Product Lifecycle Management (PLM) systems and Enterprise Resource Planning (ERP) systems have changed the landscape when it comes to part numbering. Many companies are saddled with different types of approaches when it comes to part numbering. These systems are meant to facilitate collaboration in engineering and working with the supply chain. Most of these systems have been rendered somewhat obsolete by today's PLM and ERP solutions like Agile and EBS from Oracle or Windchill from PTC. This blog will review some of the history behind these part numbering schemas and help us better understand how to optimize product development environments to work with today's technology.

Before we jump off into the history of part numbering systems and the impetus behind intelligent part numbers lets first consider the requirements for a numbering system. The goal of any part numbering system is to uniquely identify a component. Having the ability to accurately identify a component in an unambiguous manner affords correct assembly, testing, and maintenance for a given product. Further, when dealing with any properly released product, a component whose form fit or function (F3) deviates from the original released design must be assigned a new part number. A new part number allows an easily distinguishable change within a bill of material (BOM) and denotes a 'non-interchangeable' modification.

So where did part numbering systems begin and how did they develop? Certainly the need was there since the industrial revolution ushered in the means for mass production and products evolved to be more and more complex. But when did the true inception of part numbering come in to existence? The earliest reference I could find was nearly 80 years ago. In 1932 an ambitious project was conducted by a small group of students headed by Wallace Flint at the Harvard University Graduate School of Business Administration. The project proposed that customers select desired merchandise from a catalog by removing corresponding punched cards from the catalog. These punched cards were then handed to a checker who placed the cards into a reader. The system then pulled the merchandise automatically from the storeroom and delivered it to the checkout counter. A complete customer bill was produced and inventory records were updated.

So when did the idea of Intelligent Part Numbering come into existence? I could find no clear inception point in history but significant part numbering schema's (intelligent part numbers) started to appear a little over 50 years ago. This was an outgrowth of a basic requirement that unstructured information was very difficult to find and maintain. This made it necessary to define part and document identifiers with search related structures and attributes. Accordingly, the absence of manufacturing systems to govern processes forced the requirement for recognizable component identifiers.

We will review the variations of part numbering schemas that are most used today. Basically there are three:

• Intelligent - all characters represent some attribute of a part or document (i.e. material, color, dimensions, finish, resistance, etc…)
• Semi-Intelligent - some of the characters are significant but the balance is random. Useful for categorization (commodity codes), classification, or representation of families.
• Random - all characters are completely random

The first two systems require some level of design and as such must take into account system capabilities and other constraints. The random part number only requires a random number generator that assures uniqueness.

For Intelligent Part Numbering most systems allow manual input but typically do not have an interface to define or guide the numbering sequence. The reason is the possible combination and permutations of variables governing an intelligent schema are virtually impossible to build an administrative interface to capture the knowledge and logic required. If you knew the exact requirements for the schema a custom interface could be developed but at a cost and it would not likely be scalable to other customers. When dealing with the human element and relying on a manual entry the errors due to the number of characters in the part number increases exponentially as the length grows. There is a reason that phone numbers are 7 characters in length (not including the area code). Studies have found that 7 numbers is typically the limit for short term memory in most people. In addition clerical accuracy and speed should be considered. The longer the sequence of numbers the slower the input. Also it is human nature to have a higher incidence of transposing numbers when working with multiple sets.

For Semi-Intelligent Part Numbering most systems now allow for static prefixing and post-fixing (or suffixing) of characters for categorizing and classifying part numbers. The most common is a prefixing commodity code followed by a random sequence of numbers used by many companies. This allows users to select from a list of predefined prefixed and/or post-fixed numbers where the balance of the number is randomly generated sequentially.

Finally for Random Part Numbering most systems have random number generation allowing you to define the length of characters. Because of the random number assuring uniqueness of component ID's there is no need for design. However, as stated for intelligent part numbers the factors of human memory should be considered during the configuration of the system.

In the next blog we will look at the good the bad and the ugly of each system described above.

[clear-line]