Ask the Structural Engineer!

In My Humble Opinion
61

A general question if I may.

This summer my friends and I are planning on rebuilding an old garage. I’ve been doing some general design research and just wanted to understand if there’s a significant difference between a flat roof (simple and easy for us to buld) or a typical slanted roof. Because of winter, should we really be against flat-roof structures?

Basically, the question is, what is best? We’ll be working mostly with wood (I’m more of a machinist than a carpenter so I don’t know much about wood strength) with a simple 4 (or 6) post building. I envision driving 4 posts into the ground, throwing on some wood for a roof and sealing it all up against the weather.

Of course this is a very general plan. Other than the red tape of permits, we’ll probably have to get someone with some SE experience (maybe root around the University of Toronto) to get some ideas but I thought I’d throw out the question here.

The building foot print will probably be in the 14 x 14 foot range (around 4 meters each side).

If you can picture the building I envision, it’ll be simply 4 posts 10 feet high, 4 cross beams with 45 degree 3-4 foot long supports, and lots of wood on top (a typical child-fort if you will).

Thanks for any help! And before anyone freaks out how dangerous this might be, I’m really just interested in design ideas. I figure legally we’re going to need to hire some real help.

62

Flat roof = water leak. Flat roof as a term of art is a mis-nomer. All roof should have positive drainage. The steeper the roof, the less problem with leaks.

Additionally, as you noted, flat roof get more snow load. However, to eliminate the snow load you need a very steep roof, which cost more. So the question is economical. Is is cheaper to build a strong flat roof or a weaker, steep roof. Either will work.

A working definition of an engineer is someone who can do for $1 what anyone else can do for $3.

Keep in mind that small wood-framed structures have been successfully built without SEs or computer analysis for 100s of years.

63

How about a post and beam structure for a shed?

I plan on building a post and beam shed with a 10/12 roof.

I live in NJ, and like the interior layout a post and beam shed offers. Are their inherent disadvantages of post and beam over common framing?

like this: http://www.pineharbor.com/how_to_build_a_shed_1.htm

64

Cost only. Framing with dimensional lumber is by far the cheapest type of construction for a building of this type. OTOH, the post-and-beam is very attractive.

That is the key to almost all structural engineering. We can design just about anything using any material. The question is, do you want to pay for it?

65

Snow load is a factor even on a pitched roof. Because snow sticks, the pitch has to be rather steep before a designer can disregard snow load. Unfortunately, I can’t remember the exact pitch but I think it has to be greater than 10:12 slope. This then brings other factors into play such as wind loading which must be taken into account. Wind loading can exert a great deal of stress on a building.

Despite the small size of your building, you’ll probably have to go to the expense of hiring a registered Professional Engineer of Ontario to get your building design stamped and ready to go through the whole building permit application process. Municipalities in Ontario are increasingly loathe to accept any sort of responbility for design approval. This is due mostly to issues stemming from a few high-profile lawsuits in which municipalities were found to be liable for structural failures on jobs that had been reviewed by city inspectors. As a result, they are frequently demanding an engineer’s approval of the design and site reviews for even the smallest of jobs. Somebody with “experience” will not likely meet the city’s requirements.

Given the small size, and simplicity of the building you should be able to get a simple design from an engineer for reasonably low cost. Also, a tip for you might to approach an engineer or firm outside of Toronto. It’s been my experience that Toronto based engineers are a great deal more expensive than those in surrounding areas. However, if the city requires a site visit any cost savings would be eaten up by travel time.

66

Jinx:

According to the intro course that I teach, the underlying theory of FEA is over 100 years old, and was used for hand calculations in analyzing the integrity of suspension bridges and steam boilers. This theory was developed into the finite element method for structural analysis during the 1950s and 1960s. FEA packages started appearing in the mid-60s to early-70s, partially due to the development of the space program.

As late as the mid-80s (going by my friends’ and co-workers’ ages on the time period here), all the finite element model information was contained on a deck of punch cards. There was a card containing information for each node, each element, each material, each real constant (shell thickness or beam cross section information usually), etc. This deck had to be kept in order, so you sure as hell didn’t want to drop it, and run through a mainframe. Generally the engineer would drop off the deck and pick up the results later. Generating the node information was a bitch; you had to either specify the coordinates manually or, later on, sketch the mesh on vellum and scan it in. Nowadays, node generation is automated by geometry meshers.

Back in those days, FEA wasn’t used nearly as often as it is now, for obvious reasons. More often, the engineer would make assumptions and approximations, erring toward conservatism, regarding a components geometry and the load distribution through that geometry, and performed hand calculations. If a certain geometry was analyzed frequently enough, it was common to write a computer program to solve these equations.
Thanks to massive improvements in solver technology as well as computer speed and memory, finite element solution speed has improved immensely over the past few years. Back in the mid-90s, a linear static model consisting of just a few thousand nodes would take hours to run on my UNIX box. Nowadays, I frequently run models containing over a hundred thousand nodes in much less time on my PC, and I’m not even top of the line here.

canvas:

I agree with you here, especially with the development of more designer-friendly FEA packages and the push by the higher ups to show growth and increase the stock price by targeting designers, since they represent a much less limited pool than analysts. I frequently get tech support calls from and train people who have no business doing finite element analysis. The “designer-friendly” packages have the potential to become a headache, since they’re advertised as “push-button” which is something that FEA never has been, nor ever will be. Regarding this, a customer of mine said it’s “like handing a running chainsaw to a four-year-old”. I always make my opinion known that any analysis work performed by a designer should be just a quick check for minor design decisions and that it should be performed under the supervision of a qualified analyst, with the analyst performing or at least OKing the final analysis. I also push the sales people to promote our mentoring services and our fundamentals of FEM training course when they’re selling to designers.

67

Interesting thread, this. I’ve got a couple of questions too. (These questions relate to a book I’ve been planning to write for ages, but don’t know if I’ll ever finish it. If you don’t want to answer, that’s cool)

Following up on Jinx, and Straingers’s answer: if you designed a bridge (or other structure) without FEA, would that imply that you drew the main outline, then calculated by hand the tension and stress on each element, then design each element by hand by using tables and hand calculation?

What did a structural engineer do back in the seventies/beginning of eighties to get started in the job? I’m thinking about an engineer starting to work for a engineering company that designs structures such as bridges etc. Would you be part of a team, would you start drawing simple stuff on spec, would you help in checking calculations, or would you possibly be expected to calculate the specs for beams etcetera for the complete structure?

Following on Spectre’s question and culture’s answer: Given the modern tools for designing structures, what is the role left for the engineer? It looks as if such CAD-like tools make it easier for an architect to design a building or so without help of an engineer, except for final safety checking. Wouldn’t that have been different in the past, when the architect was more dependent on the SE? (I saw culture’s response to Oxymoron go into this partly).

(Over here there is a case where apparently (according to newspaper reports) the architect basically designed a whole bridge with computer programs, then had it only checked for strength by engineers, but with the result that the structure proved to be unstable in heavy wind. I’ve heard people say that this wouldn’t happen if someone with a solid engineering background would have designed it: the architect allegedly had only looked at prettiness and didn’t have as second nature to take note of solidity. )

Is there much difference between the various modern FEA programs? Do they have a CAD-like interface, or are they more like in the past, when you had to enter structures as sets of numbers without any graphical representation?

What is the job like? Do you have much interaction with the client, or are you basically sitting behind desk/computer/drawing table? Are drawing tables still used, anyway? While computers can do a lot, I would guess large drawings are still difficult to produce automatically (I’ve seen large plotters, though).

How many people work on a large structure? I’m guessing a garage or small house might be done by a single SE, but building a bridge or sky scraper would need a large team.

Besides the growing role of computers, have there been major changes in the business/work in the last few decades?

Do you have to be knowledgeable about legal issues such as safety regulations and building limits/licenses, or are these seen as specs provided by others?

Well, I’ll stop for now.

68

**TTT[/E]

I’m sure the others will chime in shortly, but here’s my two cents worth:

The general trend in (non-residential) building design over the last several decades has been toward a fragmentation into more and more specialties, and not the return of the architect as the Master Builder. Software has allowed engineers to do more in less time, but, like all software is subject to GI/GO problems. You still need that person who knows what all the correct inputs (loads, support conditions, materials, etc.) are and has a good idea of what the answer will be before pressing the button. There are always goofy conditions that a program can’t handle and require you to sit down with a pencil and a calculator. Most jurisdictions in the US will require an SE’s seal on a project of any size. Most small things, like conventional houses and small outbuildings are done entirely by architects.

I think I work in a typical design office. I spend most of my time at a computer, either doing analysis, drawing on CAD (drafting tables are mostly a thing of the past), or shuffling paperwork. I do go out to see projects as they are under construction pretty regularly.

Typically, I’m the only engineer on a project (we do mostly low-rise buildings in the 100k-200k square foot range). On bigger or more complicated projects there might be four or more engineers. Really big projects (stadia and the like) tend to have temporary offices set up with lots of staff from several different firms.

Most of my interaction is with the architects and other engineers of various flavors, occasionaly with the client.

Most FEA programs have a graphical interface, and some can swap information with CAD programs.

A good portion of our job has to do with knowing the building codes very thoroughly.

69

Canvas? Where the hell did that come from? I meant to attribute the quote to culture.

I’ll get to you on your questions in a little bit, TTT, as best as I can, anyway since I’m an ME, not a CE.

70

TTT, I’ll go ahead and answer your questions from a mechanical engineer’s point-of-view. Mileage will probably vary for the CEs.

As far as the design goes, that’s the same regardless of whether I do FEA or hand calcs. If hand calcs, I’ll get the information I need from the designer (usually just drawing printouts) and use the dimensions as needed in my stress calculations. If FEA, I’ll either import the geometry into the FEA software or build it within the FEA, mesh it (i.e. fill it with nodes and elements) load it and solve it. From there I either OK the current design or recommend changes. There’s a lot of back and forth between the designer and the analyst.

Pretty much the same as now. Coming out of college, you generally work with experienced analysts at first. You’re generally off on your own, but there’s a lot of consultation up front about how to approach the problem. A lot of stuff is repeated so you may refer to past reports on similar products as well. You also ask a lot of questions along the way. As you learn more and more, you ween yourself from this supervision although you’ll ask questions to colleagues and senior team members here and there pretty much throughout your entire career.

Plenty. Even though there’s software that targets designers, it’s meant more for a quick look and is devoid of advanced functionality, such as nonlinearities, dynamics outside of model analysis, etc. It’s intended to reduce the back-and-forth between the designer and the engineer. There will be some cross over in tasks, but the engineer will always be necessary as far as finalizing the analysis, performing more advanced analyses, and growing the technology.

Once you learn one finite element package, it isn’t too difficult to pick others up. There are some fundamental difference between the programs that make various programs strong in some areas, but weak in others. For example, one may have great nonlinear analysis capability but be pretty lousy as far as doing a dynamic analysis whereas the opposite may be true for a different software. Each software has its own level of kludge. While FEA packages have solid model ability, they don’t have nearly the design capability of CAD systems, which is why we frequently import geometry from CAD. The reason for this is that the FEA packages are geared toward analysis, and didn’t add the solid modeling capabilities until later on. Although there are a number of designer friendly analysis tools coming out the work in conjunction with a CAD model, even though the FEA and CAD are two different packages.

Since I train other people in FEA applications, I do get up and out a lot and meet with a good number of people. However, when I was a stress analyst for a large company, I only saw the customer at design reviews and the occasional status meeting. Most of my communication was with the design and project engineers. Draft tables have gone the way of the dinosaur. Designers do it entirely on computers now.

71

Here is a good collection of geeks to ask. My college professor for structures and FEA at Duke was Robert J. Melosh (R.I.P). The annual FEA papers competition is the Robert Melosh Competition, same guy. My partner and I won his annual undergraduate class project/competition called “Excalibur” using crude FEA computations in 1987. I recall he was one of the “fathers” of FEA, or of modern computational methods, or some such claim to geekdom fame.

Google is letting me down, can anyone corroborate this?

72

giggle!! I WISH I’d quoted the quote you were quoting. (or something like that).

73

Is it just me, or is there a serious lack of talented engineers out there? More and more, I keep running across things which are capable of doing the task that they’re designed for, but could be vastly improved in their design (and produced at a lower cost than what they currently are). In some cases, it’s simply a matter of moving a part a few 1/16ths of an inch over to one side.

Admittedly, I’m not an engineer, but I am studying to be a machinist and have read engineering texts for fun, so I do have some understanding of engineering principles.

74

What do you consider to be the current “dream” projects for structural engineering? That is, projects that would push the field to the absolute limits and beyond but would still be feasible using current day or near future technology?

I have heard the Gilbritar bridge being mentioned as one such project. Any others?

75

I’m full of questions. My Idol (Frank llloyd Wright) once drew up plans for a 5200 foot skyscraper building (you can see the model in his 1957 book “TESTAMENT”. My questions:
-is such a building feasible?
-how much steel would be required to build it?
-and Wright designed the building like a tree-a central column with the floors being supported by the horizontal “branches”. Would such a design work? (Wright was not an engineer, but his designs worked pretty well-the tower at the Johnsons Wx Building (Racine, WI) was designed in this way.
Finally, Wright did all of this without benefit of CAD-has anybody subject his design for the "mile-high building to a modern analysis?
I’d LOVE to have an office on the top of this monster!

76

I’ll leave the other questions to the pros, but this type of cantilever design is quite feasible. A prominent example of just such a building is the Citi-Bank building in New York City.

77

…Until there was a fire…

78

Technologically, yes. There have been several mile-high towers proposed over the years, most recently, IIRC, by Donald Trump for Chicago. No one has yet put together the economic side, however. Trump’s project has been scaled back considerably.

Lots. I don’t have anything more specific than that.

The current paradigm for high-rise buildings is a central core (usually of concrete) and a perimeter of closely spaced columns interlaced with beams such that the perimeter acts as a giant tube. You really make it easy on yourself when you make the structural system as wide as possible. It’s stiffer and more stable. FLW’s towers work for mid-rise, but would probably be difficult at BTW, Citicorp is not cantilevered in the same way that FLW’s towers are, that is, not from the central core.

I don’t know if anyone has analyzed it. If you like his buildings, you can now stay at Price Tower in Bartlesville, OK. www.pricetower.org

79

Funny you should mention. The Gibralter Bridge is also my favortite. As a more pie-in-the sky favortite, check out Paulo Soleri’s book Arcology. He was the guy who invented the term. I would love to do the structural design on one of these babies. It would be the most awesome engineering project of all times.

80

Aside from the pun on “finite element,” is there any other meaning attached to “finite elephant”?

Where I live, I have a flat roof deck, approx. 17’ x 25’. How zealous do I have to be about “shovelling” (I have a roof-friendly “Sno-Brum”) snow off the roof?

This past winter, I got rid of a large snow accumulation because we were expecting rain to follow, before the snow could melt. Authorities did encourage area residents and business owners to do this where possible, because the snow can act as a giant sponge, absorbing the rain… and sure enough, a couple of flat-roofed buildings did have collapsed roofs after the rains. Is there a structural engineering rule of thumb ratio for snow drift height:expected rainfall:size of flat roof span (assuming standard contstruction standards)?

On behalf of my poor aching back (post-shovelling), thanks!