What does "engineering accuracy" mean to you?

I have seen the phrase "engineering accuracy" in many different forms and
contexts and I thought it might be amusing to read what the builders,
engineers and nerds here interpret the phrase to mean.

For example, in my field, engineering accuracy for a solution of the lifting
surface equation is 6 figures, and methods like VLM can achieve that sort of
accuracy for rectangular wings. The same method fails, often miserably, for
other planform shapes.

I have no idea what the phrase means for actual boats and ships, except for
rowing shells where < +/- 0.5mm in the local beam is very desirable.

So, what does "engineering accuracy" mean to you when used for your vessels,
sails, props etc? Tales of woe and despair are particularly welcome.

 

Leo

For me, it is means “how real”.

For example, calculating the loads (oddly enough which I am now currently doing as I write this) of an item on a boat, when the vessel is pitching and heaving in a seaway..what are the loads on the foundations?

The accuracy does not require to be absolute. Since if I have a 1tonne weight on a boat say 150 tonnes, and pitching in a Hs of say 3.0m, if the answer I get is say 500N, then does this sound real…..er…nope. If the answer is say 50kN, does this sound real…well..it sounds far more likely than 500N. If the answer is say 500kN…does this sound real…er nope.

Thus the absolute nature of a result, is not important to me, but the “realism” of the answer is. Why?...because there are so many unknowns in a boat and so many things that can change during the life of the build..and also during the in-service life of the boat. Thus, best guesses are far more important that engineering absolutes, real or otherwise.

At the design stage…so much is unknown..despite endless data available..and so many correct theories to calculate the XX’s and YY’s. But all you need is one of the variables to be incorrect and the errors can be rather large as can the subsequent design deliverables. A boat doing 25knost compared to 28knots…liquidated damages can be huge..thus, I need “real data” to calculate. Thus I use “trends” as my guide…rather than absolutes…what ‘feels’ about right. Of course I then use the theorems, where required, to calculate my answer, but again, still with a pinch of salt

The end result, doing what it is designed to do…i.e. satisfying the SOR..is important. So long as you can satisfy the SOR, how you arrive at the solution..is almost irrelevant, so long as the methodology is consistent.

 

It means working to an accuracy that is needed.

No need to work to a mm if half a metre is good enough.

It's a waste of time, effort and energy.

But, if working to 0.0001 mm is necessary, then this is the degree of accuracy that should be attained.

Perfectionists will work to fine detail all the time when, chances are, it's only really necessary 5% of the time depending on what they are working on.

Engineered accuracy is an efficient use and application of time and attention.

 

In my research group, it's my job to translate what the researcher's experimental stated goals are into some kind of hardware that performs a function. I should be able to grasp the functional intent of the researchers (fulfilling the client's SOR using this forum's parlance) and execute it.

Therefore, if I've done my job right, I've executed a design with enough engineering accuracy so that this alone does not affect the desired functionality. (Typically, engineering is applied science and other times, it can be 'art'.)

Practically, it means that the issues and problems that eventually arise therefore are not directly related to a lack of accuracy on my part, be it prediction, tolerances for machining/manufacturing that I've specified, materials that I've specified, design/implementation of diagnostics (critical), assembly procedures I've specified, initial testing/trouble-shooting procedures, etc.

The accuracy required is always dependent on the specified function.

 

Consistent with the previous answers: Engineering accuracy is the accuracy needed to support making a decision or answering a question with the desired confidence level.

 

Do you mean accuracy in comparison to the exact mathematical solution, or accuracy in comparison to experimental data?

6 figure accuracy implies accuracy to 1 part in 1 million. I strongly doubt there are any wind tunnels which can deliver that level of accuracy in measuring the force and moment coefficients of a lifting surface during a test. On the other hand there may be some which routinely deliver that level of precision.

 

Submarine Tom hit it well. " Engineering accuracy" is relative to the accuracy needed to meet the requirement of the job. I worked on one job that required electrical parameters of ordinary components like resistors, capacitors and vacuum tubes to 8 decimal places in order to meet the requirements of the project. That was the first Trans-Atlantic submarine coaxial cable. End to end loss was in the thousands of Decibels and nothing on the bottom of the ocean could be adjusted.

Ever do a square root to 8 decimal places with a handcrank calculator that can only add and multiply?

Engineering accuracy for internal components in a modern auto engine is orders of magnitude tighter than for most chassis components.

I recently built a mold for wooden International Optimists with some tolerances of 2.5MM. That was ***** and an obvious attempt to discourage wooden boats from competition. Not appropriate "engineering accuracy" for the job at all.

 

I think it is often what the smallest possible measurement in the application is. For example, if I am using a measuring tape that has 1/16" markings, it makes no sense to make a specification to thousandths.

 

There is also the effect and limits of cost, use & life of an item. There are things like minimum breaking strength, sea margin, fouling, corrosion allowance, unit cost vs design cost, and criticality/fail-safe.

I was once in a design review where a contractor was discussing needing more money for CRAY time because there was an assemetery in the ship that wasn't in the model. They said the difference in the answer could be as much as 50 lbs out of 800 lbs for the part. I did a conserative 1/2 rho V^2 calc on my pad and came up with 900 lbs. I looked at the sponser and said "We set design load at 1000 lbs" Sponser replied "Concur, I'm not spending one more dime on that, next topic..."

As I said in a recent topic on cfd, you need to be satisfied that the accuracy of the answer warrants your belief in its precision. "Engineering accuracy" is when the precision is such that the combined accuracy and precision fall with the design margin envelope... just like tolerancing a bolt hole.

 

DCockey nailed it. It takes into account the uniformity of the materials, reliabilty, durability, environmental vagaries, duty cycle, manufacturing techniques. It means spending a sensible amount of resources on a particular decision knowing the relative importance of it to the final item. Design spiral is a good example.

It's also why when something proves really well made and demand skyrockets, the mfg immediately changes it. Vasque Sundowners, Nike Lava Domes, etc.

 

For Kerwin and by Newman from MIT, and Tuck, "engineering accuracy"
for the numerical solution of the lifting surface integral equation
was 6 significant figures. As I said, it is easy to get that accuracy
using VLM. Mark Drela's code gets close without being pushed, and,
because VLMs are very stable for rectangular wings, he could probably
get an extra digit by extrapolation from coarser grids.

I can't imagine any wind tunnel ever being able to get close to that.

As for the tolerances and precision required for real objects, I can't
find fault with any of the replies so far. Except nobody has told us
tales involving ridiculous demands, or outrageous failures to meet
targets.

 

Of course. And you obviously wouldn't use that same tape for measuring tappet
clearances, or cylinder diameters.

Remind us, do you do survey work on boats? If so, to what sort of precision
do you measure the length, beam and draft of boats?

 

No, but I have seen many early grant applications that specified a "girl" (I kid
you not) to do the calculations for the projects.

 

funny, I had a few engineering professors when I was in collage in '77-82 that would tell of the days before computers where they had rooms full of girls working the old manual adding machines. The engineers would set up the calculations as a series of number matrix tables, and they would feed it into the que and than get their calculations back the next day. This was done when slide rules would not give you the precision you needed. The early space program engineering was all calculated this way. The first computers were put to use on the space program to replace these rooms full of girls that would only do number crunching. In fact that expression "number crunching" came from the noise coming out a room of mechanical adding machines grinding away at endless stacks of calculations. that horrible mechanical sound when you pulled the lever back and it would go "crunch" and print out the answer on a paper tape.

 

Sounds like real world stuff, & if they ask for "perfect" the response is "how about ideal?" Jeff.