Latest / Mechanical Engineering Made Simple / (15) Infinite Modes & Sudden Shocks \ Shock & Vibration Engineering.

Transcript

(15) Infinite Modes & Sudden Shocks \ Shock & Vibration Engineering. · Mechanical Engineering Made Simple

• 0:00
  Welcome back to mechanical engineering made simple.
• 0:02
  We are really diving deep today. We're pushing past, you know,
• 0:05
  the basic single degree of freedom stuff.
• 0:08
  Yeah, moving into more complex territory.
• 0:09
  Exactly. Our sources point to two really
• 0:13
  critical areas for real world design.
• 0:16
  OK. What are they?
• 0:17
  First, how we model structures where the mass and stiffness
• 0:20
  aren't just points, but are spread out continuously.
• 0:23
  Think beams, plates. Like bridges, Airplane wings,
• 0:27
  that sort of. Thing precisely and 2nd, how
• 0:29
  these structures react to really sudden sharp impacts or loads
• 0:33
  things that aren't smooth or repeating.
• 0:35
  We call that the transient response.
• 0:37
  Got it. So our mission today really is
• 0:39
  to equip you, the listener, with the core ideas and the tools
• 0:44
  engineers use for these advanced vibration problems.
• 0:47
  We want you to grasp why we need these methods for complex real
• 0:51
  world scenarios. Right.
• 0:52
  It's about understanding how engineers actually analyze and,
• 0:56
  importantly, stop things from shaking dangerously.
• 0:58
  If you've ever wondered how a skyscraper is designed to handle
• 1:02
  wind gusts, or how a car chassis absorbs a sudden bump, well,
• 1:06
  this deep dive gets into the thinking behind those
• 1:09
  calculations. OK.
• 1:10
  So let's start with the foundation.
• 1:11
  We need to talk about the difference between our standard
• 1:13
  models and, well, reality. You mean lumped versus
• 1:18
  distributed systems? Exactly.
• 1:20
  In basic mechanics we often use lumped parameter systems.
• 1:24
  We pretend the structure is simple, like rigid blocks
• 1:27
  connected by perfect massless springs and dampers.
• 1:31
  It's an idealization, right? And approximation a very.
• 1:33
  Useful one. It gives us a finite number of
• 1:36
  degrees of freedom or DOF. Makes the math manageable.
• 1:40
  But the real world isn't made of massless springs.
• 1:43
  Definitely not in reality. Take any beam or plate.
• 1:46
  Its mass and its elasticity, its stiffness are distributed all
• 1:50
  the way through it continuously. So less like weights on a
• 1:52
  string, more like a a metal ruler that flexes.
• 1:55
  That's a good way to put it. And that continuous
• 1:57
  distribution, that's where things get, mathematically
• 2:00
  speaking, much more involved. A distributed parameter system
• 2:04
  fundamentally has an infinite number of ways it can move an
• 2:08
  infinite number of degrees of freedom.
• 2:10
  Infinite. OK, that sounds complicated.
• 2:13
  What's the big consequence of that?
• 2:15
  The really profound outcome is that these real systems have an
• 2:18
  infinite number of natural frequencies.
• 2:21
  Whoa, infinite frequencies? How can you possibly design for
• 2:24
  that? Doesn't that mean resonance is
• 2:27
  always a risk? It sounds like a nightmare, I
• 2:29
  know, but you don't track all infinite frequencies.
• 2:32
  You focus for every single one of those natural frequencies.
• 2:35
  The structure has a specific way it likes to vibrate a particular
• 2:38
  shape. We call that the normal mode or
• 2:40
  natural mode. Can you give us a quick picture
• 2:42
  of a mode shape? Sure.
• 2:44
  Think of a simple beam, maybe supported at both ends, like
• 2:46
  with hinges. OK, its lowest natural
• 2:49
  frequency, the fundamental one, makes the beam bend up and down
• 2:52
  in a smooth curve, like a single half sign wave.
• 2:55
  Just one big arc. Exactly.
• 2:57
  The next frequency up might look like a full sine wave bending up
• 3:01
  in 1/2 and down in the other with a point in the middle
• 3:03
  sting. Still, that's the second mode
• 3:05
  and it goes on getting more complex.
• 3:07
  OK, I see. So infinite mode, infinite
• 3:09
  frequencies, but solving the the full math, the partial
• 3:13
  differential equations, that sounds tough for real designs.
• 3:16
  It often is very complex, sometimes impractical, so
• 3:20
  engineers lean heavily on approximation methods.
• 3:23
  And if you just need that crucial first frequency, the
• 3:25
  fundamental 1, what's the go to method?
• 3:28
  That would be Rayleigh's method. It's incredibly useful, really
• 3:32
  elegant actually. It uses the conservation of
• 3:34
  energy principle for a system of vibrating freely without
• 3:38
  damping. The energy balance trick you're
• 3:40
  saying the maximum potential energy stored in the flex?
• 3:43
  The maximum elastic strain energy, Yeah, when it's bent the
• 3:45
  most. Has to equal the maximum kinetic
• 3:48
  energy when it's moving fastest through the zero point.
• 3:50
  Precisely. You equate those two energy
• 3:52
  peaks and the frequency just pops out of the equation.
• 3:55
  It's brilliant. What's the catch?
• 3:56
  The catch is you have to assume a shape for how the structure
• 4:00
  deflects, and your assumed shape needs to be reasonably close to
• 4:04
  the real first mode shape and meet the boundary conditions
• 4:07
  like how it's supported. So if your guess about the shape
• 4:10
  is way off, your frequency calculation could be off too.
• 4:13
  How do we get more accuracy or find those higher modes?
• 4:17
  Right, that's where Ritz's method comes in.
• 4:19
  You can think of it as a more sophisticated, more flexible
• 4:22
  version of Rayleigh. How is it more flexible?
• 4:25
  With Ritz, instead of assuming one fixed shape, you assume a
• 4:28
  shape that's built from a combination of simpler shapes,
• 4:31
  maybe a series expansion, and this assumed shape includes some
• 4:35
  unknown parameters, some adjustment knobs.
• 4:38
  If you will adjustment knobs OK so you can tweak the assumed
• 4:41
  shape. Exactly.
• 4:42
  You use calculus basically to find the values for those
• 4:47
  parameters that make the calculated frequency as low as
• 4:49
  possible a minimum value. Why minimum?
• 4:52
  Because mathematically minimizing the frequency
• 4:55
  calculation with respect to those parameters pushes your
• 4:57
  assumed shape closer to the true mode shape.
• 5:00
  It gives a much better estimate for the fundamental frequency,
• 5:03
  and you can even use it to get pretty good estimates for the
• 5:06
  higher frequencies too. It's about making a smarter,
• 5:08
  adaptable guess. OK, that makes sense.
• 5:10
  So we figured out how to find these critical frequencies, even
• 5:14
  with infinite possibilities. Now let's talk about fixing
• 5:17
  problems. What if a machine's operating
• 5:20
  speed lines up perfectly with one of these natural
• 5:23
  frequencies? Big trouble, right?
• 5:25
  Potentially catastrophic. Yes.
• 5:27
  Resonance. And what if you can't easily
• 5:31
  redesign the main structure? Maybe add damping directly into
• 5:34
  it? Is there another way?
• 5:36
  Absolutely. That's when we bring in
• 5:37
  auxiliary mass dampers. Sometimes they're just called
• 5:39
  damped absorbers. So you bolt on a smaller system
• 5:42
  to help the big one. Essentially, yes.
• 5:44
  The idea is, if you can't damp the main system directly, you
• 5:47
  attach a small, carefully designed secondary system,
• 5:50
  usually just a mass, a spring and a damper to soak up the
• 5:54
  vibration energy. How does that actually work?
• 5:57
  How does this little attached thing calm down the big
• 5:59
  structure? It's all about the force the
• 6:01
  auxiliary system exerts back onto the main structure.
• 6:04
  It pushes and pulls on the attachment point.
• 6:06
  Now, if there's no damping in the auxiliary system, that
• 6:09
  reaction force just kind of shifts the main system's natural
• 6:13
  frequencies around. It doesn't really kill the
• 6:14
  vibration. But when you add damping to that
• 6:17
  auxiliary mass, something crucial happens.
• 6:21
  The reaction force it generates gets a component that is 90° out
• 6:24
  of phase with motion, the displacement and acceleration of
• 6:28
  the main system at that point. 90° Why is that the magic
• 6:31
  number? Because a force acting 90° out
• 6:35
  of phase with velocity, or in this case displacement
• 6:37
  acceleration, which are also 90° apart in harmonic motion does
• 6:41
  work. It actively extracts energy from
• 6:44
  the main system over each cycle. It damps the vibration.
• 6:48
  It's like applying brakes at just the right moments in the
• 6:50
  cycle. That's a decent analogy, yes.
• 6:52
  It's actively fighting the motion by being out of sync in
• 6:55
  that specific. Way and designing these
• 6:57
  absorbers, I understand there's a clever trick involving fixed
• 6:59
  points on the response graph. Yes, this is a really neat piece
• 7:03
  of theory. When you plot the amplitude of
• 7:06
  the main system versus the excitation frequency.
• 7:09
  For a system with a damped absorber attached, there are two
• 7:13
  special frequency points, usually called A&B.
• 7:15
  And what's special about A&B? At these specific frequencies,
• 7:19
  the amplitude of the main systems vibration is completely
• 7:22
  independent of how much damping you put in the auxiliary system.
• 7:26
  Whether you have a lot of damping or very little, the
• 7:29
  amplitude curve always passes through these same 2 points.
• 7:32
  Wow, OK, why is that so useful for design?
• 7:35
  It decouples the problem. It means you can first choose
• 7:38
  the mass and spring of your absorber, which sets the tuning
• 7:41
  ratio alpha to place these fixed points A&B strategically,
• 7:46
  usually aiming to make their heights equal.
• 7:49
  This optimizes the tuning without worrying about the
• 7:51
  damping value yet. So you fix the tuning first,
• 7:53
  alpha taut G just based on these fixed points.
• 7:55
  That simplifies things a lot. Immensely.
• 7:58
  Once you've got the optimal tuning set by the fixed points,
• 8:01
  then you choose the optimal amount of damping Zeta Opt to
• 8:03
  minimize the peaks of the response curve that occur
• 8:06
  between and around those fixed points.
• 8:08
  Got it. Tune first using the fixed
• 8:10
  points, then optimize the damping.
• 8:12
  Are there other types of absorbers too?
• 8:14
  Oh yes, you can sometimes use distributed masses like long
• 8:18
  rods, but that's usually only practical at very high
• 8:20
  frequencies. A more common one, especially
• 8:23
  for rotating machinery, is the pendulum ablorber.
• 8:26
  It uses the inertia of swinging pendulums to create torques that
• 8:29
  cancel out unwanted torsional vibrations and things like
• 8:32
  engine crankshafts. Fascinating stuff.
• 8:35
  OK, we've covered continuous systems and how to tame them.
• 8:39
  Let's switch gears completely. What about sudden shocks, Bangs.
• 8:42
  Impacts Study starts or stops? Right now we're moving into
• 8:45
  transient response. We're looking at loads that
• 8:48
  aren't smooth and continuous, but act suddenly, maybe just for
• 8:51
  a short duration. Think a step input, like
• 8:53
  suddenly applying a constant force or a pulse like a brief
• 8:56
  blast or impact. So the goal isn't finding a
• 8:58
  steady vibration amplitude anymore.
• 9:01
  Exactly. Now the goal is to find the
• 9:03
  entire time history of how the system reacts to that sudden
• 9:06
  event. What's the peak displacement?
• 9:09
  How long does it vibrate afterwards?
• 9:11
  One thing our sources really emphasize here is a kind of
• 9:14
  mathematical elegance. How different kinds of shocks
• 9:18
  can be described similarly? That's a really powerful
• 9:20
  insight, whether the shock comes from a force hitting the object,
• 9:25
  the vac TD or the ground suddenly moving underneath it
• 9:28
  while TD, or even the ground suddenly accelerating like in an
• 9:31
  earthquake. Precisely the actual
• 9:33
  differential equation describing the system's response, whether
• 9:36
  that response is absolute motion solid, or motion relative to the
• 9:39
  ground delta, or even absolute acceleration, can often be
• 9:43
  boiled down to the same general form plus nua placed down.
• 9:46
  So one equation structure covers all these different physical
• 9:49
  situations. Essentially, yes.
• 9:51
  A NUA represents the response quantity you care about and
• 9:54
  represents the effective excitation based on the input.
• 9:58
  You solve the same type of equation just plugging in the
• 10:00
  appropriate sellers for your specific shock.
• 10:02
  That must save a lot of work. It does for instance, take a
• 10:05
  simple undamped system at rest. If you suddenly apply a constant
• 10:10
  force equivalent to a step input B sex, the solution is 1 Omega
• 10:14
  T, It doesn't just move to the position in stock.
• 10:17
  It oscillates. Yes, it oscillates around
• 10:19
  cyclids. And notice the cosine goes from
• 10:21
  plus one to Magda. That term goes from zero to 2.
• 10:24
  Meaning it overshoots. It actually reaches twice the
• 10:26
  amplitude of the step input. Exactly.
• 10:29
  The peak response is $2.00. That dynamic overshoot is
• 10:32
  critical in transient analysis. OK, so how do we actually find
• 10:36
  these time response functions, especially for more complex
• 10:40
  pulses? There are two main classical
• 10:43
  workhorses here, which are. First, there's Duhamel's
• 10:46
  integral, which is also called the convolution integral.
• 10:49
  What's the idea behind Duhamel's?
• 10:51
  It's quite intuitive, actually. It thinks of the complex input
• 10:54
  force as being made-up of a whole sequence of tiny,
• 10:57
  infinitesimally short impulses. OK, it's like a series of tiny
• 11:00
  hammer taps. Exactly.
• 11:02
  And we know how a simple system responds to a single impulse.
• 11:06
  Duhamel's integral just adds up or superimposes the responses to
• 11:11
  all those tiny individual impulses over time to get the
• 11:14
  total response to the complex input.
• 11:16
  It gives you a good physical feel for how the response builds
• 11:19
  up. That makes sense.
• 11:20
  What's the other main method? The other big one is using the
• 11:23
  Laplace transformation. Laplace transforms.
• 11:27
  Very powerful math tool. What's the advantage here
• 11:30
  compared to Duhamel? Laplace transforms excel at
• 11:33
  turning differential equations into algebraic equations.
• 11:36
  You transform the whole equation, including the forcing
• 11:39
  function into the Beals domain. Making it easier to solve.
• 11:42
  Usually yes, but the key advantage, especially for
• 11:45
  transients, is that the initial conditions, the displacement and
• 11:47
  velocity the system had right at the start of the shock get
• 11:51
  automatically incorporated into the algebraic equation in the
• 11:54
  South domain. Dumo might be trickier if this
• 11:57
  system wasn't starting from rest.
• 11:59
  It can be. Laplace handles initial
• 12:01
  conditions very cleanly. You solve the algebra, then
• 12:04
  transform back to the time domain to get your mileage
• 12:06
  apart. So maybe Duhamel for physical
• 12:09
  insight, Laplace for mathematical efficiency,
• 12:11
  especially with initial conditions.
• 12:13
  OK, so we get the full time history, but engineers often
• 12:16
  need just one number, right? The worst case scenario.
• 12:18
  That's right, they're usually most interested in the absolute
• 12:22
  maximum response the system reaches during or after the
• 12:26
  pulse. That's called the maximax
• 12:28
  response, denoted Amplek. And how do they visualize that?
• 12:32
  Using response Spectra. A response spectrum is a plot.
• 12:35
  It shows how that maximax response ample top changes
• 12:38
  depending on the properties of the pulse in the system.
• 12:41
  Often it plots a ratio of the pulse duration to the system's
• 12:45
  natural period. Tallow dollars, so.
• 12:47
  So you can see how systems at different natural frequencies
• 12:50
  will react to the same pulse. Exactly, and the Spectra reveal
• 12:54
  crucial patterns for many common pulse shapes like rectangular or
• 12:58
  half sign pulses. The biggest amplification, the
• 13:00
  highest unknown, usually happens when the pulse duration shout is
• 13:04
  somewhere around half to one times the natural period dollar.
• 13:07
  So when towers between say .5 and one.
• 13:10
  Roughly yes. And critically, that peak
• 13:13
  maximax response is often about 1.5 to 1.8 times the peak value
• 13:18
  of the input pulse itself. It's.
• 13:20
  Wait, so the dynamic effect amplifies the peak input load by
• 13:23
  50% to 80%? Potentially, yes.
• 13:25
  That amplification factor is vital for safe design against
• 13:28
  impacts or blasts. You can't just design for the
• 13:31
  static peak load, you have to account for this dynamic
• 13:34
  overshoot. That's a huge safety factor just
• 13:35
  from the dynamics. And for those step inputs you
• 13:37
  mentioned, the peak occurs after the input has reached its steady
• 13:40
  value. Yes, for step like inputs, the
• 13:43
  system is still accelerating. When the input force levels off,
• 13:47
  inertia carries it past the new equilibrium.
• 13:49
  That maximax response and MMSDA is actually the sum of the final
• 13:53
  steady displacement plus the amplitude of the vibration after
• 13:57
  the step is complete, which we call the residual response
• 14:00
  amplitude Nur SO1 error. Plus it's still ringing OK.
• 14:04
  One last point on transients damping.
• 14:06
  We said for the auxiliary mask dampers, damping was crucial.
• 14:10
  How important is damping for surviving a single short pulse?
• 14:14
  That's a really important distinction.
• 14:16
  For steady state vibration near resonance, even a tiny bit of
• 14:19
  damping makes a huge difference. Right?
• 14:20
  But for the Max Max response to a single pulse or step, normal
• 14:24
  amounts of viscous damping often have relatively little effect on
• 14:27
  that first highest peak. Why is that?
• 14:29
  The shock happens too fast. The system reaches its peak
• 14:32
  displacement so quickly that the damper just doesn't have enough
• 14:35
  time or cycles to dissipate a significant amount of energy
• 14:38
  before that first peak occurs. You'd need very heavy damping
• 14:41
  for it to substantially reduce wind.
• 14:43
  Interesting, so damping is key for fatigue life under
• 14:46
  continuous vibration, but less critical for surviving the peak
• 14:49
  load from a single shock unless it's extremely high damping.
• 14:53
  That's a good summary. It really highlights the
• 14:55
  different design philosophies for continuous versus transient
• 14:59
  loads. This has been fascinating.
• 15:01
  We've gone from continuous beams to sudden shocks.
• 15:04
  It feels like the real challenge is putting it all together.
• 15:07
  Right, Absolutely. Think about a complex structure
• 15:09
  like a whole airplane wing or a bridge deck getting hit by a
• 15:12
  sudden gust or impact. That's a distributed system
• 15:16
  experiencing A transient load. So you need both sets of tools
• 15:19
  we discussed. Precisely, the advanced approach
• 15:22
  involves using the normal mode analysis, like we discussed with
• 15:24
  Rayleigh and Ritz, to break the complex distributed system down
• 15:28
  into a set of equivalent independent single degree of
• 15:31
  freedom systems. Decoupling the complexity using
• 15:35
  the modes. Exactly.
• 15:36
  And then you apply the transient analysis methods like do, ML or
• 15:39
  the plus to each of those independent modal equations to
• 15:43
  find out how each mode responds to the shock.
• 15:45
  Then you add those responses back up.
• 15:47
  Wow, that's the real synthesis, combining mode shapes with
• 15:52
  transient solutions. It sounds computationally
• 15:55
  intensive for complex. Structures it definitely can be.
• 15:58
  And for doing that kind of analysis efficiently, especially
• 16:01
  for systems with many, many degrees of freedom, engineers
• 16:04
  rely on more advanced computational techniques.
• 16:07
  Methods like the phase plane delta method or the transfer
• 16:09
  matrix method are designed specifically to handle that
• 16:12
  complexity for transient problems in multi degree of
• 16:15
  freedom systems. That's where the heavy duty
• 16:17
  computing comes in. A glimpse into the next level of
• 16:19
  analysis. This has been incredibly
• 16:21
  insightful, bridging the gap between simple models and these
• 16:25
  really complex dynamic realities.
• 16:26
  Until the next equation, we will see you design junkies on the
• 16:29
  next episode of Mechanical Engineering Made Simple.