How much trial-and-error at your work location? 7

[DHambley]

I'm just curious as to the prevalence of trial and error in different industries. Below, I describe two situations with two different design methods. It may be sucessfully argued either way that one is better than the other.
Method 1:
Try a 1/4 inch shaft on the new conveyor belt design. Put the max load on it. It broke. Hmmmm, I wonder why. Let's try a 1/2 inch shaft. Install the 1/2 inch shaft. Put the max load on it. It worked for a few hours, then broke. Hmmmm, I wonder why. Let's try a 1 inch shaft. Install the 1 inch shaft on the new belt. Put the max load on it. It didn't break! Cool. We'll use a 1 inch shaft for production. Total design time: 8 hr.
Method 2:
An engineer models the shaft and coupling. It takes him about 12 hr to set up the CAD model. This model includes worse-case extremes of shaft misalignment, temperature, shock load, speed, etc. It takes him about 4 hours to go through the model with different shaft diameters. He concludes that a 7/8 inch shaft is the minimum acceptable to gaurantee the life of the product, so he adds a margin to increase the diameter to 1 inch and specifies a 1 inch shaft for production. Total design time: 16 hr.

How much of method-1 do you see at your work location?

 

[1gibson]

Most here are engineers, so probably never use method 1 (except maybe at home in their garage.)

P.S. I would go with method 3, use an equation for shaft strength and assumptions for worst case load, then calculate; about 2 hours work maybe.

 

[MintJulep]

I agree with gibson's suggestion of method 3.

 

[KENAT]

Q: How much trial-and-error at your work location?

A: More than I'd like given we're meant to be engineers. Sadly the way schedules are set and progress reported to management it often seems it's better to get to half a$$ed hardware more quickly by minimizing analysis etc. rather than spending more time on design but taking longer.

In fairness some of our stuff is cutting edge, new applications... so at least some can be difficult to predict but still, I reckon we could do better sometimes.

(Oh and I don't mean just throwing every model into FEA etc. by default, I agree with 1gibson that sometimes the middle ground is most appropriate.)

Posting guidelines faq731-376

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

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[TheTick]

What does method 1 contribute to any knowledge of why the pin broke? How does one determine fatigue life in method 1? Who is going to explain method 1 to customer service when all of the pins breakl in the field?

 

[byrdj]

If the words "let's try" are used during a discussion of a problem, you will be escorted to the gate

 

[zdas04]

There are laboratory problems where you are so far on the bleeding edge that trial and error is the only way to develop algorithms to predict material behavior.

Outside of that, most of us will use Option 3 with varying safety factors. I'll usually calculate whatever limiting parameter is being discussed with a zero safety factor. Then do it again with a non-zero safety factor and compare the results and make an Engineering Decision on what to use. In my work if something breaks in the field it can cost millions of dollars to fix it and get back on line. We use large safety factors.

David Simpson, PE
MuleShoe Engineering

"Belief" is the acceptance of an hypotheses in the absence of data.
"Prejudice" is having an opinion not supported by the preponderance of the data.
"Knowledge" is only found through the accumulation and analysis of data.

 

[btrueblood]

I do a lot of product development work, on fluid dynamic devices, and so a lot of time is set aside for "cut and try". But not for easy problems, like burst pressure predictions, but more for tweaking flow rates to the last few percentage points, or adjusting a regulator to avoid chatter in certain piping runs/ flow conditions. And if you say, "why not use CFD", we can have a nice chuckle and go back to work.

 

[zdas04]

Star for

"why not use CFD", we can have a nice chuckle and go back to work

David Simpson, PE
MuleShoe Engineering

"Belief" is the acceptance of an hypotheses in the absence of data.
"Prejudice" is having an opinion not supported by the preponderance of the data.
"Knowledge" is only found through the accumulation and analysis of data.

 

[TheTick]

I've been that engineer. First solution takes many hours. Once the model is established and verified, following solutions take fifteen minutes each.

Option A is to keep breaking things without learning anything useful.

 

[hydtools]

I suspect method 1 is too optomistic for time. Changing the shaft requires changing mating components. Other parts may also be destroyed when the shaft fails. We don't all have an inexhaustable supply of incremental parts and material.

Method 3 works well for me. Spend a several hours on hand calculations. The drawings/modelings are near final product. Build a test model to prove the design. Make adjustments. Test to prove. Done. This has served us well for real time to market with a product we won't see come back very soon because of design failure.

Ted

 

[handleman]

Method 4: Use the same diameter as the output shaft of the motor or gearhead.

-handleman, CSWP (The new, easy test)

 

[EngineerErrant]

Most here are engineers, so probably never use method 1 (except maybe at home in their garage.)

What he said. Engineering is thinking about things before you build them, not throwing different-sized rocks at a problem and hoping it goes away.

What does method 1 contribute to any knowledge of why the pin broke? How does one determine fatigue life in method 1? Who is going to explain method 1 to customer service when all of the pins break in the field?

Also what he said. If you build things without understanding, it's just a matter of time before you aren't allowed to build things anymore.

Testing is essential for good design - good final design. It's not a replacement for thought, and trying to use it as such on every nut and bolt along the way will bankrupt your company.

(As an aside: if it takes you 12 hours to model a shaft/coupling in CAD, you either need to get better at CAD or hire someone who is.)

"Engineers like to solve problems. If there are no problems handily available, they will create their own problems." -Scott Adams

 

[tomwalz]

A fair amount.

We do a lot of looking for new applications for new materials. Thus we have to do some trials because the data is usually not available.

Secondly, the data we do get has often been through a marketing filter. The engineers supply figures but marketing takes whatever looks best and publishes that. We run our own tests because we want it to break in testing not at the customer’s.


Thomas J. Walz
Carbide Processors, Inc.

http://www.carbideprocessors.com

Good engineering starts with a Grainger Catalog.

 

[MiketheEngineer]

Our guys in the field do it all the time -

Famous words of their wisdom...

I have been doing this for 20 years and it always worked.

It worked last time...

My responses vary from ...

And where did you get your engineering degree from..

After 20 years - you should know better OR you have been doing it wrong for 20 years!!

 

[ajack1]

I also agree testing is the key. Nothing demonstrates this better than F1 racing for me.

They have top quality engineers who specialise in a specific discipline, they have cutting edge technology, workshops, materials, simulation software, analysists and testing facilities at their disposal and huge budgets, but still things do not perform as expected and it all comes down to time on the test track.

 

[dhengr]

Your method #1 isn’t engineering. It’s classic DIY’er. and shows a lack of being able to apply any engineering knowledge or experience. And, your method #2 doesn’t show much engineering knowledge or judgement either, since most experienced engineers on that type of problem wouldn’t bother with FEA on that problem, at least at first, and would probably only spend a few hours to develop his/her initial design. As mentioned above your time comparisons are way out of line also for the reasons mentioned. Unfortunately there are too many people and companies using method #1 and selling that as engineering to the public, and they are hiring people to do their engineering who don’t have enough real tech. or engineering ability or knowledge to do anything other than your method #1 in arriving at a solution. Engineering has become a bastardized term, the way many people and companies us the term today.

Most good design is done with a considerable amount of technical, materials and engineering knowledge, experience and judgement in the mix. There will likely be knowledge of a previous product or project to refer to for a starting point. There is also just a good general engineering education and background which gives you some direction as you start a project. The process generally isn’t a hit-n-miss endeavor. That doesn’t mean there is no place for testing, but that generally isn’t the starting point of most general engineering design and product development work.

 

[zdas04]

"Testing" is not "Blind trial and error". The formula 1 example above is illustrative. The people working on a multi-million dollar car are not going to "try a 1/2 inch shaft and see if it breaks" on a race car. They are going to apply engineering principles and balance material properties vs. weight and pick an answer within the matrix of acceptable results. Then they'll test a sound engineering design to see how the design fits with the system and performs with the thousands of other cogs in the machine. That is about as far from the OP's "option 1" as it is possible to get.

David Simpson, PE
MuleShoe Engineering

"Belief" is the acceptance of an hypotheses in the absence of data.
"Prejudice" is having an opinion not supported by the preponderance of the data.
"Knowledge" is only found through the accumulation and analysis of data.

 

[TheTick]

I have been fortunate to work in a couple places that value engineering (not just DIY hacksmanship). Engineering isn't just calculating, it's acquiring knowledge for future use. This means a combination of real data and calculations. An engineering model needs to be refined for the appropriate amount of precision and for valid assumptions and simplifications.

It can take some effort to refine one's "math model" to match reality, but once they sync up, great things are possible. A well-refined model saves many precious hours not trying things that are destined to fail. A well refined model also can be used to predict things beyond experience.

 

[SNORGY]

At least *try* to do a T/J = t/r = G@/L before destroying the first few shafts...