|
|
Posted by D Sherman [72.47.9.228] on Monday, November 07, 2011 at 11:19:20 :
In Reply to: Mechanical Engineers Question posted by Paul(in NY) [12.64.158.56] on Monday, November 07, 2011 at 08:51:15 :
It's all computer models these days, and in the old days it was all calculations with pencil and paper and slide rule. You don't "proof test" civil engineering like you do a rifle. What you do do is generally design for a 5 to 1 safety factor. An easy way to do that is to run all the calculations assuming the concrete and steel are only half as strong as you know they really will be. On major projects there is also lots of "proof testing" of component parts, especially batches of concrete.
The other big thing that's changed in bridge design besides the use of computer modeling is a federal requirement that the bridge must survive the failure of any single component. That's why you very rarely see a new steel truss bridge. Although they're very economical of materials, there are single-point failures in a traditional truss design where if one pin or top or bottom chord member fails, the whole thing will collapse. What I don't know is how they define "element" in a reinforced concrete arch structure. If every piece of rebar in a tension member is an "element" for regulatory purposes, it would be easy to add enough extra so that if one bar breaks the rest will carry the load.
There was probably some testing on models done in the old days before materials and calculations were as refined as they are today, but of course there were also major failures. Bridge failures are usually due to pretty extreme deterioration or damage from collisions.
The only explicit model testing I'm aware of was extensive wind-tunnel testing of the second Tacoma Narrows bridge, after the first failed in a fairly moderate wind due to resonant flutter. After that spectacular failure (recorded on film) raised everyone's awareness that wind loading matters, engineers tested models of other structures in wind tunnels. I don't know if they still do that or they now rely on computer models for the aerodynamics as well. Modeling the static performance of a structure is way easier than modeling the fluid dynamics of wind loading.
Simple old bridge designs like a simply-supported beam (log across the creek), or a suspension bridge or pinned-truss bridge with only vertical loads considered are easy enough to model with pencil and paper, slide rule, and trig tables. If you rivet or weld the joints but calculate as if they were pinned, you get a greater safety factor but waste some steel. In modern designs where all the joints are rigid, or there are no joints at all, and loads can come from all directions (seismic as well as wind), there's no way to work it all out except by finite-element ana.lysis on a computer. The early concrete bridge designers had to make a lot of simplifications in order to reduce to something they could calculate by hand.
So, yes when you drive across the bridge you're trusting your life to the computer programmers and the guys who created the computer model. On the other hand, there's still that 5:1 safety factor. A commercial airplane might be designed to a 2:1 safety factor at best, and a military fighter might be designed to 0.8:1 and then any parts that break during testing are beefed up just a little bit, since weight matters more than anything there. Paying passengers will soon be riding in Boeing's new glued-together plastic airplane, which of course was all "designed by computer". We'll see how that goes.
Follow Ups: