Alec Washington @alecwash477 - Twitter Profile

10 days ago

Topology optimization? Of course you thought it would solve your problems. You fed the solver a single static load case and watched it chug on von Mises stress like it was Gatorade, all while completely ignoring principal stresses and actual mechanical properties. Let me guess -- you just finished the SolidWorks Simulation or Autodesk Fusion generative-design marketing deck (or maybe the Ansys webinar), and now you think organic lattices are the future because the pretty red-to-blue plot said so. You’re going to believe that right up until the first prototype hits the machine shop and they call you, asking why the flanges are thinner than the paper your college degree -- the one you Chegged your way through -- was printed on. Then the part buckles laterally under a moment that wasn’t even in your load case, crumples, and suddenly it’s your problem. While you’re dealing with that, you’ll finally notice that the “optimized” geometry has internal corners that act as stress risers, which the solver didn’t catch because the mesh was too coarse. Whoops. Now your impossible-to-reach toolpaths are forcing the shop to either use EDM or scrap the whole batch because you gave zero thought to how the part would actually be fixtured, inspected, or survive any real manufacturing process. The QC guys aren’t even turning the CMMs on; they’re just laughing the drawings back to the office. Then you’ll finally open Hibbeler’s Mechanics of Materials for real this time -- not just to regurgitate the von Mises formula -- and you’ll read the sections on lateral-torsional buckling and realize that the critical moment drops with the cube of the flange thickness and the square of the unbraced length. Your solver maximized both penalties while smiling at you. It never checked geometric nonlinearity. It never asked whether the material was ductile enough for the equivalent-stress assumption to hold. It never cared. After that, you’ll crack open Timoshenko or the AISC specs, stare at the LTB equations, and it will hit you: topology optimization isn’t failing because the algorithms are bad. It’s failing because you treated a single-parameter stiffness game like it was engineering. Anyway, Dave from a subsystem four teams away needs to move a bolt pattern over a quarter inch. Do you want to throw your CAD model away, or should I do it for you?

emm0sh's tweet photo. Topology optimization? Of course you thought it would solve your problems. You fed the solver a single static load case and watched it chug on von Mises stress like it was Gatorade, all while completely ignoring principal stresses and actual mechanical properties.

Let me guess -- you just finished the SolidWorks Simulation or Autodesk Fusion generative-design marketing deck (or maybe the Ansys webinar), and now you think organic lattices are the future because the pretty red-to-blue plot said so.

You’re going to believe that right up until the first prototype hits the machine shop and they call you, asking why the flanges are thinner than the paper your college degree -- the one you Chegged your way through -- was printed on. Then the part buckles laterally under a moment that wasn’t even in your load case, crumples, and suddenly it’s your problem.

While you’re dealing with that, you’ll finally notice that the “optimized” geometry has internal corners that act as stress risers, which the solver didn’t catch because the mesh was too coarse. Whoops. Now your impossible-to-reach toolpaths are forcing the shop to either use EDM or scrap the whole batch because you gave zero thought to how the part would actually be fixtured, inspected, or survive any real manufacturing process. The QC guys aren’t even turning the CMMs on; they’re just laughing the drawings back to the office.

Then you’ll finally open Hibbeler’s Mechanics of Materials for real this time -- not just to regurgitate the von Mises formula -- and you’ll read the sections on lateral-torsional buckling and realize that the critical moment drops with the cube of the flange thickness and the square of the unbraced length. Your solver maximized both penalties while smiling at you. It never checked geometric nonlinearity. It never asked whether the material was ductile enough for the equivalent-stress assumption to hold. It never cared.

After that, you’ll crack open Timoshenko or the AISC specs, stare at the LTB equations, and it will hit you: topology optimization isn’t failing because the algorithms are bad. It’s failing because you treated a single-parameter stiffness game like it was engineering.

Anyway, Dave from a subsystem four teams away needs to move a bolt pattern over a quarter inch. Do you want to throw your CAD model away, or should I do it for you?

132

6K

434

2K

334K

AlecWash477 retweeted

MERICA MEMED

@Mericamemed

6 months ago

This is by far the best edit yet.

64

14K

2K

5K

567K

AlecWash477 retweeted

vids that go hard

@vidsthatgohard

11 months ago

This is what the internet was made for

7

504

82

114

41K

AlecWash477 retweeted

LAKE CENTER CHRISTIAN BASEBALL PROGRAM @lccsHARDBALL

about 1 year ago

2025 STATE RUNNER-UP: REMEMBERING our past players, HONORING those who are moving on, and FORGING AHEAD with our returners...

lccsHARDBALL's tweet photo. 2025 STATE RUNNER-UP:

REMEMBERING our past players, HONORING those who are moving on, and FORGING AHEAD with our returners... https://t.co/Wp43o5qowV

0

17

4

0

3K

Who to follow

Seeing the World through 1s and 0s

AlecWash477 retweeted

Memes

@PardonMyMeme

about 1 year ago

13

2K

71

58

72K

AlecWash477 retweeted

MA5K CEO @Nobbie_OCs

almost 2 years ago

The scene where the X-Wings and U-Wings pull up to Scarif and immediately start wiping out the Imperial Walkers are some of my favorite 30 or so seconds of cinema ever