One business practice we get asked about frequently by our customers is what to do about part numbering and how PLM systems work with various part numbering schemes.
I won’t hide the fact that I am very biased about part numbering methodology. There are two primary forms of part numbers – “intelligent” and “sequential”.
Intelligent numbers have meaningful significance within the number – and in that they serve as a quick means for visual recognition of part type, product class, supplier, operation process, production phase, or other similar breakdowns. There is often some “sequential” component within the part number. The intelligent part number is a favorite in the small -mid sized businesses that have had to operate predominantly manual engineering practices and have a mid-ranged ERP system that is not particularly restrictive about part numbering. So the part number itself acts as a “database” of the part information. The problem that frequently arises is that a part may change “classification” but not really change form fit or function – such as switching from manufactured to procured so the company faces a process decision – change the part number just to keep the data “correct” or leave the number as is and live with the incorrect meaning behind the number.
Many years ago Hewlett Packard conducted a study on the role of intelligent part numbers in maturing businesses. The conclusion of the study was that as a business matures and grows – the viability of maintaining intelligent part numbers as a business practice became increasingly costly and in the end the companies with sequential numbers were able to more easily adapt and evolve as business entities. Businesses that have grown by merger and acquisition find the intelligent part numbers mire the business down as each entity brings a different “decoder” to the numbering scheme – making things more and more complex.
The benefit PLM brings is that part numbers are a unique identifier – the availability and access to all of the classifications that were being manipulated within the part number are actually captured as their own data fields. Changes to those data sets that are not changing form fit or function are not forced to remain faulty nor does a new part number have to be generated. Thus the sequential number is easy to leverage. Most PLM vendors will allow the enablement of using intelligent part numbers (in part because the customers demand it) but they will all acknowledge that intelligent part numbers are not a recommended business practice. In fact to implement automated generation of “intelligent” part number schemes in PLM systems often requires custom rule sets or even custom SDK code to deploy within some PLM systems.
The challenge exists for companies “converting” from intelligent numbering systems to sequential numbering systems. Some companies have hard coded integrations that parse the numbers for other systems – and that means to go to a sequential system some integrations may have to be reworked. But for many companies – it is not a requirement to renumber legacy parts – but rather to pick a time and move sequential from that time forward. For cultural acceptance it is often easiest to mark that “time” as being when the company deploys PLM for the first time – setting a best business practice in place along with the tools rather than forcing the tool to perpetuate a bad practice.
Copyright 2007, LR Hirr, All Rights Reserved
i am interested in the hewlett packard study, but have been unable to find any other reference to it… does a link exist to find it online?
thanks
paul
Comment by paul — January 24, 2008 @ 6:06 pm |
Hello Paul.
I just discovered your question about the HP numbering system study. As it turns out, I was part of the task force that conducted the study mentioned in the post you saw. The study was done in 1971-72 and I tossed my copy of the final results during a move about 10 years ago. Now I wish I had kept it because I am finding more and more resistance to the idea of avoiding the use of intelligent numbering systems.
Back then my role was that of a systems analyst designing the Bill of Material processing systems HP was in need of implementing as they were replacing their manual procesures with computer-based systems. The “intelligent” part numbering systems were soon to run out of part numbers, so we had to figure out what to do. As the posted message notes, we concluded that the only rational solution was to remove the intelligence from the numbering system and use computer based indexing keyed to the descriptive data associated with each part. Very soon after we completed our study, I was promoted to a new position that sent me overseas so I never had to deal with the excruciating pain HP went through to convince all the old-timers to give up their beloved intelligent numbering. I am no longer working for HP, but it would not surprise me to find out that they are still trying to eliminate the vestiges of intelligence in their numbering system.
Right now I am trying to convince the genealogical research world to divest themselves of the goofy genealogic “intelligence” in the myriad of numbering systems that have evolved over many decades. These systems were helpful back when computers were not common place for managing genealogy databases. But now that we have had nearly 20 years of personal computing technology applied to database management, it is high time to abandon the numbering systems that are useless now days. Because there are so many “old timers” involved in genealogy research, I have discovered that what I escaped at HP back in 1972 has finally caught up with me with a vengeance.
Richard Rands
HP Veteran
Mountain View, CA
Comment by Richard Rands — September 16, 2008 @ 2:48 pm |
Richard -
It’s funny how “the more things change the more they stay the same”. I was wondering if you could recall if the study had been written up as an HP Tech Brief or was it specific to the division you were at or was it at HP Labs? I too have long misplaced the report yet still have to fight the fight on a regular basis. I’d be up for approaching HP to see if it could be dredged up but need to know where to start.
I recently had someone point out that sequential numbers have embedded intelligence as well (that could be construed negatively)in that they are time based. Some systems now are using “random number generation” to obsfucate the timeline, but to assure the unique meaningless form of numbering.
Comment by Laila Hirr — September 18, 2008 @ 2:41 pm |
Like many others, we ran into the same problem in trying to decide on using significant or non-significant part numbers. There are pros and cons to each, and as a very small business we were more worried about what was happening today. While we didn’t want to paint ourselves into a corner ten years from now when/if we got to be a big company, we finally decided that was our biggest concern.
We finally ended up deciding on the part numbering service from part-numbering.com (http://www.part-numbering.com). It let us set up our significant numbering formats with a shared sequence across all of our number formats. Our hope is that when we get big, or the “significant” format becomes to much burden to support, we can just switch over to a non-significant format. Since all of our part numbers are drawing from the same sequence, they’ll already be unique across all formats. So the conversion should be a snap (we hope).
Cheers,
Steve
Comment by Steve — November 19, 2009 @ 12:44 pm |