who doesn't use manual computations anymore? 5

[pattontom]

More than half of local structural engineers I know doesn't use manual computations anymore. They reason it's long process to get for example the biaxial interaction diagram, moments of each columns and beams and their interactions especially if they are designing very tall stories. So they rely on ETABs for example which is complete package and can even produce complex spectrum seismic analysis that would take 3 weeks for just one projects. As you know ETABs can automatically calculate all the moments and shears of each element and their combinations that you may miss out manually. So who amongst you also use such package and no longer do every process manually? And for those who do manually and spend 3 weeks for just one project writing dozens or hundreds of handwritten paper, what do you think the ETABS folks can miss out when they don't do each member manually?

 

[patprimmer]

pattontom

I am not a structural engineer and I never designed a static structure bigger than a garage and then I had a structural guy check my beams and footings, however a computer is a computer and a calc is a calc and some rules are universal.

I have successfully designed a few structures that have wheels and are towed along roads at some speed, all by hand and rule of thumb and dead reckoning. None ever catastrophically failed and all where substantially lighter than commercially available products. I had them built and running before a computer only jock would have finished the drawings.

I have also used computers since they occupied several floors of a large office block which was obviously some time ago.

I have also been around long enough to appreciate that electronic devices can and do make mistakes. I mean give me a break, just how many silicon layers at a nano thick are in a computer chip and how many chips and possible routes are there for an electron to travel around all the boards and through all the chips. It's actually incredible that they don't all get lost or diverted to the wrong spot or the right spot at the wrong time.

When was it that Intel had the floating point problem.

I know I personally had a simple pocket calculator way back when that repeatedly gave a very obviously incorrect result to a particular simple calculation. Lucky is was wrong enough so that a simple thought process identified the error. It was the type of thing like say working out the area of a circle say 3.something inch dia and the answer came out to thousands of sq inches.

I repeated it at least 5 times and gave it to several colleagues to check. We all got the same patently wrong answer from the same input. A simple change of one character in the input gave correct results, but going back to the original calculation even after other correct but different calcs still produced the same error.

As a result, I always check my calcs by several methods including a rough in the head to test for ballpark accuracy. Anyone who does not is well on the way to landing face first in the mud. One of your mentors actually needs to give you a little trip and a firm push so you learn the hard way as that seems the only way you will learn.





Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[BigInch]

I use transient hydraulic programs and never see any calculations done by hand. I wouldn't believe one if I did. It's not about the process you use to do a design, it's how you use the tools at your disposal to maximum effect in reaching a solution in which you have confidence. I'm not interested in producing e-paper just so I can sell it by e-ream weight. I sell the solution.

"People will work for you with blood and sweat and tears if they work for what they believe in......" - Simon Sinek

 

[Ussuri]

This is why ETABS and STAADs are so popular because they are just formulas put on graphical interface

I think you will find that these programs are not just a collection of formulae. How many would they need to cover every single eventuality?

Pretty much all analysis programs give wrong results. They lie to you. You just have to determine if the results are close enough for your purpose. That is where your judgement comes in.

 

[mtu1972]

I am 62 years old and try not to do any calculations by hand if I don't have to. I use VisualAnalysis and TEDDS and Excel spreadsheets. Even simple beams and columns can be done by modifying a "saved" model by changing the size, lengths, loads, etc. I print out load, shear, and moment diagrams along with the final design view showing the allowable design ratios. I am proud to be a "Black-Box" Engineer.

I have done way too many hand calculations in my day. But that experience usually gives me some insight as to what my first trial sizes should be. We don't to 30 story structures, but we do model large industrial facilities with 2 to 3 floors and lots of differing equipment loads.

I too have been using computers since 1970 - of course they were main frames with punch cards. My first matrix analysis and finite element classes were in 1973. My biggest pet peeve is people that refer to matrix frame analysis programs as Finite Element Analysis. Two different animals.

My next pet peeve is trying to model everything: girts, purlins, stair stringers, etc. That is absolutely nuts.

So I try to be computer literate, but I still design with ASD and the 8th Edition AISC manual on my desk. It sits on top of the 13th Edition which gets opened far fewer times. But since the programs design the members per the Codes I specify, it is not necessary to dig too deeply in the manuals any longer.



gjc

 

[jtx]

Just my two cents...

I guess you can consider me a "youngster." I'm in my early thirties, but I'm no spring chicken, either. I echo the sentements of many of the engineers that have commented, including the engineers who are over 50.

I use computer analysis software almost on a daily basis. But, I always back up the results it spits at me with some form of hand calc. Why? Because very smart and experienced engineers taught me not to trust them.

When you are working with a large structure (or any structure for that matter), how do you know your analysis is correct? How do you know if you have proper load paths? How do you know if your structure's dynamics is in the ballpark? How do you know if your software is giving you a "correct" deflection for a beam? Do you even know what your software is doing or how its doing it? I'm not saying that you need to do everything twice. What I am saying is a responsible engineer would verify his/her work to ensure that what the computer spits out is accurate, even if it's just a bunch of scribbles on a piece of scrap paper.

There is nothing wrong with soley including computer inputs and outputs in your calculation package. I do that all the time. I make sure that I am 10,000% comfortable with what I am putting in.

Just a tip, the engineers who are over 50 (as you describe them) can offer you insights to engineering in ways that no code book or textbook can. Instead of insulting someone because they are "old," I suggest you show a little respect and listen to what they have to say.

 

[Archie264]

One pitfall that I've seen with over-reliance on modeling software is improperly fixing joints that should be pinned. That is, modeling a joint in the software as fixed when in reality it behaves liked a pinned connection. It can make for some very squirrely designs...

 

[IDS]

My biggest pet peeve is people that refer to matrix frame analysis programs as Finite Element Analysis. Two different animals.

Not really important, but I'm just curious why that peeves you. The way I see it, a beam in a frame analysis is just a particular type of finite element, so it is finite element analysis, just as much as models with plate or brick elements, or any combination of the three.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[JAE]

I agree with IDS - a finite element can have two joints, three joints, etc. All are solved by matrix analysis no matter how many joints there are per member/element.

 

[Ron]

A humorous post at the least!!

I am an "older engineer". As Mike McCann noted, he has been using computers since about 1966...I'm not far behind him. Early 70's for me.

In high school, I taught my math class to use a slide rule....went through engineering school with one until my senior year. Amazing how many times over the years I've calculated things in my head that were within a few percent of the answer obtained by more accurate means. Sorry you younger ones have missed out on such thought provoking processes. Did it just today with my younger protege.

I'm also no stranger to sophisticated computer analysis....but my prior "training" has taught me to be suspicious of the output and to look at it critically. Computers are tools of the profession....they don't provide the engineering judgment...just the mathmatical manipulation. If you depend on them to solve engineering problems, you are using them incorrectly. If you depend on them for expedience in the mathmatical process...then you have the right idea.

Though an expensive camera can show a scene in its proper context, it takes a photographer to compose a photo that captures the nuances of the scene. A piano can render all the notes of a musical scale, but it takes a composer to assemble those notes into a palatable melody. The engineer must compose on the computer.....his instrument, his tool, his canvas.

 

[TXStructural]

Several years ago, it was found that one software program mentioned in this thread improperly applied and distributed parapet wind loads (this was for a high-rise project that was in design.) Shortly thereafter in a different firm, we found that another program improperly computed base shear. These were only found because we took time to verify our results with hand computations and/or comparison between various packages. It is not at all unusual for certain inputs to result in strange outputs. And it is always possible that a mis-entered value is not what one intended to type. While it seems easy enough to verify the software itself, you have not lived until you had to run a verification routine for each piece of software, on each computer, each day it is used, and keep a log of the date, time, user, and version, along with a copy of the relevant printout. And then found that minor version revisions resulted in slight differences in results in some computations, but not in the computation routine run for the verifications.

Hand calcs and other design aids, such as books and tables, allow rough design to be done as a sanity check. Randomly verifying things like single beam moment and shear values using quick hand computations help identify where problems might be. Simple tributary area and loading computations, combined with span information make sure that loads were applied properly. If things don't add up, you need to find out why. Blind reliance on the blackbox is not a good idea, but neither is doing every calc twice by hand, to verify the verification. In safety or reliability analysis, the human is the weak link vs a machine, so don't make the mistake of blindly accepting hand calcs, either.

 

[BigH]

My fear is that engineering will become a slave to computer programs and that there will be those engineers who cannot, even if they wanted, to be able to cross-check the computer output - even for a ball-park figure. I've seen slope stability analyses in a few cases by others that would be laughable if I wasn't crying so much - results that never could be in nature. Another comment is that any computer/calculator device may provide one a false sense of precision . . . as Ron knows, 3 sigfigs or 4 is about the best one can expect - but far too many just post/report the numbers without questioning precision. I once saw a pile capacity - based on the Davisson "failure criteria" (graphical on a pile load test) - reported to the nearest 0.01 kN. Pretty precise, eh? By the way, I was only 2 when Vic Wertz was robbed by Willie Mays in the World Series.

 

[SAIL3]

having come to the conclusion that computers/computer programs are here to stay(as long as the power grid holds up), I chucked my 24" slide rule a few months back...feel vulnerable w/o it..anyway, engineering is more of an art than a science, imo....

 

[Archie264]

Ron, BigH, Sail3,

Regarding slide rules, my mentor told me that whe he was starting out his mentor had a slide rule with a broken cursor - apparently it was only able to read to two digits worth of accuracy. He didn't replace it because, as he explained, that's all you need. My mentor told me that that left quite an impression on him. It's a shame those dudes didn't realise they need 14 digits worth of "accuracy".

 

[dcarr82775]

In the recent past a certain very popular PT slab program overstated punching shear capacities at corner columns by a fair amount. Computers are always correct (except when my old SegaGenesis cheated at Madden Football), it is the programming and input that are the problem.

 

[mtu1972]

Doug - it bugs me because lots of small things bug me and they really aren't the same. Yes FEA does use Matrix Methods to solve for its answers, but the fundamental set ups are different.

For structural members, the stiffness matrix is comprised of the terms that relate the internal forces in the member with joint displacements - such as PA/L, 12EI/L^3, 4EI/L, etc. Then the structure is solved like a complex spring where [P] = [K][d]. Joint displacements are solved for first and then member internal forces are found using the assembled stiffness matrix and the applied loads.

For finite element analysis my interpretation is; the relative displacements of the nodes on a particular element are defined with respect to one another (i.e. it's internal element stiffness); an overall stiffness matrix is developed, the displacements of nodes are solved for; and then the internal stresses of the elements are solved for using the internal stiffness fomulations that were used to define the element.

Here is a much better answer from thread earlier this year (thread507-323635)

PowersPE80 (Civil/Environmental) 10 Jun 12 9:59
Hello, Can someone please explain to me, without getting overly technical, what exactly a finite element analysis is?

Thanks.

. . .

JAE (Structural)
10 Jun 12 12:33
Powers - very tough to do that well without knowing what you do know.

Are you are familiar with frame analysis programs where you have beams and columns represented by "stick" members - a member that has a joint at either end? You connect all these into a frame or other representation of a structure and by knowing the deflection-force relationship between the two joints you can analyze the structure.

These two-joint "sticks" are two-joint finite elements. They are essentially a member/entity that creates a relationship between two joints. If you move one joint, the other responds with a defined force.

Each joint, typically - in 3D models - represents six degrees of freedom - X, Y, and Z direction translation and X, Y and Z direction rotations.

Now if you expand that two-joint member into a three or more joint member, the member now becomes either a planar flat element (triangle or rectangle) or perhaps even a solid element with multiple joints that all are related to each other in terms of displacement and force.

The analysis/solution is similar to the two-joint member except there is a problem. The two-joint beam type members can be analyzed such that both equilibrium and deformation compatibility are satisfied (Sum of forces/moments = 0 and the deflection of element A at a common joint with element B is the same). However, with more than two joints, you can't satisfy both of these - only one. Most all finite element solutions choose to satisfy the deformation compatibility and then by keeping the size of the elements small, minimize the sum of forces error in the model.​

And then the mesh can be refined to determine if you are honing in on the approximate answer.

gjc

 

[MikeHalloran]

I haven't done a calculation on paper since I got a TI-55 programmable calculator, a very long time ago.

I haven't stopped checking calculating machines' output since I discovered that some of the TI-55's user registers were also used by the machine for intermediate results of some of its built-in functions.


In the meantime, I've learned to program microprocessors right down on the silicon, and gotten pretty good at it, I think.

I still don't trust the damn things.



Mike Halloran
Pembroke Pines, FL, USA

 

[msquared48]

All they are is a tool. You still have to think, or should at least.

Mike McCann
MMC Engineering

http://mmcengineering.tripod.com

 

[mikesg]

Is it impractical to combine TXStructural's solution (two splices instead of one) but weld instead of bolting - this way you will have the beam look like it should and work like it should?
If welding is impractical at all, even in places with smaller forces, can slip-critical bolts be used in order to avoid the reduction of capacity due to bolt holes?

 

[patprimmer]

mikesg

Wrong thread I think.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[haynewp]

Yeah but you see the Atari had wood grain so that trumps your BIM/Revit/ETABS/Wii U.