½Shop-Floor Work Instruction+ û how important they are? 2

[dmytry]

Hi All,

My name is Dmytry Mykhaylov and I’m a software engineer. The thing is... I sort of «inherited» one project and I need to make a qualified decision about whether to shut it down or continue up to some final product. The objective of the project is to create an on-line service for «Process Plan» or the «Shop-Floor Work Instruction» generation. It may be considered as a light-weight web-based CAPP system for small workshops or, if my understanding of MES is correct, it may be considered as a part of MES.

Unfortunately, I’m absolutely unfamiliar with the problem area of the project. I mean... I do know how to do this project as a software developer, but I have no idea what it is all about! This is the first time in my practice when I have an absolute zero of knowledge, but have to use only my common sense. This is why I started to surf the Internet and eventually founded this wonderful forum.

Why does my common sense keep me searching? One moment about the core algorithm of this application tells me that it might be a really helpful and useful for somebody else:

• The algorithm doesn’t depend on any specific equipment, tools, instruments, fixtures, coolant, etc. etc. – anything what might be needed to «perform» the proper work. Technically, if stone axe will be put in the «knowledge base» of the system – it will be treated as a regular «tool» and used appropriately when needed.
• First implication: user never refers to any «shop-specific» data when describes the future process (may be only except information regarding material, which may be in the shop-specific format or classification).
• Second implication: the same «process description» may be run against different sets of equipment (or «knowledge bases» of different shops) and every time the «shop-specific» process plan or work, instruction will be generated.

Limitations:

• It has nothing in common with ISO 10303, STEP/STEP NC etc. It doesn’t require shop to have any specific equipment on the floor. It can work with _any_ equipment/tools/fixtures.
• It doesn’t generate CNC-code. It is not the goal. The goal is to provide _detailed_, readable «work instruction» for shop-floor personal and generate the proper data about equipment/tools/fixtures usage on every step of this process (after that these data may be transferred to the appropriate MES or MES+ERP systems for further analysis).

The purpose of this post is to ask one question about how useful such kind of service might be for today’s manufacturing. Is it important to have a formal description of the process on the shop floor?

As an illustration to my question, here is the meaning of some columns in this document:

For example, the record "R" of the document contains the following info:

• "PI" - # of the position
• "D or ?"- design size of the diameter or width
• "L"- design size of the working stroke
• "t"- cutting depth
• "i"- number of cuts
• "S"- feeding
• "n"- spindle’s speed
• "V"- cutting speed
• "? aux"- auxiliary time (min)
• "? main"- main time (min)

The terminology and grammar of the document might be slightly inaccurate, but I hope it demonstrates the general purpose of the service.

I’ll be thankful for any advice, comment or information of any kind,

-dm

 

[MikeHalloran]

The documents you seek to generate are called Operation Sheets, or something similar.

The small shops of my acquaintance don't use them, or don't distribute them if they exist, for several reasons:

- The workers are expected to design their own process on the fly, in very little time, and have a lot of experience doing so.
- The workers have already memorized the process.
- The workers have evolved their own process.
- The process changes dynamically on the shop floor.
- Only the workers know the real process, and they damn sure aren't going to write it down, for fear of being replaced.
- All documented processes don't work, and nobody wants to document the laborious process that actually is working, for fear of being fired for spending so much money.
- If the documents describe a process that actually is working, they have commercial value to competitors, and no manager would entrust them to his own workers.
- Nobody who knows the process knows how to do the math, and nobody who could do the math knows the process.
- Documenting the process might uncover some inefficiency that is intentional, i.e. expose some larceny or corruption in high places.
- Someone uses knowledge of the process as a power base, and does not wish for the knowledge to become distributed.

WRT to the record you have provided, it is associated with a turning or milling operation. A _lot_ of software exists for computing this stuff, including some built into machinery and some available from tool manufacturers, at least for single operations in isolation. Machinists generally ignore it, and adjust the process by the shape and kinetics of generated chips, by sound or by smell.

It's relatively easy to optimize any single operation, e.g. to balance tool life and metal removal rate. Big cost savings, and big quality risks, come from readjusting the process flow, e.g. moving a particular operation from one machine to another in order to increase the total shop throughput. The risks come from things that are hard to quantify in advance, e.g. losing control of a hole location because of too many chuckings.

Frankly, I think you're handicapped in this instance by a complete lack of overlap of experience and vocablulary between you and your intended user population. Additionally, the people who could benefit most from your product are typically not authorized to spend money to buy or use it, and/or are extremely frugal.



Mike Halloran
Pembroke Pines, FL, USA

 

[dmytry]

Mike, wow... wonderful response! Thanks! I'll print it in bold and put on the wall of my cubicle

But, I'm still wondering... Did you just describe what people call a "tribal knowledge"? I understand that it is a part of the reality, but is it a big problem which better to be solved ASAP? How do you measure your efficiency in all this chaos and mutual distrust? How do you plan in such kind of environment where "Nobody who knows the process knows how to do the math, and nobody who could do the math knows the process"? And what about ISO and its requirements to the documentation?

It looks like you do _need_ some solution for all of what you've just mentioned, or you won’t survive. I'm far from the thought that "I'm the solution", but something must be done. The picture you just drew looks horrible.

But thank you again; it is like a snapshot from the front line.

 

[MikeHalloran]

The core of ISO, as it appears to be understood by management in America, is that if you document your process thoroughly enough, you could hire idiots off the street as production staff, greatly decreasing your labor costs, and you could abuse them at will, because they are interchangeable.

They forget that you'd need to hire _literate_ idiots, e.g. lawyers, so your labor costs would actually go up.

Operation sheets for JIT, called Method Sheets, include graphics sufficient to eliminate the literacy requirement, which actually has the side benefit of making them language- independent, a big virtue here in polyglot SoFla.

But, the whole idea enshrined in ISO, that you have to _tell_ people how to do their job, is itself fallacious, and symptomatic of distrust. The idea that a document of any kind _could_ replace training and experience is just silly.

Succesful small business owners rely totally on "tribal knowledge". E.g., they expect their welders to know how to weld. They expect their machinists to know how to drive a German Precision Hand Milling Machine, which you might mistake for a file. And they expect them to crank out work efficiently and rapidly and with acceptable quality, so that the business survives and everyone gets paid.

As a business grows, you do have to start measuring and checking things. Somebody who can do the math has to come out of the office and learn the process and get the chaos under control. It happens that I'm doing that right now. But trust and respect has to be part of that process.








Mike Halloran
Pembroke Pines, FL, USA

 

[dmytry]

quote:
---------------------------------------------
...
But, the whole idea enshrined in ISO, that you have to _tell_ people how to do their job, is itself fallacious, and symptomatic of distrust. The idea that a document of any kind _could_ replace training and experience is just silly.
...
---------------------------------------------
hmm... I understand it in slightly different way. IMHO ISO wants to make sure that the quality of any product will not depend on manufacturer if manufacturer + product are certified by ISO. So, if you want to certify something by ISO you have to document the way you've manufactured it. With the help of this info your product may be replicated by somebody else, probably with the same level of quality. (you are probably smiling right now yes, I understand that it will help to make illegal copies of your products, BUT the copies will be more reliable and, potentially, less harmful for customers. Lack of documentation never stops pirates. It challenges them!

So, the document doesn’t substitute training or experience. It just accumulate experience and increase safety of products. May be it is like in drug-manufacturing. If you got certification for some new drug – follow the process which had been certified. I understand that this is rather extreme example and this approach will be inefficient in general manufacturing, but I’m using it as an illustration only.

quote:
---------------------------------------------
...
Successful small business owners rely totally on "tribal knowledge".
...
---------------------------------------------

Of course they are! The half of my life I’m working in startups or small companies. And _every_ time when the small company tries to get some relations with bigger one, the bigger company starts to ask questions about “procedures”, “best-practices”, “measurement”, “processes” etc. because they want to see their potential partner (even the small one) as a _transparent_, and _predictable_ for their business. Do you have the similar situation on your area?

If you own a small shop and want to get some contract from somebody who doesn’t know you. Will it be helpful for you to provide for the first meeting something more tangible then just “tribal knowledge”? I saw many times when simple statement, like: “Don’t worry! We are very experienced developers! We’ll do your project in no time!” just doesn’t fly. But on the other hand, if on the first (!) meeting you are ready to provide any king of formal interpretation for the potential project and demonstrate how it fits with your methodology... you may not get the contract, but you definitely get the attention!

Something tells me that software technology and manufacturing one have a lot in common. Do you think?

 

[MikeHalloran]

An ISO certification guarantees nothing, beyond full employment for ISO auditors and legions of consultants who help you prepare for the process.

Survival of the certification process pretty much requires simplification of haphazardly evolved systems of management that might then work better, not as a result of the ISO process, but as a result of the attention given to them.

The documents created as part of the certification process represent an accumulation of intellectual property that would not otherwise be aggregated for easy redirection. The magnitude of the commercial risk involved is widely underappreciated.

;---

Sunshine promises work about as well on hardware as they do on software. Idiots keep making them. Idiots keep believing them.

;---

I wrote software full time for about ten years. I like to write, and I like to build machinery. Software is like a machine, that you write. They suffer analogous quality defects, express analogous behaviors, and respond to analogous diagnostic tactics.



Mike Halloran
Pembroke Pines, FL, USA

 

[dmytry]

Thank you, Mike. I wish you the best.

-dm

 

[dmytry]

Mike, do you know any forums where I could talk with shop owners or any kind of consultants who work in manufacturing?
I would really appreciate it.

Thank you,
-dm

 

[PSE]

Mike brings out many good points on the uses and limitations of documented processes. I utilize them only if needed to supplement a component or assembly drawing to capture what needs to be done for the operation to be complete. Method cannot be captured by a work instruction nor can experience. This we recognize by the need to include training and supervision as part of the "quality system". ISO cares not if you make good or bad product, as long as you do it consistently by the procedures.

Regards,

 

[mrainey]

"do you know any forums where I could talk with shop owners or any kind of consultants who work in manufacturing?"


There are some really sharp people here:

http://www.practicalmachinist.com

Manufacturing Freeware and Shareware

http://mrainey.freeservers.com

 

[Ashereng]

Mike brings out several good points.

ISO simply means this:
"You document what you do, and you do what you document."

Nothing more, nothing less. A good quality program does not necessarily mean a good product/services.

If you document a poor procedure, and follow it, it is still a poor procedure. Yes, you are ISO compliant, but, it is still a poor procedure.

Lots of ISO certified companies produce bad products. Lots of non ISO certified companies produce good products.

 

[dmytry]

Thank you for all your answers. Yes, I've mentioned "ISO" in one of my posts, but "ISO certification" is not a part of my concerns. I think my initial post doesn't work very well for me, because I over emphasize the role of "documentation" or "instructions". My fault. The instructions are one of the "nice features" which the application I'm thinking about may have, but they are not its primary objective.

The primary objective is to provide the functionality which allows disconnecting formal design and formal manufacturing processes. Or at least will make them less coupled. I think the way to do it is through describing manufacturing process without mentioning any specific equipment/tools/fixtures/etc. - _anything_ what you will need on the shop floor to make job done. If you can describe the process this way and run this description (lets call it meta-description or meta-process) on the set of equipment which represents the particular shop and get the “process sheet for this particular shop” as a result – mission accomplished.

Lets imagine for a sec. the application which works like this:

- you install it
- you enter the information about your equipment/tools/fixtures/shop structure into the database. if you remove/update/buy some equipment, then you should update the database and reflect all changes
- you’ve got the order on something and you decided that you may need a "Process sheet" for this
- when you are entering information into the application, _you never tell it how to make_ (I mean in which order what equipment/tools/fixtures to use). You just enter the info about _what should be done_ (80% is geometry and you can get it straight from the file with drawing)
- you press the button and you get the process sheet
- if you disagree with the process (the result), you edit the data (you make imperative suggestions to the system) and system will regenerate the process. In fact, you can tell to the system: "Idiot! use this one!"... and it will ;-), but you will do it only in 5% of cases

Does anyone could refer me to the existing system like this?

Thank you,
-dm

 

[looslib]

What you are looking for was at one point in time referred to as Generative Process Planning. Back in the early 80’s there were 2 camps on the thought of Computer-Aided Process Planning. The Generative approach and the Retrieval approach.

The retrieval approach work from a classification of your part and then standard plan for the part shape would be retrieved from the database and you could edit it to fit your exact part. These edited plans could then be stored to help refine the database to allow the search and retrieval system to get a better plan the next time through.

The generative approach relied on steps in a plan being stored and the sequence of certain steps being controlled by the system logic. It also relied on part characteristics to get the base shape and manufacturing steps from the database. If your part had a tapped hole, the system would know that you needed a center drill operation, then a drill operation and finally a tapping operation..

CAM-I developed both types of systems in the 70’s (retrieval) and 80’s (generative). Alex Houtzel took the retrieval system and developed MI-Plan from it. He now has a semi-generative system available from his own company. Dr. Ham of Penn State and Dr. Allen of BYU, both thought that the generative system was better. It was harder to design and implement, but it could also take into account manufacturing quantities. Would you build a one-off prototype using the same methods as a 10,000/year production run?



"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli
Sr IS Technologist
L-3 Communications

 

[dmytry]

Hi Ben,

Great contribution! Thanks. Before to answer I want to read more about the projects you've mentioned. From the terminology prospective I would classify ‘my’ system as a: "pure generative by nature, which you can use as retrieval if you like".

Just for clarification purpose, two questions:

1. Alex Houtzel is it for "HMS Software"

http://www.hmssoftware.com/?

2. Quote: "Would you build a one-off prototype using the same methods as a 10,000/year production run?" - what do you mean? "It is impossible" or "It is useless" or something else...

thank you,
-dm

 

[mrainey]

What adds to the fun is that one part can have fifty features and another can have exactly the same fifty, plus one. That single difference can sometimes force the two parts to be manufactured with very different processes.

Manufacturing Freeware and Shareware

http://mrainey.freeservers.com

 

[looslib]

dm,

Yes, that is the same Alex Houtzel.

If your were building one set, lets say a wheel for a toy, you would make them on a manual lathe. If you had to build 20 sets, you may goto a CNC lathe. If you had to build 100 sets and the design allowed it, you may use a stamping process. If you had 10,000 sets to make per year, you would look at die casting. The manufcaturing process chnages as the quantity changes. And as mrainey says, one additional feature may change the whole complexity of the part and the related manufacturing processes.

Most CAPP systems have a Group Technology coding system front end to help identify similarities between part shapes. Does or will your system also do part classification and coding? Many papers written in the late 70's and early 80's and presented at SME conferences on these and CAPP concepts and theroies.



"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli
Sr IS Technologist
L-3 Communications

 

[dmytry]

Ben,

http://www.sme.org/search/jsp/SearchResultController

with results for "generative" is definitely my place to read. Thank you.

quote:
----------------------------------------------------------
If your were building one set, lets say a wheel for a toy, you would make them on a manual lathe. If you had to build 20 sets, you may goto a CNC lathe. If you had to build 100 sets and the design allowed it, you may use a stamping process. If you had 10,000 sets to make per year, you would look at die casting.
----------------------------------------------------------

As far as I understand generative approach _before_ my long reading of SME site, the major problem is in “manual lathe”. If you can handle process for manual lathe, the rest is the question of scalability of your data structure. Here is the scenario of conversation between user (U) and system (S) I can image:

U: “Who can handle ‘wheel for a toy’ of this _shape_?”
S: “We can! Manual Lathe, CNC Lathe, Stamp and Die Cast Machines raise their hands”
U: “Ok, here is the _dimensions_”
S: “Stamp Machines lovers its hand. It can’t do so small items (for example)”
U: ”I need 500 of them”
S: “Manual Lathe lovers its hand. To large bunch to work on. The Master will be unhappy...”
S: “System will pickup CNC Lathe because it came first from the database” (or it may be more intelligent choice – it doesn’t matter right now)

Is it how it suppose to work?

I’m not brave at this point to answer on mrainey comment...

-dm

 

[looslib]

Yes, the logic built-in to the system will define the optimal manufacturing process based on the input parameters.

Getting all of the logic and maintaining it for your company is the hard part. The software has to be flexible to handle the changes without a major rewrite of the ubderlying code. In the case of BYU, they used a program called DCLASS to process decision tree structures. Generative CAPP from CAM-I used decision tables.


"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli
Sr IS Technologist
L-3 Communications

 

[dmytry]

Hi mrainey,

quote:
----------------------------------
What adds to the fun is that one part can have fifty features and another can have exactly the same fifty, plus one. That single difference can sometimes force the two parts to be manufactured with very different processes.
----------------------------------

Could you provide an example? I understand it this way:

- you have a piece of metal with 3 holes. 1 Drill Machine will be used for that. (but Mill is an option)
- you add 1 pocket
- now 1 Mill will be used for holes & pocket (Drill + Mill will be bad choice)

does it work as a _very_ basic example for your scenario?

thanks,
-dm

 

[dmytry]

Hi Ben, I hope you are still around...

Well, it is really educational reading, but it looks like a “sort of” pure theoretical approach. It looks like nobody actually using “generative planning” for planning or estimating. IMHO software for estimations should use it as a “must have” functionality, but I didn’t find to much references on it in the Net. Do you know some commercial products which successfully implemented generative approach?

I looked through Visiprise Inc. (former HMS) site... no visual indication that they have something like “semi-generative” it looks like they are using dialog-based system for entering data for planning (but difficult to judge – they don’t have downloadable demo-version)

Thanks,
-dm