Drawing changes during R&D phase - too many changes and pushing drawings out too early

[ves39]

How do you guys handle assigning drawing numbers and tracking/pushing changes during R&D and prototyping of a new product or part?
How and when are part numbers assigned?
How are parts released when a product or part is actively being designed and changes can be frequent and quite drastic?

I'll use an example I encountered on a project just recently. We're designing an overhead panel that has switches, comm equipment, plumbing, etc. and various components and brackets located on the inside. We have a prototype completed that is 90% of the way there, but there are still a few final components being designed and finalized.

One of these parts was a bracket to hold some hose clamps for the plumbing in the panel. The first drawing of the part got sent out to sheet metal and they started cutting material to make the bracket.

Later in the day, a change was made to give the bracket more adjustability. The holes for mounting the clamps were changed to slots, so they could be moved up and down and accommodate different clamp sizes. The material size wasn't affected, so the sheet metal worker continued on with manufacturing the part.

At the end of the day, a final change was made. The plumbing had a protective conduit added to it, which changed the clamp to an even larger size. The bracket now had to be made wider to accommodate two hoses and the larger clamps. A third drawing was given to sheet metal, but now they had to cut new material because they wouldn't be able to make the part from what they had prepared.

So now there were three different drawings floating around all with the same drawing number. They didn't indicate the version of the part or drawing. We don't do that because changes are so common and frequent. Changes and drawings often get deleted as the design progresses, so we don't really track the evolution in the design of a part and there are times where it's difficult to go back to an earlier version if we end up going down the wrong road.

All of this obviously leads to a lot of wasted time, money, and materials.

I think the bigger issue here is our design and manufacturing process. I think we need to revamp our design process and be more thorough with checking that all requirements are met. We're very trigger happy about sending drawings out early. Many changes can be made in the hours that follow and some of the work may end up getting completely scrapped because of a change in the part. I think we need to take a look in the mirror as an organization.

Anyway, I was looking for some advice on where I could begin to start tackling this problem and maybe getting some advice on how manufacturing and changes are handled during the R&D phase of a project in a well organized company with systems in place.

 

[CWB1]

Part numbers are typically assigned as new CAD files are created. Regardless of whether you use PLM, a spreadsheet, or other means of tracking parts, every part should have a release status to signify where it is in the lifecycle. At a minimum I'd expect four options - new/concept, released for prototype, released for production, and cancelled. Each status generally has different review/approval requirements and different audiences for the print. When a part is released to production its generally assigned the next sequential revision level, when sitting at some other status each iterative "baseline" is tracked by date with no revision assigned. Also, please be aware that every revision is assumed backwards-compatible with prior revs.

 

[ves39]

So is a part number assigned right at conception of the part during development? What if the requirements of the part change, or it's part of a subassembly and the higher or lower level component changes affect its design and form or functionality?

Our engineering department is tackling some new projects that are being developed from the ground up. The legacy way of doing things just isn't working and is leading to a lot of problems with manufacturing and getting projects completed in a timely and cost effective manner.

 

[IRstuff]

It seems to me, based on your description, that no one is in charge. The manager of the project should have the chutzpah to say, "Pencils down!" and fire anyone who doesn't listen.

Do you have no design review process? The manager should be adhering to some semblance of a design process; "December 15 is the drop-dead date for design changes for this iteration, next iteration is Feb 15," or something like that. Every design change should have at least a Design Decision Record, or ECO, that spells our what the change is for, and why it's important, and ought to be signed off by the relevant managers, particularly at the stage of the development as you described

I've seen organizations that were so undisciplined that they would make design changes, find out they didn't work, undo them, and the come back days later and implement the same design change they already rejected because they didn't keep track of anything.

Or worse yet, a company that had developed a state of the art semiconductor process, released to production, and then some rando process engineer decided to change the process to make his life easier without doing any sort of review, resulting in a massive yield loss that wasn't even detected until years later, when a product that was specifically sensitive to that part of the process was being fabricated. What was truly annoying is that the process change was known to be deleterious 10 years earlier and was well documented in the literature.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[CWB1]

Yes. Part numbers are assigned as you create the model/print files, so essentially before the first line is drawn. If the part is unreleased new/concept status then a bit of common-sense needs to be applied regarding whether/not to pull another p/n for design changes. If you need a different type of part then pull another new p/n (lever vs shim, lightbulb vs switch, etc). If you need a different size/material/etc of the existing part and are unlikely to end up with two versions in production, just use the existing p/n.

 

[truckandbus]

R and D parts are tricky in that the usual rules for revision control for form fit function are not applied. This means that you need to be explicit in how to handle previous revision parts - remove and replace from existing builds, rework part or use new going forward, scrap existing and use new goingforward.
For "I wonder if this will work" exercises, use sketch part-numbers/marked prints to get one off parts made for proof of concept.
If developmental parts are being put into the standard process flows for supply chain and quality need to be able to clearly identify 'not for production' to avoid the three quotes, 16 week lead time with tooling and paper submissions.
Pump the brakes with proper process for drawing and design reviews

 

[Jaycen]

ves39,

As a process-oriented guy, you're correct that the biggest part of the problem is the lack of organization at the beginning of the project. Just like with software or even building a report from a database - the less you know about what end result you're trying to accomplish, the more iterations and wasted time and effort you must run through in order to get to the finished product. The folks in the mix of all this are constantly harried, frustrated, and they feel the internal friction of the constant push-pull of these changes and lack of method and control.

Project kick-off meetings are crucial. Identifying as many of the deliverables (product specifications) up-front will save you a ton of time on the back end. A loose development structure is needed, which you can refine as you try out the new method. This should include the initial kick-off with as many critical stake-holders as possible, such as the end-user and/or the guy who has to install it. Having a physical meeting (or at least a Zoom call) allows you to quickly brainstorm or iterate through questions that can be quickly answered without trying to write a book in an e-mail full of irrelevant questions and possibilities that nobody will bother to read.

You should then have a set of project milestones where the same group (or most of them) check-in from time-to-time in order to catch problems and changes. I always do some kind of "project scope" document that lives in the folder with drawings, etc, so that anyone can peruse it to see "What were we thinking when we started this? What was the goal? Did we foresee any major issues? Were their issues we should have seen but missed, because of a gap in our process?"

I try to keep things high-level with bullet points and sub-bullet points where detail increases as you dive in. This way, the creatives can peruse the info without getting mired in details if they don't want to, but the details are there for those who want/need them.

These are just some basic organizational structures to get you started. Your local realities will dictate different directions you'll want to take as you see what does and doesn't work for your company, but with each project, you should be learning things which allow you to shore up the system and push more of the details toward the start of the project, so that fewer gyrations are required at the end.

Good luck!

Obstacles cannot crush me; every obstacle yields to Stern Resolve.