Mechanical engineer: how to work faster? 11

[EighthBen]

I am a mechanical engineer in a wood sawmill company, I am designing various machines. I use solidworks for my design. The problem is this: I am not satisfied by the speed I design things. Here are some aspects I analyzed and noticed:

1. I always try to make the model as "full" as possible - including all bolts, chains, and anything else. This is essential for generating BOMs correctly, and also avoiding design mistakes due to things that were "imagined" differently than they really are. Any decrease in model detail would result in more mistakes (previous engineers, like 3-4 people have perfectly showed how not designing something leads to the need of fixing the final product with angular grinder in the workshop, making much more mess than use). I don't think I can save anything here.
2. I use all the productivity tools: keyboard shortcuts, best-practice design tools and strategies, PDM, fast hardware computer, two monitors, 3d mouse, I know the software deeply (including that I have passed several SW certificates). I don't think I can increase anything the speed here.
3. I have a fair amount of experience in the things I design - I don't have to spend too much time on the phone consulting the suppliers, or the machinery guys/welders to make the design better, and so on. I mean, I don't get stuck often during the design. I don't think I can do anything faster on this.
4. Of course, there are times when I have to do the research: check the solutions of the competitors, consult the suppliers about the things I haven't used before, perform FEA simulation, etc. Also, I sometimes have to explain things to the welders/machinery guys/assembly guys - about what have I designed and what is the intent of those solutions. But - obviously - I will never know everything, and so these communications will always be needed. I don't think I can rapidly save any time here.

Besides, I think I can say that I make fairly little amount of mistakes. I check, and I check, and then I check things again. I must admit I lose some time here (absolutely not significant amount of time, but it adds up a little), but design mistakes would make much, much longer to fix during the manufacturing process compared to the model, so I wouldn't like to risk the increased amount of the mistakes for winning a few hours on design checks.

That being said, I feel stuck in the situation. Even a fairly simple machines takes 1-2 weeks or more to design, and when I look at the finished model - and I can't believe it took that much time. I have no idea how can I make it faster.

If anybody here, the more experienced engineers, could share their productivity strategies, that would be great. I am pretty sure I can not win anything on the software level (as I said I think I can say I know Solidworks fairly well), but maybe some global design strategy could be used to save a few percent of the overall design process.

Thank you for your advice

 

[Gumpmaster]

A graph of percent of design completion vs time would show an exponential curve with the amount of time to fully finish a design approaching an asymptote of infinity (obviously not quite there, but conceptually close). Completing the first 90 percent of the design takes 50 percent of the overall schedule. Completing the final 9 percent of the design takes the last 50 percent of the schedule. The trick to saving overall schedule is to know what level of completion is good enough. Do you really need to be 99 percent complete to start manufacturing or is 95 percent good enough? Note that you can virtually never have a fully optimized or complete design. This is evidenced by the fact that products that have been manufactured for years still have periodic refinements. There's always something you could have done better or more completely.

What do you and your boss think a good amount of time to design one of these machines would be? With your current design process what level of complete are you at that time? Is the machine able to be manufactured at that level of complete considering an acceptable amount of questions from the shop?

Another thought, are there other examples in the industry of similar items being designed in the time frame that you and your boss are hoping for? If yes, then either they begin manufacturing at a different level of completion than you, or they have a better design process then you. If there aren't any other examples, then maybe you and your boss have un-realistic expectations for yourself.

 

[3DDave]

Books of general interest:

The Mythical Man Month (Brooks)
Death March (Yourdon)

 

[gruntguru]

For me the concept design stage is usually a bunch of pencil sketches on scrap paper. Allows plenty of iterations without wasting time putting detail into 3D models.

je suis charlie

 

[drawoh]

gruntguru,

It depends on what you are designing. A mechanism or action can easily be shown as a pencil sketch. Most of my messy design problems come from having to stuff something into limited space. I start designing with whatever scale tool I have, be it drafting board, 2D[&nhsp;]CAD or 3D[ ]CAD. A preliminary 3D[ ]model can be whipped up very quickly, and then inspected and reviewed. Usually, it is worthwhile to model several design concepts, especially if fabrication is going to cost a lot more than engineering.

If you want to do DFMA or any other fancy engineering technique, you need to start off and investigate several design concepts.

--
JHG

 

[CWB1]

In my experience one of the most common wastes of time is simply unnecessary model iteration. I too usually start with a series of pencil sketches and hand calcs to identify areas of interest or concern, then I set up the necessary spreadsheets and simulations, and finally use the results to drive my solid model. Yes, engineering is an iterative process of design, test, repeat...but if you need to iterate your solid model 5x before test simply to make everything fit and work then you missed a step. At the risk of sounding negative, to me thats the difference between a true engineer and the proverbial "CAD jockey."

 

[tbuelna]

"In my experience one of the most common wastes of time is simply unnecessary model iteration. I too usually start with a series of pencil sketches and hand calcs to identify areas of interest or concern, then I set up the necessary spreadsheets and simulations, and finally use the results to drive my solid model. Yes, engineering is an iterative process of design, test, repeat...but if you need to iterate your solid model 5x before test simply to make everything fit and work then you missed a step."

I agree with your approach of using sketches/layouts/spreadsheets/simulations to optimize a new design prior to expending engineering resources constructing a high fidelity digital CAD database. Modern digital CAD databases use a complex system of relationships and links between files, and if the parent models are not constructed with care it can cause lots of problems and rework later on.

However, in my experience a process of "design, test, repeat..." is not normal practice. Testing costs money and takes time. Instead the design is validated by analysis, and then if required the analysis work is validated by qual testing. It's more like, "design, analyze, repeat until the analysis confirms the design meets requirements.....then if required conduct qualification testing to validate analysis work."

 

[GregLocock]

That's right, the tests verify the analysis. Check out the V model for product development.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

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

 

[drawoh]

justalearnin,

A lot of model iteration is communication with co-workers.

"That's not what I meant."

"Ooh, that's cool, but how about this?"

"Oh, is that really what it winds up looking like?"

"Wow, you can do that? That lets me do this..."

Many years ago, a manager asked me something, and I quickly free-hand sketched up a mechanism. It worked, and the manager announced that we should do all design like that. I was horrified. Usually, I am stuffing lenses, mirrors and beamsplitters into a tiny space. The available space usually is my primary design problem, followed by figuring out how people are going to get at stuff.

I am designing furniture here at home. Sometimes, I start with a free-hand sketch. Sometimes I go straight to LibreCAD. It all depends on what I do not understand yet.

--
JHG

 

[btrueblood]

Drawoh,

Sketches? CAD? Pah.

I needed to make a camera filter adapter, and walked out to my garage, and found a piece of acetal that looked about right. Started whittling on it with my lathe, messed up the threads twice, ended up with a too-short piece of acetal (thus verifying the cut-twice measure once principle), redid the whole thing in PVC, realized that I'd not be able to turn the i.d. without clamping on my just-completed threads (home mini lathe jaws too short) and had to come to use the big lathe at work. Turned the i.d. at work based on memory of what the diameter needed to be to slip over my scope, came home and checked, found was off by about .01", back to work, home again, there we go.

Design by messing around, I think it's known as "puttering".

 

[CWB1]

However, in my experience a process of "design, test, repeat..." is not normal practice. Testing costs money and takes time. Instead the design is validated by analysis, and then if required the analysis work is validated by qual testing. It's more like, "design, analyze, repeat until the analysis confirms the design meets requirements.....then if required conduct qualification testing to validate analysis work."

I believe for the first part we are disagreeing mainly on semantics as analysis is usually considered a significant part of the overall design process.

A lot of model iteration is communication with co-workers.

I would suggest that one of the biggest challenges in life is communication. Realistically though, if you are losing more than an hour or two a week (~2-5% of your time) early in the design phase then I would recommend you do what is necessary to address that issue. Both lean and agile philosophies have some useful tools from daily standup meetings to task boards not to mention all manner of project management software to assist with this. Nobody lives in a perfect world, but miscommunication shouldn't drive "a lot" of work on anybody's part.

 

[drawoh]

justalearnin,

I am not arguing that my co-workers are uncommunicative goofs. I am arguing that as the mechanical designer, I do not know everything. The requirements of electronics, optics (in my case), manufacturing and the end user must be taken into account. This means communication.

My book on DFMA (Boothroyed, Dewhurst, Knight) has a graphic that shows that engineering affects about 70% of the cost of the product. In reality, this value should vary wildly. In the case of the OP, it might be on the low side, given that it is one-off plant equipment.

How about we estimate the total cost of manufacturing all of the product, and budget some percentage of that for engineering? If the total manufacturing cost is high, the designers should take the extra time, do extra analysis, and try several alternate designs, and look for ways to be clever. If the manufacturing is relatively cheap, then engineering must work fast. It would be a good idea to select relatively expensive, reliable components to eliminate as much debugging as possible.

--
JHG

 

[KENAT]

Like drawowh, I spend a lot of my time playing geometry games trying to fit everything into the available space, then allow for realistic manufacturing tolerances (not that many of my colleagues both with that part), usability, manufacturing ... so I do tend to jump to CAD pretty quickly when I know what I'm going to do. However, I also have plenty of sketches around too from when I get to to a bit more brainstormning.

The guys I admire, and really should try and emulate more, are the ones that manage to refine the basic problem to some mathematical relationship(s) and crunch through that. Now like I said a lot of my problems are softer geometry related issues but still should do better.

Something to look out for is applying the wrong/inappropriate model of project management or design process or whatever as I alluded to early.

Some philosophies that may make sense for say software when the '0' & '1' can be re juggled relatively quickly with almost immediate results may not make sense for hardware projects where you need to design, document, produce tooling for the part then make the part....

Philosophies like fail fast, fail option may perhaps make sense used correctly. However, when failing for entirely predictable reasons like tolerance stacks or similar the I'm sorry but it's no longer engineering by any measure - more the puttering Btrue mentioned.

Posting guidelines faq731-376

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

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