Some weeks ago I wrote a post about non-intelligent part numbers (here) and this was (as expected) one of the topics that fired up other people to react. Thanks to Oleg Shilovitsky (here), Ed Lopategui (here), David Taber (here) for your contribution to this debate. For me, the interesting conclusion was that nobody denies the advantage of a non-intelligent part number anymore. Five to ten years ago this discussion would be more a debate between defenders of the old “intelligent” methodology and non-intelligent numbers. Now it was more about how to deal/wait/anticipate for the future. Great progress !!
Non-intelligent part number benefits
Again a short summary for those who have not read the posts referenced in the introduction. Non-intelligent part numbers provide the following advantages:
- Flexibility towards the future in case of mergers, new products, and technologies of number ranges not foreseen. Reduced risk of changes and maintenance for part numbers in the future.
- Reduced support for “brain-related connectivity” between systems (error-prone) and better support for automated connectivity (interfaces / digital scanning devices). Minimizing mistakes and learning time.
What’s next?
So when a company decides to move forward towards non-intelligent part numbers, there are still some more actions to take. As the part number becomes irrelevant for human beings, there is a need for more human-readable properties provided as metadata on screens or attributes in a report.
CLASSIFICATION: The first obvious need is to apply a part classification to your parts. Intelligent part numbers somehow were often a kind of classification based on the codes and position of numbers and characters inside the intelligent ID. The intelligent part number containing information about the type of part, perhaps the drawing format, the project, or the year it was issued the first time. You do not want to lose this information and therefore, make sure it is captured in attributes (e.g. part type/creation date) or in related information (e.g. drawing properties, model properties, customer, project). In a modern PLM system, all the intelligence of a part number needs to be at least stored as metadata and relations.
Which classification to use is hard to tell. It depends on your industry and the product you are making. Each industry has its standards which are probably the optimized target when you work in that industry. Classifications like UNSPC might be too generic. Although when you classify, do not invent a new classification yourself. People have spent thousands of hours (millions perhaps) on building the best classification for your industry – don’t be smarter unless you are a clever startup.
And next, do not rely on a single classification. Make sure your parts can adhere to multiple classifications as this is the best way to stay flexible for the future. Multiple classifications can offer support for a marketing view, a technology view (design and IP usage), a manufacturing view, and so on.
Legacy parts should be classified by using analytic tools and custom data manipulations to complete the part metadata in the future environment. There are standard tools in the market to support data discovery and quality improvement. Part similarity discovery is done by Exalead’s One Part and for more specific tools read Dick Bourke’s article on Engineering.com.
DOWNSTREAM USAGE: As Mathias Högberg commented on my post, the challenge of non-intelligent part numbers has its impact downstream on the shop floor. Production line scheduling for variants or production process steps for half-fabricates often depends on the intelligence of the part number. When moving to non-intelligent numbers, these capabilities have to be addressed too, either by additional attributes, immediately identifying product families, or by adding a more standardized description based on the initial attributes of the classification. Also David Taber in his post talked about two identifiers, one meaningless and fixed, and a second used for the outside world, which could be built by a concatenation of attributes and can change during the part lifecycle.
In the latter case, you might say, we remove intelligence from the part number and we bring intelligence back in the description. This is correct. Still human beings are better in mapping a description in their mind than a number.
Do you know Jos Voskuil (a.k.a. virtualdutchman) or
Do you know NL 13.012.789 / 56 ?
Quality of data
Moving from “intelligent” part numbers towards meaningless part numbers enriched with classification and a standardized description, allow companies to gain significant benefits for just part reuse. This is what current enterprises are targeting. Discovering and eliminating similar parts already justifies this process. I consider this as a tactical advantage. The real strategic advantage will come in the next ten years when we will go more and more to a digital enterprise. In a digital enterprise, algorithms will play a significant role (see Gartner) amount of human interpretation and delays. However, algorithms only work on data with certain properties and reliable quality.
Conclusion
Introducing non-intelligent part numbers has its benefits and ROI to stay flexible for the future. However consider it also as a strategic step for the long-term future when information needs to flow in an integrated way through the enterprise with a minimum of human handling.
1 comment
Comments feed for this article
February 1, 2016 at 3:51 pm
bradwindisch
My company will be implementing Microsoft Dynamics AX with Bluestar PLM. We have just achieved acceptance that we will use non-intelligent item numbers in the future, and are now working on classifications and metadata to support the future system. Next will be to decide what to do with the legacy item numbers. We will either 1) migrate the old item number into the new item number field and only take out new non-intelligent numbers for new items, or 2) we will renumber all old items prior to migration and set up a cross reference field to hold the old number. We are a manufacturing company so many items have solid models and drawings that support them, and those legacy files contain only the legacy number. Would appreciate any advice or recommendations from you or your readers about the best method to migrate legacy items.
Brad hello. Without knowing the details I would use option 3. Leave the old numbers the old numbers. Make sure that you update the old parts with the right attributes and classification. In theory the old part numbers need to become irrelevant for people in your company. See also my answer to Tony Foulon in the list of comments. I hope this helped.
Success Jos
LikeLike