How to identify over designed components based on FEA or test data?

[kdaenerys]

I am trying to create an algorithm to identify bottleneck components or over designed components based on the FEA data.
These components from a larger system are already in field. While trying to create a newer version of these or for other purposes like weight/cost reduction, what methodology/approach would you suggest?
I am currently deciding based on Six Sigma Max stress in FEA data Vs Yield Strength of component and few other parameters like fatigue life.

 

[JStephen]

Is this in relation to anvils or rocket engines or buildings or pump jacks or fence posts or what? Some kind of context would help here.
Standard size of components based on availability and compliance with design codes will cause a lot items to be overdesigned if evaluated by FEA data, but that's not any motivation to go changing the design, either.

 

[rb1957]

you're trying to redesign something that exists because you want to … (fill in the dots)

are you trying to reduce the price (or weight) or the ongoing maintenance costs ? 'cause these have different approaches.
The former (reducing a design value) would say "what items have the most of this property?" (eg what components cost the most) as a way to identify the biggest hitters.
The latter would ask "what breaks down the most? is it expensive to fix, or merely a nuisance?"

another day in paradise, or is paradise one day closer ?

 

[kdaenerys]

I am considering heavy duty vehicle structures in scope. e.g. axles/frames/functional components
I want to propose a method to identify over designed components because many of the vehicles I work on have very traditional designs (50 years old or even older). If a design engineer want to start working on a new program the next day, he should where he needs to start.
There are all the parameters to be considered like weight, material cost, manufacturing cost, joining feasibility, new application feasibility etc.

I am trying to redesign something that exists because I want to improve the design in early phase of a program before looking at various program targets like weight, cost etc.

 

[rb1957]

I guess I don't like your use of "over designed" to mean components that are perceived to be heavier than required.

Yes, a designer starting out should look at existing designs but then there are a host of questions …
How well has the component performed in service ?
What design restrictions lead the original designer to this solution ?
What new methods (in design and manufacturing) could ease these ?
What have we learnt since the original design ? maybe the original design guide was overly conservative ?
What new design constraints apply now ? (maybe we want a longer service life ? maybe there are new design certification requirements)
How completely does the design specification detail ALL of the reale world requirements ?

All this should be in any basic text on designing things.

another day in paradise, or is paradise one day closer ?

 

[DrZoidberWoop]

I'm not sure about your FEA experience or program availability, but topology optimization has become very advanced. That might be a route you want to investigate if you're going to be investigating technical details. But I don't know the scope/direction you want to take. Maybe you're looking to take a management route to achieve your goals, rather than research.

Here's a link to what Abaqus can do these days:

https://www.3ds.com/fileadmin/PRODU...MULIA-Abaqus-Topology-Optimization-Module.pdf

 

[Tmoose]

"many of the vehicles I work on have very traditional designs (50 years old or even older). "

So you are working on 1970s vintage equipment still in the service, and with spare parts currently being produced in India ?.

===============.

"These components from a larger system are already in field."

So, do you need seamless backwards compatiblity to fit directly into the "large system"?

================.
It appears there a suppliers of new modern heavy equipment in India, so I'm guessing sales opportunities of new equipment of old design might be rather limited, and becoming more so.

I'd start by quizzing the sales department about their top sellers, regarding price and quantity sold annually, keeping an eye on the prospects of the market for repair parts drying up as the economy moves toward the purchase of new modern equipment.

If a component is never replaced, then improving the design will serve no purpose.
If the service life is considered adequate, and you are supplying the spare parts regularly, I'm thinking your company is not looking to reduce sales, but might appreciate a higher profit margin or to gaining a price or performance advantage over your competition.

If you are having the parts made by suppliers then their manufacturing capabilities may or may not result in economical production of a new design.

 

[kdaenerys]

As DrZoidberWoop said, I am looking for a management route where the method deployed in excel or other software scans through FEA data for 20000+ components and tells top 20 or so components more suitable for optimization. Consider this as identifying the bottleneck in the system in terms of apt design. Topology optimization and such softwares or design changes would come after we have identified the components.

 

[rb1957]

are you (and this is not meant as a slight) in management ?

IMO, you're not looking for a "bottleneck" but rather trying to prioritise components for improvement (maybe it's an ESL thing ?).

I would rank the 20,000 components by the parameter you are trying to optimise, by weight or cost or ?, as the heaviest and costliest components offer the most room to improvement. Then you may have 10 components to look into in detail.

another day in paradise, or is paradise one day closer ?