Engineering number formats
Engineering number formats
(OP)
How do you guys like your numbers to display when doing math?
Most of the numbers I work with are between 0 and a million. So I end up with numbers like "589,484", or "12,587". Ideally I'd like those to display with 2 or 3 sig figs, ie "590,000" or "12,500". But there is no option to easily display numbers like this in any software I use.
The "engineering notation" option in MathCAD and other software doesn't work for me; it is hard to quickly read for most people. 589x10^3, 12.5E3 ... ugh.
Thoughts?
Most of the numbers I work with are between 0 and a million. So I end up with numbers like "589,484", or "12,587". Ideally I'd like those to display with 2 or 3 sig figs, ie "590,000" or "12,500". But there is no option to easily display numbers like this in any software I use.
The "engineering notation" option in MathCAD and other software doesn't work for me; it is hard to quickly read for most people. 589x10^3, 12.5E3 ... ugh.
Thoughts?





RE: Engineering number formats
RE: Engineering number formats
I want to see how other engineers like/would like to display and communicate numbers, and how many engineers use and like the "engineering notation" format.
RE: Engineering number formats
Now if you're teaching a class where you're trying to use simple, easy-to-understand examples which only need to be precise enough to get the point across, perhaps, but if I'm performing some sort of legitimate calculation and I get a complex result, I tend to use it as is and not try to imply any sort of 'tolerance' or 'safety factor' until I get to where I was headed since sometimes it not always that easy to discern early in a problem's workflow as to which parameters will be 'additive' or 'subtractive' with respect to the final result. I would rather have a result I can have confidence in than one that might be easier to remember or manipulate when it's reused. If that's an important consideration, I'd rather make that call ONCE and know that it's the ONLY accommodation which I've made in the entire process.
John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA
http://www.siemens.com/plm
http://www.plmworld.org/museum/
To an Engineer, the glass is twice as big as it needs to be.
RE: Engineering number formats
- Steve
RE: Engineering number formats
RE: Engineering number formats
Implying that "precision" in data reporting is somehow better "accuracy" is simply wrong. When I was in school (the HP calculators were a couple of years old), I had a Mechanics of Materials teacher stand up and say that if the data would support an answer of 252 and your calculator read out as 252.1123, and you wrote that down with the decimals you were 100% wrong. A lot of folks didn't take that to heart and after the first test most of those folks were gone. He didn't care what numbers you wrote down on the intermediate calcs, but the answer had to be supported in the input data and the quality of the algorithms you used. He was right then. He would be right today. We seem to have lost any comprehension of matching the answer to the input and it drives me crazy.
David
RE: Engineering number formats
I remember in my uni days we had a lecturer who was just as pedantic on the issue of "unjustifiable precision" as your mechanics lecturer.
I particularly remember he once set us an assignment to determine the water supply requirements for a city "with a population of 1 million people" - the important point being that one of your key input parameters had just 1 significant digit of accuracy. He would deduct 25% marks for every unjustifiable digit in your final answer. If your calculations arrived at a figure of 53.265 GL (say), the "correct" answer was 5 x 10^1 GL (or 5 x 10^10 L). Even writing 50 GL implied two digits of precision, and you would be down 25% straight away, while 53.3 GL would lose 50%, and so on.
While I am not as pedantic as he was, it was a lesson well learnt. In my own field of structural engineering, many of the design loads are probably only known to 1 or at best 1 1/2 digits of effective precision. (We may use a design wind speed of 47 m/s say, but do we REALLY know the wind speed with that level of accuracy, let alone the pressure coefficients on all surfaces of a building, and so on.) I let my calculations carry full computer / calculator precision, but I will generally format my final solution to show no more than 3 digits of precision (for which my lecturer would have knocked off 50% marks!)
Somehow, while deep in my gut, I know that we really only "know" the wind load effects etc to about 1 digit of precision, I have to report the rafter stresses as "125 MPa" say - I just can't bring myself to report the bending stresses as "1 x 10^2 MPa" or "2 x 10^2 MPa" for a rafter that is working quite hard.
RE: Engineering number formats
Will
Sheffield UK
Designer of machine tools - user of modified screws
RE: Engineering number formats
johnbaker is right. What we were taught was to carry the digits on through the calculation (when necessary) so that problematic rounding errors would be avoided.
I still dismiss anyone's information as ill informed if it is reported in too precise a number. It is a sure sign of ignorance.
RE: Engineering number formats
Surely there are some other engineers out there that like to make things as simple as possible when communicating numbers.
If I tell someone "120,000lbs", there is little room for misinterpretation.
"1.2x10^5" will get a "huh?" from the average person.
"120x10^3" not much better.
"118,427" looks unprofessional imo, and for lack of a better word, bore people. =)
"60 tons" is ambiguous.
"120k", "120kip" isn't always understood.
Yall see where I'm going with this range of numbers?
/not a very good geek. :D
RE: Engineering number formats
Accuracy needs to fit the task.
RE: Engineering number formats
Therefore David, the gauge you mentioned if I looked at it would have an accuracy of +/- half of 250 = 125 which makes a reading of 119.547 outside the minimum accuracy of the gauge.
I have used the rule of thumb all of my career when I read any gauge.
rmw
RE: Engineering number formats
If you don't know what the mirror is for (and care) it is:
David
RE: Engineering number formats
RE: Engineering number formats
I was taught about engineering notation in college in Canada back in the '70s. I like it a lot, and I wish spreadsheets and programming languages supported it.
I have written a C subroutine that outputs numbers in engineering notation.
RE: Engineering number formats
The surveyor says "4.0", because he only believes his level rod to the nearest tenth.
The engineer says "4.00", because he only believes any answer to 3 significant figures at most.
The CAD drafter says "3-63/64"...
The mathematician says "4".
The lawyer says "What do you want the answer to be?"
RE: Engineering number formats
- Steve
RE: Engineering number formats
RE: Engineering number formats
Will
Sheffield UK
Designer of machine tools - user of modified screws
RE: Engineering number formats
Accountants should be barred by law from practicing anything but accounting.
I have no idea how this guy has managed to stay in business, but all his contracts are for high dollar cost plus jobs and he has so far apparently been able to get away with the insanity.
RE: Engineering number formats
RE: Engineering number formats
I like to try to keep things to sigfigs or at least something reasonable. A lot has been lost with the advent of computers and calculators - slide rule precision baby - that's what it is about in my type of engineering . . . especially when "stuff" like unit weights are guesstimated (and sometimes badly).
RE: Engineering number formats
Tobalcane
"If you avoid failure, you also avoid success."
"Luck is where preparation meets opportunity"
RE: Engineering number formats
RE: Engineering number formats
If I saw one of your reports that implied 9 digit accuracy I would get out my magnifying glass to see where else you were lying to me. Sorry to be blunt, but the the output of a +/-10% calculation with inputs that are no better than +/-5% is not 123.456789012, it is "about 125", or "in the range of 100-150". I have never seen a report where an engineer reported unreasonable precision that he didn't also have other errors that were material. Every time. People who don't bother to consider the precision of their inputs and equations tend to also be sloppy engineers who put too much reliance on what the computer tells them, often without bothering to verify that equations they are using are even germane to the problem they are trying to solve.
You are right that you would rarely have a case where you would make a different decision with and answer of 123.456789012 or 125, but my issue is that while 125 might be defensible in the data, the other number isn't. Ever. I suppose that makes me a "purist". So be it.
For intermediate steps I always use the full precision available. Rounding the intermediate steps does not add value (we did it in the slide rule days, but they're gone). When you get to the end, an engineer is obligated to look at the result and assess his confidence in it. I do that by tossing implied, unsupported precision.
David
RE: Engineering number formats
RE: Engineering number formats
Absolutely. My problem is the squishy definition of "right". In the world I live in (mostly fluid mechanics stuff) there are very few closed-form solutions to problems, and the few that exist were developed based on a page or two of assumptions. Consequently, I rarely do or see equations that support much better than +/-10% (and most of them are closer to +/-30%). So if I take the most precise data that the human mind can devise a tool to capture and feed it into an equation where I'm basically pulling a result off of a poor Xerox of a 9 scale log-log chart, then I get uneasy saying 125 instead of "over a hundred". I couldn't bring myself to say 123.456. I'll carry 9 decimals in the intermediate steps (my Picket is in a box as well), but I'll make the answer reflect my confidence in the data.
If I had a problem with a closed form, general equation that didn't have to contain any assumptions, then if I had great data I'd take the answer to the precision that was supported by the data. I just never have that.
David
RE: Engineering number formats
Or vice versa.
old field guy
RE: Engineering number formats
That is the perfect example. Multiply a constant times an approximate value and change the uncertainty from +/-0.5 miles (804.672 m) to +/-50 m (164.042 ft).
10.0 miles could easily be called 16.1 km since then the uncertaingy is +/- 0.05 miles (80 m) and +/-50 m which are certaintly in the same ball park.
It is not hard, but it seems to be a lost concept.
David
RE: Engineering number formats
Exactly. Especially when the figure is written by a reporter who wouldn't know a "significant digit" if it bit him in the ass...
old field guy
RE: Engineering number formats
----------------------------------
If we learn from our mistakes I'm getting a great education!
RE: Engineering number formats
David
RE: Engineering number formats
----------------------------------
If we learn from our mistakes I'm getting a great education!
RE: Engineering number formats
RE: Engineering number formats
David