PLM Data, Identification and Part Numbers

I’d like to follow my yesterday post about PLM data modeling and talk about one of the issues that in my view are very important in PLM systems and implementations. The issue of identification or how it sometimes called “numbering system” is fundamental when you start thinking how to organize you product data. This is not a new problem, in my view. It comes all the time in the beginning of each implementation, when you start thinking about how to identify literally everything in your system. It normally starts from Part Numbers but spread out later.

The identification is a very complex problem. In the beginning, you can easy underestimate the size of this issue. However, as much you will be going forward you can easy come to the conclusion that this is one of the most important issues to decide before doing any implementation. I’d like to put few of the challenges that I think important to mention when you think about identification.

Multiple Systems
In the situation when you run many systems, you need to synchronize numbering and identification schema between them. This can be a not simple task and require significant effort and time.

Global Design and Manufacturing
Product development is going global these days. You want to design, build and support your system on a global scale. Product design and manufacturing are often happening in different countries and locations. In many cases, your local manufacturing facilities will be using local ERP system with local numbering and identification schema. At the same time, global product design will be interested to rely on the single identification worldwide.

Manufacturing and Supply Chain
The product development activity can be split between different parties – OEM’s and suppliers. OEM and suppliers are using separate and often different systems. To synchronize or coordinate numbering systems between them is another challenge on a global scale.

Company Mergers and Acquisitions
This is another type of the activities when you will face identification problems. Mergers and acquisitions happen and, in this case, you need to make an effort to create a single common identification schema from two or more separate systems.

This is not a full list, but figure out the most critical aspects that need to be taken into account. Recently, I came across a very interesting write-up about Part Numbering on the ZeroWait State blog. I think you can get some ideas about possible Part Numbering options such as – intelligent, semi-intelligent, automatic.

What is my conclusion today? I think the problem of the identification will become more urgent very soon. Most of the systems in product development and manufacturing were designed 15-20 years ago and considered problem of the identification as a number in a local database. Growing exchange in design and manufacturing information on a global scale will introduce new types of identification problems. In my view, enterprise systems in general, but PLM specifically will need to learn some lessons from internet systems development to find a right solution to this problem.

I’m interested to hear about your practice and experience with implementation of identification systems in your organization and during the implementation you made.

Best, Oleg


56 Responses to PLM Data, Identification and Part Numbers

  1. Colin Clifford says:

    I have seen Part Numbering cause stand up arguments is some companies, it is always good for a long discussion when developing a PLM solution.

    Not sure I am happy with the term Best Practice, I prefer Good Practice, and in this case PLM Good Practice is to use an unintelligent system mastered by a single global PLM system.

    I have seen this implemented in a company across multiple geographies with multiple ERP integrations, and the benefits of rationalising part numbers has more than offset the effort needed to adjust reports and other systems.

    The PLM data model should already be tracking all the intelligence which would be embedded into an intelligent system (as well as legacy numbering systems), and people using intelligent systems rely more on judgement than the system to avoid errors, and this I would not call Good Practice.

  2. Colin, Thanks for your comments! In my view Part Numbering (or identification) solution becomes very complicated when you are trying to rationalize it across the global system with multiple ERP. Since Part Numbers in many cases define interconnection between different pieces of data, they are the first things to agree about. Such discussions about “Good Practices” take time and resources. Unfortunately… I think, we need to come with more stable identification schema for product information. We may need to acquire some ideas from the www system organization. Best, Oleg

  3. Dave Opsahl says:


    Interesting post. Couple of thoughts.

    My experience has been that numbering systems in many companies were designed in one of two ways – either the system reflected the manner in which the ERP system was established, and therefore mimicked whatever defacto information archtecture had been established at that point, such as work breakdown structure (WBS) or something similar; or, the system was set up to ascribe some significance to each element that would mimick the assembly structure. This excludes of course the ‘totally random’ form of naming indidviual elements whatever you want.

    There is a lot to be learned today IMHO from the world of Master Data Management. MDM orignally emerged from the need to establish a source of reference for the 150+ attributes that were required by UCC and its European equivalent to support inter-company transactions between the manufacturers, distributors and big-box retailers. In the work we did on MDM, in no case did we ever find a single system which had even a majority of those attributes – indeed, there was always a significant portion that were not found in any system whatsoever and therefore needed to be added manually. Many of the issues you describe are structually similar to this problem.

    In working now for several years with SharePoint and people’s SharePoint implementations, we’ve also seen where many implementors of SharePoint, usually internal IT personnel with little or no outside help, have underestimated or ignored the importance of information architecture, which takes on a whole new level of importance when dealing with documents.

    Those organizations that have implemented enterprise content management (ECM) systems have found out how important that is, and when you add the complexity of relationships between a bill of materials item (which is essentially what we are talking about when we discuss numbering systems) to the many documents which reference that item, the two problems become intertwined and the need to consider both when designing an architecture becomes clear, particularly when one begins to think about “findability” – the ability to locate relevant information in the search process.

    There is an interesting, albeit lengthy, presentation on this topic here:

  4. Dave, Thank you for sharing your thoughts and for the link to SharePoint video! There are two specific issues I like in your comments – (1) importance of information architecture and (2) findability. With regards to the first one, I think all efforts around DW or MDM are basically trying to find, agree and map on the “single data model” schema. In the end, these are very expensive and un-practical. But, I haven’t seen an alternative to this approach in today’s practices. With regards to the second point (findability), I think, a good numbering system is very helpful to find everything you need. However, this is not enough in lots of situations. Users are utilizing multiple criteria to find information and this is not limited to Part Number only. Best, Oleg

  5. Oleg,

    Great blog. And some insightful comments. Part numbering or identification numbering is a challenge. It becomes even more challenging when filenames get thrown into the mix, but I digress.

    I really like the concept of master data management mentioned by Mr. Opsahl. Your architecture should enable potentially many-to-many links so that ERP items could be linked to BOM items and engineering parts, etc. I liken it to the old Sears and Robuck cross reference. It enabled repair personnel to take the Sears model number and quickly find the engine model number regardless of whether it was Kohler or Tecumseh.



  6. Michael, Thanks for your comment! As I mentioned before, MDM has interesting ideas. However, today’s implementations are even more complex than PLM :)… Also, the idea of having “single moderated schema” to keep all information in the organization in sync, is somewhat I’m afraid won’t work or will be tooooo expensive to implement and manage. Maybe I’m wrong, but these days, I see only very big customers have a luxury to think about DW and MDM solutions. Just my thoughts… Best, Oleg

  7. Awadhesh Parihar says:

    While I have seen this issue (part numbering) being discussed ad nauseam before(and after) PLM implementations – but would like a fresh take on the subject.
    Numbering systems, as applied to any business object is for the purpose of unique identification of the object in a given business context. Similar to our names as in the records, as an employee or a citizen or a student and so on..but we all can describe ourselves as ‘me’ as well and function the same way as we do.

    In PLM parlance, part numbering is to be able to carry out ‘transactions’ concerning the part. And is a necessity born out of manufacturing/procurement convention than ‘engineering’ convention

    So, why should we not try and decouple the concerns here- through intelligent classification (attribute) schemes concerning PLM objects, and identification based on geometric features or the cliched ‘form’,’fit’ and ‘function’.

    How to do it, would be a seperate debate! Why do it, is the point I want to raise here

  8. Awadhesh, Thank you for comment and idea share! I like an effort to make a definition more granular on the level of attributes. This can be an interesting approach. However, the implementation in the systems available today, can be complicated. Have you had chance to think about how to implement that? Thanks, Oleg

  9. Dave Opsahl says:

    Awadhesh brings up an interesting point – which is, do number systems fundamentally need to have any implied intelligence whatsoever?

    Back in the ’80s, there was a (short) period where a set of applications called “group technology” were all the rage, and there were essentially numbering schemes that classified parts and assemblies according to form, fit, and function. They were, IMHO, solutions which were defined by the available technologuy – in this case the predominance of hierarchical databases such as CICS – rather than defined by needs or requirements (at the time, relational databases were only starting to come into production use).

    They failed to gain wide acceptance for two primary reasons: one, it required that everyone using them be skilled in the nuances of the classification scheme (which was not trivial); and two, because it created a set of constraints that prevented common use, such as being able to organize and present information by another means, such as supplier.

    I’m not sure if intelligent classification, if by that we mean intelligent numbering schemes, introduces the same problems in a modern context or not, and along with that, similar constraints.

    Intelligent classification, if by that we mean using software to derive and associate form/fit/function from the data itself, that I think has real merit. Then, the MDM concept can provide a real solution, because you can associate such information with data contained in multiple LOB systems to help users answer common questions, such as – I need to know what parts are available from which approved suppliers that have these form/fit/function characteristics, and can be delivered in this timeframe, within these cost parameters.

    That’s a powerful decision support tool for engineers and managers alike.

  10. Dave, I think technical history is making spirals sometime. The RDBMS won over Hierarchical and Network models because of simplicity and performance. The schema proposed by Mr. Codd was more efficient in implementation and mode flexible. To define primary and secondary keys were way simpler. However, nowadays, I’d expect some renaissance in the network based systems. The computational power can help to resolve previous problems… Don’t you think so? Best, Oleg

  11. Awadhesh Parihar says:

    Thanks Dave/Oleg for explaining some brilliant ideas/concepts.

    However, by intelligent classification scheme, I didn’t intend the nomenclature scheme, but a classification system intelligent enough to identify part (or any object for that matter) based on its meta information or geometrical/analytical characteristics. A number would only make sense to someone who is well versed with the nomenclature scheme, but ‘purpose’ ought to be primary and nomenclature secondary

    Dave brought up Group Technology/Cellular Manufacturing concepts – which did drive the classification systems in the past, but cannot be compared with an elaborate classification scheme such as in SAP CS. PLM domain lacks it (have been a great fan of Part Family concepts in SDRC/UGS PLM systems, but that really needs to be evolving further) not as technology but at conceptual level

    Data model disucssions is really imminent here, but I am purposely staying away from it, as am doing some homework on concepts such as ‘big table’ and would come back with applicability or necessity in PLM domain. But to help further this discussion of Part numbering, my take is, we should look at classification systems first, aspects or views of business entities (remember Material Master in SAP and its zillion views all connected to MARA table), and scheme whereby part announces its characteristics (like CAD drawings telling what they do, in XML form) and global characteristics coordinator kinda concepts)

    Have seen many discussions and interesting applications such as Geolus from UGS, experienented/suggested Apache Lucene for global indexing in PLM and so on. Will be happy to hear if some of these make any sense to you all

  12. Awadhesh, Thanks for your thought! I think such topics as classification and granularity are tightly dependent. In my view, this is one place where there is the lack of power in today’s data modeling solutions from PLM/ERP vendors. Best, Oleg

  13. Menk Slot says:

    Hello Oleg,

    Numbering system was always a discussion and will always be. People get use to a numbering systems over the years and they recognize parts by this system.

    The discussion always start with: Should it be a random system or more intelligent? Before we used computers, the only way to find back information was by the number. So at that time the only solution was an Intelligent system. The problem was/is to combine Identification and classification of a part makes the numbering systems very big and complex.

    Now these days, it is possible to use more metadata or base the search on relations between objects. E.g. Give me all parts used in Project A with Commodity Code XXX.

    So at this time it is not necessary anymore to use intelligent numbers. It is better to split Identification and classification.

    I have seen a lot of Companies starting a PLM Project, because of the classification system within this system. I didn’t see many succeed. People don’t have time to classify the parts and if they have it still is very difficult.

    Lately I see some Pilot projects with classification systems integrated with a 3D CAD System. The system can classify based on features, descriptions and can extract the dimensions from the CAD System. That is what was written in the brochure. In real life it is not so simple.

    I hope that those kind of systems will be developed more and need to have deep integrations with CAD and PLM



  14. Menk, Thanks for bringing up issue of dependencies between numbering and classification. I think, this is really important. I think, existing solution are lack of flexibility and this is the reason why lots of people still keep developing numbering system to support classification. And in case of multiple enterprise systems it becomes even more important. In the end of the days, numbering schema is the only thing that keeps all PLM, ERP, SCM etc. systems consistently together. What is your take on this? Best,Oleg

  15. François says:

    Hello Oleg

    Ambitious topic for a blog!

    When you start from “Part identification” and you go to “Part numbering” you already make an hypothesis which is not true for the full part lifecycle. Part identification start usually without a number, and sometimes continue for some activities without part number (exemple CAD activity).

    But OK to discuss part numbers, in fact IDs. There are many reasons to provide an ID for a part, simply at the beginning shortening a description because it is simpler to remind it if it is short. It leads to inteligent numbering. Intelligence is in this wording “people intelligence” because people know how to interpret an ID made from different attribute values.
    Bus as soon as the community is growing (different people communities or locations, people using different systems) using the same part, there is the tentation to unify Ids across applications. And usually then start dogmatic thinking about this merging. And I have seen situations where it is not possible to keep the same number for the whole lifecycle of a part, because people are not talking about the same thing, but beleive they are. In fact many times the question is not the number itself only, but the evolution of it, and its revision index. So people need to take care comparing two objects in different systems, possibly having the same ID, and ask themselved several times: are we talking about the same thing?
    On the other hand, yes, unintelligent numbering schemes are ok for people using a system, because of the existence of attributes to complement the ID. Don’t forget there are other people using those numbers only, but having not a system to use (shopfloor, stocks), who may need some intelligence in the ID.

    Next time, I guess you will open a post about file naming, right?

  16. Francois, Thanks for your comment! From my standpoint, there are three separate topics I want to talk about – 1/ PLM data (or product data), which includes everything – product characteristics, documents, drawings, part numbers, product structures, etc.; 2/ Identification and 3/ Part Numbers. I see them dependent, but separate. The identification mechanism is helping us to keep PLM data ordered. It can be just sequential ID, but it also can be DNS/URI or whatever else. Until the data is in one system, this is fine and there are multiple options to organize and identify information. However, as soon as you come to the situation when data is not belonging to single system it becomes very complex. The number of existing systems, organization internal and external information creating very complicated situations related to product data identification. Part Numbers are, in fact, one of the mechanisms used by customers for “global product data identification”. The majority of enterprise systems are not built with a global state of mind. Therefore, customers are trying to create cross-identification solutions (i.e. Part Numbers) to resolve it. BTW – I see file naming as one of the identification mechanisms. I already posted about files –> Best, Oleg

  17. Guy Alroy says:

    Hi Oleg,

    You touched a very sensitive subject…

    I must comment that discussing the Part Numbering schema as part of the PLM project scope is something that I am not sure I’ll recommend to the customer nor to the implementer, unless we are talking about a very young company with very minimal data (I assume all are with me on the same page here, this means changing the ERP schema, references in documents/drawings, pricelists, customer references, etc.)

    I have done numerous PLM implementations where the requirement was to adhere to an intelligent part numbering schema. Usually those were pretty complex, to the level of an hierarchy of classification, incrementally building the prefix to the part number and ending with a sequential number. For example: if the component is an electronic part > passive > resistor > variable it will get a prefix of 112022 and then the next available number in this schema.

    There are several issues that a company should consider choosing to go in the intelligent part numbering direction, especially when going to very detailed levels, especially if in an industry or technology that continues to evolve. Such a ‘classification based’ part numbering schema must be maintained, if you don’t maintain it according to the technologies that the engineers acquire, you will get to a mess, a mess, where the users will not use the intelligence anymore as they can’t count on it. So if you are not up to the challenge of keeping up to speed with the technology and it’s embracement by your R&D and engineering, DON’T GO THAT WAY. If you still want the intelligence keep it simple.

    I would always argue why there is a need for an intelligent part numbering schema and not just sequential numbers issued by the PLM system (and pushed to the ERP). Usually it will be related to a situation where there is an analysis of the PN (or usage of the part number) where no access of the PLM/ERP/SCM/MES is there to get the additional data. It can be in context of a paper drawing situation or a shop floor not equipped with automated processes and heavily using paper. My successful implementation of simple, sequential PNs was in organizations that had a strong MES system (or related ERP module) and openness for a change (as most of the companies did have some level of intelligence in the PNs). Even then, engineering and manufacturing did have some hard time to adjust, but if communicated and trained well, got used to the new idea.

    A PLM and ERP technology that will not allow you to access the part details by the dumb numbers and sort and export the data according to a certain sort would usually be the dinosaurs, the modern systems all do that.

    Then there is the subject of handling methodologies where the part number ‘overloaded’ with revision tags, variation tags and similar, and that might be very problematic when it comes to the alignment between systems, especially when people used those to indicate steps in the route. Here there is no ‘out of the box’ solution as it requires an analysis of the business processes as implemented in all the related enterprise software involving the numbering scheme and logic.

  18. Guy, Thank you for sharing your thoughts! I agree, Part Number is complicated topic, and, unless you are coming in the very early beginning of the company’s life, you already have some Part Numbering schema in the organization. Then Part Numbering becomes complicated. Companies are trying to dance around multiple systems. Each of them comes with their own approach. These are the biggest challenge and problem, in my view. As the number of systems and external dependencies will grow, customers will ask for more reliable solutions for “identification” and “numbering” .The urgency to create a proposal for global identification system will grow. Systems that will think “global”, “flexible” and “open” with regards to the identification of data will survive in the future. Just my thoughts… Best, Oleg

  19. Paul says:

    Ambitious is the right word. I too have seen stand up arguments about numbering. Specifically in my case when to re-number a part. The Design view is ‘fit, form, function’, but when your PLM (or config system) also holds (for example) Condition of Supply because it is now a Manufacturing Definition (not just a design) is a Revision sufficient (e.g. you need to keep them in different locations in stores as each ‘version’ is manufactured differently).

    I have also had the ‘just use the next number’ argument. Which is great if you can lookup the part on a system, but when manufacture and maintenance is frequently carried out remotely from the system, a number that tells you (for example) it’s a fuel system pipe, not a hydraulics pipe (because the numbers don’t match) is very useful and efficient.

    Someone mentioned using the www – perhaps the trick here is to have multiple ‘public’ identities for a thing ‘,‘ (or ‘kinked fuel pipe’, or ‘pipe, fuel, left, bottom’) and a single internal ID ( or PIP123456). This might help all the communities ID things how they want, whilst maintaining a single, consistent data model? (DNS for PLM anyone?)


  20. Paul, Thank you for comment! I agree, there are lots arguments for “numbering”. Smart, Dumb, Semi-smart :)… The name doesn’t matter in my view. The real problem is co-existing of identification in multiple systems. Part Numbering schemas are trying to resolve it. My take is that PLM need to think about a problem of “global identification”, so DNS is not a bad idea at all. Best, Oleg

  21. Awadhesh Parihar says:

    Very thought provoking arguments! On second thoughts, if DNS is the argument, are we saying all part numbers with a specific form, fit and function and of defined input and output (like h parameters to a black box), will have same number, across the industry?

    And in the transition phase, organizations can have bidirectional translators, till they completely phase out their legacy nomenclatures?

    If STEP can do this for the entire product definition, standardization of part numbers (other than STANDARD parts, which already have standards such as DIN)shouldn’t be a big deal???

    And if you should distinguish your part from that of you competition, you can have a universal suffix to the part numbers, such as FuelPump@ABC Motor Company..

    There are million details to it (much the same as would have been before evolution of any standard), but if it helps get departments talk to each other, and companies talk to their supplier, vendors and customers, it would be a giant leap towards industry wide interoperability? And of course benefits can be monetized.

    Inventory, Shipments and Services will undergo a metamorphosis.Regulatory compliance, ecological footprints etc. will be implemented with relative ease? Of course there would be arguments as to why it cant be done, but there is merit in exploring why it should be and can be done? Eager to listen to your views on this

  22. Awadhesh, I’m trying to see if STEP (or any other standard) can come with the idea of global identification… Similar to some of your thoughts. Prefix, suffix, global naming, numbering… DNS or any other ideas can be very beneficial in my view. I’m not sure about translators you mentioned. In my view, if we can eliminate translators, life will become easier… Thanks for your comments! Best, Oleg

  23. Brian says:

    Lots of great comments here. I’ll add a couple.
    1) You can’t outgrow a non-intelligent part numbering system and you don’t need to “maintain” it either. 2) Where you need classification information (and you will) use the Part Description which can be as structured as you like (built from a Hierachical context if possible).
    Most of the comments againt non-intelligent systems revolve around not having access to the computer system. In most of these cases (engineering drawings on the production floor, field repair etc) the user will have a Description as well as the part number anyway.
    Part numbers (just numbers not Alpha and special characters) are the easiest to get right every time you enter a number into a database and are the fastest for data entry operators.
    We are operating in a world where more data is becoming available from mobile devices so the number of places where an organisation “can’t” get to their computer system to look up all the additional information that may be of interest should be going down dramatically.
    Intelligent or Semi-Intelligent systems let people make assumptions about the parts without looking up the system for confirmation. Lots of times you will be right but sometimes you will be wrong and cause more trouble than looking the information up in the first place and then someone will be wondering why the company spent so much time and effort setting up systems to capture all this PLM/CRM/ERP data and make it available to you in the first place.

  24. Brian, Thank you for your insight and comments! I think, intelligent numbering system was born before computer systems came to the office. So, originally, they helped to provide a way to find a document, parts and assemblies and classify it in the meaningful way. Lately, intelligent part numbering system got lots of the supporters because of multiple systems. Part Numbering systems helps to organize and identify data in multiple systems. I think, in order to overcome the complexity of current systems, we need to invent better identification systems that can operate on a global scale and not dependent on the specific enterprise platforms, applications and solutions. Still lots of issues to how make it. However, to learn from existing Part Numbering system is a good idea to start from… Best, Oleg

  25. Awadhesh Parihar says:

    I introduced the ‘translator’ bit just to alley the cultural change management related fears when organization move from their homegrown part nomenclature to Global Schemes. No matter how much we want canonical definitions to be adhered to, as long as there are imperfections, there will be translators :-) (didn’t mean it to sound biblical, apologies if it came out so)

    Brian’s comments on part description built from hierarchical context appears very logical and well thought. But thats secondary problem, mostly related to ‘adoption’ and ‘familiarization’ with ‘A’ nomenclature. Part number scheme, still remains the inevitable evil!

  26. Awadesh, The “translator” issue is the evil and need to disappear. How? I still don’t know. However, enterprise systems kill themselves on the issue of translation between one representation to another… Just my thoughts… Best, Oleg

  27. Brian says:

    Actually Awadhesh the Part Number scheme goes away entirely if you have non-intelligent numbers. The next number is just the next number in the simple sequence. Full stop.
    Where you have to put the effort is into the Hierachical Description structure to ensure good classification of parts once, when the part is defined. Of course if you get the classification wrong and come back later to correct it you might change the part description but you won’t need to change the part number since it didn’t mean anything in the first place. Changing part numbers is hard in most systems since they are the key field for relationships.
    If you don’t have a Description based on Hierachy (fairly common even if it is a bit hit and miss) you can still apply classification Meta Data to a part to allow you to search for all of the parts with a given set of attributes. Once again, if you get the classification a bit wrong in the first instance it really shouldn’t be a problem as all you will be correcting are the search terms (tags) that apply to the part to allow people to find it more easily in the first place. In fact the more I think about it, allowing meta data fields to be linked to the part number rather than using a Hierarchical map should be less effort and require less maintenance. It should also be relatively easy to allow new meta data tags to be added by (component specialist) users when and only when a new tag is required.
    It’s very hard to move a company from a historical intelligent/semi intelligent number to a non-intelligent one but provided you do have “translators/lookups” then you can find your way from one system to the central global ID which is what you have all of your data attached to.
    An interesting thing with the decision between intelligent and non-intelligent part numbers is the comment that people can figure out what the part is from the information in the part number. This is only true for people who use the parts/system all the time. For everyone else the intelligent number is just as unintelligible as the non-intelligent number and usually a lot harder to remember.

  28. Brian, I agree with you on the topic of Part Number vs. Classification. Actually, Intelligent Part Number is kind of Classification. So, if you have a classification schema you can use to identify your parts, you can provide 1-0, A-Z Part Numbering. Thanks for sharing your thoughts and discussion! Best, Oleg

  29. Awadhesh Parihar says:

    Hi Brian
    Attribute based part identification (and classification system), extending to even geometry based identification is where I started with, in this blog

    But like most our friends opined, the complexity is more owing to multiple systems and departments. Think of it, if ERP also has started a serial numbering (just the next number), need for a uniform nomenclature too gains prominence. If you extend the logic to suppliers/vendors (the extraprise), problem is as big as resolving IP addresses using a DNS, or services using IORs

    So, to conclude what I said,
    a) Classification systems
    b) Global Part UID – and by the way, it would drive standardization in a) as well
    c) Translators only in the interim (Oleg’s views around the translators, graciously accepted! But consider this as necessary step towards ‘adoption’ of b)

    are some thoughts that merit further exploration.

  30. Awadhesh, Frankly saying, I hate translators. However, this is today’s reality. If you can go away from translation between different systems, life will be much easier in the enterprise, in my view. Thanks for your comments! Great discussion… Best, Oleg

  31. Shaoping Zhou says:

    The history of part numbering scheme and allocation method is probably more interesting than the debate of which scheme is better. I suppose that I am hopelessly a history nut :)

    Back in “stone” age where industrial companies grew their empire and expanded product offerings consisting of millions of components and SKUs, some really smart people “automated” the process of allocating blocks of numbers identified by the first 2 or 3 digits to specification staff in different diciplines, development centers, and countries. They keep track of this with number allocation cards.

    A large number of big companies in West and East today still have this practice, albeit, in digital form with advanced databases and cPDM systems.

    The debate around the pros and cons of numbering scheme to a certain extent reflected the lack of appreciation of part classification and attributes in text and images. Once classification and intelligent search become the norm, I think that the numbering scheme debate will largely disappear in my humble opinion.

  32. Shaoping, The history is important, in my view. It presents two reasons why an intelligent numbering system was created. 1/to support classification; 2/to support cross department and cross-companies’s identifications. The main problem I see these days is in existence of multiple systems (this is the reality) and absence of the efficient system-independent mechanisms for data identification and classification. In my view, PLM vendors are trying to impose practices of “singularity” or “single PLM data storage” to resolve this problem. I’m not sure, this is the most efficient way. However, this is the industry trend for the last 10 years. Thanks for your comments! Best, Oleg

  33. Awadhesh Parihar says:

    Hi Shaoping
    My thoughts are the same as well, and if you see my posts, in this inexorably long discussion (Oleg, need some innovation in this blog as well, to enable representation of chains of thoughts :-)), I did mention geometry and purpose leading to identification than just a dumb number or classification scheme thats based on common denominator of attributes.

    Increasing complexity of the system in doing so, is the flip side, we need to be vary of. My thoughts are captivated by Oleg’s blog on complex societies and am trying to think simple as we speak :-)

  34. Awadhesh,
    It is interesting to see what are the acceptable practices of data classification in the companies and how PLM implementations can support it. Maybe it will turn to the next blog topic ;)… Thanks for your great contribution to the discussions!
    PS. Yes, I’m aware about the problem of comments/threads. I’m working to resolve it. For the moment, DISQUS is the leading option I’m thinking about. Howeger, it means to take the blog out of WP.COM where it hosted now. WP.COM is not supporting DISQUS, unfortunately.
    Best, Oleg

  35. Paul says:

    Oleg, you talk about multiple systems, but there is another issue here of multiple customers. I work in defence and the customer has his own numbering convention – NATO stock number (NSN). Plus different disciplines have their own Standards to adhere to, so for example, a standard part may have an internal catalogue number (used across projects), a DIN, EN, BS standard number. Then it gets an NSN (but this is very late as the numbering takes time and it is expensive). however, if it is used multiple times in the same product, it attracts several Logistics Control Numbers (LCNs) – one for each unique usage – dictated by DEFSTAN0060 or MILSTD1388 or whatever….

    I completely agree that it should be possible to allocate one ID within the enterprise (assuming that enterprise doesn’t change to radically during the project….) but the question then becomes who’s ID – design (that is the same valve here, here and here) or support (that’s the valve here, that’s the valve there, and that’s the same valve somewhere else) ?

    Also, if you ‘lose’ some of the classification from the the identity (e.g. use dumb numbers) then the position in the structure becomes important in identifying the specific part (as above, in design, you can only identify the left valve, from the middle, from the right from their parent, whereas support identify each one uniquely). Given the discussions about multiple BoMs (or views) then these become even more vital as they form more of the classification/identity of a thing.

    One last thought, if we every get a Universal Part Number…. how will OEMs hide the mark-up they put on a pattern part if both versions have the same identity???? :-)


  36. Paul, I agree- there are lots of interesting aspects related to multiple classifications. Your example with multiple customers is correct. The similar one is when A Supplier works with multiple OEMs. This is in my view a typical case for industry. In my view identification is a unique (you can call it a universal part number – UPN :)). However, rest of the things is “a classification” or “classified characteristics”. So, in your case NATO stock number will be specific characteristics applied to the UPN. Best, Oleg.

  37. Awadhesh Parihar says:

    Hi Paul
    The point you brought up about part identification and usage of the part in the structure are two different problems in my opinion.
    Its not so important for an automotive as its for a ship or aircraft, that the positional data be available as meta information.
    Distinction is in the production process – if factors of production move to the job (fixed layout), or job moves through factors of production (assembly line).

    Spatial arrangement, sections, zones etc. are thought of in PLM metamodels (as add-ons though) for fixed layout.
    But, to think of it, part identification still remains the primary concern as opposed to the usage of part in the structure (BOM, BOM views)

    However, one problem, in my experience, where the singular part changes it identity (for the purposes of stock keeping) where @(at the rate) part number is used. If a crankshaft leaves the organization for heat treatment and comes back, they denote the part number differently (say CrankShaft@Operation Sequence 10, or CrankShaft@Operation Sequence 20). That in my view is a dimension that isn’t quite reported in the literature. Any views that the audience may have, would help further this discussion

  38. Paul says:

    Awadesh, your point about evolving part numbers is a good one. We have not mentioned that the ‘thing’ is generally marked with it’s identity – part marking. As you say, how you manage a change in part number a as physical thing evolves is far from easy – Billet, Part machined, fully machined, treated, installed, modified, repaired, re-treated, damaged etc…. Also, given the extension of responsibility of OEMs (e.g. into in-service and disposal aspects) that identity (plus a serial number?) becomes ever more important.

    My issue with structure position relates to the different requirements of (say) design vs support. For design, once a pump meeting the requirements has been identified, e.g. pump123 (from supplier C), that is what appears in the BoM wherever it is used. By definition, there is no distinction since it is the answer to the same requirement in each case. However, in support, the specific pump is relevant as the application may change maintenance requirements (e.g. use as primary vs back-up will drive different servicing intervals). If your global identity is the design identity, you have to have context to know which support data applies – easy whilst it’s fitted, not so easy on a bench although you can ‘lookup’ that serial number and see form its metadata or structural callup, which application it had.. If, however, you use the support identity, even out of context you can identify the application and hence, the correct service interval.

    I know turbine manufacturers have been serialising everything for many years for traceability, perhaps the move to true PLM will drive universal serialisation – then the part number identity won’t matter, the serial number will be unique and everyone can refer to it how they want. That would solve your issue Awadesh – Serial atOp10 vs Serial atOp20


  39. Brian says:

    I agree with the trend towards serialisation. If you “care” about the logistics of a part enough to want to track things about it then it needs to have a unique serial number. This doesn’t change the fact that it is “Part xxx” as specified at design.
    I think in the end the key is, as Awadhesh has said, it Attribute classification. This, however, is IMHO separate to the Part Number that you give something.
    As the use that the particular physical part is put to becomes clear (primary vs backup etc) then different meta data tags (attributes) can be attached to the specific item, by serial number, in the specific Logistics system that drives that recording.
    This leads to your “basic” meta data attributes which exist in the PLM system and apply to all of that Part Number plus additional meta data tags as the part is serialised and used in production then logistics/repair etc.
    If you don’t “care” about the end use to that degree then you don’t serialise it.

  40. Brian, thank you for this clarification! I think, it is a very good example of how “identification” can “overlap” with multiple characteristics. And Part Number is one of them from my view. Best, Oleg

  41. Paul says:


    this starts to drive the ‘serialise everything’ view – as you cannot retrospectively serialise something 2 years into service.


    perhaps as you imply, part number is just an attribute of the thing. maybe those horrible OBID/GUID system gobbbledegook codes finally have a use for normal people! Not sure I’d like to stamp a bracket with a 32/64 character random string though! And Tech Pubs would be a nightmare….


  42. Paul, I think what you call “horrible gobbledegook” codes need to be kept for system. You need to be able seamlessly get an access to the information with meaningful Part Number and other properties/characteristics. Does it make sense? Thanks for your comment! Best, Oleg

  43. banyan08 says:

    PLM data modeling article is really good. The concepts about multiple systems, Global Design and Manufacturing, Manufacturing and Supply Chain, and Company Mergers and Acquisitions found to be informative. The explanation on each category is useful. Global concepts of design, manufacturing and supply chain is very explanative.

  44. Banyan08, Thanks for your comment! Best, Oleg

  45. […] PLM Data Identification and Part Numbers This simple topic raised huge amount of attention and interest. I found that despite the fact all manufacturing and engineering organization are handling Part Numbers, there are not much agreement on this topic. Many opinions with very controversial proposals about how to do so. […]

  46. […] The discussion raised lots of opinions and comments. You can track them on the following link. Today I want to discuss multiple aspects related to the identification of […]

  47. […] The discussion raised lots of opinions and comments. You can track them on the following link. PLM marketing and sales speeches are rarely discussing this topic. It assumed as a solved […]

  48. […] you remember my old blog about Part Numbers- PLM Data, Identification and Part Numbers? I think, in the context of File Names, the same questions are coming again. To use smart numbering […]

  49. […] you remember my old blog about Part Numbers- PLM Data, Identification and Part Numbers? I think, in the context of File Names, the same questions are coming again. To use smart numbering […]

  50. […] other side, identification can be more complicated story. I posted before about document numbering, part numbering and identification before. These topics can be complicated and requires implementation skills and […]

  51. […] other side, identification can be more complicated story. I posted before about document numbering, part numbering and identification before. These topics can be complicated and requires implementation skills and […]

  52. […] encourage you to refresh your memory and read one of my blog post back one year ago – PLM Data, Identification and Part Numbers. I found some interesting facts specially reading all 51 comments to this post. Finally, the […]

  53. […] encourage you to refresh your memory and read one of my blog post back one year ago – PLM Data, Identification and Part Numbers. I found some interesting facts specially reading all 51 comments to this post. Finally, the […]

  54. Dave Champlin says:

    All “intelligent” part numbering systems develop dementia with age. Not original – one of the few times an outside consultant said something remarkable. The lure of adding meaning to a part number is very strong but it is a siren song. Unless the designer of the numbering system can predict the future and every possible bit of infomation that might want to be “communicated” the system will fail at some point. In anything but the most trivial environment the part number can only be an index to other systems (PDM/ERP/etc) that maintain the metadata associated with the part

  55. David, Thanks for your insight. I think, the interest of people to provide a semantically understandable classification and taxonomy is natural. An “intelligent” part numbering system is just one of the examples where people are trying to make it understandable. If future PLM/PDM systems will provide an easier solution here, the “intelligent” part numbers will eventually die. However, it is still not happened, and we keep thinking how to improve. Best, Oleg

  56. Harrogate VoIP systems|Harrogate Telephone systems|IP telephone systems Harrogate…

    PLM Data, Identification and Part Numbers « Daily PLM Think Tank Blog…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Get every new post delivered to your Inbox.

Join 290 other followers

%d bloggers like this: