Techniques in developing products

Hi all,


I have quietly, sometimes not so quietly, monitoredthis websiteover the past number of years for interesting nuggets of information. I wanted to get some opinions regarding the techniques some users are suggesting to others to solve designissues.I have read posts suggesting techniques and features like toridial bends, tweak replace, mirroring geometry and copied and translated geometryas "first choice" features and techniquesto design products.


I have always considered these types of techniques as cheaters to be used when all other, more robust and finitely controllabletechniques don't quite make it. Most of us develop engineered products. In other words, we need to document the critical features of our designs. A living hinge for example is an engineered feature with definite radii / material / range of motion relationships. A toridial bend is not the technique for this design issue. But, yet I see it suggested as the way to get it done. Sure it's fast and the boss thinks you rock. But do you really? Have you sweat the details to ensure the product will work?


I have used Pro for fifteen years. I have not painted myself into a corner or had to throw a model away and start over for almost that long. This is because I learned the proper techniques and I do not cheat when it comes to building a model. I use robust and efficient techniques.I take into consideration I will not be the last person to use the file. I take into consideration that my customers will live with that model or assembly for years to come. I take into consideration when an issue occurs I can quicky resolve the model without having to undo the bandaid technique to get to the root cause. I never have to tell a customer that I have to rebuild the model to fix what should have been done correctly in the first place.


I constantly see users who "blaze throughit" and make a pretty model.I review models that have little or no finesse in its construction. In my opinion this is notDesign Engineering in as much as it is CAD pumping. Basically pretty models with little or no value outside of aestectics. I think this type of data creation is a detriment to the future of our design jobs. I believe when cheating techniques become the norm we all will suffer. I would like to know your opinions.


thanks.

Tell me about it . I don't know how many times I've had a model blow up on me because a previous user's dimensioning and feature creation sheme resembled a house of cards. I would suggest a training program on robust modeling techniques but, as we've seen in some past topics, who can agree on the right structure? I've heard they're adding a new functionality in WF 3.0 that allows for freezing failed features but I guess this is another example of cheating.

I agree with you guys 100%.


Way I see it, in order to achive this. You need two thing.


Number one, you need a good working standard. This might vary from user to user but I would think we pretty much know what they are.


Second is a support from upper management.
I've ran into a few cases where models were done pretty badly and I pointed that out but management gave an'Okay' to release.
Reasons were 'We are already late with the project.', 'We will come back and fix it.', 'This is a one time deal, we won't use that design again.' ...etc


My point is without managements back up, 'Cheaters' will win.
Because, to the management's view, 'The drawings are done' (even though it may be in bad shape)


(perhaps this should been in Rant & Rave)
Charles

dlongmi, I think you oversimplify the case. The techniques that you mention may or may not be bandaids/cheats depending on how used and where and when.


I agree wholeheartedly on using proper modeling techniques. I can see a real distinction between the robustness of my models and that of others, and, as you say, it is because of good modeling practices.


However, I have found it is just like making a physical part: how you do it is determined by the nature of part, it use, it's life expectancy, what the customer wants (which is usually indicated by what they paid), etc.


For instance, I have established EDM assemblies and drawings to model our electrodes in. These are the foundation that we build on for the electrodes for every mold built. I don't use replace surfaces on them (partly because there's not really any need to). But when I am actually designing the electrodes themselves for a specific job, time is critical, and that has an impact on how I design it. Idon't usereplace surfaces as a first choice becuase it is to easy to fall into. I seen guys use a hundred replace surface features and I (re)did it with 8 regular surface features. But sometimes, replace surface is the way to go. If I spend 6 hours to do a 1 hour job just becuase I refuse to use replace surfaces, who am I benefitting? You better blaze through, it's a one-off one-time design. Get it done and get to the next one.


Replace surfaces can be redefined and rerouted succesfully.


Copy and transform features are very useful and robust, even though you can't redifine them you can usually reroute them.

Robust use of the Pro/E software? Where is it written? Where are the guidelines? Is it better to create half a part and mirror the other half? Are relationships/parameters a pain to most users? Do we use Pro/E's "default" sketching plane which arbitrarily chooses a datum/surface as a sketching reference? Do we finish off sketches by making sure all constraints/dimensions are strong? Should assemblies have all components on a layer? The questions are endless.


You will always have a difference of opinions amongst designers. You will always have varying degrees of competency between designers, each with their own direction and actions to solve the same problem. 20 years from now no one will care except the poor soul who gets my models

Donha,


There are plenty of resources to reference (books, people, user groups, conferencesetc...) that will give a basic roadmap of best practices. They are out there for the good designers and engineers that strive to learn everything about their trade tools.


Are relations and parameters a pain? NO. If you know what you are doing.


Pro/E doesn't forceYOU to do anything. Unless you don't know what you are doing.


Robustness of Pro/E files is attainable. If you know what you are doing.


Sketches should be contrained. We use Pro/E not SDRC. Remember your mathematics...there isa solution...if not, then it's indeterminant.Sketches should be considered in the same way. Leaving things to chance in engineering is foolish.


Layering is one of the oldest and most fundamental best practices in all CAD modeling, hands down.


Do you name your features? Do you give yourself (and others) your model's intent map? Do you use start parts and assemblies? Do you use the tool as an engineering software or a drafting tool?


I didn't say I was answering the endless questions. I questioned the means with which users are attaining the answers. I now know how Einstein and Tesla felt. They saw the answers...simply and plainly...without excess baggage. Efficiency at is most pristine. Many posts I read here are the same old brute force methodology issuesover and over. Pro/E is an elegant piece of software like a instrument...nottheaxe manypeople use it as.


What is sad is your statement "20 years from now no one will care except the poor soul who gets my models" is the exact problem Ielude to.Complacent designers/modelers that do notcare. I'm sure glad you said it. You so elegantlyprovided the evidence for mycase...thanks.



Edited by: dlongmi

Rcamp,


I don't think so. Over simplifying is saying something like...every model's first feature has to be a protrusion...if not then the model will fail. You have described a thought out methodology to attain your solution...to your credit. You didn't just give up when faced with an issue and copy an old part off each time and start over.Many people do this. I absolutely know a model's construction technique is best implimented when the intent, use, budget and longevity are considered. However, it seems many people do not do this. Many people "use up" their arsenal with the first attempts and then resort to the "cheats"...covering up a cut with a protrusion, lopping off the right half of a part and building something else in its place...silly thing like that. There is more and moreevidence each day that we are losing the craftsmanship and care that used to be design skill.


But as all mathematical laws are general in nature to encompass the nature of math itself...Pro has the same characteristics. Many users never delve deep enough to understand the laws...they skim the surface and touch only some of theapplications.


I certainly enjoy the exchange of opinions.


Cheers.



Edited by: dlongmi

<DIV>I tend to agree with dlongmi.Here are few things that bug me:</DIV>
<DIV></DIV>
<DIV>1) Unnecessary use of surfaces (say used to create protrusions, when a single protrusion will do) high in a tree and nigh impossible to modify without completely re-defining/re-routingor even re-modelling the part.</DIV>
<DIV>2) Overcomplicated sketches, particularly those with sketched rounds & chamfers!</DIV>
<DIV>3)The inability of some people touse Insert Mode, and so you get buried features and worse those tiny protrusions to "correct" geometry(or similar cuts)which always fall down.</DIV>
<DIV>4) And a variation on (3) rounds and tweak featuresscattered throughout the tree and referenced madly - a veritable maze of references(If you need to export to ANSYS they need all the rounds and small features off...which, with a badly modelled part, can create hours of work).</DIV>
<DIV></DIV>
<DIV>It is not that"tweak" features and mirrored geometrydo not have a place, they do. It just some users go out of their way use 'em unnecessarily.</DIV>
<DIV></DIV>
<DIV>(At the moment I'm updatingsomeone's casting , which has protrusionsburyingrounds, protrusions, cuts and drafts. There is also lovely draft on top of a draft)</DIV>
<DIV></DIV>
<DIV>That's my twopenneth!</DIV>
<DIV></DIV>
<DIV></DIV>

whilst i agree with dlohgmi i have my own opinion such as rcamp; that is it depends on th efinal use of the data





i work in consultancy and teh dliverables/data received varies greatly; ie:


i may create a surface model/block modelquickly for use to visual to sell and idea. This may have no relation to the final cad assembly just used to gain a contract/sell an idea.


working with far east maunfacturers, i fingd that sometimes no mattter how well i design a product for scalability/robustness (skel models, etc) the vendor will choose to build right on top of my model/re-create the data in a differing (cheaper) CAD system.


workin with blue chip companies mostly on surface reference data, which is generally used by engineers who detail the internals - to keep the design intent/industrial design this form should have been arrived at through negotiationbetween mechanical constraints and style, if a ID is bought into buy sa customer - itshould not change through development without good reason.


the other are of work i get involved with is please fix this......................


this is usually some companyup against the wall to a deadline and the engineerscannot/will not deliver the solution. Usually these models are a nightmare cut and shut; ithas to be said some of these models arecreated by so called proffesionalswhogoon numerous training courses, but fail to grab thebasics either through lack of use (how many hours in a week do you actually spend in front of the CAD screen?)or plain being ignorant, lazy or stupid.


lets face it these situations happen, in consultancy we have our own `bodge it' merchants, i myseldf may have been guilty at one point or the other - im sure in big projects (automotive etc) modelling is critical. But if i have a simple moulding to send to cheeeky boy