Download: Pro-G (75) - 56,96 KB

History:

• 2008/6/8 : v 1.0
• 2008/4/15 : french translation of LightBiz from ThemePorter

Dry pasta material data can be found in [1] :

• density : 1.5 kg/dm³
• Young modulus 3600 N/mm²
• elasticity and yield limit : 16.778 N/mm²

Using CosmosWorks, we made a statics study of the bridge under a load of 10N (~1Kg) on 0.1m x 0.1m square placed on the center of the deck, and obtained the safety factor which gives the maximal load of the bridge as well as the location of maximal stresses, which indicates the points to reinforce

Référence:

1. Luis Alberto Segovia González, Inácio Benvegnu Morsch, João Ricardo Masuero, “Didactic Games in Engineering Teaching - Case : Spaghetti Bridges and Building Contest “, 2005, Proceedings of COBEM 2005 18th International Congress of Mechanical Engineering

Installation and usage:

1. Download wp-processing (94) - 1,46 KB , unzip and and upload the full wp-processing folder to your server it in folder wp-content/plugins
2. Activate the “Processing” plugin in WordPress admin console
3. Copy your Processing content on sur votre serveur in a folder named /processing, preserving the file structure used by Processing’s IDE. While looking for application “AppName”, the plugin will search for file “/processing/AppName/applet/AppName.jar”
4. Insert a Processing applet in a page or article by using the following syntax : [ processing=AppName ]

History:

• 2008/04/13 0.1 alpha

ToDo / A Faire:

• paramètres pour taille et path

• press [up] and [left] cursor keys to alter the ellipses ratio
• press [left] and [right] cursor keys to alter the ellipses tilt angle

Source code: Galaxy

Built with Processing

GLua was especially designed as a base library for the Demoniak3D real-time engine, as it provides a consistent framework with GLSL shaders. ]]> http://projets.ch/wp/blog/43/feed/ http://projets.ch/wp/blog/10/ http://projets.ch/wp/blog/10/#comments Thu, 20 Mar 2008 16:10:52 +0000 Philippe Guglielmetti http://projets.ch/wp/blog/2008/03/20/librairie-glua/ Contents:

• vec3.lua : 3D vectors and arithmetic
• mat3.lua : 3D matrix and arithmetic
• vec4.lua : 4D vectors and arithmetic
• glsl.lua : generic functions following glsl prototypes as defined in book “OpenGL Shading Language” by Randi J. Rost
• test.X.lua : unit test of module X

Usage & Samples:

Usage of the classes and functions is pretty straightforward.

Check the test.X.lua modules for more details or examples

Download:

GLua is available on http://luaforge.net/projects/glua/ under LGPL licence.

Implementation details:

• the classes are based on class.lua, described on http://lua-users.org/wiki/SimpleLuaClasses
• glsl.lua offers generic function that work on LUA numbers and tables, and therefore all vector / matrix classes. It makes extensive use of functional programming to achieve this:
  • the “apply” function is defined as follows:
    --- applies a function to a table of parameters
-- @param f : function to apply to each element in v
-- @param v : (vector of) parameter(s) to f function
-- @return : (vector of) result(s) of f(v)
function apply(f,v)
if type(v)=="number" then return f(v) end
if type(v)=="table" then
local res={}
for i,x in ipairs(v) do res[i]=f(x) end
return res
end
error(”apply “..f..”(”..type(v)..”) not implemented”)
end
  • then, functions can easily be defined to support numbers, vectors, or matrices:
function sin(rad)
return apply(math.sin,rad)
end
• dot product is implemented in 2 different ways :
  • through the “power” ^ operator in classes
  • as a generic dot(p1,p2) function in glsl.lua

J’ai rapidement modélisé une roue de train approximative en acier de construction standard avec SolidWorks et calculé avec CosmosWorks l’effet d’une charge de 12 tonnes correspondant environ au poids d’une motrice divisé par le nombre de ses roues. Le résultat à l’arrêt est ci-dessous : la contrainte au contact du rail est approximativement la moitié de la limite élastique de la roue :

A 55 tours par seconde, la force centrifuge change donc complètement la donne:

La contrainte due au contact avec le rail est à peine plus élevée que celle due à la force centrifuge (on parlera du signe plus bas) . Le coefficient de sécurité correspondant n’est plus que de 1.6 au lieu de 2.1 à l’arrêt : en fait la roue est plus près d’éclater à cause de sa vitesse que de s’écraser sous le poids de la loco !

Si on regarde ce qui se passe dans la zone de contact avec le rail, on voit que la contrainte est en compression (zone bleue) alors que dans toute la roue les contraintes sont en traction sous l’effet de la force centrifuge

la compression due au 12 tonnes de poids du train est extrêmement localisée : à quelques milimètres de la zone de contact, c’est comme s’il n’y avait plus de rail. Chaque point du bandage de la roue passe 50 fois par seconde d’une forte traction à une forte compression et vice-versa en moins d’un millième de seconde : combinés avec les vibrations et les chocs, ce sont des conditions idéales pour la fatigue des matériaux…

Un facteur de sécurité de 2 sur les roues du TGV à 575 km/h est suffisant pour un record, mais pas pour un train censé rouler des heures par jour à cette vitesse. Si les trains commerciaux atteignent un jour cette vitesse, ils auront des roues très “high-tech”, ou plus de roues du tout.

History

It all started form a “challenge” posted by Cliff Huprich on alt.machines.cnc newsgroup: find the shortest path going through 400 random X/Y points. Some people tried to solve the problem with existing (expensive) tools or simple heuristics.

We took the challenge by considering the problem as a general operations research problem, made some web searches and proposed to use a “Travelling Salesman” approach in some posts summarized below:

This is known as the “travelling salesman” problem.
see http://www.ing.unlp.edu.ar/cetad/mos/TSPBIB_home.html for a complete reference.
What you need is a program that would “sort” your list of points along the shortest path.
—
Made some more research on the “travelling salesman” problem (TSP) that
reminded me good old times at university…

Basically, TSP is a “NP-complete” (even “NP_hard”) problem, which means the absolute optimum path can only be found by trying all possible paths, which results in exponential complexity :  if it takes a time T to solve for N points, it will take N*T to solve for N+1 points. On a powerful PC, if you have more than about 100 points, forget it!

Luckily, there are several approximate methods and algorithms around. See http://w1.859.telia.com/~u85905224/tsp/TSP.htm for very nice, animated web-based solvers that illustrate the most common methods.

The efficiency of the various approximate methods depends on how the points are distributed : they work faster on points that are grouped in clusters or spread along a curve compared to random points.

http://www.math.princeton.edu/tsp/concorde.html is apparently the most efficient code available (check their “benchmarks” link ) but it still takes 7 mins for 1000 points, one hour for 1300, and days for 2000…
So to solve your problem, the first question is : do you accept sub-optimal solutions ?
The second is : how many points do you have ?
And the third : how are they spread in space ? (please send me a sample by e-mail)
—
I could compile concorde on Windows and it works pretty well in 2D now, about 5 secs for 1000 points on my dual AMD 1800.
Will modify it for 3D and ask Princeton if they allow me to release the executable
for free, because their license restricts use to academic research.
—

The “travelling salesman problem” (TSP) in general considers a “cost” for each pair of points (cities).
all you have to provide to the algorithm is simply this cost function.
If the function gives the (euclidian) distance between the points, it will solve the problem discussed here.
In the case where you have highway or routes, you assign them for example the travel time as cost (and even give different costs to the A->B and B->A route), and infinite (or very large) cost to pairs that have no direct route between them, which actually speeds the algorithm considerably up by limiting the search width. This is why TSP algorithms are tested/benchmarked precisely on “symmetric euclidian” problems like the one we have here.

As you see, the TSP approach is very powerful, but I still have a difficulty with the problem posed here as TSP solvers give the shortest *closed* path through all points. For an open path with a given start point, there should be a simple thing to do but I’m not sure which…

Links:

1. http://www.ing.unlp.edu.ar/cetad/mos/TSPBIB_home.html everything on Travelling Salesman Problem (TSP)
2. http://w1.859.telia.com/~u85905224/tsp/TSP.htm very nice, animated web-based solvers that illustrate the most common methods.
3. http://www.math.princeton.edu/tsp/concorde.html concorde solver
4. Kenneth Castelino Roshan D’Souza and Paul K. Wright “Tool-path Optimization for Minimizing Airtime during Machining” http://kingkong.me.berkeley.edu/~kenneth/research/pubs/airtime_minimization_joms.pdf

[/english]