothly, so you get high power density but less
• 10:48
  noise. OK, let's switch over to chains.
• 10:50
  Now. Chains often pull double duty
• 10:52
  right power transmission but also used a lot for conveying
• 10:55
  things. When does a chain absolutely
• 10:57
  need replacing? What's the tell tale sign?
• 11:00
  The definitive limit is all about chain elongation, but
• 11:03
  specifically the elongation caused by wear in the pins and
• 11:06
  bushings where the joints articulate.
• 11:08
  For a chain used in power transmission, that stretch
• 11:11
  measured over a length must be less than or equal to 1.5% of
• 11:17
  its original length. For a conveyor chain, you get a
• 11:19
  little more leeway. Up to 2% elongation is
• 11:22
  acceptable. Go beyond those limits and the
• 11:24
  risk of the chain jumping teeth or just breaking goes way up.
• 11:28
  And that where is why lubrication is usually so
• 11:31
  critical, and also why we have specialty chains designed to
• 11:34
  minimize that need, right? Absolutely.
• 11:36
  Take something like lambier roller chains.
• 11:38
  They're designed to be Lube free, they use special burnings
• 11:41
  made of centered metal that are impregnated with oil, and the
• 11:44
  pins have special coatings. The result, especially at lower
• 11:47
  scooths where keeping standard Lube in place can be tough, is
• 11:50
  potentially much longer, where life may be up to 30 times
• 11:53
  longer than a standard chain in some cases.
• 11:56
  What? About really tough environments
• 11:57
  like around water or even saltwater.
• 11:59
  Yeah, for corrosion resistance, especially against water, you
• 12:02
  have options like WP roller chains.
• 12:04
  These use a special mechanical coating process.
• 12:07
  The key thing here is that it's a mechanical process, not
• 12:10
  chemical plating that often involves heat.
• 12:12
  Avoiding high heat means the chain parts keep their core
• 12:15
  hardness and the tight fits between pins and plates, which
• 12:18
  are crucial for maintaining the chain strength and wear
• 12:20
  resistance. Heat treatment during plating
• 12:23
  can sometimes compromise those properties.
• 12:25
  OK, so wrapping this up, what's the big picture from this deep
• 12:28
  dive? We've covered a huge range, from
• 12:32
  using algorithms to literally grow optimal shapes, to
• 12:35
  controlling motion down to the atomic level with flexures and
• 12:39
  then ensuring power gets from A to B reliably with the right
• 12:42
  belts and chains. It's quite the spectrum.
• 12:44
  It really is. And I think if there's one core
• 12:46
  take away for you, the designer listening, it's about synthesis.
• 12:49
  It's understanding that the abstract math, whether it's a
• 12:52
  topology optimization result or that cantilever deflection
• 12:55
  formula, it's only truly valuable if you can also control
• 12:58
  the smallest physical details on the ground.
• 13:00
  You could have theoretically the most perfectly optimized part
• 13:04
  shape imaginable, but if you fail to control the stress in a
• 13:08
  flexure keeping Sigma much much less than Sigma yield, or you
• 13:13
  ignore the proper thread fit tolerance on just one critical
• 13:17
  fastener, that one small oversight can undermine the
• 13:20
  entire beautifully designed system.
• 13:23
  Precision engineering really means mastering the micro level
• 13:26
  details to make the macro level performance actually happen.
• 13:29
  That's a really powerful thought.
• 13:30
  The perfect theory needs perfect execution, right down to the
• 13:33
  smallest detail. Go forth and keep exploring
• 13:35
  this. Fascinating, interconnected
• 13:36
  world of recision engineering. We'll catch you on the next Dee
• 13:39
  dive.