Part Numbers are hard. How to think about data first?

Part Numbers are hard. How to think about data first?


One of the topics that usually raises a lot of debates is Part Numbers. One of my first takes on the complexity of Part Numbers was here – PDM, Part Numbers and the Future of Identification. Ed Lopategui reminded me about that topic in his GrabCAD post – Intelligent Numbering: What’s the Great Part Number Debate? few days ago. He speaks about four aspects related to handling of Part Numbers – creation, readability, uniqueness and interpretation. The conclusion is complex as well as the topic itself. Here is the passage, which outlines the conclusion Ed made.

Balancing all these diverse factors is difficult, because no solution is optimal for every company. Here are some final tips to help you make prudent decisions: 1/ Understand your PDM/PLM system part number generation capabilities; 2/ Understand the limitations of any other systems that interact with your parts; 3/ Go through every activity that requires interpreting part numbers and understand what system access is available, and how the interfaces work. This will provide a good basis for your interpretation cost; 4/ Understand how easy/difficult it is for a new employee to interpret a part number.

These tips made me think again about Part Numbering, data and different data and process management tools involved into the process of Part Numbers handling. Most of approaches are focusing on systems and functionality to handle part identification and classification. What we do is trying to align our need to identify and classify parts with what multiple systems can do. The hardest part is to find Part Numbers that will make all systems involved into the process (CAD, PDM, PLM, ERP, SCM, etc.) to work smooth. Honestly it is too complex and too costly.

So, how to manage that complexity? Is there a reasonable way to resolve the complexity of Part Numbering and made everybody happy? Thinking about that I came to conclusion that companies should start thinking about data first. From the longevity standpoint, data must have much higher priority compared to any data management system. In some industries companies are obliged to keep data for decades. Thinking about that, I want to outline some principles that will help you to do so and will allow to create some standardization around parts and data identification.

1- Disconnect Part Numbers and classification from specific applications. PN should not be dependent on requirements and capabilities of data and process management systems. Data has much longer lifespan compared to applications and systems. By defining PN independently you will keep data and processes in your company clean and well organized.

2- Generate PN based on classification, business needs and processes.  Develop independent service to make it happen. This service is most probably should be independent from existing data management systems and converted in some sort of URI based notation.

3- Use  independent service to convert independent PN into system specific identification. You can convert for any system you have at your disposal – PDM, PLM, ERP, SCM… What is important is to be able to control the process of conversion and adapt it each time data and/or process management system changes.

What is my conclusion? Product data is one of the most expensive assets in manufacturing companies. It represents your company IP and it is a real foundation of every manufacturing business. Think about data first. It will help you to develop strategy that organize data for longer lifecycle and minimize the cost of bringing new systems and manage changes in existing systems. I think, some services should be developed to make the process of part numbering easier for manufacturing companies. As manufacturing is getting global to maintain part numbering systems becomes a huge problem. Just my thoughts…

Best, Oleg


Share This Post

  • Paolo Zotti

    I believe there is a big unspoken assumption beyond the long lasting querelle about intelligent or not intelligent part numbers: people fear that with non-intelligent part numbers they won’t have sufficient information about the part… but that doen not need to be the case.
    In PLM systems, you can easily get all the information you need about the parts, just hoovering over it or selecting it, so this assumption is valid only for the people that handle the physical items in downstream systems or on the shop floor, and maybe for manufacturing services suppliers.
    Well, if you introduce a non-intelligent part numbering system, you do need to provide sufficient information about the part to the downstream users, but this can be done in many ways – basically in the same way that you used to communicate the intelligent part number… in the end, if you physically stamp the P/N on the part, why not stamping the Description field as well?
    The most important thing is that the information provided is correct. PLM can have a big role in this, if implemented well, regardless whether the information is in the unioque ID or not.

  • beyondplm

    Paolo, thanks for this insight! I agree – many people put additional semantic into PN just because they don’t trust data management (PDM, PLM, ERP, etc.) to do it in a complete form. This is sort of fear of incomplete data management. Another aspect – fear of complexity of data management solution and classification. Indeed, PLM may have a big role here, but these implementations are still too complicated.

  • A couple of thoughts here – I agree with the concept of generating numbers independently in principle, however in practice many an ERP system tends to choke hard if parts numbers aren’t just so.

    Regarding Paolo’s comment below, it’s not about embedding all the information (Code all the things!) Some try and that does indeed get out of hand. It’s about the user experience. It makes a difference when you’re dealing with numbers that have just enough structure for basic identification and readability in contrast with a generic sequence. Just on a day to day man-machine interface. Even if the system is seemingly ubiquitous, which in most cases it isn’t. Description won’t necessarily help, because description still needs the PN to be unique, so now there’s two things to remember.

    For those adamant on taking the machine determined approach, I challenge them to spend 2 continuous weeks in a liaison position on a factory floor where you’re not just looking up parts, but handling, phoning and/or yelling at people about them. You’ll find that perspective is enlightening in this debate.

    Now when the day comes when the factory floor is nothing but robots building other robots, then things will be different. But until then, you’ll need more human-centric PN’s as long as you keep those pesky meat bags around.

  • beyondplm

    Ed, good thoughts, thanks for sharing! I believe the need to have meaningful PN is clear. The question – how to make it? This is probably where future PLM innovation can come. Maybe to have PN generated based on related Part information? Who knows…

  • Pingback: Existing data prevents companies to improve Part Numbers?()

  • Pingback: Beyond PLM (Product Lifecycle Management) Blog » Why to use intelligent Part Numbers in 21st century?()

  • Pingback: Why to use intelligent Part Numbers in 21st century? | Daily PLM Think Tank Blog()

  • Pingback: Intelligent Part Numbers might be a good idea in “connected world” | Daily PLM Think Tank Blog()