standard: static design definition, fabricated or purchased,
stocked or not stocked
altered: fabricated from standard, not stocked
custom: job-specific design, fabricated or purchased, not
stocked
Our machines are assembled using a mix of each of these classes of
components. Machine component counts are typically 300-1000.
Questions:
1) Do you have similar component classes? How do they differ?
2) Do you use different information fields or formats for
numbers of each class?
(e.g. standard, aa-bbbbb-ccd; altered, abbbb-cccd;
custom, aa-bb-ccccc-dde)
3) What categories of information do you embed in the numbers
for each class?
(e.g. machine serial number, assembly location,
function, revision level, handedness, material type, etc.)
4) Do you have any significant problems with your current
numbering system?
5) Can you envision a better system? What would be different?
6) How do you handle numbering of components of a welded
assembly and the assembly itself?
Thanks in advance for anyone's help.
And if you continue on this part numbering path you will always be
actively involved.
> standard: static design definition, fabricated or purchased,
> stocked or not stocked
>
> altered: fabricated from standard, not stocked
>
> custom: job-specific design, fabricated or purchased, not
> stocked
Your ERP system should handle most of this information. If you don't
have an ERP system, get one. Inserting such significance in your part
numbering system will only lead to problems down the road. Significant
part numbering has inherent flaws, you will eventually run out of
numbers for instance. Or, what happens if you number a part with an
"NS" for non-stock, and then later decide to actually stock the part?
Would you change the part number? And if you do, now you've got two
identical parts with different numbers. Eek!
>
> Our machines are assembled using a mix of each of these classes of
> components. Machine component counts are typically 300-1000.
>
> Questions:
>
> 1) Do you have similar component classes? How do they differ?
> 2) Do you use different information fields or formats for
> numbers of each class?
> (e.g. standard, aa-bbbbb-ccd; altered, abbbb-cccd;
> custom, aa-bb-ccccc-dde)
> 3) What categories of information do you embed in the numbers
> for each class?
> (e.g. machine serial number, assembly location,
> function, revision level, handedness, material type, etc.)
> 4) Do you have any significant problems with your current
> numbering system?
> 5) Can you envision a better system? What would be different?
> 6) How do you handle numbering of components of a welded
> assembly and the assembly itself?
>
> Thanks in advance for anyone's help.
>
>
If you want to differentiate bewteen "platforms", or be able to sort
certain types of parts, use a "semi-significant" system. My example,
used here at DTM is shown below:
XXXX-XXXXX-XXX
XXXX = numerical prefix denotes either "machine" (we have three), or is
used as a designator for a type of OEM part - 2100 series numbers
for example are electrical connectors, 2850 series for fasteners, etc.
XXXXX - sequencial numbers with the first digit being specific to the
type of part - 2 for mech. assy's, 3 for elec. assy's, 5 for purchased
modified parts, and so on..
XXX - the revision level of the part.
So..
9204-2XXXX-000 = 2500plus (our machine)-mechanical assembly-rev level
2850-03478-000 = mechanical fastener-the next unused number-rev level
We will have problem #1 some time in the distant future - we will run
out of numbers, but we've been assigning numbers for 10 years, and
we're only up to 03700 on our OEM parts. We've got some time.
If it's reporting you're after, the ERP system that you have or choose
will offer things like "group codes" or "product classes" to help sort
out your stock or non-stock needs, as well as provide reports for
anything from "where-used" to "on-hand" to "supplied or manufactured".
Now of course everything I've said here can (and will, I'm sure) be
argued to death. Significant numbering systems are useful for some
companies (look at Fafnir or Dodge for example), but for the most part,
these systems will do nothing but bog you down in the future.
Just my (long-winded) opinion.
Richard Doyle
Sent via Deja.com http://www.deja.com/
Before you buy.
A number prefix for Standard parts: 123-123456
An "AD" prefix for main assembly drawings: AD12345RAA
An "SA" prefix for sub-assembly/weldments drawings: SA12345RAC
A "CD" prefix for components we make: CD12345R03
The nice thing about this system is, you can tell what a file is
immediately. You can also search by the prefix to get in the ballpark. We
have recently dropped the rev number off all parts. Since we archive the
paper copies of each rev, it rarely paid to keep electronic files of each
rev. The existing file is always the latest rev.
We don't actually use an altered part number, we just use a "CD" number.
If you think about it, every part you make is an alteration of a purchased
part.
Also, if we create a purchased part in SolidWorks, we use the
manufacturers part number as the file name. That's the number we would have
if we were finding parts from a catalog.
The difference between letter rev's and numbered rev's is: letter rev's
are production release and numbered rev's are in a prototype stage.
Revision control is much faster, more simple, and easier on prototypes.
One reason we make money is, we can turn a product extremely fast. Most of
our products are original, custom designs. We sell a lot of one-offs. From
design-to-door is about 3-weeks for these. (Depending on the load on the
shop floor and complexity of the product, of course.) It's hard to do this
with piles of paperwork and meetings.
7 digits, no significance.
>
> altered: fabricated from standard, not stocked
>
7 digits, no significance.
> custom: job-specific design, fabricated or purchased, not stocked
7 digits, no significance.
>
> Our machines are assembled using a mix of each of these classes of
> components. Machine component counts are typically 300-1000.
>
> Questions:
>
> 1) Do you have similar component classes? How do they differ?
I'd say everyone has exactly the same types of parts you listed.
Fabricated, modified, and purchased. How they are used is really
irrelevant (i.e. stock, custom, etc).
> 2) Do you use different information fields or formats for numbers of each
> class?
If anyone does, they shouldn't. There is no such thing as an
intelligent part numbering system. No one can forsee all the various
component types classes and categories, and therefore, to try and
pigeonhole each component is a waste of time, and will onyl cause
problems down the road.
> 3) What categories of information do you embed in the numbers for each class?
None. See above.
> 4) Do you have any significant problems with your current numbering system?
Aside from the fact that it exists solely in my own head at this point?
No, no problems. My current employer uses the 'we have no system'
system. It's horrible. My last employer had a rigid (un)intelligent
part numbering system. It was horrible. Smart, enormous companies such
as GM use non-intelligent 7 digit part numbering systems (or bar
codes). If they haven't run out of part numbers, it's unlikely anyone
else will.
> 5) Can you envision a better system? What would be different?
See above.
> 6) How do you handle numbering of components of a welded assembly and the
> assembly itself?
Every component that requires operations to be performed on it prior to
later use in an assembly or weldment should have a part number.
The drawings are irrelevant, the Bill of Material is king. It is the
only document that anyone should ever use to find information about an
individual part, weldment, subassembly, assembly, machine, or project.
Everything springs forth from the Bill of Material.
Jim S.
>
> I'd say everyone has exactly the same types of parts you listed.
> Fabricated, modified, and purchased. How they are used is really
> irrelevant (i.e. stock, custom, etc).
>
One of the problems is not so much how they are used, but how they are
created. Existing parts will have a part number in the erp system, but
newly designed custom parts do not. With 20 designers all creating
custom parts separately on different projects, the thought was to
generate a number specific to that job with perhaps a job number
embedded in it. That way each designer knows what the next available
number is. Then if the part were to be made into a standard part later
it would be given a standard part at that time.
> > 2) Do you use different information fields or formats for numbers of
each
> > class?
>
> If anyone does, they shouldn't. There is no such thing as an
> intelligent part numbering system. No one can forsee all the various
> component types classes and categories, and therefore, to try and
> pigeonhole each component is a waste of time, and will onyl cause
> problems down the road.
>
That's why we pared it down to three so far. These three
categories should would have significant conflicts would they?
1) Standard Fab parts that don't change and are locked in the system. If
they do change, the new version would be given a new number.
2) Fab parts that are specials. Either reworked standard parts or
totally from scratch new parts.
3) Purchased parts. We do a lot of custom work with an ever expanding
list of purchased components. It is still undecided whether we would
assign our own number to each purchased part when it already has a
vendor's part number to use in the system.
A documentation nightmare waiting to happen. The Engineers must be
disciplined to assign part numbers at the proper time. The proper time
is when you are creating the drawings. No part needs a drawing until it
is being put to paper. Build the Bill of Material, as you create the
drawings. The computer system should be capable of telling you when
someone has taken a part number you were going to use. If so, simply
change your drawing. Such changes will be very rare and very localized
if you get the Engineers to follow good practices. Engineers (myself
included) often get too concerned with making the Bills of Materials
neat (i.e. all the part numbers in reasonable order etc...). It just
doesn't matter when a part number is just that: a number.
>
> > > 2) Do you use different information fields or formats for numbers of
> each
> > > class?
> >
> > If anyone does, they shouldn't. There is no such thing as an
> > intelligent part numbering system. No one can forsee all the various
> > component types classes and categories, and therefore, to try and
> > pigeonhole each component is a waste of time, and will only cause
> > problems down the road.
> >
>
> That's why we pared it down to three so far. These three
> categories should would have significant conflicts would they?
>
> 1) Standard Fab parts that don't change and are locked in the system. If
> they do change, the new version would be given a new number.
Any drawing that changes the form fit or function (i.e. backward
compatibility) of a part should receive a new number. Regardless of
whether it is standard, special, or purchased. Part numbers are dirt
cheap. Mistakes made due to erroneous information on drawings or in
Bills of Material are costly because they don't become evident until
after manufacture, or assembly, or delivery to the customer.
>
> 2) Fab parts that are specials. Either reworked standard parts or
> totally from scratch new parts.
>
Why the need to have 'specials' at all? It makes more sense to me to
delegate certain Bills of Material as 'standard' as opposed to
individual parts.
> 3) Purchased parts. We do a lot of custom work with an ever expanding
> list of purchased components. It is still undecided whether we would
> assign our own number to each purchased part when it already has a
> vendor's part number to use in the system.
Each should be assigned a part number in your system. There is far too
much room for error when you have to enter a part number like:
1624E024SL+16/8 262:1
as opposed to
3124567
You will end up with hundreds of duplicates simply because people may
describe things differently. At my previous employer it was common to
see this in the computer system:
O-RING,NITRILE,1.250IDx1.375x.063WALL
O RING,NITRILE,1.250IDx1.375x.063WALL
O'RING,NITRILE,1.250IDx1.375x.063WALL
This brings me to a very important point I forgot to mention in my first
reply. Spend your time developing a consistent way of describing parts,
as oppsoed to trying to be clever with part numbers. Compose a document
that provides a template for describing the parts most likely to be
duplicated in the system. As new templates are needed, assign an
individual the task of amending the master template. By standardizing
the description, you eliminate the need for intelligent part numbers
because people will search by description, as opposed to using the first
few digits of a number. If you decide to change the template, it is
easy to determine all the instances that must be changed, simply because
they will already be logically grouped in the system.
The points I've raised don't come out of the blue. It's really the
result of attending a single seminar put on by a company that
specializes in this sort of thing. Sitting in the seminar, I listened
and discovered that my employer was breaking about every good practice
mentioned, and it became painfully obvious how bad our system was.
The company is called R.D. Garwood, and I would strongly suggest that
you contact them, or at least order/read the book (Bills of Material:
Structured for Excellence) before embarking down a road that you may
regret later.
Good Luck,
Jim S.
Now does anyone want to talk about drawing control and Vault access.
NOOooooooo......
Frank
"Mark R." wrote:
> I am actively involved with projects to implement SolidWorks. A
> project team recently began the official process to design a new
> component numbering system that will function well with SolidWorks.
> The team is currently researching component numbering systems to glean
> useful ideas. The system would handle these component classes:
>
> standard: static design definition, fabricated or purchased,
> stocked or not stocked
>
> altered: fabricated from standard, not stocked
>
> custom: job-specific design, fabricated or purchased, not
> stocked
>
> Our machines are assembled using a mix of each of these classes of
> components. Machine component counts are typically 300-1000.
>
> Questions:
>
> 1) Do you have similar component classes? How do they differ?
> 2) Do you use different information fields or formats for
> numbers of each class?
> (e.g. standard, aa-bbbbb-ccd; altered, abbbb-cccd;
> custom, aa-bb-ccccc-dde)
> 3) What categories of information do you embed in the numbers
> for each class?
> (e.g. machine serial number, assembly location,
> function, revision level, handedness, material type, etc.)
> 4) Do you have any significant problems with your current
> numbering system?
> 5) Can you envision a better system? What would be different?
> 6) How do you handle numbering of components of a welded
> assembly and the assembly itself?
>
> Thanks in advance for anyone's help.
Frank DeWitt | mailto:fr...@lbpinc.com
Mechanical Design | Phone: 716 624 3052
LBP INC. | Fax 716 624 1038
2365 Cox Rd. |
Bloomfield NY 14469 |Web http://www.lbpinc.com
Mark
If you are issuing a do-it-yourself system some tips.
Try and avoid issuing numbers sequentially. Better to pull from a pre scrambled
list, that way no one looks for hidden meaning in the number sequence.
Consider including a check sum digit in the number. This helps to validate
key-board entry and confirms that you've entered a legitimate number. There are
a variety of check digit systems used so you may look around.
Which is correct if both are the same part? Is it run-run-branch or
run-branch-run? See Branch run. Run branch run. What's the difference
between an adapter a connector and a union. Define these things and the
order they are listed. Screws get real interesting.
<bee...@my-deja.com> wrote in message news:8rivfh$ubp$1...@nnrp1.deja.com...
> I agree completely with Jim. With an ERP system or PDM system in place
> keep the part numbering system simple. 7 digits. When you look
> something up 99% of the time you will search by description. Doing
> part numbering with hyphens and prefixs and suffixs is archiac.
>
>
How would revision levels be tracked? Through the database? If each time
a part is revised and given a new rev. level (assuming a rev. level
system is to be used) you would have to leave a part number trail
somehow through the erp system or some database to track all that. Or,
would it still be recommended to tack on a rev. level to the end of the
component number?
How would right hand and left hand components be handled? Would each
hand have a separate number, not necessarily right next to it's mate.
How would you track the number of it's opposite hand mate for revisions
etc. Would that be handled the erp system or database? Now you would
have two revision trails to follow when a revision was to be made.
In article <39DD3159...@lbpinc.com>,
Frank DeWitt <fr...@lbpinc.com> wrote:
> Been there, done that, Trust me, you won't even get a tee shirt.
> Management seems to have some built in desire for "Intelligent
numbering
> systems". These lead to a sketch system, or fixture system, or no
numbers
> at all. "Leave them blank, we will put the numbers in later." Don't
go
> there. As others have said Numbers are cheep. Just pull the next one
off
> the pile and enter the info into the data base.
>
Have you or anyone else read the book "Manufacturing Data Structure:
Building Foundations for Excellence With Bills of Materials and Process
Information"
By, Jerry Clement With Andy Coldrick With John Sari
I see it on a few online bookstores and it looks to be along these same
lines as well. I'm thinking of getting that one as well.
In article <39DD1B72...@abraxis.com>,
Jim Sculley <nic...@abraxis.com> wrote:
> The points I've raised don't come out of the blue. It's really the
> result of attending a single seminar put on by a company that
> specializes in this sort of thing. Sitting in the seminar, I listened
> and discovered that my employer was breaking about every good practice
> mentioned, and it became painfully obvious how bad our system was.
>
> The company is called R.D. Garwood, and I would strongly suggest that
> you contact them, or at least order/read the book (Bills of Material:
> Structured for Excellence) before embarking down a road that you may
> regret later.
>
> http://www.rdgarwood.com
>
> Good Luck,
>
> Jim S.
>
"Mark R." <donamar...@mobynet.com> wrote in message
news:39dcbd15...@news.netasset.com...
> I am actively involved with projects to implement SolidWorks.
{snip}