Drawing numbering system 14

[Adrian2]

Dear Folks;

Back in the old days, a number for each drawing sheet was quite sufficient. Now with seperate drawing, part and assembly files I was wondering if anyone can share a good numbering/recording system for keeping track of drawings.

At the moment I record my drawing, part and assembly numbers in a spreadsheet which allows me to search or sort my work for particular information. But of course this is not an optimum solution for large numbers of drawing files.

Best Regards

Adrian D

 

[JJCPE]

Food for Thought

When I was a new Engineering grad, wet behind the ears, I was tasked to define a significant numbering system for screws. So I started defining the categories, subcategories, dash numbers and the like. I wrote a simple one page description of the system, and was pleased with the result. I could design a product, and rather than go search for the part number of a screw, could just define it and know what the appropriate number was.

One of the categories was for ‘special’ screws, ones that wouldn’t fit into the ‘standard’ categories. Various thumbscrews were good examples of specials.

The other Engineers and Engineering Managers were impressed with my work, and I was pretty pleased with myself.

Of course I was then asked to seek out all of the various screws currently in use and write an Engineering Change Notice to convert them from their existing part numbers to their new, improved part numbers.

Turned out we had about 27 different screws in our inventory, 7 of which were ‘special’.

Then I calculated the number of numbers the system used.

The result was over 7 MILLION.

The total number of different parts in our inventory was less than 4,000.

A list of existing screws could fit on the same page as the description of the numbering system, and it had the advantage of telling us what we already had in our inventory.

Years later, at another company going through the conversion of the inventory control system from one software package to another, the IT department had a high speed printer that created a set of books on a monthly basis which listed the inventory sorted by part number, by description, and by ‘type’ code (keywords).

This became a multi-dimensional ‘index’ of what we already had. If you knew the part number, you could locate the description. If you knew what you wanted, you could find out if it already existed.

The point is this:

significant numbering systems tell you what could be, not what is.

The key to part numbers is to find what you have (and be able to add to that). There are many ways to create and use indexes, and the use of computer systems has only made indexes easier to create and use.

I could go on and describe my experience at other companies, but the end result is that a simple insignificant number log, good naming conventions (keywords), matched up with powerful search capabilities is the heart of any good numbering system. The number log keeps the process of assigning a number simple, and the keywords and search capabilities make the system simple to use.

A significant system complicates the search (because you have to remember, or search for all of the rules before you can use it), and complicates the number assignment (because you have to find the right log).

With an insignificant system, you will wind up with situations where different parts share the same set of keywords, but they will have different part numbers, and a where used search will help you distinguish between them (or looking at their models or drawings based on part number).

I’m currently developing a medical device, and I’m at the stage of estimating the number of parts the device will have. The document control department has ‘given’ me a block of numbers that is about 150% larger than I will eventually need. I’m ‘cherry picking’ numbers I’ll use for assemblies, custom parts, and off the shelf parts. This allows me to use what I call a ‘soft’ significant system. It has no significance beyond my decision to choose which number I’ll use for a given part, assembly, etc.

When the product is released, the only significance is the serendipity that all ‘new’ numbers for the product fall within the range of the block of numbers ‘given’ me. They’ll all share the same first set of digits. The number that happens to have the largest number of zeros in it will just happen to be the top assembly, which will lead to the Bill of Materials, which is just another useful index.

The last bit of advice is that you should include both the part number and keywords in the electronic file name. How to handle revisions and versions is trickier, but if you plan to use a PDM system, keep them out of the file name and deal with it some other way.

 

[SnowCrash]

This is an interesting approach. What I have noticed about the various systems in place is that there are two part numbers involved.
One is a catalogue number for any given stock part which the sales rep might use to describe it for an order. It is familiar to anyone who evere speced a valve or cylinder for example.
The other is a sequential build number for that part which is simply a factory control number used to provide manufacturing instructions.
Catalogue numbers are unique for a given design as each one produces a discreet design. Build numbers can be duplicates but it does not matter.
A screw catalogue number might have only certain designs associated with build numbers, so not every possible screw gets documented, only the ones needed in production.
You could call this a part number and use it for the file name as well. Then you use custom properties to link to the catalogue number.
This is just one of many ways to control production.
--
Crashj

 

[kcarpenter]

We use a 7-digit part number scheme where I work. We also name drawings and the associated parts/assemblies the same, i.e. 1234567.SLDXXX is the file name for the drawing and prt/asm. We also use a part number generator (in MS Access) which allows us to reserve or assign single part numbers or in bulk. Thinking more about the problem being discussed I see two possible solutions if you were to use a simple non-significant numbering scheme as we do. 1.) Use a simple number generator like we do with fields for part number, description, supplier, etc. Whatever you want to store and once it's in the database you can then query the database for the information you're looking for. 2.) Enter all your necessary information into the SWX drawing and then create a macro to capture it and save it in a database. Either of these solutions should be easy to do and each has its pros and cons.

In any case I think you'll find that you're trying to do something with a tool that it's not ment to do...be an ERP/PDM/PLM system. The information is stored in the file itself with no means of easily querying for it, which is why you’re trying to create an elaborate naming scheme for your SWX files. The problem isn’t how to name the drawings/parts/assemblies, it’s how do I know what I’m looking at/for. I honestly think you're going to want to have this information outside of the SWX files themselves and let some other software control the myriad of specifics on particular drawings/parts/assemblies.

K.I.S.S. – Keep It Simple Stupid (one of our VP’s favorite lines).

PS - we do use SmarTeam.


Kevin Carpenter
CAD Systems Specialist
Invacare Corp.

 

[JJCPE]

Kevin,

What you describe (part number generator) is something I’ve always wanted to do.

Unfortunately I’ve always worked within a company that didn’t have the smarts to go that route.

I did however get involved in a project where I had to learn how to build an Access database, and then move data in and out with Visual Basic.

You are so right when you said that “I think you’ll find that you’re trying to do something with a tool that it’s not meant to do. . .”

If all you are doing is create an indexing system to find stuff (and not keep track of all the cross-references) then it would be fine. Way back in my AutoCad/AutoLisp days, I built some powerful tools that prompted the user to fill in all of the title block information when ever a drawing with a border was started.

Back in those days, all you had were drawings. I had an epihiny when I realized that a CAD file (with ‘zero-thickness’ lines positioned with 13 digit accuracy) was an idealized model of a part.

With Solid Modeling, the idealized model of a part has become so much more complicated. It’s a better model, but it isn’t a drawing, and if more than one part is included, it’s no longer an idealized model of a part; it’s an idealized model of the relationships of parts.

What I think we need is something similar to the ‘title block’ approach. When ever a new SolidWorks file is created, provide the user (if they want) with a set of attributes to define/fill out. Then make use of those attributes by building an index.

ERP/PDM/PLM systems do many, many things. When I was in school, I worked for a company that developed a document management system called ARMS – Automated Records Management System. The system involved many, many, technologies – from computers to networks to storage systems to imaging systems (it was one of the reasons Nuclear Power Plants cost so much!). I will never forget the day that one of the architects of the system waved me over and showed me a document. It was a simple description of an index. This system architect drove home to me that fact that this very complicated system pivoted around the index. It was the key, he said, for a user in Baltimore, Maryland, being able to call up an image of a memo generated by an inspector in Diablo Canyon, California, describing the results of an inspection of a weld in a pipe, stored in some unknown third location..

The system, as sophisticated as it was in its day, could not function without a simple means for everyone using it to find what they were looking for. The system handled about 300 million records!
K.I.S.S. also stands for – Keep It Simple, Smart

The moral is that it doesn’t matter if you are looking at a memo, spreadsheet, model, part or assembly (or an image drawn on the back of an envelope). They are all records, and they all should have one thing in common – a unique identifier that can be used as the pivot point for an index.

John

 

[rokahn]

I'm setting up PDM/Works following the advice from this and a few other threads which seems most appropriate for our company. I'd appreciate it if someone would clarify a few points.

1. If we use PDM/Works, should we also have this external database to query for part numbers or does PDM/Works take on this task?

2. Does anyone have a macro they'd be willing to share to automatically export fields from SW/PDMW to an external database?

kcarpenter wrote:
I see two possible solutions if you were to use a simple non-significant numbering scheme as we do. 1.) Use a simple number generator like we do with fields for part number, description, supplier, etc. Whatever you want to store and once it's in the database you can then query the database for the information you're looking for. 2.) Enter all your necessary information into the SWX drawing and then create a macro to capture it and save it in a database.

 

[JNR]

I would share our fabulous part/drawing numbering system with you, but it has been stinking up the place for over 20 years now and I still get the technicolor yawns over it occasionally.

What I am really getting at is you are on the right track spending some effort to get it right first time. Stupid little issues with this can really leave teeth marks in the sit-upon later. I encourage you to do some test simulations before you commit.

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

There are only 10 types of people in the world - those who understand binary and those who don't.

 

[macduff]

All,
My 2 cents is that you should never build logic into a number. A number is just a number. I know some of you might disagree with that, but that's how I feel. Just like kcarpenter said, K-I-S-S. I also know a lot of us our lock into the part sysems at work, but if you are starting from scratch, keep it simple.

Now building logic into the way to keep track of the drawings, parts and assemblies.....that's a different story. I spent a lot of hours re-doing the custom properties in our SW templates. I even went as far as having a custom properties for each and every "Next Assy" and "Used On" blocks in our drawings. That way you can search where that part is being used in our products using PDMWorks.

I know that a lot of you do not have a PDM, PLM or Smart Team, but you have to plan for the future. Keep a basic system upfront and have everyone follow it. Sooner or later you'll be on a somekind of PDM system. This will allow you to have a smoother transition when you convert your files over.

One more thing....never use the revision letter of a drawing, part or an assembly in your filename. Your asking for trouble down the road. Food for thought...

Best,

macduff

 

[JNR]

I agree with macduff. A "non-tracking" partnumbering system is best in the end. Ours is a hybrid!!! Problems you encounter with tracking numbering systems are you tend to you run out of numbers much quicker. Also when new "classes" of parts show up or worse - you decide to break some exisiting ones down to fine granulation - lots of confusions start. And when someone uses the wrong part number coding you can never find the part again!! (Believe me they will - and you will have gray areas too.)

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

There are only 10 types of people in the world - those who understand binary and those who don't.

 

[PakRat]

Great thread on an ever-present problem....Now for my 2-cents.

I am of the opinion that the less information that is encoded in a part/drawing number, the more a company has to rely on "tribal knowledge" to identifty a part/drawing that is known to exist, but whose timeliness has faded. In extreme cases, it is easier to (reluctantly) recreate the part than invest the time to find it. As absurd as it sounds, I have seen this happen.

And to take things a step further...When you couple a random part number system with no nomenclature standard, terminal velocity is quickly obtained.

I do not, however, believe that a part/drawing number system should contain infinite detail. Just enough so that: given a part number, the system and sub-system should be readily identifiable, and, knowing the system/sub-system to which a part is related, the search parameters are significantly narrow enough to ensure the part will be quickly found.

 

[kcarpenter]

PakRat...I have to ask when is enough enough (as in descriptive part numbers)? I mean where do you draw the line in making your part number descriptive enough to allow the users to quickly identify what they're looking for/at?

We have 850+ different types of screws (then throw in nuts, washers, etc.) we use at my company. How does someone differentiate 60+ 1/4-XX screws? Head type, length, coatings, material, UNC, UNF, etc. It's impossible to create a numbering scheme that would allow a person to quick identify a specific type of fastener they're looking for. M$ would have to quadruple their current 255 character limit on filenames.

I know you're thinking, "I said, 'I do not, however, believe that a part/drawing number system should contain infinite detail.'" But what good is it if I did narrow it down to those 60+ screws? I still need to open up every one of them to find the one I need. Also, you still need to train new personnel on the "system/sub-system". Hence the use of a part number generator/database. Anyone can take a simple class or read a book and create a useful MS Access generator/database. Point previously said new person to the generator/database and search away.

My point is still the same, and that is you almost certainly need to store this information outside of the file itself.


Kevin Carpenter
CAD Systems Specialist
Invacare Corp.

 

[TZellers]

I can't resist. Significant part numbers are a nightmare, plain and simple. I would avoid it at all costs. This is such a long thread, but someone pointed out that computers are dominant these days and you can have an non-significant p/n and a significant drawing number, material code and so forth. Why? Any p/n that is over 10 characters promotes error, as stated before. Also, why put your company at risk by giving a customer a BOM with significant p/n? For example, your p/n is E265B70 - and description is o-ring. Well, that must be a -265 o-ring of Buna 70 duro. Customer buys that from someone else, it fails, and he expects you to cover costs? HE may not do it on purpose, but it happens. Most customers shouldn't know more than they need to.

For those really interested in this, do a search for a Technical Research Report entitled "Non-Significant, Self-Validated Part Identification Numbers" by G. Harhalakis, M.E. Bohse and B.J. Davies (ISR T.R. 86-15). This was published by the Institute For Systems Research and their website is

http://www.isr.umd.edu

If you want to hear nightmares about significant p/n, let me know. Also, sequential drawing numbers are a joke (a habit it seems many older companies used).

 

[PakRat]

kcarpenter:

My answer is no answer. You have to define what "enough detail" is in your specific case. I was being purposefully vague because this is a very subjective topic, and, in the end, a sucessful system is the one that meets the company's minimum part/drawing number requirements without tying the hands of future generations.

I will add one other consideration. One must consider the needs of the company's ERP/MRP system. For most, this is the life's-blood of the company that allows for meaningful production planning and inventory control. The part numbering system must consider that information is required by the MRP system. A poorly implemented MRP stsem is an ugly thing.

Case in point...My last company used a part/drawing numbering system that made no distinction between the same part produced from different materials (galvanived or cold rolled), size (18GA or 20GA) or color. Of course inventory was a mess and when they tried to turn on the MRP system, maximun chaos was achieved in short order. A simple statement of the obvious (to most) is that different things can't have the same name.

And just to complete the picture, their numbering system was 40yxxx where x indicated the sheet size of the drawing and yyy was a sequential number starting at 1 and headed to infinity. Furthermore, any dash numbers (opposite hands and variations) were tabularized on the face of the drawing and suffixed to the root drawing number.

Faced with these problems, I designed a new numbering system that worked with the 15 digit limitation of the MRP system. It followed the pattern: PP-MMxxx-DDD-ZZ where PP indicated the system (about 25 standard systems), MM indicated the subsystem (average of 50), xxx was a sequentially assigned number with 000 reserved for the top-level product drawings. DDD indicated the dash numbers and ZZ carried the material/finish code.

I know this may seem overly complex, but when you consider that the company was extremely vertically integrated (coils in the back door, finished product out the front), this system met the following needs:

- Could identify WIP in various stages.
- Could handle the standard product lines.
- Could handle minor variations to the standard product line (custom colors).
- Could handle specials (I won't bore you with those details).
- Met the requirements of the MRP system: could handle a highly-optioned product line and drive from the configurator to a dedicated BOM.
- Allowed all work-content to be captured (very important).

I doubt this numbering system would work for anyone else. I detail it simply to demonstrate that its is the result of examining the the company's primary needs.

Likewise, how you subdivide your 850+ collection of hardware is something I would be ill-qualified to address. Keep it simple, but keep it in balance. Also, don't worry about how everyone will adapt to a new numbering system. If they see that it makes sense and is a positive for the company, they will adapt.

Good luck.

 

[tluxon]

...never use the revision letter of a drawing, part or an assembly in your filename. Your asking for trouble down the road.

I remember how nice it was working for companies who spent time outside of quadrant one (urgent, important), but I am with a company now that doesn't encourage taking the time to sharpen the saw because there's too much wood to cut. Believe it or not, we don't even have a MRP system here, which I find nearly impossible to work without.

We don't (and may never) have the luxury of a PDM system and 98% of the drawings here are old AutoCAD drawings that aren't even to scale (original regime scaled the parts to fit the sheet rather than scale the sheet to fit the parts - another result of lack of training). When I was brought in here just over 3 years ago to be the Design Engineer, I also brought in SolidWorks, but it's almost exclusively used for new product development - not the 100's of constantly evolving old designs.

Even though I have a ton of background with databases (compiled many with Clipper almost 20 years ago), I really don't have the time to create and/or maintain one to keep track of the SW file properties like Author, Description, Revision, and Product. What I've done instead is an ad-hoc system where I DO include the part number, revision, and very basic description in a long filename for parts and assemblies, while using just the basic number and revision for drawings.

While I'll concur that it's frequently on the virge of getting out of hand and difficult to manage with multiple users, I really haven't figured out a better way to keep track of revision history and design evolvement than including the revision in the filename. In the design process, I must be able to branch out from a given base point of each part into several variations simultaneously before settling on a single "upgrade" path. I'm wide open for suggestions on how to accomplish this without including differentiating information in my filenames.

Thanks,

Tim

 

[Adrian2]

Dear Folks;

I have read with great interest all the comments people have made about part numbering systems. I wanted to share some more information here because many people have brought up MRP systems in addition to PDM and I think we’ve got 3 separate threads going at once. Let’s try an make it 4.

One thing that cannot be overlooked is the usefulness of a good standardized bill of material. Even without an MRP system, a good bill can eliminate much confusion. Back when I was involved with designing cartoners and other packaging machinery, our company had a good standard bill of material. Given a customer order, a designer could practically build an entire cartoner by checking off standard parts lists in a bill of material. Should he need a new assembly he would create a new parts list, which would be tacked on to all the existing checked off lists under the machine number.

Every day designers would give me their new parts lists and I would update the standard bill of material when time permitted. After many lists were created and indexed it became obvious where duplication was occurring and the designers were encouraged to use existing parts lists. This made it easy to use and reuse standard parts rather than creating something new. A word processor was the only office automation used to maintain the bill.

The bill of material is what drives every MRP system. It is said that in order to achieve proper function, bill of material accuracy must exceed 98%in an MRP system. Any less and you just get scads of work orders printed out for stuff you don’t need. Personally, I feel the idea of a unique, non-significant number is still best. This is also true for the basic identifier of any part in an MRP system.

Many have mentioned the use of numbering systems with fasteners. MRP pundits will tell you that nuts and bolts, o-rings and clevis pins don’t need to be inventoried or classified with an MRP system. Regardless of whether they have differing materials or head types they do not represent a significant dollar value when compared to high value cast and machined parts. Nor do they usually have very long lead times. An estimator, purchasing agent or junior draftsman can review assemblies released to the shop for net requirements and have them in stock long before the higher level assemblies require them.

One way to control fasteners is with just-in-time kanban based inventory control such as the 2 bin system. 2 bins of screws are kept one behind the other in a rack, both are shrink wrapped. The moment the back bin is brought out and the shrink wrap broken, that’s the time to reorder. Only one careful individual is required to prevent stock outs. No one has to worry about getting min/max quantities and reorder points into the MRP system.

I still think the best way to proceed is with unique non-significant numbers. It’s been 17 years now and I can still remember a lot of the part numbers out of the old bill of material. Cant say the same about some of the vendors Air Cylinder or Valve part numbers though….. was it 75-AX-fghte=28vac-!!!^&(*****) or something like that ?


Best regards,

Adrian Dunevein

http://www.aaadrafting.com

“K.B.O.” – Winston Churchill

 

[dogarila]

TZellers, I do want to hear nightmares about significant part numbers?

 

[Standing]

Netshop21,
I agree with Adrian.
We used significant part numbers until our last layoff. Now we have no one to maintain the system. We have switch to non-significant part numbers. I have worked for over 10 companies in the past. Some companies had significant part numbers and some without significant part numbers.
I was looking for a filter that goes into the exhaust port on an air value. I found it in electrical filters after many minutes. Now we have 3 places to look. Miscellaneous, air filters and electrical filters. By the way our miscellaneous numbers are full. I believe the reason that significant part numbers are such a problem is the fact that the managers almost always hire a documentation clerk to assign part numbers. Only one place I worked hired a component engineer, all the rest could not tell the difference between a diode and a resistor. The solution from the management was to make rules. The engineers are responsible to make sure there are no duplications. Just try to get 9 different engineers and designers to standardize names. To get a significant part number system to work, you have to teach 9 different engineers the workings of the system, not just one person. Management makes one component engineer responsible. He quits after 6 years, the new guy does not think like the last one. What to do? Non-significant part numbers is the way to go, anyone can put the next part at the end of the list. With computers we can search for manufactures numbers for duplicates. We can sort numbers anyway we need too.


Bradley

 

[dogarila]

Bradley, I agree with you, non-significant part numbers is the way to go. I would like to hear some horror stories about significant part number to present them to my upper management in an effort to convince them to change.

We are an automation company and the numbering system we are using right now is:

1234.5.66.77.88.999

where: 1234 - project number
5 - system number
66 - cell number
77 - station number
88 - subassy number
999 - detail number
All subassy numbers have detail number 000
All stations numbers have subassy number 00 and detail number 000 and so on.

 

[Standing]

netshop21,
If I may quote from the young kids at our church; “That’s so old school”. The reason I say that, is because your numbering system is very similar to the system I used when I first started in the 1970’s. Before computers, we had to do something. The system worked great until we started using parts from project 1234 on project 4321and on project 5596 and on project 5569. Then we made a rule all parts on a project had to have the project number. Which meant 4 drawings. We ended up with 4 exact same plates with different numbers. Then we found a mistake in a hole location. We changed 3 but forgot the fourth plate. A year later we order 10 of the fourth plate and found that we had to rework the parts. Back then we had no choose. Today we have a choose.

Bradley

 

[kcarpenter]

OK....I'm going to share our PN generator to those who wish to see/use it as an EXAMPLE, nothing more. It is a non-significant generator which can create/store PN, Descrip., ECN num., user name. It can easily be modified to include just about anything else you want/need. Show it your bosses, co-workers, etc.

If you would like me to email you the MS Access 2000 DB (quit small once the data was removed) please send me an email at: kcarpenter_removespam_@invacare.com

I need to do something useful here to help.

Kevin Carpenter
CAD Systems Specialist
Invacare Corp.

 

[TZellers]

Our part numbering system is primarily 7 digits, but can be as long as 13. Want to see the madness? Here goes:

Digit 1 is the Pump Type and Material. So, for our model 1000 in Noryl, the digit is 1. For our model 3000 in Noryl is digit 3. 1 for 1000 and 3 for 3000 – makes sense, right? Well then we added polypro as a material for the 1000. So, what to do. How about a 2 – that’s not taken. If metric ports were required and Noryl was used, let’s use M. What happens if we want metric and polypro? Haven’t done that YET.

Digit 2 is the Impeller. We have standard numbered impellers. So a #1 impeller would be represented by 1 in this slot. 2 for 2, 3 for 3 and so on. The material of the impeller is supposed to follow the pump material, but polypro doesn’t chemically bond well, so we often need to mix and match (and no, there is no digit for that!). Now, if the customer wants a trimmed impeller, that would require a “T” to follow the impeller digit. Whoops, that slips over into the 3rd slot, but let’s forget that for a moment. If the customer wants a semi-open impeller, use add an ‘S’ after the impeller digit. If they wanted a trimmed semi-open, it would follow that a TS or ST would be added. #9 is the catch all special impeller. Now which digit are we on again?

Digit 3 is the Motor HP and # of poles. It goes from A (1/6HP and 2 pole) to S (3HP 4 pole) and even a T (1/10 HP 2 pole). There is also a 9 for special and a 0 (zero) for pump end only (or if supplied with a power frame and no motor). There’s also a spot in the 9th spot, but we’re jumping ahead.

This goes on and on. The elastomers are supposed to follow the mechanical seal spec, but don’t always. What started as an easy way (pre-computers, btw) to work with p/n’s became a nightmare to figure out. Plus, why give your competitors a keycode to decifer your part structure? The moral of the story: you can never guess where your company is going to wind up, so why try with significant p/n structures?

I worked for a division of I-R that used the non-significant numbering system illustrated in the white paper I listed before. It was IDP, a pump company. They were a merger of I-R and Dresser Industries. Dresser had bought out Worthington. Now, there are 3 companies over 100 years old, each with their own system. So, corporated developed a CCN (corporate number) to identify all of the unique parts, as there were some same numbers between the older companies. How about a part with 2 or 3 part numbers? Moral of that story: no matter what you do, it ain’t bullet-proof.