Lean Product Development - A Technology Approach

One issue I've come to appreciate over the years is the difficultly of transferring design information to manufacturing. This difficult task has been further complicated by the introduction of a disruptive technology - 3D computer modeling.

After leaving the relatively defined structure of large corporations, I was involved in setting up an Engineering department at a design consultancy. The lack of complete internal drawing standards was aggravated by the impression of the (mostly young) engineers that the 3D model described the part they wanted built. At that time most of the smaller prototype shops were still using paper drawings with limited CAM.

To address this problem, I developed a white paper "Minimum to Manufacture Documentation" which can be found on my web site. It was used to ensure our prototype shops knew what they were getting and our engineers knew what was expected of them.

My next revelation came while working at a CNC contract machine shop. We would receive design documentation in a variety of forms from 3D models to paper drawings. While a paper drawing usually contained the information required to produce the part, it often required us to build a 3D model to efficiently fixture and process the part. In addition, larger companies often used internal specifications whose meaning were not spelled out. By itself, the 3D model was more efficient but usually did not give us enough information to adequately design the fixtures and process.

I updated the MDM document and also developed a presentation on "Machining Your Part" to discuss some of these issues. The presentation is also on my web site and currently focuses mostly on CNC machining, but I hope to expand it to other processes.

The last piece of this story is my recent attendance at a seminar on Lean Product Development. At the time I was also checking files and drawings on a drive train project. The goal in documenting the dimensions and tolerances is to ensure that the manufacturer has enough information to use the optimum process to produce the part.

It occurred to me that in most instances the desired process had been developed by manufacturing and engineering through years of experience. In fact I was often working backwards, taking what I knew about the desired process and ensuring the dimensions and tolerances were appropriate for it. Looking at it this way the effort to determine the correct dimension and tolerance scheme is not value added, since the real goal is to use a process that gives a quality part.

Let's take the example of a shaft that is press fit into a bearing at each end. In this case I needed to look up the tolerance of the bearings from the vendor. Then I needed to look up the recommended fit of the shaft from the standards. I also needed to determine the geometric tolerance of the bearing surface and shoulder on the shaft from a similar part. When manufacturing receives the files they will look at the drawing and then compare it to a similar part they are making to determine what process is capable of making the part within tolerances.

What we really wanted is a power transmission shaft that would perform adequately and was cost effective to produce. We already knew how to do that but had disseminated the information across several documents in multiple departments.

Why couldn't the designer just select a library feature for a shaft that fits the specified bearing from a CAD pallet and apply that to the part? Embedded in that feature would be the manufacturing process(es) that were known to produce it. The CAM program could recognize the feature and the required process and generate the tool path.

To implement this would require cooperation between engineering and manufacturing - always a good thing. The technology is already there in the major CAD/CAM systems. It would be better if this was developed as a standard so that it could be implemented by any software.

Your thoughts? Reply to this post or email me: david at dmajic dot com

CNC Machining - Tolerances

How much does a Zero cost?

I have recently given several presentations on how a part is processed in a CNC shop. The talk discussed the decision making process for selecting the right machining process. It also discussed some rules of thumb and showed how to make a design more machinable.

The one issue I struggle with presenting is tolerance. Most engineers understand that a .001" (.025mm) tolerance is going to be more costly than a .01" (.25mm) tolerance. Sometimes,a small change in the tolerance can have a large affect on the part's cost. This is because of process tolerance; that is, each way of doing something will have an intrinsic tolerance.

For example, a total tolerance of .005" (.13mm) is easily obtained by drilling. However, once the tolerance decreases to .002" (.050mm) reaming is required. The time to produce that hole is effectively doubled. If you estimate a $1/minute machining cost this can add $.25 per hole.

It seems that there is no standard way that companies and engineers tolerance their drawings. This is inexcusable because ANSI B54.1 completely describes how this should be accomplished. There are plenty of resources on the Web that reference parts of ANSI B54.1, but for less than $100 you can purchase Machinery's Handbook. I purchased mine on CD-Rom and have the files on my computer, so that it's handy whenever I'm developing a new part.

Later I plan on trying to tie specific tolerance classes to manufacturing cost.