The National Stock Number (NSN) System - Part I

 Why NSNs? As I mentioned in previous articles, almost every product sold in the world, at any store, has a reference number. This number can be a Universal Product Code (UPC), a Unique Product Identifier (UPI), a Model Number, a Manufacturer Part Number, Drawing Number, SKU, UPC, EAN, ISBN, etc.

Manufacturer Part Numbers (MPN) are assigned by each manufacturer using their own criteria. There is no enforced standard for the MPN, although there are some industry conventions.

For example, one manufacturer might produce an item and assign it part number A123, while another manufacturer might produce a completely different item and also assign the same part number. Additionally, different manufacturers implement different business rules regarding the significance of the part number itself. For example, the use of spaces or dashes might be significant to one manufacturer but completely irrelevant to another.

For this reason, the US, NATO, and other governments adopted a means of assigning the same catalog number to items that have common technical characteristics. This is done using the NSN.

Let's take for example, the rivet pictured above.  It has the following Reference Numbers from different manufacturers or organizations that all refer to the same item:

AF3213-6-02                   
AF3213-6-2                    
CR3213-6-02                   
CR3213-6-2                    
HC3213-6-02                   
HC3213-6-2                    
MIL-R-7885/2                  
M7885/2-6-02                  
NAS9301BNS-6-02               
10111712

So, what the federal logistics information system does is to assign one stock number to designate all those part numbers that share the same "form, fit, function".

That rivet is assigned NSN 5320-01-435-5075 because all those part numbers share the same technical characteristics, such as size, material, type, etc.

The catalog of all these stock numbers is part of the Federal Logistics Information System (FLIS) maintained by the Defense Logistics Agency (DLA) in Battle Creek, Michigan.

Mandates and pragmatism

Although it is not possible for me to trace this particular branch of my family tree, our family has always assumed that we have some Taino ancestors. I grew up listening to my Dad tell me the only Taino tale that I know and I often think about it when faced with seemingly impossible situations.  Lately, I have thought about it more.

Since I like to keep food on my table, I have to be careful what I write about. Therefore, for this article, I will share my Dad's tale of a Taino Cacique (chiefstain) and let your imagination do the rest.

Mao, the young Cacique of his Taino tribe and his new wife are expecting their first child any day. Mao is loved by his people and he has brought them prosperity. As it is tradition, Mao and his wife ask the tribe's healer and spiritual leader to visit them prior to the child's arrival.

The healer arrives at Mao's hut and, when crossing the threshold, notices the Cacique's spear above the door.

You see, Mao's father, Atahualpa, had pronounced an edict that every hut must display a simbol of the household's occupation, similar to how the surnames "Hunter" and "Carpenter" came to be in Anglo culture. Mao, being the Cacique, proudly displayed above his door the ceremonial spear as the chief warrior of his tribe.

The healer, however, is not pleased. He begins to chant loudly and drops to his knees crying.

Confused, Mao asks the healer what is wrong.

"Your child will die right here", he says.

"Oh no!" said Mao as his wife runs in to learn the terrible news herself.

Everyone is devastated.

The healer begins to get a better vision and explains to the distraught couple, "just before your son learns to walk, he will be playing by the door and your spear will fall and pierce his head".

The news could not be worse.

Just as the couple begins to accept their awful future, Mao's mother pays a visit.

"What is wrong?", she asks and Mao shares the bad news.

"So the problem is that my grandson will die because that spear will fall and kill him. That spear that is only there because of my late husband's edict?" says the matriarch.

"Yes", responds Mao between sobs.

"Bring down the spear and do away with the edict - it is your own mandate that is creating the problem my son".

Product Identifiers and Catalog Systems

In this article, I will attempt to set up some of the background for a series of articles for an eventual discussion on unique item management.

Part Numbers

The part number system is one of the most misunderstood parts of a system of material identification. It is almost entirely based on best practices, although there are some rules for certain types of material. Perhaps the best article I have seen explaining part numbers is the Wikipedia page at https://en.wikipedia.org/wiki/Part_number, which is a miracle it ever got written so well with exactly zero references.

Example UPC Lookup Results

Universal Product Cataloging Systems

Beyond the wild wild west of the part number system, attempts have been made by industry to standardize product identification. What is interesting is that we are exposed to these every single day. Even if a product has a part number, if it is sold somewhere, it is almost guaranteed to have a universal product number or equivalent code.

One example is the ISBN system used for books and other media. ISBNs are nationally assigned, meaning that each country has an authority that manages them. You can look up an ISBN freely many different ways, such as via http://www.isbnsearch.org/. If you are a developer, there are Application Programming Interfaces (APIs) such as Amazon Web Services that allow seamless integration with your application. There are also databases that participate in the open-data initiative, such as the Google Books API (https://developers.google.com/books/docs/v1/using).

The Universal Product Code (UPC) is probably the identifier that we see the most. It is in almost every product we buy. The UPC is assigned by the product manufacturer or designer, although that is not always necessarily the case since any vendor with a vendor code can procure a UPC. It is a catalog identifier at the product level, meaning, for example, that a UPC for a 16 oz. bottle of Nestea Pomegranate Passion Iced Tea would be the same UPC found on every single bottle of the same product, regardless of what store is selling it. The UPC system is managed by an international non-profit organization called GS1. Just like the ISBN, the UPC can be freely searched via many databases such as the following: https://www.barcodelookup.com/.

There is also a system called EAN, which is the European Article Numbering system. The EAN is also managed by GS1 and it is functionally used in an identical way to the UPC.

You might have also heard of GTIN (https://en.wikipedia.org/wiki/Global_Trade_Item_Number), which is the Global Trade Item Numbering system. This is an umbrella construct, also managed by GS1, that includes the UPC, EAN, and other variations of universal product catalog identifiers.

There you have it. Just about every product that is sold in stores throughout the world is covered by an ISBN, UPC, EAN, or GTIN, and all those identifiers are publicly and freely searchable.

Amazon Web Services (AWS) Product Lookup API

SKUs

Now we get down to the retail level. Stock Keeping Unit (SKU) numbers are routinely confused with other product catalog identifier numbers. The key thing to keep in mind is that SKUs are internal to a supplier and do not mean anything to anyone else other than the entity managing the stock. However, the confusion is so prevalent that many retailers (particularly when interfacing with customers) use SKUs interchangeably as product identifiers. The Amazon Web Services (AWS) API even allows developers to search for SKUs. Amazon being the giant that they are, a SKU search in their API is pretty close to a universal SKU search.


Next article: the Federal Stock Number system.

Intelligent Products

I confess that I like reading research papers. It is an excellent way to stay up to date on the profession and also a good way to understand what is a fashionable trend (i.e. flavor of the day) vs. what is a trend toward progress. I always liked to read copiously, but it wasn't until I started working on a doctoral degree that I learned how to read efficiently in order to achieve a research goal.

When an interesting research topic comes to mind, I usually start by searching for literature in Google Scholar (https://scholar.google.com/), which many people surprisingly don't know it exists. But I digress.

I have been reading, off and on, for years about unique item management. This goes back to my brief time as a civil service employee in 2007 where I ended up working on a project that would have required individual parts in a submarine to be managed at the item level in order to enable remote management.

The Unique Logistics at  Ikea

At the time, one of the striking things that I found as I was researching the topic of managing items by unique identifier rather than by catalog identifier was how much of the research was coming from Finland and Norway.

It turns out that Scandinavia is a hotbed of logistics innovation for some reason. One of my go-to papers when I need to introduce someone to the possibilities is titled "Intelligent Products: a Survey" (Meyer, Gerben G., Kary Främling, and Jan Holmström. "Intelligent products: A survey." Computers in industry 60.3 (2009): 137-148). I highly recommend it and, if you like it, I also recommend that you follow Holmström's work.

 Back in 2007-2008, when I was working on that project, the idea was thought to be so far out that it wasn't even given the same level of respect as magic and was only discussed in hushed tones in very small groups. So, 8 years later, most of my appliances, and even my doorbell and light bulbs are on the internet. People have actually accepted this technology quite easily, by the way - much more so than the logistics community.

I have a love-hate relationship with research from the Gartner Group, mostly because I usually have to pay for it and I am too cheap. However, many years ago, when working on my Master's thesis, I read a research report titled "Data in its Place" in System Software Architectures Doc. No. R-401-107. It introduced me to the concept of the "Distributed Data Paradox". To put it in my own words, it means that there exists a tug of war between business functionality and technology, as information throughput improves with advances in technology. The paper goes on to suggest that researchers and practitioners need to understand when sufficient technological advancement has taken place to prompt a "reconceptualization" of a business function.

So perhaps this is the right time to consider the idea of unique item management instead of stock keeping unit (SKU) management, particularly in DoD logistics.  It may not fit the bill in every case, but I believe it is worth exploring.

Probabilistic Counting and RFID

Probabilistic Counting Methods

Lately, thanks to some questions that I have received, I have been thinking more about the feasibility of RFID for inventory control.

Back in 2004, I led a pilot test on the use of RFID for inventory management. The technology performed well; however, we were left with two clear challenges: 1) tag read reliability, and 2) determining localization. Back then, one of our recommendations was that inventory accuracy policy should be re-examined and adapted to counting methods where the material is not visible by the "counter". We also suggested a "time window" policy, where detection of an item during a set time window, even if detection is not successful at other times during that same window, would constitute a successful count. I could go on, but I will concentrate today on the counting method.

How do we count?
We don't often think about this, particularly those of us who have been counting material for decades.  DoD policy set in DLM 4000.25 assumes a deterministic method of counting (as opposed to probabilistic). See, when we count items of inventory, we physically contact the item (visually or physically) and count expecting no error in our count method. It is not until we validate our count against the inventory record that we may conduct a recount if the numbers do not match.

Using a jar of marbles as an example. A deterministic model would have the counter empty the jar and count each marble individually in order to come up with a number of marbles.

Trivia: There is a DLM 4000.25 which is in two volumes, the second of which is the Supply Procedures Manual. There is also a DLM 4000.25-M which is a separate document called "MILSTRAP Manual" which is essentially the same content as Vol II of the first document, but the chapter numbers are off by one. This is the cause of much confusion.

Other ways to count.
There are, of course, other ways to count the marbles. Let's say that we know that each marble weights approximately 10 grams and that the jar weights just over 1 Kg. Then we can infer that there are "probably" 100 marbles in the jar. This is a probabilistic method, and one that we could use even if we could not see what was in the jar at all, as long as information about the contents of the jar was available to us:

The DLM 4000.25 actually implements a simple probabilistic counting method for certain items, allowing for count tolerance of as much as 10% for certain items (see Table C6.T1 of DLM 4000.25 or Table C7.T1 of DLM4000.25-M).

What does this have to do with RFID?
Back when I conducted my RFID pilot, the best tag read rate we experienced was 94% under near-perfect conditions. The average read rate was 74%.

That does not even answer the question about location. Were these items in their correct bin location per the accountable system? Also, the system only detected the presence of a tag, not that the material associated with the tag was actually there and in the proper condition.

There are RFID systems that implement localization by means of triangulation or "smart bins". However, those are still probabilistic in nature.

These are the kind of considerations that make RFID unacceptable as a valid means of certifying accountable record inventories per DoD policy.

What do we need to move RFID adoption forward?
The technology is there and, whether functionality improves or not, the fact remains that the count method represents a significant procedural shift at DoD. What DoD needs now is a consistent policy that takes into account probabilistic counting methods such as RFID tag detection or other means of automated counting. I will be writing a more extensive paper on the subject.

P.S. The other elephant in the room is IUID. That will be another article.