This can also be resolved by directly measuring the 3D shrink of any part, so for repeat production that shrink factor, in 3D, by feature and direction, can be accurately known and accounted for. It is likely Not what the material provider estimates it to be. We can help, we do this all the time.
I'm frequently amused by how often the solution to a product nonconformance is simply to revise the drawing to match the parts. It looks like, in this case, there were two problems with the design: first, shrink factors weren't taken into account properly; second, tolerances were much tighter than needed. Fortunately, these two problems cancelled each other out. If the clearances hadn't been so big compared to the tolerances on the part, this could have been a much more difficult problem to solve.
If you want to avoid the costs involved with drawing changes, it's important to take the realities of the manufacturing process into account early on. Tolerances should be reasonable, based on both what the process can achieve, and -- much more importantly! -- what the design actually needs. Specifying tolerances which are unnecessarily tight is a great way to increase cost, but for most engineers, that's not exactly a key design goal.
Of course, by widening the tolerances later on, you can make yourself look like a hero by finding a "cost reduction"!
Good comments by Bill and Dave. Isn't this what Engineering is about: making something new that works the first time. I'm reminded of this every time I watch a show about a new airplane and see the footage of its first takeoff. What happens to the crew if it doesn't lift off? In defense of the process engineers, they were young and relativly inexperienced at the time. This was probably a good lesson for them, and they hopefully took the time and effor to learn from it. I've also seen a few instances over the years where designer/draftsmen have created subassemblies for expensive equipment that were terribly deficient for one reason or another. I've concluded that men qualified in general to do this work are often inexperienced in the new industry that just hired them. I think this is the reason why ISO 9000 and other quality specifications require that Design Reviews be performed on new products.
In 1988 Motorola won the first ever Malcolm Baldridge National Quality Award for its launch and execution of Six Sigma manufacturing practices. Of course, in 1988 very few knew what Six Sigma actually meant. One thing all the design engineers knew for certain was that all the products were going to get huge, because all the tolerances calculated required huge ranges in order to meet the calculated 3 parts per million mandated by Six Sigma. One of the job assignments I got was to analyze hundreds of dimensional instances out-of-spec and determine how they could be salvaged in accordance with Six Sigma.Like the story told above by Bob Salter, Six Sigma compliance doesn't necessarily mean changing anything; in a majority of instances, it was achieved simply by opening the tolerances on the print to account for the wide range of process variation.I could make three choices for dimensions out of spec: (1) correct the tool or process to match the print, (2) revise the print to match the processed parts, or (3) open the tolerance on the print to allow for the variation.Since the first option was the most expensive (tool revision) and the second option meant compromising your design, guess which option was selected most often-?Simply opening a +-.005" tolerance to a +-.030" allowance, and plugging those new numbers into the statistical equations resulted in the CpK values of 2.0 or greater equating to Six Sigma design.If only Malcolm Baldridge knew the real specmanship behind that perceived quality !
> Of course, in 1988 very few knew what Six Sigma actually meant.
One reason for the dominance of Japanese manufacturing from the 60s is that Statistical Analysis was a given at ALL levels. The average Japanese shop floor worker could not only do the maths but, more importantly, understood the significance of the numbers to HIS output and quality.
I was privileged to take a small UK company to ISO9001 in less than 12 mths; improving quality, productivity and morale in the process. Probably my most satisfying achievement.
We later did an OEM project for Yamaha who told us that this was the first time they had encountered a Western company for which AQLs etc were more than just numbers and who used those numbers to improve quality rather than to apportion blame after the fact.
It's not the numbers that are important but their significance. EVERYONE has to understand them, including the WHY. Design Engineers are often as bad as Management, who are the worst of all in their ignorance.
Since the clearance was in the order of inches it would make little sense to hold the tolerance very tight on that surface. One real part of engineering in cost reductions is reducing the tolerances to what is required for the application, rather than holding the tolerance to the best that the best process can produce. Of course it would provide nice bragging rights to be able to hold the dimension to +/- 0.0002", or better, but when the closest surface was 6 inches away it does not make sense. Holding to tight toleranes where there is no need just adds to the part cost and reduces the production yield, aside from having the added cost of much tighter inspection gaging. Understanding which clearances are critical and which ones are not is the thing that makes some engineers seem "luckier" than others. Years ago I got price quotes for a fixture plate that were much higher than I had anticipated, and it was unclear why until it became clear that the detailer had demanded very tight olerances on all of the dimensions, completely missing thge design intent. I went through the drawing and set all of the tolerances correctly and got a requote, which came back as less than a quarter of the original quoted price. When the fixture plate arrived it went togather with all of the other parts and worked as designed. Everybody was happy.
William K.'s comments on specifying tight tolerances where required and allowing loose tolerances where they are acceptable are very correct, especially when a person wants his company to be profitable so he can keep his job. On the other hand, I was thrown a curve ball by the manufacturing engineers on another jet engine seal. This was a rotating part made from 17-4 PH material, and their problem was with some axial dimensions. The overall width of the part was toleranced minus .005 inches as this was necessary for the stackup of several parts mounted next to each other on the main shaft. The other axial dimension toleranced the same amount was the location of the sealing part of the rotating seal from its abutment face. This was to insure that the rotating and stationary parts of the sump seal alligned properly: makes sense, right? Not dimensioned or toleranced, because it was not important, was the axial length from the sealing part to the other end face. Its resultant value and resultant tolrance can easily be calculated from the two existing dimensions. The resultant tolerance would be plus zero, minus .010 In.; however, the manufacturing process engineer advised that he had to hold this dimension to a .001 In. tolerance in order to achieve the minus .005 In. tolerance on the location of the sealing part. Unfortuneately, I moved on to another position before this conflict was resolved. This does show that design engineers and manufacturing engineers see parts in different ways, although, I suspect that they have to work with the machine tools that are abvailable, and this could have been the root of the conflict.
It appears that the specific dimension was not critical to the product, but vital to the process of machining the product. Those conditions do exist on occasion. So it probably would have been valuable to provide both dimension and tolerance information on that dimension as well.
Living with production capabilities is one of the challenges that often challenges product design engineers, as the article winds up implying. Understanding the manufacturing capabilities is a big part of "DFMA", designing for manufacturing and assembly. That comes close to home for me, since very early in my mechanical designing career I did design an assembly that could not be built. All of the parts would fit and work with correct clearances, but the thing could not be assembled. So changing the shape of one machined cutout by almost an inch allowed for assembly. The good news was that it was a one-off design and so the extra time for the rework was not a big deal. But it was very educational.
@ricardo;Sounds like you had a great experience in your UK assignment, being both professionally gratifying and educational. Your viewpoint that all persons in the program need to understand -- from assembly technicians to top management –are spot-on. I agree with your comment, "EVERYONE has to understand them, including the WHY. " ... as a model that everyone should pursue.And while more and more companies today are endeavoring to follow that path, I do grimace in the reflection of the white-washing we offered in 1988. You're exactly right. It was about the perception vs. the actual.Truly a facade of real value.
What should be the perception of a product’s real-world performance with regard to the published spec sheet? While it is easy to assume that the product will operate according to spec, what variables should be considered, and is that a designer obligation or a customer responsibility? Or both?
Biomimicry has already found its way into the development of robots and new materials, with researchers studying animals and nature to come up with new innovations. Now thanks to researchers in Boston, biomimicry could even inform the future of electrical networks for next-generation displays.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.