Latest / Mechanical Engineering Made Simple / (#39) Three Pillars of Precision

Transcript

(#39) Three Pillars of Precision · Mechanical Engineering Made Simple

• 0:00
  Welcome back design junkies, to Mechanical Engineering Made
• 0:02
  Simple. Today we're diving deep, really
• 0:05
  deep into a whole stack of material covers everything from,
• 0:09
  you know, basic handbook stuff, tolerances, fits, hardware right
• 0:14
  up to the cutting edge like conceptual modeling and even
• 0:16
  nano manufacturing ideas. Yeah, it's quite the range, a
• 0:19
  huge amount of technical knowledge, but our mission
• 0:22
  really is to boil all that dense info down into something
• 0:26
  actionable for you. We're going to focus on what
• 0:28
  makes or breaks modern design today.
• 0:30
  The three pillars, if you like, which are high efficiency,
• 0:33
  mastering ultra precision kinematics, and crucially,
• 0:37
  sticking to real world manufacturing standards.
• 0:39
  You can't ignore the reality of how things get built.
• 0:42
  Absolutely. OK, let's kick things off with
• 0:44
  that first pillar efficiency. It feels like in modern
• 0:47
  engineering, we're not just aiming for strong anymore, but
• 0:49
  smart, right? Exactly.
• 0:50
  Smart and lean. The big conceptual shift,
• 0:53
  especially if you look at aerospace or automotive, is all
• 0:56
  about topology optimization. OK, so topology optimization,
• 0:59
  what's the the the core goal there?
• 1:01
  What's the nonnegotiable objective it?
• 1:03
  Sounds simle almost elegant. Really maximize the structural
• 1:06
  erformance while, and this is key, minimizing the amount of
• 1:10
  material you use aggressively. Like nature does bones or trees,
• 1:14
  that kind of thing. Precisely.
• 1:15
  It's bio mimicry in a way. Put material only with the load
• 1:18
  path actually needs it and nowhere else.
• 1:21
  And the payoff, is it just saving a bit of weight here and
• 1:23
  there or something bigger? Oh, it's systemic.
• 1:25
  Way bigger than just saving a few grams.
• 1:27
  The immediate benefits are well, first, significant cost savings.
• 1:31
  Less material, less cost. Then there's the reduced
• 1:34
  environmental impact. Less material means less energy
• 1:38
  to produce it transport. It and better performance too.
• 1:41
  Lighter usually means faster or more agile.
• 1:43
  Enhanced functionality? Exactly.
• 1:45
  Yeah, light weighting is huge. So yeah, it's driving
• 1:47
  sustainability and cost effectiveness in any sector
• 1:51
  using complex parts. So how does the software
• 1:53
  actually find these, you know, perfect shapes?
• 1:56
  Is it just fancy FEA? It goes way beyond standard FEA
• 2:00
  simulation. These conceptual models use
• 2:02
  pretty advanced optimization techniques.
• 2:04
  Think genetic algorithms, particle swarm optimization,
• 2:08
  that sort of thing. These let the software explore
• 2:11
  literally millions of potential material layouts.
• 2:14
  It finds solutions that honestly, a human designer might
• 2:17
  never even think of. They look strange sometimes, but
• 2:21
  they're structurally ideal. Minimizes prototyping too, I
• 2:24
  imagine. Big time cuts down on costly
• 2:26
  physical prototypes. Right, so this sounds incredibly
• 2:29
  powerful. Why isn't every single company
• 2:32
  using this for every single part starting tomorrow?
• 2:35
  What's the holdup? Ah, that's the $1,000,000
• 2:37
  question. The real friction point is
• 2:40
  standardization, or rather the lack of it.
• 2:42
  Well, think about it. Aerospace has different material
• 2:45
  needs and failure modes than say, medical implants or
• 2:48
  automotive chassis. Each sector has unique
• 2:50
  requirements, unique materials, unique constraints.
• 2:53
  So integrating these optimized models into existing
• 2:56
  standardized manufacturing workflows gets really
• 2:59
  complicated. You often need custom post
• 3:01
  processing, extra verification steps.
• 3:03
  It's not quite plug and play yet.
• 3:05
  So the ideal digital model meets the messy reality of the factory
• 3:08
  floor. That's it exactly.
• 3:10
  Bridging that gap is the challenge.
• 3:12
  OK, fascinating. So for macro efficiency, let's
• 3:15
  pivot hard to the micro world, ultra precision when you need
• 3:19
  motion control down at the angstrom level, we're talking
• 3:22
  nanotechnology, semiconductor lithography, those traditional
• 3:26
  sliding joints with friction are just out right.
• 3:29
  Completely out friction, backlash, stiction, They're
• 3:33
  killers at that scale. That's where flexure bearing
• 3:36
  systems come in. They're the go to solution.
• 3:38
  Because they bend, not slide. Right motion comes from the
• 3:41
  material elastically deforming, so you get incredibly smooth,
• 3:45
  fine movement, no friction, no backlash.
• 3:47
  They are pure kinematic systems in essence.
• 3:50
  But getting that Angstrom level repeatability, that sounds
• 3:54
  incredibly difficult. You can't just bend any old
• 3:56
  piece of metal, surely? What's the checklist?
• 3:58
  No you definitely can't. There's some non negotiable
• 4:00
  conditions you absolutely have to meet 4 main ones.
• 4:03
  OK, first the material itself needs low hysteresis.
• 4:05
  You don't want it remembering previous stresses.
• 4:07
  That's why you see specialized materials, sometimes even single
• 4:10
  crystal silicon, in extreme precision ABS.
• 4:13
  Right. Minimizes internal friction.
• 4:15
  Exactly. Second, no dislocation motion in
• 4:18
  the material's crystal structure.
• 4:20
  That means your operating stress has to be significantly less
• 4:22
  than the yield stress. Keep it purely elastic.
• 4:25
  So way below yield. Got it. 3rd, the load has to be
• 4:28
  super repeatable. Same magnitude, same direction
• 4:31
  every single time. Any variation throws off your
• 4:34
  position. Makes sense?
• 4:35
  And the fourth, and this one is critical, linking back to
• 4:37
  manufacturing reality again. Perfect assembly.
• 4:41
  Any tiny bit of micro slip where parts join or any unintended
• 4:45
  friction introduced. When you bolt it all together,
• 4:47
  boom, your angst from repeatability is gone.
• 4:50
  Wasted. That really drives home how
• 4:52
  design and manufacturing are inseparable at this level.
• 4:55
  Absolutely. OK, let's sit back for a second.
• 4:57
  The fundamental framework for all motion is kinematics.
• 5:00
  Constraints remind us of the core principle there.
• 5:04
  It boils down to a really simple but powerful equation degrees of
• 5:07
  freedom. Let's call it r = 6 minus the
• 5:10
  number of constraints C. So R equals six C = F. 6 being
• 5:15
  the possible motions in 3D space, translate XYZ, rotate
• 5:19
  around XYZ. Correct, and by applying
• 5:22
  constraints you remove freedoms. The goal is usually to get R
• 5:26
  down to one for predictable single axis motion, and if you
• 5:30
  want that predictable motion, especially linear motion from
• 5:32
  rotational constraints, you must follow Blanding's rule of
• 5:36
  complementary patterns. OK, don't just name the rule.
• 5:39
  What happens if you break it? What's the engineering
• 5:41
  consequence? Bad things happen.
• 5:43
  Blanding's rule basically says that for pure rotation, the
• 5:46
  constraint lines must all intersect the desired rotation
• 5:49
  axis, and for pure translation they must be parallel.
• 5:53
  If you violate that, say your constraint lines don't intersect
• 5:56
  the rotation axis properly, you're not just getting a bit of
• 5:59
  error, you're introducing over constraint.
• 6:01
  Meaning. Meaning the part has to deform
• 6:03
  and unpredictable ways just to move it all.
• 6:06
  You lose predictability. You introduce parasitic motion
• 6:09
  for angstrom level work that's fatal.
• 6:11
  You build chaos into your system.
• 6:13
  Wow. So from the clean, rigid world
• 6:15
  of kinematic points and lines, we instantly hit the physics of
• 6:19
  the material itself, Elasticity. How do we bridge that?
• 6:22
  How do we mathematically ensure our flexure is stiff enough but
• 6:25
  still compliant enough? That bridge is compliance or
• 6:28
  it's inverse stiffness. We use formulas linking the
• 6:31
  geometry and the material properties.
• 6:34
  Take a simple example, a cantilever beam deflection.
• 6:36
  The deflection delta is proportional to the source F
• 6:40
  times the length cubed l ^3 / 3 times Young's modulus, E times
• 6:45
  the area moment of inertia I so delta SC dot L33.
• 6:50
  OK, I remember that one. What that formula shouts at you
• 6:52
  is that precision design is always about two things, the
• 6:55
  materials inherent stiffness. It's Young's modulus E and the
• 6:58
  shape's rigidity capture by I the geometry.
• 7:01
  You have to engineer both, right?
• 7:03
  Material and shape, you got it. OK, so from the theoretical
• 7:06
  purity of a flexure, let's jump right into the heart of
• 7:09
  manufacturing reality, the humble, ubiquitous bolted joint.
• 7:13
  We need them stiff, especially preloaded ones.
• 7:16
  Now here's a classic engineering question.
• 7:17
  Maybe even a trap. When you calculate the
• 7:19
  equivalent stiffness of that preloaded joint, the bolt
• 7:22
  stretching, the flange compressing, are those
• 7:24
  stiffnesses acting in series or in parallel?
• 7:26
  I yes, the classic trap. They look like they're in
• 7:29
  parallel side by side, but for the stiffness calculation they
• 7:32
  act in series. Why?
• 7:34
  Explain that. Because the defining
• 7:36
  characteristic of elements in series for stiffness is that
• 7:39
  they share the same force. The preload force is the same in
• 7:42
  the bolt causing stretch and in the flange causing compression.
• 7:46
  But the displacements are different.
• 7:47
  Exactly same force, different displacements means they combine
• 7:50
  like springs in series. If you calculated assuming
• 7:53
  parallel, your prediction for the joints dynamic response will
• 7:57
  be completely wrong. OK series it is.
• 7:59
  Good clarification now another common bolted joint
• 8:02
  misconception. Sheer resistance.
• 8:05
  If I play load sideways perpendicular to the bolt axis,
• 8:09
  is the bolt Shank taking that shear load?
• 8:12
  Usually no, or rather, it's designed not to.
• 8:15
  In a properly preloaded joint, the shear load is almost
• 8:18
  entirely resisted by the friction between the clamp
• 8:21
  member. Friction again.
• 8:22
  Yes. The critical factors are the
• 8:24
  coefficient of friction between the surfaces and crucially, that
• 8:27
  preload force F preload clamping them together.
• 8:31
  The bolt's job is primarily to provide that clamp load.
• 8:34
  So if you really need positive shear location, relying on
• 8:38
  friction isn't enough. Definitely not reliable enough
• 8:40
  for precision location. For true positive shear
• 8:43
  resistance you absolutely need dedicated features like
• 8:46
  precision dowel pins or shoulder bolts.
• 8:49
  They physically prevent sliding. Right.
• 8:52
  And getting that preload, that clamp force correct is vital for
• 8:56
  the friction to work. But you mentioned measuring
• 8:58
  stretch directly is often impractical on an assembly line.
• 9:02
  Strain gauges are expensive. So how do we typically measure
• 9:05
  preloads in practice? We usually default to the least
• 9:09
  precise method. Ironically, for one of the most
• 9:12
  critical parameters, we measure the applied torque on the bolt
• 9:15
  header nut. Torque wrenches.
• 9:17
  Torque wrenches? Automated drivers.
• 9:18
  Yeah, it's fast, it's cheap, it's easy to implement on a
• 9:21
  production line. But.
• 9:22
  But it's notoriously not ultra repeatable.
• 9:24
  Why? Because the actual preload you
• 9:27
  get for a given torque depends heavily on the friction under
• 9:30
  the bolt head and in the threads, and that friction
• 9:33
  varies with surface finish, lubrication, dirt, everything.
• 9:36
  So we're accepting a fair bit of uncertainty there.
• 9:38
  A huge amount in many industries.
• 9:41
  It's a calculated risk, a trade off between precision, cost and
• 9:45
  speed. It works most of the time, but
• 9:48
  it's far from perfect. And because of that
• 9:50
  imperfection, standards become absolutely critical.
• 9:53
  They're the safety net defining how parts should fit together,
• 9:56
  even with these variations. Exactly right.
• 9:59
  Standards are the bedrock of reliable assembly.
• 10:02
  Think about screw threads, unified inch threads, metric
• 10:05
  threads. They have rigorous
• 10:06
  classification systems like 1A2A3A for external inch
• 10:10
  threads, 1B2B3B for internal, or 6G6 out for common metric fits.
• 10:16
  These dictate the quality and clearance.
• 10:18
  And beyond threads the fits between shafts and holes.
• 10:20
  Same story, governed by standers, ANSI ISO.
• 10:24
  They define tolerance zones and allowances to create different
• 10:26
  types of fits. You have clearance locational
• 10:28
  fits for parts that need to assemble easily, transition fits
• 10:31
  where it might be a slight clearance or slight interference
• 10:33
  for snug location, and then the interference fits.
• 10:35
  Force fits Shrink fits where you need significant force or
• 10:38
  temperature difference to assemble them.
• 10:40
  Ensuring parts made anywhere in the world can actually connect
• 10:42
  as intent. That's the goal, predictable
• 10:45
  interfacing. OK, let's wrap up by looking at
• 10:48
  actually transmitting power and motion belts and chains.
• 10:52
  We often see synchronous belts like timing belts compared to
• 10:56
  traditional V belts. What's the niche for synchronous
• 10:59
  belts and what's their main trade off, especially noise?
• 11:03
  Synchronous belts. You mentioned Power Group GT3
• 11:06
  HTD is another common type. Their big advantage is positive
• 11:10
  non slip engagement. They're for timing applications
• 11:13
  where you need precise positional control.
• 11:16
  The trade off is typically noise.
• 11:18
  That meshing action of the belt teeth engaging the sprocket
• 11:20
  teeth generates noise especially as speed and belt with increase.
• 11:24
  Is anything being done about? That Oh yeah, manufacturers are
• 11:27
  constantly improving tooth profiles.
• 11:28
  The Gates GD3 profile, for instance, was specifically
• 11:31
  designed for better, quieter meshing compared to older
• 11:34
  designs like HTD. Using larger sprockets also
• 11:37
  helps reduce noise. And since they rely on that
• 11:40
  precise tooth meshing, alignment must be super critical, right?
• 11:43
  What happens if things are slightly off?
• 11:45
  Oh, misalignment is an absolute killer for synchronous belts and
• 11:49
  V belts too. Actually, even slight angular or
• 11:52
  parallel misalignment causes high tracking forces on the belt
• 11:55
  trying to pull it off the sprockets.
• 11:57
  This leads to edgeware, increased noise and eventually
• 12:00
  catastrophic failure of the tensile chords inside the belt.
• 12:03
  Are there hard numbers for acceptable misalignment?
• 12:06
  There are. For something like a standard V
• 12:09
  belt or micro V belt, the general rule is miss alignment
• 12:12
  shouldn't exceed about 14 of a degree angularly, or maybe 116
• 12:16
  of an inch parallel offset per foot of distance between the
• 12:19
  shafts. Exceed that and you're
• 12:20
  drastically shortening the belt's life, guaranteed.
• 12:23
  Wow, that's tight. And getting the installation
• 12:26
  tension right is obviously key to.
• 12:28
  Why is that initial tension so important?
• 12:30
  It does 3 crucial things. First, it provides the grip
• 12:34
  needed to transmit the torque without the belt ratcheting or
• 12:36
  jumping teeth under peak load. Second, it has to compensate for
• 12:40
  the natural tension decay that occurs as the belt seats itself
• 12:42
  into the grooves right after installation.
• 12:44
  You over tension slightly initially.
• 12:46
  And 3rd it helps pre stress the dry frame and maintained overall
• 12:50
  system stiffness. How do you measure it
• 12:52
  accurately? Often it's done with simple
• 12:54
  force deflection methods. Push on the belt span with a
• 12:57
  known force and measure how much it deflects.
• 13:00
  Or for more critical drives where you know the operating
• 13:03
  torque queue, you can calculate the required static tension
• 13:07
  using specific formulas provided by the manufacturer.
• 13:09
  OK quick switch to roller chains.
• 13:12
  What are their main jobs and when does an engineer say OK
• 13:15
  this chain is done? Roller chains are workhorses, 2
• 13:18
  main functions, power transmission obviously and also
• 13:20
  conveyance, moving stuff along assembly lines.
• 13:22
  For example, standardization like ANSIISOJIS make selecting
• 13:28
  them pretty easy. And wear limits.
• 13:31
  They're surprisingly durable, but they do stretch over time as
• 13:34
  the pins and bushings wear. The rule of thumb is replace a
• 13:38
  power transmission chain when its length is increased by about
• 13:41
  1.5% due to wear. For a conveyor chain which often
• 13:45
  runs slower, you might let it go to 2% elongation.
• 13:47
  Beyond that, things get inefficient or risk jumping
• 13:50
  teeth. Exactly.
• 13:51
  Reduced efficiency, potential for sprocket damage, and
• 13:53
  eventual failure. Is there anything to help with
• 13:55
  maintenance? Specifically, lubrication chains
• 13:58
  need oil, right? Lubrication is absolutely vital
• 14:01
  for chain life. It reduces wear, dissipates
• 14:04
  heat, cushions impact. But yes, some applications make
• 14:08
  frequent lubrication difficult. That's where specialty chains
• 14:11
  come in. Things like Soubaki's Lambie Day
• 14:13
  series or similar products from other brands.
• 14:15
  They use oil impregnated centered metal bushings around
• 14:19
  the pins. So they're self lubricating.
• 14:21
  Effectively yes, for a period. They offer significantly longer
• 14:25
  wear life without needing external re lubrication,
• 14:28
  especially at low to medium speeds.
• 14:30
  Big maintenance saver. One important note though,
• 14:33
  o-ring chains often used on motorcycles seam lubrication
• 14:36
  free but are usually not great for industrial conveyance.
• 14:39
  Why not the O-ring seal lubricant in which is good, but
• 14:43
  they also add friction and stiffness to each joint.
• 14:45
  After the chain articulates around a sprocket, it might
• 14:48
  resist straightening out properly, which is bad for
• 14:51
  conveyors. Good distinction.
• 14:53
  And even with standard chains, lubrication needs aren't
• 14:55
  constant. Right.
• 14:57
  You generally need to lubricate more frequently right after
• 14:59
  installation as the parts break in, and definitely more often
• 15:03
  under harsh conditions. High speed, heavy loads, high
• 15:06
  temperatures, dirty environments.
• 15:08
  Tailor the schedule to the conditions.
• 15:10
  OK, we have covered a massive amount of ground today.
• 15:13
  It's really clear how these different areas have to mesh,
• 15:16
  doesn't it? From, you know, the high level
• 15:19
  computational optimization, trying to sculpt the perfect
• 15:22
  lightweight part down to the angstrom level precision needed
• 15:25
  for flexures, and then grounding it all in the reality of bolted
• 15:29
  joints tolerances and making sure belts and chains actually
• 15:33
  deliver the power reliably. It really does highlight that
• 15:36
  constant interplay, doesn't it, between the conceptual ideal and
• 15:40
  the, let's say, practical realities of manufacturing and
• 15:44
  assembly. And thinking back to where we
• 15:46
  started with topology optimization, aiming for that
• 15:49
  perfect mathematically ideal shape and material density, it
• 15:53
  brings up a really interesting final thought for you to
• 15:55
  consider. Given the topology optimization
• 15:57
  generates these, well, often complex and theoretically
• 16:01
  perfect forms, how does that conceptual design process need
• 16:05
  to evolve? How can it start to proactively
• 16:08
  model and actually account for the real world limitations we
• 16:12
  just spent time discussing? You mean things like?
• 16:14
  Things like the inevitable micro slip in assemblies, the
• 16:17
  practical alignment tolerances you can hold on a factory.
• 16:20
  The material history is is that limits flexure precision, the
• 16:23
  uncertainty in bolt preload from using torque.
• 16:26
  So building the imperfections into the optimization loop from
• 16:30
  the start. Exactly how do we bridge that
• 16:32
  gap, not just in post processing, but right from the
• 16:34
  initial design concept? Because closing that gap between
• 16:37
  the perfect simulation and the imperfect real world
• 16:40
  manufactured part that feels like we're a lot of the next big
• 16:43
  engineering innovations are going to happen.
• 16:45
  A fascinating challenge indeed, building the real world into the
• 16:48
  digital twin more effectively. To my fellow design junkies, I
• 16:51
  want to say thank you. The show exists because of you,
• 16:54
  the ones who chase precision, who think in torque curves and
• 16:57
  tolerances instead of trends. If you enjoy the show, please
• 17:01
  let us know with a like and a comment.