Latest / Mechanical Engineering Made Simple / (#45) From Stone Piers to Supercomputers: Unpacking the Engineering Secrets of Strength, Stiffness, and Structural Collapse

Transcript

(#45) From Stone Piers to Supercomputers: Unpacking the Engineering Secrets of Strength, Stiffness, and Structural Collapse · Mechanical Engineering Made Simple

• 0:00
  Welcome back to the Deep Dive. Today we're tackling something
• 0:03
  that's, well, all around us, but maybe not always obvious.
• 0:07
  Structural analysis. Basically, the science of why
• 0:10
  things stay up. Or sometimes why they fall down.
• 0:14
  It's a really fascinating field underpinning, you know, pretty
• 0:16
  much everything built. We're digging into some Core
• 0:18
  Materials day on strength of materials and structures.
• 0:21
  Yeah, it can look pretty dense, lots of math, but the concepts
• 0:24
  are crucial. Our mission today is to unpack,
• 0:28
  let's say, 3 big areas. First, how different forces
• 0:32
  combine in structural parts, leading to this kind of classic
• 0:35
  rule for columns? Yes, a very important one.
• 0:38
  Second, we'll really nail down the difference between strength,
• 0:41
  just not breaking, and stiffness, which is about not
• 0:44
  bending too much. They sound similar, but they're
• 0:46
  critically different in engineering.
• 0:48
  Absolutely critical. And finally, we'll look at the
• 0:50
  modern tools, the computational stuff, matrix methods that
• 0:54
  engineers use for really complex structures, things you just
• 0:56
  couldn't calculate by hand. Right.
• 0:58
  It's like the journey from, you know, basic physics formulas to
• 1:02
  the incredibly powerful software that designs modern aircraft and
• 1:05
  bridges. So where should we start?
• 1:08
  Let's jump right into segment 1, the core of combined stress.
• 1:13
  So the absolute simplest starting point is direct stress,
• 1:16
  right? You push or pull on something
• 1:17
  straight along its axis. Exactly.
• 1:20
  Force, let's call it P, divided by the area it acts on.
• 1:23
  AP over A. That's your basic stress
• 1:26
  compression if you're pushing, tension if you're pulling.
• 1:29
  But like you said, reality is messy.
• 1:31
  Loads are almost never perfectly centered, are they?
• 1:34
  Especially in something like a column or what the sources call
• 1:38
  a stocky strut. Right.
• 1:39
  A stocky strut, short and thick. Basically it'll crush before it
• 1:43
  buckles sideways. And yes, the load is usually off
• 1:45
  center, even if just a little bit.
• 1:47
  So what happens then if that force P isn't dead center?
• 1:50
  Well, that's where eccentric thrust comes in, right?
• 1:53
  Physics tells us that applying that force P at a distance E the
• 1:56
  eccentricity from the center is exactly the same, mechanically
• 1:59
  speaking, as applying the force P at the center plus adding a
• 2:02
  bending moment M and that moment M is simply P times E, P times
• 2:08
  the eccentricity. So you get 2 effects for the
• 2:11
  price of one off center load. You still have the direct
• 2:13
  compression from P correct, but now you've also got this bending
• 2:17
  sort of twisting action from the moment.
• 2:18
  Then precisely, and when you combine those stresses, the
• 2:22
  uniform compression from PA and the varying stress from the
• 2:25
  bending moment, you often end up with a big problem, which is
• 2:29
  tension. Even though the main force is
• 2:31
  compressive, pushing down the bending effect can stretch one
• 2:35
  side of the column enough to put it into tension.
• 2:37
  And that's the big red flag, isn't it?
• 2:39
  Especially for certain materials.
• 2:40
  Absolutely. Think about stone, brick, old
• 2:44
  castles, unreinforced concrete. These things are fantastic under
• 2:49
  compression. They can hold up mountains, but
• 2:51
  tension? They hate it.
• 2:52
  They're incredibly weak in tension.
• 2:54
  Little bit of tensile stress and poof cracks, failure.
• 2:58
  It's dramatic. Which leads us neatly to this
• 3:01
  this golden rule you mentioned earlier, the core of cross rule.
• 3:04
  It sounds fundamental. Oh, it is.
• 3:05
  It's a beautiful piece of classic engineering insight for
• 3:09
  a rectangular column if you want to guarantee, absolutely
• 3:12
  guarantee that no tension develops anywhere in that cross
• 3:15
  section. OK, keep it all compressed.
• 3:17
  Then the loads eccentricity that distance from the center has to
• 3:22
  stay within the middle third of the columns depth.
• 3:25
  So if the depth is H, the eccentricity E must be less than
• 3:30
  or equal to H / 6. H / 6 That's specific fraction,
• 3:34
  so keep the load within that central H3 zone.
• 3:37
  Exactly. That zone is the core.
• 3:39
  Keep the lows line of action within that core and the entire
• 3:43
  section stays in compression. No tension, no cracks, even in
• 3:47
  brittle materials like masonry. That's amazing.
• 3:49
  You can almost picture like Roman engineers maybe not
• 3:53
  knowing the formula but instinctively knowing they had
• 3:56
  to keep loads centered on their massive stone Piers.
• 3:58
  They certainly understood the principle, perhaps through, you
• 4:01
  know, trial and error over centuries.
• 4:02
  It's why those structures are still standing.
• 4:04
  They kept the forces where the material could handle them in
• 4:06
  compression. And you mentioned this isn't
• 4:08
  just ancient history. This age 6 idea connects to
• 4:11
  modern techniques. Absolutely.
• 4:14
  Think about pre stressed concrete.
• 4:15
  What are we doing there? We're deliberately introducing a
• 4:19
  massive compressive force before the beam or slab ever sees its
• 4:23
  working load. OK, so you're pre compressing
• 4:25
  it. Right.
• 4:26
  We do that specifically to counteract the tension that we
• 4:29
  know will develop due to bending later on.
• 4:31
  We're essentially shifting the entire stress profile, keeping
• 4:34
  it within that safe compressive range, effectively using the
• 4:37
  same principle as the H six rule.
• 4:39
  That's a great connection. So understanding how forces
• 4:42
  combine and how to avoid tension is key.
• 4:45
  But you also mentioned strength isn't the whole story.
• 4:47
  Let's move to segment 2, stiffness.
• 4:51
  Being strong versus being stiff. Yeah, this is a really important
• 4:54
  distinction. In practice, strength is about
• 4:56
  load capacity. Yeah.
• 4:58
  How much force can it take before it yields or breaks?
• 5:00
  OK, will it collapse? Right, but stiffness is about
• 5:03
  deformation. How much does it bend OR deflect
• 5:06
  under load? A floor beam could be perfectly
• 5:08
  strong, not even close to breaking, but if it sags so much
• 5:11
  that the floor feels bouncy or, you know, the partitions crack.
• 5:14
  Then it's failed in terms of serviceability.
• 5:16
  It's not usable. Exactly.
• 5:18
  The sources make it clear stiffness, limiting deflection
• 5:21
  is often just as important, sometimes more important, than
• 5:24
  pure strength. And what governs the stiffness?
• 5:26
  What makes something resist bending?
• 5:28
  It boils down to a crucial property called flexural
• 5:31
  stiffness. It's represented by two letters
• 5:34
  you see everywhere in structural design.
• 5:36
  E times i.e. I.
• 5:39
  EI OK, what do E and I stand for?
• 5:42
  So E is Young's modulus. That's a material property.
• 5:45
  It tells you how inherently stiff the material itself is,
• 5:48
  how much it resists stretching or squashing elastically.
• 5:51
  Steel has a high E, Rubber has a very low E.
• 5:54
  Right, the materials inherent springiness sort.
• 5:57
  Of exactly, but I that's the second moment of area, sometimes
• 6:00
  called the area moment of inertia.
• 6:02
  This isn't about about the material, it's about the shape
• 6:03
  of the cross section. How is the material distributed
• 6:06
  relative to the axis it's bending around?
• 6:09
  So This is why an I-beam is so efficient.
• 6:11
  Precisely. An i-beam puts most of its
• 6:13
  material far away from the center in the flanges.
• 6:16
  This drastically increases the value of I compared to, say, a
• 6:21
  solid square bar with the same amount of material.
• 6:24
  A higher I means much greater resistance to bending.
• 6:27
  So a bigger E or a bigger I gives you more stiffness and
• 6:31
  together EI is the magic number for resisting deflection.
• 6:34
  That's the one flexural stiffness EI.
• 6:37
  When engineers calculate beam deflections, the equations
• 6:40
  fundamentally relate the bending moment and the curvature of the
• 6:43
  beam, and EI is the constant of proportionality, EI frac D2V
• 6:48
  That's the core equation. OK, so we can use calculus with
• 6:51
  that equation to find the deflection, but you hinted there
• 6:54
  might be other ways, especially for more complex setups, yes?
• 6:57
  Sometimes Drake integration gets, well, complicated,
• 7:01
  especially with beams that aren't simple or structures made
• 7:04
  of many parts. That's where energy methods come
• 7:06
  in, specifically the principle of virtual work.
• 7:08
  Virtual work? That sounds a bit theoretical
• 7:11
  imaginary. It does sound abstract, but the
• 7:13
  idea is surprisingly powerful. It states that for any system
• 7:18
  that's in equilibrium, just sitting there balanced if you
• 7:21
  imagine a tiny arbitrary virtual displacement.
• 7:24
  OK, a pretend movement. The total work done by all the
• 7:27
  real forces acting on the system during that pretend movement
• 7:30
  must be 0. External work done equals
• 7:33
  internal work, stored or strain energy.
• 7:36
  OK, how does imagining something help calculate a real deflection
• 7:40
  though? Well, here's the clever part.
• 7:42
  Let's say you want the real vertical deflection at a
• 7:45
  specific point on a beam. You apply an imaginary unit load
• 7:49
  like 1 Newton or 1 LB vertically.
• 7:51
  At that exact point. This is your virtual 4.
• 7:53
  OK, a tiny test poke. Exactly.
• 7:55
  Then you consider the real loads on the structure and the bending
• 7:58
  moments they cause internally. You calculate how much internal
• 8:01
  work these real internal moments do because of the tiny virtual
• 8:04
  displacements caused by your imaginary unit load.
• 8:07
  I see. So the imaginary load creates
• 8:09
  imaginary tiny movements, and you see how much work the real
• 8:12
  internal forces do during those movements.
• 8:14
  You got it. You equate the external virtual
• 8:17
  work, which is just one times the real deflection, to the
• 8:20
  internal virtual work, which you calculate by integrating the
• 8:23
  effects of the real moments and the virtual moments along the
• 8:26
  beam. It often avoids complex direct
• 8:29
  integration. That sounds much more efficient
• 8:31
  for certain problems. It really is.
• 8:33
  Take a simple cantilever beam fixed at one end, free at the
• 8:36
  other with a low W at the free end.
• 8:39
  Using virtual work you can quickly find the deflection at
• 8:42
  the end IS1EL3123 DI dollar. L ^3 so doubling the length
• 8:47
  makes the deflection 8 times worse.
• 8:49
  Exactly that insight. The cubic relationship with
• 8:51
  length is often more valuable than memorizing the whole
• 8:54
  formula. Energy methods give you that
• 8:56
  kind of feel for how structures behave.
• 8:58
  Powerful stuff for beans and frames, but even energy methods
• 9:02
  have limits, right? You wouldn't analyze, say, a
• 9:04
  whole aircraft fuselage or a massive offshore oil rig piece
• 9:07
  by piece using virtual work by hand.
• 9:10
  Oh, absolutely not. For that level of complexity
• 9:13
  where you have thousands, maybe millions of interconnected
• 9:16
  parts, different materials, complex load pads, you need
• 9:20
  serious computational firepower. That brings us to Segment 3, the
• 9:25
  modern toolkit. Right, Things get too complex
• 9:27
  for hand calculations or even simple energy methods.
• 9:30
  So what's the foundation for these modern tools?
• 9:34
  It really starts with matrix algebra.
• 9:36
  It turns out that the physical laws governing structures,
• 9:38
  things like Hook's law for springs, force, equal stiffness
• 9:42
  times displacement, can be represented very elegantly using
• 9:45
  matrices. OK, so matrices are like the
• 9:47
  language computers speak for structures.
• 9:49
  Pretty much, and they're incredibly efficient for
• 9:52
  handling large systems of equations simultaneously.
• 9:55
  This led to the development starting around the 1940s,
• 9:58
  mainly pushed by the aerospace industry.
• 10:00
  Needing lightweight complex structures.
• 10:02
  Exactly of the matrix displacement.
• 10:05
  Method matrix displacement method?
• 10:06
  What's the core idea there? You basically think of the
• 10:08
  entire complex structure, say a bridge, Truss or an aircraft
• 10:12
  frame, as an assembly of simpler pieces like individual bars or
• 10:15
  beams. Each piece acts like a spring
• 10:19
  with a certain stiffness. OK, break it down into simpler
• 10:22
  bits. Right.
• 10:23
  The method then connects these elements at nodes.
• 10:26
  The goal is to figure out how the nodes move the
• 10:29
  displacements. When you apply the loads, it
• 10:32
  sets up a massive matrix equation for the whole
• 10:35
  structure. The vector of applied loads P
• 10:37
  equals the overall structural stiffness matrix K times the
• 10:41
  vector of unknown nodal displacements, so PK.
• 10:44
  So you build this giant K matrix representing the stiffness of
• 10:48
  everything put together. Yes, assembling it from the
• 10:50
  stiffness of all the individual elements.
• 10:52
  You know the loads P and you solve this huge system of
• 10:55
  equations for the displacement. Precisely once you know how
• 10:58
  every node moves, you can go back and calculate the stress
• 11:01
  and strain within each individual element.
• 11:03
  It's systematic, perfect for computers.
• 11:05
  And that method evolved into what we use today.
• 11:08
  It's the direct ancestor of the finite Element method, or FEM.
• 11:12
  FEM or FEA for finite element analysis is really the workhorse
• 11:17
  of modern structural engineering and beyond actually.
• 11:20
  How does FEM go beyond the matrix displacement method?
• 11:23
  It's much more versatile. It can handle well almost
• 11:25
  anything really. Complex geometries, structures
• 11:28
  made of different materials, weirdly shaped parts, non
• 11:32
  uniform loads, even things like heat flow or fluid dynamics.
• 11:36
  Not just structural stress, it also handles nonlinear behavior
• 11:39
  much better. Material stretching beyond their
• 11:41
  elastic limit or large deflections changing the
• 11:44
  geometry. So how does it work?
• 11:45
  Still breaking things down. Yes, that's the core idea.
• 11:49
  It devised the continuous structure, whether it's a solid
• 11:51
  block, a curved shell, whatever, into a mesh of small finite
• 11:54
  pieces, the elements. Finite elements, hence the name.
• 11:57
  What shapes are these elements? They can be simple shapes like
• 12:00
  triangles or quadrilaterals for 2D problems, or tetrahedrons and
• 12:03
  bricks for 3D. The computer then solves the
• 12:05
  underlying physics equations like the stress, strain
• 12:08
  relationships approximately within each small element.
• 12:11
  And then it somehow stitches all those little solutions together.
• 12:14
  Exactly. It ensures that the
• 12:15
  displacements and forces match up at the nodes where the
• 12:19
  elements connect. By solving for the behavior of
• 12:22
  all these interconnected elements, it builds up a picture
• 12:25
  of the whole structures response.
• 12:26
  What about really curvy shapes like an airplane wing or a car
• 12:30
  body? Using flat triangles seems like
• 12:32
  would be a a rough approximation.
• 12:34
  Good point. For those cases, they use more
• 12:37
  sophisticated elements. Often they add extra nodes along
• 12:40
  the sides of the elements, not just at the corners.
• 12:43
  This allows the edges of the element to be curved, mapping
• 12:46
  much more accurately onto complex surveys.
• 12:49
  They're called higher order or isoperimetric elements.
• 12:52
  Clever. So FEM let's engineers model
• 12:55
  pretty much any shape and loading condition imaginable,
• 12:58
  far beyond what those really methods could handle.
• 13:00
  Absolutely, From intricate biological implants to massive
• 13:05
  civil engineering projects like dams or tunnels, FEM is the go
• 13:08
  to tool. We've covered a lot of ground
• 13:10
  then from that really elegant, simple H six rule ensuring old
• 13:15
  masonry didn't crack. Keep it in the core.
• 13:17
  All the way to the incredible complexity of finite element
• 13:20
  analysis modeling structures with millions of tiny virtual
• 13:24
  pieces. It's quite the journey in
• 13:26
  engineering analysis, but what's fascinating is that even the
• 13:29
  most complex femme simulation still relies on the fundamental
• 13:33
  principles we talked about earlier.
• 13:34
  How so? Well, deep inside each of those
• 13:37
  tiny finite elements, the computer is still using the
• 13:40
  basic stress strain relationships of the material
• 13:43
  like Young's modulus E and the geometric properties related to
• 13:48
  that second moment of area I. Effectively, that EI stiffness
• 13:52
  concept is still fundamentally there, just applied at a micro
• 13:56
  level across thousands of elements.
• 13:58
  So the fancy tools still rest on the old foundations.
• 14:00
  They absolutely do. The accuracy of the big complex
• 14:03
  model depends entirely on getting those basic material
• 14:06
  properties and behaviors right. Which brings us to a really
• 14:09
  interesting final thought, kind of a provocative question.
• 14:11
  Maybe we started with the certainty of that H Six rule for
• 14:14
  old simple materials. Now we have immense
• 14:16
  computational power, but what's the real bottleneck for
• 14:20
  confidence in modern design? Well, the source material
• 14:23
  actually touches on this. It points out that predicting
• 14:25
  failure especially for new complex materials like, say,
• 14:29
  advanced carbon fiber composites.
• 14:31
  Right, things that don't behave like simple steel or concrete.
• 14:34
  Exactly. Predicting how and when those
• 14:36
  materials will actually break under complex stresses often
• 14:40
  still relies heavily on physical experiments on empirical data.
• 14:44
  The theoretical models used in FEM might be great at predicting
• 14:47
  deflection or stress up to failure, but pinpointing the
• 14:50
  exact failure point for complex materials?
• 14:53
  That often comes back to lab testing.
• 14:55
  That's a critical point. So even with supercomputers
• 14:59
  running sophisticated FEM code, if you haven't accurately
• 15:02
  characterized how the actual material fails through testing,
• 15:06
  your prediction is still limited.
• 15:07
  Precisely. The computation can only be as
• 15:10
  good as the material model you feed into it, and for cutting
• 15:12
  edge materials, defining that failure behavior perfectly from
• 15:15
  theory alone is still a huge challenge.
• 15:18
  So it leaves us with a question for you, the listener, to think
• 15:21
  about. As we develop ever more complex
• 15:23
  materials and ever more powerful computers, what will ultimately
• 15:27
  limit our design confidence? Will it be the computational
• 15:30
  models eventually catching up to reality, or will the inherent
• 15:34
  complexity and variability of the materials themselves always
• 15:37
  require that final layer of experimental validation?
• 15:40
  Yeah. Where does the ultimate
• 15:41
  confidence come from? The math model or the material
• 15:44
  test? It's something engineers grapple
• 15:46
  with constantly. Definitely something to chew on
• 15:49
  as you look at the engineered world around you.
• 15:51
  Indeed, a fascinating journey through strength and stiffness.
• 15:54
  Thank you for joining us for this deep dive.
• 15:56
  We'll see you next time.