Some History

Before the advent of computers, product design and manufacturing where already big business. Much of what is currently taken for granted in terms of IT systems today did not exist and most activities had to be carried out manually.

Part numbers (and BOM) are central to any product design and in the manual days, it seemed logical to add intelligence to part numbers. Given the human capacity for pattern recognition, inspection of an such a part number could impart a lot of information about that specific part.

Consequently, intelligent part numbers became embedded in design processes, human memory and in engineering departments. But then, computer systems came along and because of search capability in databases, intelligent part numbers became redundant. So much so, the strong recommendation from PLM vendors was to do away with intelligent part numbers and use a sequential numbering scheme backed by a robust attribute set.

But, in conversations with a spectrum of customers, most still have intelligent part numbering schemes and when faced with the choice of sequential numbering, many balk. So, if this scheme of adding intelligence to part numbers is dead, why is it still very much alive?

Before we answer the question, let us look at the pitfalls of intelligent part numbering and why everyone recommends not using it.

Intelligent Part Numbering – A description

Intelligent Part Numbering is characterized as follows:

• Section or sections of the part number identifies some property or properties of the part
• What is not incorporated in the part number is described by other attributes attached to the part number object in a database

An example is given below

The part number is built up using properties and descriptions of the part in each one of the sub numbers

Sequential Numbering

By way of contrast, let us look at how sequential numbering works:

• Part Number is generated sequentially
• All properties are included in attributes

The part number is generated as the next number in the sequence. Attributes contain all the part properties.

Risks associated with intelligent part numbers

There are multiple risks associated with intelligent part numbers. Here is some of the pitfalls:

• Incorrect part numbers are difficult to change. (Steel part is accidently numbered as aluminum) This part is then released and then an engineering change notice is required to correct
• Categories can contain to many or too few designators, but once implemented are difficult to change. For example, steel could become to general and needs to be spilt into high tensile or mild. This can lead to a nightmare of renumbering, everything steel must be changed.
• Proprietary information may be included in the intelligent part number. These numbers are typically shared with suppliers and this info becomes obvious to third parties.
• Date categories can eventually run out. For example, 96 as a shortening of 1996 became a problem. (Remember Y2K)
• Mistakes are made in allocating part numbers. The scheme may be obvious to experts, but to the casual user can cause confusion and lead to errors.
• Categories become over-used (“Miscellaneous”), ambiguous (“General”), or irrelevant (“Flip Phone”).
• Stealth alternative schemes can circulate underground if the official scheme is too cumbersome
• Merging multiple numbering systems is impossible. This occurs when companies make acquisitions and need to combine engineering departments.
• New users and outside parties (suppliers) need to be educated creating ongoing training requirement.
• An upgrade to a new PLM system becomes impossible because the new system does not support intelligent part numbers
• Integration with a ERP system that uses random numbers is very difficult

Dead or Alive?

Of course, there are complications with sequential numbering and attributes, but these are much easier to resolve with database manipulation.

But, notwithstanding the pitfalls, Intelligent Part Numbering is still alive. And customers still request such schemes.

It is uncertain why this is the case, but it may be a consequence of our human capacity for irrationality. Its not that such a response is confined to Part Numbering – consider many other fields (politics, medicine, economics etc.). But this is a PLM blog and not a psychology dissertation.

So, is there a middle road?

Hybrid Numbering

If some level of intelligent numbering is required (or insisted upon), then there is a middle way. As below:

• Suffix or Prefix of the number has some significance
• This Suffix or Prefix is selected from a predefined list
• Other properties are moved to attributes
• Balance of the number is sequentially generated

Of course, this is not ideal, and the purists will be calling foul. But it does at least represent a compromise. After all, ultimately all of our affairs are compromises.

What are your thoughts?