Hacker News Reader

Comment by WalterBright | Hacker News Reader

"But they suggested that over-reliance on the computer models, a common pitfall in the profession, may have played into it."

It's very easy to be seduced into assuming a computer model's results are correct. Computer models of critical bits should always be double-checked by a separate independent calculation.

For this bridge, a simple analysis of static forces should have been sufficient for a cross-check, despite the article repeatedly implying this was a difficult/tricky computation. It is not.

View on HN

Static force analysis only works when the materials being modeled are well known. Titanium dioxide concrete is a novel material—and, from the article’s text and subsequent research—appears to have a narrow effective range before integrity issues appear.

Static force analysis is independent of the materials (well except for the weight of those materials). It is working out what the loads will be on each member.

Designing how those members resist/react to those loads is dependent on the material though.

Knowing the strength of materials is pretty darn basic.

I would disagree. System failure modes can be quite complex. Yes you may have a good approximation of a material's Ultimate Tensile Strength through lab testing (although it isn't a single number and can be affected by a lot of variables, so it's usually designated as a range). But if your material yielded in a way that didn't account for, it could affect the loading on a different member, and trigger an issue with an entirely different part. Real world systems are more complex than your freshman year Statics assignments

Actually, I worked for Boeing on gearbox design. So I am fairly aware of strength of materials and did a lot of such computations. If you're unsure of the strength, you use the minimum guaranteed value. You don't use materials for critical pieces for which you don't have any data. That would be incredible incompetence.

Again, what I am talking about is using statics to determine the minimum loadings to check against the computer models, which should predict higher values or they are wrong.

This is not difficult.

It always scares me that civil engineers and other real engineers (with professional engineer certificates) rely on software built by software "engineers" who may just be some kids out of high school that picked up coding.

It's an extreme scenario, yes, as I'm sure/hope the software goes through rigorous testing.

When I was growing up I lived in a small town and the city planner had mocked the entire town in the original SimCity. The ongoing joke was that he would test any proposals in that.

Better than most, I guess.

At least you had a city planner.

The city I grew up in is a larger one, and currently seems to be entirely in the hands of private developers - who just dump closed-off groups of blocks at random throughout the city, hoping to sell the apartments, without any consideration to things like neighborhood, surrounding architecture or road throughput.

If you did find a way to design a structure so that analysis of the static forces was not straightforward, might that be a good indication that you shouldn’t build it at all?

Not necessarily. Static analysis means a specific thing in structural engineering. Pin jointed frames with diagonal members (eg trusses) are easy to analyse with just geometry (statically determinate). But moment resisting frames without diagonals are not statically determinate and require more iterative calculations (usually computer modelled these days). Doesn't mean you shouldn't build them though.

Doing a pin joint frame static analysis is simple, and will yield a lower bound on the loads on the struts. It would serve as a good cross check on the computer model.

Reply on news.ycombinator.com