6 ms·

Very few things surprised me as much as rail design when I was studying structural engineering. Surface tension and fluid dynamics were both trippy too, but I e

Very few things surprised me as much as rail design when I was studying structural engineering. Surface tension and fluid dynamics were both trippy too, but I expected those things to be complicated and while surface tension blew my mind due to the relative simplicity of the proof, and fluid dynamics blew my mind because it was somehow 100x more complex than I estimated. Walking into rail design was hilarious. I worked on motorcycles and did some car stuff. I figured it was obvious, and sorta dismissed this assignment as a joke. Nope. My dismissive intuitions were just flat out wrong. It kinda leaves an impression on you to sorta avoid saying you know for sure before putting in some amount of work.

dharmab 5y ago
I worked on motorcycles for years using DIY guides and YouTube tutorials. Opening up real engineering books was an eye-opening and humbling experience that made me a better mechanic, driver/rider, homeowner and software engineer. (For the curious motorcyclist, I recommend "Honda Common Service Manual" as a starting point.)
- Swizec 5y ago
  > opening real engineering books was eye-opening Turns out designing a new system to fit requirements is orders of magnitudes harder than fixing a system somebody else designed. You see this in software all the time. Anyone can follow a tutorial. But can you start from scratch and build something novel? Can you build it such that others can maintain long after you’re gone? That’s hard. Same with cooking. Anyone can follow a recipe. But can you design a recipe?
  - dharmab 5y ago
    Another thing I learned: Most engineers can build something that works. It is much, much harder to build something that optimizes for cost.
    - bch 5y ago
      “Anybody can build a bridge that stands - it takes an engineer to build a bridge that barely stands.”
      - taeric 5y ago
        This brings to mind my amusement when friends would bash Ikea cabinets as too flimsy. I was struck that they were just strong enough to do what they claimed. No more.
        throwaway0a5e 5y ago
        Everyone loves to shit on value engineering but the reality is that it's a lot harder and takes a lot more craftiness than designing things under less price pressure. When you start talking about revisions to complex things where you have to both work fast and avoid revising too much of the thing (to keep costs down) it gets really crazy.
    - dougSF70 5y ago
      And it takes an architect to build a bridge that rarely stands...cf. the millennium bridge in london. Designed by architects, fixed by engineers.
      - elzbardico 5y ago
        You do know that the architects hand off their work to structural engineers on anything more complex than a single pavement building before it gets built, do you?
        sbisson 5y ago
        To be fair to the architects of the Millennium Bridge, the structural engineers used a vehicle bridge model, not one for pedestrians.
        5y ago
  - throwaway0a5e 5y ago
    Designing to fit requirements is easy. Making it scale economically is where all the math gets involved.
- 3pt14159 5y ago
  I know right? I also loved the times where I got to the math and all that internal intuition lined up completely. Torsional deformation for example. I almost blamed myself for not inventing the math myself it so obviously matched my intuitions. It's kinda fun no matter which way it goes. Black body radiation never quite sat right with me. One of the few subject areas where I just resigned myself to memorizing the formulas and moving on with life. Same with non-integer dimensional spaces for the most advanced partial differential course I took. I can visualize 2 million dimensional spaces just fine professor. But one and a half? What does this even mean?
  - goldenkey 5y ago
    It just means that the degrees of freedom aren't used fully. This is usually seen in fractals, where there is a level of redundancy that is respective of the fractional dimension missing. Think about it like this. If I have a 2d field (x,y) and I enforce every point's x value to be 0, I pretty much just made the x degree of freedom redundant, and can now call the field a 1d field. If instead, I enforce every 3rd point's x value to be 0, I've now got a 1+2/3 dimensional space. Because there is some redundancy, I no longer get the full entropy that 2 dimensions provide.
    - tambourine_man 5y ago
      That is a great, simple explanation, thanks.
    - Michaelfonzolo 5y ago
      Never encountered this idea before, does it have a formal name? Most fractional-dimension spaces encountered refer to either the Minkowski or Hausdorff Dimension.
      - goldenkey 5y ago
        It's equivalent to the measures you mentioned. Here's a paper which discusses fractional degrees of freedom: Effective degrees of freedom of a random walk on a fractal by AS Balankin · 2015 · Cited by 42 — This allows us to define the fractional dimensional space allied ... number of effective dynamical degrees of freedom on the fractal https://pubmed.ncbi.nlm.nih.gov/26764671/ https://pubmed.ncbi.nlm.nih.gov/26764671/ To hammer the relationship home, consider the holographic principle, which is based on observation of black holes, and states that our reality only needs 2 spatial dimensions instead of 3. Both Hawking and Susskind have eluded to this being the case solely because of the symmetries in the laws of physics. The symmetries cause a massive redundancy/pattern in the field values over the 3d space, such that we should theoretically be able to predict the state of the entire field given only the values at two-thirds of the volume. Therefore, you can imagine a 2d surface which contains the state of our universe, and some kind of computational (possibly geometric/algebraic) projector, which understands the redundancies, reads the 2d surface, and renders a sparse 3d volume. In the case of our universe, the projection operation might be extraordinary complex, requiring a deep understanding of the laws of physics and the redundancies they induce into the underlying state that they operate on. https://en.wikipedia.org/wiki/Holographic_principle https://en.wikipedia.org/wiki/Holographic_principle
- greyhair 5y ago
  I worked professionally as a motorcycle mechanic from mid 1970s through the early 1980s. Honda's service manuals were very good, but my favorites were the early manuals from the late 1960s to mid 1970s, mostly for their Japanese to English translations. One of my favorite lines from an early CB750 manual was "Pleased to be applying the 26mm spanner to the castellated nut. Thank you." How can you not follow instructions when they are worded like that? I learned about train wheels decades ago, when I briefly ran through train history as an interest. What I found interesting at the time was how thoroughly this was understood right at the advent of the age of steam. Rail engineers understood this from the beginning. Another thing is that since the wheels ride an a narrow portion, going straight most of the time, wheel wear distorts the conical section, which leads to the wheels rubbing (instead of rolling) in corners, which accelerates the wear. So rail wheel wear is non-linear. Once the wear reaches a point wear it causes slipping/rubbing, it leads to positive feedback on the rate of wear. A failure mode that isn't obvious until you thing about it.
  - dharmab 5y ago
    I wonder if a similar phenomenon is the reason my final drive chain and sprocket seems to wear a little after install, be stable for 10,000+ miles and then suddenly wear a lot.
  - HeyLaughingBoy 5y ago
    It blew my mind when I learned that some curved sections are actually greased periodically in order to reduce the friction.
jacquesm 5y ago
Almost anything when done at the fine edge of engineering for optimum price point without sacrificing reliability is going to be an amazing thing to do a deep dive in. Windmills are another, they seem so simple and obvious until you dig in.
- na85 5y ago
  Hey Jacques, whatever happened to your wind turbine you were giving away?
  - jacquesm 5y ago
    No takers, unfortunately. After sitting there for two years it eventually ended up scrapped which I'm still quite sick about.
    - na85 5y ago
      What a tragedy!
      - jacquesm 5y ago
        Agreed. I still don't really understand it, I imagined there would be lots of people working on off the grid houses who would be more than happy to pick up a free windmill but apparently that wasn't the case. So much work down the drain. Anyway, I shouldn't dwell on it, I've been lugging it with me from one residence to the next since 2008 so it wasn't for lack of trying.
    - beckingz 5y ago
      That is really sad to hear. It was a really cool project to see!
gerdesj 5y ago
Road curves can be quite surprising too. There are always exceptions due to topography and obstacles but optimal curves are not always obvious. A simple corner on a motorway/autobahn/highway/etc has at least three curves in plan - in, out and shake it all about. OK, in and out and the main bend itself. The idea is that you need to transition the various forces into and out of the turn as safely as possible. A right angle turn is not a simple: straight -> quarter circle -> straight. As well as that, we have to consider line of sight and overtaking and water runoff and the effect of wind on high sided vehicles and ... and ... . Oh and of course these turns happen in 3D. I'm quite a fan of Holden Hill in Devon on the A38 - get your speed right in either direction and it feel effortless but you need enough power too. Purists in the UK will probably point at Snake Pass and most of Welsh roads and the like but I know Holden Hill. Anyway, roads are sodding complicated. Why not curl up with this little number: https://www.standardsforhighways.co.uk/prod/attachments/c27c55b7-2dfc-4597-923a-4d1b4bd6c9fa https://www.standardsforhighways.co.uk/prod/attachments/c27c... That's the design standard for major UK roads. Note things like crest and sag curves, worrying about kerbs and so on.
- mycall 5y ago
  > in and out and the main bend itself How would you rate humans for doing this right, in general , everywhere they have?
  - gerdesj 5y ago
    Not sure I've looked at the entire world's road bends. I do know that with a few exceptions, English roads do what the posted signs say they will do.
- mnw21cam 5y ago
  That's interesting you mention Haldon Hill. They may have got the A38 bit right, but if you go up the other road, the A380, then there is a quite sharp blind right bend at the top that clearly hasn't been done right.
  - gerdesj 5y ago
andi999 5y ago
What proof of surface tension are you referring to?

dharmab 5y ago

I worked on motorcycles for years using DIY guides and YouTube tutorials. Opening up real engineering books was an eye-opening and humbling experience that made me a better mechanic, driver/rider, homeowner and software engineer. (For the curious motorcyclist, I recommend "Honda Common Service Manual" as a starting point.)

Swizec 5y ago
> opening real engineering books was eye-opening Turns out designing a new system to fit requirements is orders of magnitudes harder than fixing a system somebody else designed. You see this in software all the time. Anyone can follow a tutorial. But can you start from scratch and build something novel? Can you build it such that others can maintain long after you’re gone? That’s hard. Same with cooking. Anyone can follow a recipe. But can you design a recipe?
- dharmab 5y ago
  Another thing I learned: Most engineers can build something that works. It is much, much harder to build something that optimizes for cost.
  - bch 5y ago
    “Anybody can build a bridge that stands - it takes an engineer to build a bridge that barely stands.”
    - taeric 5y ago
      This brings to mind my amusement when friends would bash Ikea cabinets as too flimsy. I was struck that they were just strong enough to do what they claimed. No more.
      - throwaway0a5e 5y ago
        Everyone loves to shit on value engineering but the reality is that it's a lot harder and takes a lot more craftiness than designing things under less price pressure. When you start talking about revisions to complex things where you have to both work fast and avoid revising too much of the thing (to keep costs down) it gets really crazy.
  - dougSF70 5y ago
    And it takes an architect to build a bridge that rarely stands...cf. the millennium bridge in london. Designed by architects, fixed by engineers.
    - elzbardico 5y ago
      You do know that the architects hand off their work to structural engineers on anything more complex than a single pavement building before it gets built, do you?
      - sbisson 5y ago
        To be fair to the architects of the Millennium Bridge, the structural engineers used a vehicle bridge model, not one for pedestrians.
        5y ago
- throwaway0a5e 5y ago
  Designing to fit requirements is easy. Making it scale economically is where all the math gets involved.
3pt14159 5y ago
I know right? I also loved the times where I got to the math and all that internal intuition lined up completely. Torsional deformation for example. I almost blamed myself for not inventing the math myself it so obviously matched my intuitions. It's kinda fun no matter which way it goes. Black body radiation never quite sat right with me. One of the few subject areas where I just resigned myself to memorizing the formulas and moving on with life. Same with non-integer dimensional spaces for the most advanced partial differential course I took. I can visualize 2 million dimensional spaces just fine professor. But one and a half? What does this even mean?
- goldenkey 5y ago
  It just means that the degrees of freedom aren't used fully. This is usually seen in fractals, where there is a level of redundancy that is respective of the fractional dimension missing. Think about it like this. If I have a 2d field (x,y) and I enforce every point's x value to be 0, I pretty much just made the x degree of freedom redundant, and can now call the field a 1d field. If instead, I enforce every 3rd point's x value to be 0, I've now got a 1+2/3 dimensional space. Because there is some redundancy, I no longer get the full entropy that 2 dimensions provide.
  - tambourine_man 5y ago
    That is a great, simple explanation, thanks.
  - Michaelfonzolo 5y ago
    Never encountered this idea before, does it have a formal name? Most fractional-dimension spaces encountered refer to either the Minkowski or Hausdorff Dimension.
    - goldenkey 5y ago
      It's equivalent to the measures you mentioned. Here's a paper which discusses fractional degrees of freedom: Effective degrees of freedom of a random walk on a fractal by AS Balankin · 2015 · Cited by 42 — This allows us to define the fractional dimensional space allied ... number of effective dynamical degrees of freedom on the fractal https://pubmed.ncbi.nlm.nih.gov/26764671/ https://pubmed.ncbi.nlm.nih.gov/26764671/ To hammer the relationship home, consider the holographic principle, which is based on observation of black holes, and states that our reality only needs 2 spatial dimensions instead of 3. Both Hawking and Susskind have eluded to this being the case solely because of the symmetries in the laws of physics. The symmetries cause a massive redundancy/pattern in the field values over the 3d space, such that we should theoretically be able to predict the state of the entire field given only the values at two-thirds of the volume. Therefore, you can imagine a 2d surface which contains the state of our universe, and some kind of computational (possibly geometric/algebraic) projector, which understands the redundancies, reads the 2d surface, and renders a sparse 3d volume. In the case of our universe, the projection operation might be extraordinary complex, requiring a deep understanding of the laws of physics and the redundancies they induce into the underlying state that they operate on. https://en.wikipedia.org/wiki/Holographic_principle https://en.wikipedia.org/wiki/Holographic_principle
greyhair 5y ago
I worked professionally as a motorcycle mechanic from mid 1970s through the early 1980s. Honda's service manuals were very good, but my favorites were the early manuals from the late 1960s to mid 1970s, mostly for their Japanese to English translations. One of my favorite lines from an early CB750 manual was "Pleased to be applying the 26mm spanner to the castellated nut. Thank you." How can you not follow instructions when they are worded like that? I learned about train wheels decades ago, when I briefly ran through train history as an interest. What I found interesting at the time was how thoroughly this was understood right at the advent of the age of steam. Rail engineers understood this from the beginning. Another thing is that since the wheels ride an a narrow portion, going straight most of the time, wheel wear distorts the conical section, which leads to the wheels rubbing (instead of rolling) in corners, which accelerates the wear. So rail wheel wear is non-linear. Once the wear reaches a point wear it causes slipping/rubbing, it leads to positive feedback on the rate of wear. A failure mode that isn't obvious until you thing about it.
- dharmab 5y ago
  I wonder if a similar phenomenon is the reason my final drive chain and sprocket seems to wear a little after install, be stable for 10,000+ miles and then suddenly wear a lot.
- HeyLaughingBoy 5y ago
  It blew my mind when I learned that some curved sections are actually greased periodically in order to reduce the friction.

jacquesm 5y ago

Almost anything when done at the fine edge of engineering for optimum price point without sacrificing reliability is going to be an amazing thing to do a deep dive in. Windmills are another, they seem so simple and obvious until you dig in.

na85 5y ago
Hey Jacques, whatever happened to your wind turbine you were giving away?
- jacquesm 5y ago
  No takers, unfortunately. After sitting there for two years it eventually ended up scrapped which I'm still quite sick about.
  - na85 5y ago
    What a tragedy!
    - jacquesm 5y ago
      Agreed. I still don't really understand it, I imagined there would be lots of people working on off the grid houses who would be more than happy to pick up a free windmill but apparently that wasn't the case. So much work down the drain. Anyway, I shouldn't dwell on it, I've been lugging it with me from one residence to the next since 2008 so it wasn't for lack of trying.
  - beckingz 5y ago
    That is really sad to hear. It was a really cool project to see!

gerdesj 5y ago

Road curves can be quite surprising too. There are always exceptions due to topography and obstacles but optimal curves are not always obvious. A simple corner on a motorway/autobahn/highway/etc has at least three curves in plan - in, out and shake it all about. OK, in and out and the main bend itself. The idea is that you need to transition the various forces into and out of the turn as safely as possible. A right angle turn is not a simple: straight -> quarter circle -> straight. As well as that, we have to consider line of sight and overtaking and water runoff and the effect of wind on high sided vehicles and ... and ... . Oh and of course these turns happen in 3D. I'm quite a fan of Holden Hill in Devon on the A38 - get your speed right in either direction and it feel effortless but you need enough power too. Purists in the UK will probably point at Snake Pass and most of Welsh roads and the like but I know Holden Hill. Anyway, roads are sodding complicated. Why not curl up with this little number: https://www.standardsforhighways.co.uk/prod/attachments/c27c55b7-2dfc-4597-923a-4d1b4bd6c9fa https://www.standardsforhighways.co.uk/prod/attachments/c27c... That's the design standard for major UK roads. Note things like crest and sag curves, worrying about kerbs and so on.

mycall 5y ago
> in and out and the main bend itself How would you rate humans for doing this right, in general , everywhere they have?
- gerdesj 5y ago
  Not sure I've looked at the entire world's road bends. I do know that with a few exceptions, English roads do what the posted signs say they will do.
mnw21cam 5y ago
That's interesting you mention Haldon Hill. They may have got the A38 bit right, but if you go up the other road, the A380, then there is a quite sharp blind right bend at the top that clearly hasn't been done right.
- gerdesj 5y ago

andi999 5y ago

What proof of surface tension are you referring to?