The Gladieux Thread, August 1999

1. Sean Gladieux asks for suggestions on comparing manufacturing by fabber with “traditional” techniques.
2. Marshall Burns refers Gladieux to Phill Dickens’ work on the subject.
3. Elaine Hunt offers encouraging words and refers Gladieux to the MMS (??) Web site.
4. Phill Dickens summarizes his Flymo findings, lists seven challenging questions arising from them.
5. Elaine Hunt says she’s seen an answer to one of Phill’s questions somewhere.
6. Nick Osborn argues future manufacturing more likely to fab tooling than actual products.
7. Ben Halford cites blasting nozzle most economically made by SLA up to quantity 12.
8. Ronald Jones agrees direct manufacturing by fabber is inevitable, but starting with high-value and high-tech products
9. Marshall Burns compares impact of the Internet on distribution to impact of fabbers on manufacturing.
10. Allan Lightman compares impact of the industrial revolution to that of fabbers.
11. Sean O’Reilly describes Ford research, says fabbers (a) will never do high volume, (b) demand new design paradigm.
12. Karl Denton warns talk of using fabbers for production is premature until a major breakthrough in technology.
13. Sean O’Reilly adds sobering statistics on speed, says fabbers’ strengths are in making complex shapes.
14. Brian VanHiel argues that Burns’ vision may be overoptimistic.
15. Michael Vincek suggests this debate is a waste of time.
16. Roger Spielman reports he’s been using SLS to make end-use aerospace parts for years.
17. Ronald Jones speaks of the paradigm shift that fabbers are bringing about.
18. Phill Dickens argues with Osborn’s prediction of the continued dominance of tooling in manufacturing.
19. Karl Denton argues that Spielman’s low-volume examples do not portend real production applications.
20. Charles Overy implores Gladieux to include nonmanufacturing impact, e.g. catching design flaws, inventory reduction.
21. Oliver Cole wants to know what he can profitably make on his FDM now!
22. Brock Hinzmann tells Cole that selecting items to manufacture by fabber is tricky.

Sean Gladieux asks for suggestions on comparing manufacturing by fabber with “traditional” techniques.

Well, I am taking the leap of faith here. I have been reading this RP list for quite some time but haven't yet posted any messages. But I think it is about time.

I am a graduate student at Bradley University, Peoria Illinois and am just beginning to develop my masters project proposal for my MS in Manufacturing Enginering. I am interested in performing a cost analysis of RP part manufacture compared to traditional part manufacture. I am thinking about selecting a fuel filter base part for a heavy duty industrial engine. The part is normally die casted. The part has been manufactured with RP technology by John Howarter of Caterpillar Inc, so I know it can be done. What I plan to do is to analyize the cost benefit ratios, and cost savings of using RP technology to produce over die casting with CNC produced dies.

What I would like from anyone who has lasted this long into my message is any advice or input as to where I should go with this project. This is an open ended request. I was just wondering if anyone out there has already performed this type of analysis and if so would you share your insights with me. I would like to be able to contact you during my project's development if possible too.

Any input is always appreciated.

I can be reached via personal email at ssglad (at) cat (dot) com

Sean M. Gladieux
Industrial and Manufacturing Engineering and Technology Department
Bradley University
Peoria, Il 61629

Marshall Burns refers Gladieux to Phill Dickens’ work on the subject.

Dear Sean,

Excellent research topic! You should get a copy of Phill Dicken's paper of a similar nature based on parts used by a Brittish lawnmower manufacturer. Contact Phill at pdickens@dmu.ac.uk. I and many others will be interested in your findings.

Best regards,
Marshall Burns

Marshall (at) Ennex (dot) com
Ennex Corporation, Los Angeles, USA, (310) 824-8700
www.Ennex.com

Elaine Hunt offers encouraging words and refers Gladieux to the MMS (??) Web site.

great things are accomplished with small leaps of faith...very few things without

just finished a meeting with a grad student asking about his research....my message to you and all potential researchers.......

Don't box yourself in by looking at today's technology and trying to see where your work fits.
Look outside today and beyond tomorrow...
Consider a perfect world with a perfect material and a perfect technology and then dream of what could be done...
then do it...
what you accomplish may provide the leap of faith to even greater things...

Consider the following from MMS web site:
1. The global market for dies, molds and precision parts is estimated to be $65 billion in sales, with $20 billion U.S. market share. [Destefani, 1993].
2. Die and mold making is a dying trade. The average diemaker in the U.S is 56 years old. Soon, the knowledge will retire with them.
3. Approximately 70-80% of the production costs of a die are already determined during the design stage. [Waltl et al, 1997]
4. Only 6% of American die/mold companies use 5-axis milling.

The new semester begins!

Elaine
*******************************************************************
Opinions, suggestions, and other controversial matter VOID where prohibited.
******************************************************************
Elaine T. Hunt, Director
Clemson University Laboratory to Advance Industrial Prototyping
206 Fluor Daniel Bldg. Clemson, SC 29643-0925
864-656-0321 (voice) 864-656-4435 (fax)
elaine.hunt (at) ces.clemson (dot) edu
http://rafiki.vr.clemson.edu/credo/persall/persall.html

Phill Dickens summarizes his Flymo findings, lists seven challenging questions arising from them.

Sean,

I saw your message to the rp mailing list earlier and the comments that you got.

This is a subject that is dear to my heart.

You have started a very important thread here. The original work we did was with a company called Flymo (part of the Electrolux group), they make garden care equipment. We wanted to see if it was economically feasible to manufacture end use parts by RP instead of injection moulding. We found that RP cost about 100 times more and injection moulding was about 100 times faster. With the development of newer machines such as the SLA 5000 and then the SLA 7000 these numbers will have changed. However, what we also showed was that it was economic to make even small simple parts in numbers up to 6,000 before injection moulding became cheaper!

We have just installed an SLA 7000 so it would be interesting to redo the numbers.

I am 100% convinced that RP will move into direct manufacture of end use parts with numbers into the thousands. This raises lots of other issues.

1. If we use RP then assuming the material properties are okay how would you change the design to make it better. Remember an RP machine (of some sort) can make any geometry.

2. How often would you change the design if you are not limited by tooling.

3. Do other factors limit you e.g. physics of the process in the product or other parts e.g. filters in your case.

4. How will manufacturing change. Is everything Just in Time. Why make to stock?

5. Can you eliminate parts e.g. if you sinter the part could you also make the filter as well at the same time.

6. How will this freedom affect part variation. Is greater variety a good thing? We have spent a long time reducing part counts. What are the pros and cons?

7. If you use RP instead of die-casting which process would you use? Would you try and make a sintered metal part to get near to what you have now? Why do you have a die cast part now? Would a plastic RP part be okay? Would you need to change the design to make it okay in plastic?

These are the issues that spring to mind now. I am sure there are lots of other issues. You could do your project in any one of these. Choose the one that is of most interest to you. If you would like us to make you some parts for this project let me know. We have a Thermojet, SLA 7000, FDM 2000 and a new EOSint M extended (metal sintering) will be installed in September.

I am in Atlanta at the moment visiting Georgia Tech and then onto Texas for the SFF Symposium and then onto University of Arizona and so I will not be able to email the paper until I get back to the UK on the 19th. Please email me then to remind me.

COMMENT TO EVERYONE ELSEThere must be lots of other ideas on this. Lets start sharing. Have I missed some important points? Is this all rubbish? Don't just sit there. Get contributing.

Prof. P. M. Dickens
Dept. of Mech. & Manuf. Eng.
De Montfort University
The Gateway
Leicester
England LE1 9BH

Tel: +44 (0)116 257 7689
Fax: +44 (0)116 257 7025
Mobile: +44 (0)468 827039
Email: pdickens@dmu.ac.uk

Elaine Hunt says she’s seen an answer to one of Phill’s questions somewhere.

Phil,

>2. How often would you change the design if you are not limited by tooling.

I saw the figures somewhere in the last month detailing how many design changes can be made before you begin to increase product costs due to changes. I usually bookmark and lay aside the book but may have trashed it. I'll keep looking........

Guess I should make you walk up here from Georgia........

Elaine
*******************************************************************
Opinions, suggestions, and other controversial matter VOID where prohibited.
******************************************************************
Elaine T. Hunt, Director
Clemson University Laboratory to Advance Industrial Prototyping
206 Fluor Daniel Bldg. Clemson, SC 29643-0925
864-656-0321 (voice) 864-656-4435 (fax)
elaine.hunt (at) ces.clemson (dot) edu
http://rafiki.vr.clemson.edu/credo/persall/persall.html

Nick Osborn argues future manufacturing more likely to fab tooling than actual products.

"Prof. P. M. Dickens" <pdickens@dmu.ac.uk> 06/08/99 01:33:18
> We found that RP cost about 100 times more and injection moulding was about
> 100 times faster.
> ....
> However, what we also showed was that it was economic to make even small
> simple parts in numbers up to 6,000 before injection moulding became cheaper!
> ....
> I am 100% convinced that RP will move into direct manufacture of end use
> parts with numbers into the thousands.This raises lots of other issues.

Phil

At the Euromold Pre-Fair Press Conference in Frankfurt recently the head of 3D GmbH gave a presentation which included references to SLA machines being used for rapid manufacturing (RM), which I have a problem taking too seriously at the moment.

Surely the future for RP is to work towards "instant tooling" which is fast and cheap enough to negate many of the problems associated with production tooling today?

In this way, when coupled with mass production processes & techniques (e.g. injection moulding) that have already undergone many decades of development (compared with RP's one decade to date) the result is a hybrid solution which recognises the relative merits of each technology.

I concede that (for instance) an SLA 7000 building trays full of widgets can outstrip conventional injection moulding on cost (initially anyway) - it will be interesting to see your new figures when they are available - but how many widgets have you seen recently that are made from UV resin? Just look at the mechanical performance differences between a prototype part that is machined from solid thermoplastic vs. the same form but moulded condition. OK, so in some circumstances a high grade SLA material may suffice, but I would argue that, even for complex parts, building ONE high res. RP master to make ONE (fast, low cost) rapid tool is a more efficient route.

I would be very interested to work with you to provide proof of the above statements for a range of different part sizes, geometries, etc. Please let me know if you have any interest in this offer.

Regards
Nick

Nick Osborn
Managing Director
------------------------------------------------
Swift Technologies Ltd.
140 - 144 Station Road, March,
Cambs. PE15 8NH, UK
"Real Parts Real Quick - Because Time Is Money"
Tel: +44 (0) 1354 650 789
Fax: +44 (0) 1354 650 799
www.swiftech.co.uk

Ben Halford cites blasting nozzle most economically made by SLA up to quantity 12.

Sean

By using RP many of the elements and operations associated with the design and production process can be eliminated bringing significant benefits to the customer. To justify this a complete overview of the design, manufacture and operation of the unit needs to be provided so that this approach can be accurately quantified. Time scale, cost and functionality implications associated with this should also be analysed and where possible compared with a conventionally fabricated system.

Traditionally designers have been forced to adjust product form to permit manufacture using conventional processes. This creation of components from assembled parts introduces complexity, reduces functionality and renders some designs uneconomic. Layered manufacturing offers the potential to eliminate these compromises by building individual integrated forms cost effectively. Until recently the relatively poor mechanical and dimensional properties of RP models has prohibited their use directly in real world applications. Progressive development of various RP processes has now reached the stage where these systems may be considered as viable production platforms for the manufacture of selected, limited run, components.

To take the case of a Cryogenic Blasting nozzle, we found that the break even point for an additive manufacturing approach (SLA 350) was just 12 components. With reference to functionality these SL 5190 models were run at 6.5 Bar dynamic pressure, with nozzle velocities of 350 m/s and with frozen CO2 at below -85 DegC going through them - there was no wear and only moderate thermal stress fractures. The models were used to optimise 9 design iterations at once and proved to be representative when compared with control production spec components. In this instance I wouldn't necessarily use this SLA resin for service but might consider specifying a high performance epoxy system or an SLS Nylon as the end material. The only concession required when using SLS powders rather than liquids would be the post reaming of the injector nozzle to reduce frictional losses.

This case study is just one of a number of examples where we are using RP models as functional components and I see a huge niche, expanding in line with accuracy and material performance, for the Free Form Fabrication 'Press & Play' approach. Having said this it is important to remember that we must use the strengths of each manufacturing approach (including rapid tooling) in parallel to achieve the most effective performance / cost result. Using the Cryogenic Blasting nozzle as an example again, where the design featured a straight barrel it is best to buy a section of pipe rather than build it in layers. Conversely one of the designs we trailed featured rifling inside the barrel and this could be economically produced via the additive route. Good luck with your Masters.

Regards

Ben (Halford)
PERA Technology
England
ben.halford (at) pera (dot) com
0044 (0)1664 501501

Ronald Jones agrees direct manufacturing by fabber is inevitable, but starting with high-value and high-tech products

Prof. Dickens and All Solid Imagers

The points that you raised in your message to Shawn are insightful. They address issues that our company has made a focus of its research. The issues all relate to "Where is RP going?" If you are in this field (equipment supplier, material supplier, service provider, user or researcher) you need to have a pretty good understanding of this.

There have been many who have put forth the goal of RP as we know it today to transform into "the manufacturing method of the 21st Century"! The barrier to this evolution is both economic and technical. I agree with Prof. Dickens that the recent advances in throughput speed as evidenced by the latest 3D System SLA product release yields sufficient economic benefit to realistically forecast freeform fabrication as a viable manufacturing method. Similarly, the advances in materials via DTM and others to get closer to intended materials, especially metals suggests that direct fabrication in final materials is not only possible, but probable. Contrary to most consultants in the field, I cannot see a near-term evolutionary path to "home based manufacturing" or even "Freeform Manufacturing" as being wide spread or universal. I believe the path must first pass through the world of "high value products" and advanced technology components that take advantage of the unique build capabilities that several RP systems currently offer.

Prof. Dickens is also correct concerning the impact on product design. We are already seeing evidence of this in key markets. Apple Computer for example has redefined with both its new desk top and lab top PC's what a PC should look like. Due to their success in embracing complex geometry using RP, Apple is putting pressure on the rest of the industry. Another case in point. Recently, during a conversation with a major shareholder of a power yacht manufacturer, he decided to over rule the decision by his plant manager to reject RP. His view was that sooner or later one of the firms in their market would redefine what a luxury motor yacht should look like using RP and he was determined that it this was not going to catch his company "flat footed". Like the good professor, I would also like to hear from others on the list on this important subject.

Ronald Jones
President
Shared Replicators, Inc.

Marshall Burns compares impact of the Internet on distribution to impact of fabbers on manufacturing.

Dear RP-world,

It looks like we've got some debate going in here on the merits and feasibility of using fabbers ("rapid prototyping") for manufacturing. I had dinner last night with some friends in the Internet business who had an interesting insight on this subject. I'll be interested on the comments of people on this list to these ideas.

One of the reasons the advent of the Internet has been so impactful is what it does to the VALUE CHAIN for the distribution of products. One might draw the modern, pre-Internet value chain as follows:

Manufacturer --> Wholesaler --> Distributor --> Retailer --> Customer

What the Internet does is that it opens the opportunity for direct interaction between the manufacturer and customer, so that the value chain can be reconfigured:

Manufacturer --> Customer --> Manufacturer

The last link above reflects the fact that with the Internet the customer has a greater opportunity to interact with and influence the manufacturer. If I were writing this posting in a graphical medium, the above would be a cycle from manufacturer to customer and back. The reduction of intermediaries in the relationship between manufacturer and customer eliminates costs and time from the distribution process, and improves communication. The result is greater satisfaction of the customers' needs, faster, and at lower cost.

The optimal structure is not always this fully collapsed value chain, but there can be opportunities for an Internet intermediary that adds value to the product or to the customer/manufacturer interaction:

Manufacturer --> Web portal --> Customer

But instead of going into details on this, let's change back to OUR subject of the use of fabbers in manufacturing, and let's look at it from this point of view of its effect on the value chain in the design and manufacturing of products.

The idea is that what the Internet does to the value chain for the DISTRIBUTION of products, the fabber does to the value chain for the products' DESIGN AND MANUFACTURING.

So the value chain for manufacturing in the industrial era might be viewed as:

Concept --> Prototype --> Tooling --> Production --> Distribution --> Product

What the fabber does is eliminate the need for tooling and mass production AND ALSO distribution because the customer (who may be a business or an individual) may very well be operating the fabber on-site or at a local 3-D Kinko's. Also the concept of a prototype becomes fluid because the customer can iterate the product, try out each iteration, and either keep iterating or stop iterating, depending on when satisfactory performance is achieved. Which iterations are prototypes and which are products? Such semantic distinctions are not important. The value chain becomes:

Concept --> Iteration --> Concept

where the last link, as in the above Internet example, would be shown as a cycle if I were writing in a graphical environment.

Now I'm sure a lot of people on the list are getting ready to respond to this by arguing that for most products a complete and ready-to-use product cannot be made on a fabber because of limitations in materials, the need for assembly of mechanisms, and the high cost of operating fabbers. To all those people, I ask you to think back to the days of the Model T Ford and ask yourself if you could have foreseen interstate freeways (autobahns) and suburban shopping malls. Think back to IBM's first computer, the 650, and ask if you could have foreseen the Internet. Think back to Goddard's first suborbital rocket launch and ask if you could have foreseen people walking on the moon. Of course today's fabbers cannot make even a small fraction of the catalog of modern products enjoyed by people around the world. But tomorrow's fabbers will.

The next step in this discussion could be to look at what happens to the total product value chain when you combine the effects of both fabbers and the Internet. I'll leave that for another posting.

Best regards,
Marshall Burns

Marshall (at) Ennex (dot) com
Ennex Corporation, Los Angeles, USA, (310) 824-8700
www.Ennex.com

Allan Lightman compares impact of the industrial revolution to that of fabbers.

Ron, Phill, Sean and all RPers,

You have started an interesting thread - prognosticating on the role of RP for the future. I have abreviated my message by cutting off the previous discussions - most have probably already seen them once or twice.

A long, long time ago (as measured in Internet time or about 10 years in calendar time), when the future of RP was but a gleam in the eyes of the few people working in this field, a sagacious engineer commented to me, "We stand on the threshold of the second industrial revolution."

The first industrial revolution involved the use machinery for production rather than using hand tools. This in turn lead to standardization, assembly line production, and the development of tools for mass fabrication of standard parts. With the concurrent cost reductions and improvements in quality, all designs were accommodated to the standard components that were now readily available. The engineer, viewing RP at its onset, saw an opportunity to be able to make custom parts, each individually tailored to the requirements of the task for which it was to be used, while deriving the cost and quality benefits of 'mass' production. To date this has not been realized. Instead, RP has been used in a manner in which it directly competes with the other fabrication techniques used for mass produced parts and assemblies. In part this is a result of the mindset of designers and engineers who have been educated under the framework of the benefits derived from the first industrial revolution.

With the recent development of higher speed RP equipment, the availability of new materials with improved 'strength,' are we now in position to consider design/fabrication that is customized for the application (one or very few of a kind) and use a design that is suited to the materials available? There are many instances in which either the need is individual or where standard components just do not fit well enough. Can the added cost of RP be justified by the customization provided? These will probably be higher cost, high value added products. Medical applications are an area in which we have seen some gain by RP. Hopefully the health insurance companies won't snuff it out. Part of the obstacle that needs to be overcome is that this represents out-of-the-box conceptualization, which requires a mindset adjustment. Is anyone looking along these lines?

Allan Lightman
> ------------------------------
> Allan J. Lightman, Ph.D.
> Senior Research Scientist, Research Institute
> Associate Professor, Electro-Optics
> University of Dayton
> 300 College Park
> Dayton, OH 45469-0150, USA
>
> Voice: +1-937-229-3966
> FAX: +1-937-229-3433
> E-mail: Lightman (at) udri.udayton (dot) edu
> WWW: http://www.udri.udayton.edu/rpdl

Sean O'Reilly describes Ford research, says fabbers (a) will never do high volume, (b) demand new design paradigm.

Phil:

Several years ago we did a study on the relative costs of producing plastic (as opposed to metal) parts. To do the study we had to make several assumptions:

1.) Materials - we ignored the difference between materials (for example, many automotive components are made with glass-filled nylon or talc-filled polypropelyne - both VERY abrasive. Most of the processes we looked at could NOT produce parts made of these materials - at least at that time.)

2.) Geometry - we assumed a fairly simple geometry - no undercuts, etc. - something that could be made in a simple core/cavity mold.

Based on these assumptions, we looked at a number of processes:
- Free From Fabrication ( I still refuse to call it "Rapid Prototyping"). Materials limited to the particular process used (SLA, FDM, etc.)
- RTV molding (from an FFF master)Materials limited to 2-component urethanes and some castable epoxies.
- Kirksite spray tooling (sprayed on to an FFF master). Materials included "benign" plastics - No Glass-filled nylon, for example.
- Traditional machined steel tooling - Any material

We were interested in cost as measured by per-piece cost. The costs included "tooling" cost amortized over the expected life of the tool, labor costs measured at a constant hourly cost, material costs (much higher for SLA for example)and so forth.

It is no surprise that the higher the number of parts required, the more cost effective traditional methods became.The most significant factor turned out to be the tooling costs. We estimated about 40 pieces per silicone rubber mold, about 200 - 300 pieces from spray metal tooling, etc.

While interesting, the study really didn't help much because of the REAL material requirements for REAL parts, together with other attributes such as surface finish, weight, etc. I have tried to find the study, but have been unable to do so. I'm on my third new computer since we did the study ( around 1994) and it appears to be long gone.

Two additional OPINIONS:

1.) Rapid Fabrication or Free Form Fabrication will never compete with traditional production processes unless there is a vast increase in the kinds of materials available and a dramatic increase in process speed/throughput. This statement in made in the context of HIGH VOLUME ( say 50,000 units per year or higher)

2.) The true strength of FFF lies in its ability to produce geometry of ANY complexity (have you ever played with the Cubital "Brain Gear"). Fundamentally, it can produce ANYTHING that can be modelled in solid CAD. Virtually all of the "Design Rule Knowledge" that is implicitly at work in the design world today imposes subconscious constraints on designs based on the capabilities and limitations of existing manufacturing processes. The consequence of this is that until we can break the creative barriers in peoples' brains, the true advantages of FFF will never be realized. For the most part, we have compared FFF to traditional processes, measured it by the traditional measures and tried to force it into traditional engineering and manufacturing processes. We really need an entirely new paradigm.

Regards,
Sean O'Reilly

Karl Denton warns talk of using fabbers for production is premature until a major breakthrough in technology.

Marshal and others,

Unless there is a change in the equipment that makes it more affordable, more accurate, more repeatable, have better tolerance.... Then the thought of using these technologies for manufacturing will simply not happen. Here at Williams we have tried on several occasions to use the SL technology in a production environment and it simply falls short. That is not to say that it will always fall short but that it does now!

On every occasion that we have tried to use the Sanders equipment it too has fallen short! These and ALL of the RP (fabber if you prefer) technologies are simply not ready to take on the "production" task whether it be short run or long haul! We are still a young industry and as such every new facet that leads us to the end product will be the "hot ticket" at the moment and the one that actually does propel us to the next level may be years in coming. We keep hearing that the manufacturers are building bigger better faster machines but what we really need is a change in technology that is so different that it will equal the initial introduction of RP to the world. Bigger faster better will not get us to production part producing equipment.

We are on the way and as with children we must crawl before we can walk and then run, we still need to crawl for a while!

Karl R. Denton
Lead Engineer
Williams International

Sean O'Reilly adds sobering statistics on speed, says fabbers’ strengths are in making complex shapes.

Phil:

One additional dimension on the FFF as a manufacturing process. In my experience (which, of course is limited to manufacture for the auto industry) one of the critical items is process throughput - how many parts/pieces per hour. I don't know how representative of the entire injection molding universe the following numbers are but, here goes:

For a small, simple part like an instrument cluster lens, the entire cycle - close the press, inject the plastic, allow to cool, open the press remove the part - can be accomplished in under a minute depending on tool material, tool cooling and part material. In addition, most IM molds are 2-, 3-, or 4-cavity molds meaning 2,3 or 4 parts are molded in one shot. The cycle time will go up for this kind of mold, but not a whole lot. For large parts - say 3 ft. long or bigger, the cycle time can be in the 5 - 10 minute range.

For the sake of estimation, let's say that a "small", simple part can be made in 30 seconds and a "big" part in 8 min.

It's hard to imagine any FFF process (which will use an additive, layer-by-layer building process) being able to build a full sized part in 30 seconds. I spent some time several years ago, measuring the "cycle" time of various FFF machines. The fastest (at that time) was the Cubital machine which had a per-layer "cycle time of about 1 min.

Cubital had the advantage of forming the layer in one single flash of its UV lamp and as a result the cycle time was independent of the size of the cross-section currently being built. Most of the others (SLA,SLS, and FDM) saw an increase in the per-layer cycle time as the current cross-section got bigger. (One paradox was that the bigger -meaning more material- a layer was, the faster the LOM could build it - recall that the LOM "removes" material - in a sense) The worst was about 4 minutes per layer.Now, let's assume a space-age FFF cycle time of 30 seconds per layer. At layer thicknesses of. say 0.005 in. our hypothetical FFF machine will take about 200 min. to complete a build of 1 inch. Even if we assume multiple parts per build, the "throughput" will be VERY FAR from 30 sec per part.

At the risk of sounding repetitive, the REAL strength of FFF is its ability to build astoundingly complex parts at about the same rate that it builds "simple" parts and the ability to build anything solid that can be modelled in CAD. Most "complex" products today are made up of a bunch of "simple" (to manufacture) parts.

The idea of a truly complex part being built in a single "operation" is what FFF offers. To take advantage of that we need a new mindset.

Regards,
Sean O'Reilly

Brian VanHiel argues that Burns' vision may be overoptimistic.

Marshall,

I couldn't resist....

I ask you to think back to the days of the atomic bomb and ask yourself if you could have foreseen fission powered-vehicles. Think back to all of your old Popular Mechanics, and ask if you could have foreseen the heli-car. After the moon landing how long did you think it would be before we had a lunar luxury resort? What happened to my interactive 3d television? Where is my 'paperless office'? It's 1999! Why arn't I living in an orbiting colony? Where is my personal robot? Why arn't I talking on a holographic videophone?

Technology doesn't always go the way we expect it to. Fabbers will move towards your envisioned destiny. Maybe they'll get derailed by some unforseen constraint, maybe they won't. In the meantime I'm not buying any more 3D stock.

-Brian
*** Brian VanHiel - Mech. Eng. - Nordson Corp - bvanhiel (at) nordson (dot) com ***

Michael Vincek suggests this debate is a waste of time.

All

You guys have way toooo much time on your hands.

Mike

Roger Spielman reports he’s been using SLS to make end-use aerospace parts for years.

Marshall, Karl, Sean, et.al.

[Marshall Burns wrote:]
> Concept --> Iteration --> Concept......Parts? (love it)

[Karl Denton wrote:]
> Unless there is a change in the equipment that makes it more affordable,
> more accurate, more repeatable, have better tolerance.... Then the thought
> of using these technologies for manufacturing will simply not happen.

For the past several years, I have been working closely with some very high caliber people who seem to spend large amounts of time and effort figuring out how I can't do the things I have been doing. True, I can think of thousands of things that will be a great challenge with the current technology, but for every thousand of "can't do" parts, there are a handful of "can-do" parts that I can make a significant difference on. I just acquired two new DTM systems based on ROI's exceeding 200% (first year) on this "handful" of parts. And yes, we ARE making real end use parts.

Consider the advantage of focusing on improvement of what we have and fine tuning the process rather than discounting the virtues because it won't fit your immediate application.

Building aerospace parts as we speak

Roger Spielman
Rocketdyne

Ronald Jones speaks of the paradigm shift that fabbers are bringing about.

Phill, Sean, Allan, Marshal and anyone else who wants to wade in,

Some time ago I became acquainted with the concept of paradigms and paradigm shifts. I believe part of the explanation as to why the manufacturing world views RP as a way to improve upon the mechanized industry model and fields such as medicine, forensics, architecture and fine art see RP as an improved method for customized fabrication relates to the fact that most paradigm shifts originate from the "fringes" of a field, not from the center. In fact, I see most of the breakthroughs to occur in the next few years will be outside of established industry. Then through a reverse process, manufacturers will adapt to an increased market demand for "custom" products that can only be cost-effectively produced using equipment that had its origins in what we call RP today.

But this whole conversation reminds me of a technique that Heinlin, a science fiction writer developed called "future history". Using this technique, writers create a future world, but recognize there must be a logical progression for the evolution of the culture as well as technology. Given that notion, you can use your imagination to envision a future world where products are mostly custom; complex systems like spacecraft and aircraft carriers are self-replicating and even self-improving. And picking up on Marshal's thoughts, I agree that e-commerce will become the distribution backbone of the world economy and production will adapt to anything which provides competitive advantage using that medium. Perhaps within the next few years, we'll see literally thousands of new service bureaus dedicated to new custom applications that are directly marketed to customers through the internet. By default, manufacturing as we know it begins to diminish and we return instead to the cottage industry that preceded the industrial revolution but updated to the 21st Century.

Phill Dickens argues with Osborn’s prediction of the continued dominance of tooling in manufacturing.

Nick,

Please find comments below:

> Surely the future for RP is to work towards "instant tooling" which is fast and cheap
> enough to negate many of the problems associated with production tooling today?

This may well be the case for many situations especially when you want lots of the same part. However, we need to remember the limitations of the different Rapid Tooling processes in terms of durability and also the different mechanical properties of the mouldings due to different cooling rates than in metal tools.

Also I would argue that tooling is the last thing industry needs! If a manufacturing process needs a specific tool to make a product then it has instantly become inflexible. The design becomes locked whether it is good or not and there is resistance to change because of cost, risk or lost production.

I am not saying that we will use RP to make everything. We will still need conventional processes and tooling where numbers are high or where we cannot get what we want with an RP type process.

> I concede that (for instance) an SLA 7000 building trays full of
> widgets can outstrip conventional injection moulding on cost
> (initially anyway) - it will be interesting to see your new
> figures when they are available - but how many widgets have you
> seen recently that are made from UV resin? Just look at the
> mechanical performance differences between a prototype part that
> is machined from solid thermoplastic vs. the same form but
> moulded condition. OK, so in some circumstances a high grade SLA
> material may suffice, but I would argue that, even for complex
> parts, building ONE high res. RP master to make ONE (fast, low
> cost) rapid tool is a more efficient route.

In the end it all comes down to money. If an RP type process can make what you want at a lower cost then use it if not use another process. We also need to not get trapped into present thinking. There is a variety of RP processes now producing parts in a variety of materials. We have only just started along this route and I am sure we will see more processes and many more materials in the coming years.

> I would be very interested to work with you to provide proof of
> the above statements for a range of different part sizes,
> geometries, etc. Please let me know if you have any interest in
> this offer.

I am not sure that concentrating on direct cost comparisons for existing designs is the best thing to do (although we will need to do some of this). Lets make use of the advantages of the RP type processes and free up the designers and the manufacturing organisation.

We will re run the three parts on the Flymo study and check the new numbers. When we have done this I would be very grateful if you could make a tool with your process and we can do a straight forward break even analysis.

Many thanks
Phill

Karl Denton argues that Spielman’s low-volume examples do not portend real production applications.

Roger,

Rapid Prototyping is not and will not be "the" manufacturing process of the future! There are several very small pockets of companies that are using these technologies to help save them time and money of prototyping and yes maybe small lots of parts. This is a LONG WAY from production runs particularly for the automotive, toy and consumer electronics industries where they produce millions of parts per year! I too work for an aerospace company and if we could fine tune the process a bit more we could conceivably use the SLA machines for production pattern makers. But that is only because we produce a small number of a given part per year. And this would only be on parts that "fit" the tolerance and accuracy range of the SLA machines. They would not come close to producing our bladed products with anything that resembled accuracy, Sanders equipment is way to slow and immature to be considered production! When the fundamental processes change so that they will work and work en-mass producing the same parts the same way then maybe we will come close to production.

Karl R. Denton
Lead Engineer
Williams International

Charles Overy implores Gladieux to include nonmanufacturing impact, e.g. catching design flaws, inventory reduction.

Sean,

Reading many of the replies on the list (great thread!) It seems to me that you need to carefully define your terms. As you have chosen a part that has already been designed and built you are eliminating the whole design process. If you ignore the reduction in design cycle time and soft costs of error reduction etc. from rapid concept builds then you will almost definitely ignore most of the benefits of current RP machines. My take on the market is that most commercial machines are sold to generate strategic advantage and soft cost reduction - these costs are difficult to quantify which is my opinion on why the market for the machines is slack despite keen interest - however the costs are real.

So if you consider just the cost to manufacture will you consider amortizing the machine costs, how will you deal with labor costs, tooling changes etc.? The problems of accounting for these in a large shop like Caterpillar is very different from a small shop.

Also when you consider just manufacturing you ignore the whole back end of the product/part life cycle. The costs here are numerous and large including inventorying and shipping of original parts, replacement part inventory, cost of carrying inventory, inventory taxes, short run or redesign if necessary to support premature failures and above planned replacements, inventory disposal ( which is often more than the value of the parts - especially if you work for the government or government contractor!), parts distribution network including associated escalating IT costs to track and manage etc. etc. etc.

Overall, I would argue that if you just look at the straight cost of goods and direct charges (tooling, labor and machine time etc.) the result is a foregone conclusion for a "standard" (whatever that means) part in volume. However if you consider the entire life of the part and its relationship to the product and even future products in the line - then the figures will change substantially.

Perhaps you can even convince some machine vendors that a large market exists outside of the realm of "strategic advantage" (read price premium) and in the realm of numbers that corporate controllers and yes, even small business people can justify. If you really want to get people excited try going out of the mechanical design shop with your friend at Caterpillar over to inventory or accounting - tell them that the economic order quantity on just 5% of the parts in inventory is now 1 and lead time is 24 hours.

Good luck
P.S. Don't really take any of my suggestions - It will make your life hell -

Charles

Laser Graphic Manufacturing
Precision Models for Architecture and Development

Oliver Cole wants to know what he can profitably make on his FDM now!

Dear Fabbers,

I was very excited to see all your opinions re-RP being "the manufacturing process of the future", as an independant RP bureau we are however compelled by the financial realities of life to look at the situation as we have it now, today.

I was really interested to see Roger Spielman's contribution and I like the idea of looking for those "can-do parts". Perhaps Roger could elaborate a bit on the types of parts, materials and quantities that he is producing?

Trying to produce small batches of high value parts on RP systems does look like an interesting possibility. What other industry sectors have applications that could prove viable?Do we (RP bureaux) need more feedback from end users to identify potential niches?

Which RP process is most suited to this? how important are material properties, cycle times part costs etc? (We have Stratasys FDM technology that produces ABS plastic parts)

Are their any UK RP-MLers reading this that would like to investigate potential applications? If so why not get in touch. Iam sure that this is a side of RP that we should look at carefully.

Yours,

Oliver Cole
Technical Director
RP Solutions Ltd
Tel: 01203 632 120
Fax: 01203 632 131
Email: rps@dircon.co.uk

Brock Hinzmann tells Cole that selecting items to manufacture by fabber is tricky.

Oliver,

In response to a request from Elaine Hunt for product ideas (If I Had a Machine...) from some years ago, I wrote a poem that was eventually made part of a paper on the Personal Factory (I think a copy still resides on Marshall Burns's Web site at Ennex). That same paper, however, is full of all sorts of caveats. While each of us can probably think of a few things we would like to fabricate for ourselves, it would really make more sense for someone else to make them for us, using input from us. It's up to you to find those >>fringe<< customers willing to deal with the current limitations of the technology.

Brock Hinzmann
Technology Navigator
SRI International

[Editor's note: Hinzmann's paper on the personal factory can be found here.]

The Hunt Thread, November 1999

1. Elaine Hunt asks who’s used fabbers to make “a model that was the final product,” and what the results were.
2. Greg Pettengill says fine art applications are the best examples.
3. Andy Christensen says medical modeling is one of the few production applications of fabbers.
4. Marshall Burns says he’s working on articles for Rapid Prototyping Report on this subject.
5. Carsten Freyer expresses confidence that fabbers will be used that way one day.
6. Nick Osborn gives examples of Sanders tooling and masters, production human busts by LOM, and others.
7. Dan Eifert tells of using SLA and Actua to make molds for production wiring harnesses and hand grips.
8. Ron Clemons cites hundreds of end-use SLS parts, e.g. utility pole guard, gives four reasons such projects come up.
9. Glenn Whiteside echoes Freyer’s excitement, wonders why the vendors aren’t leading the charge on this.
10. Robert Welther says many customers use DSPC castings in production, e.g. motorcycle engine barrels, explains benefits.
11. Tom Richards says they often ship 25 to 75 castings made from fabbed masters; he assumes they're used in production.
12. Nuno Reis refers to a publication and a conference that highlighted this subject.
13. Tom Richards mentions Molecular Geodesics is also doing fascinating work in this area.
14. Imamura asks Richards for clarification of his casting terminology.
15. Tom Richards explains his casting terminology.

Elaine Hunt asks who’s used fabbers to make “a model that was the final product,” and what the results were.

How many of you users have used a RP technology to produce a model that was the final product? What was the product and how well did it perform? Are any of you considering RP as a manufacturing process for any product you make? Why not?

Elaine
*******************************************************************
Opinions, suggestions, and other controversial matter VOID where prohibited.
******************************************************************
Elaine T. Hunt, Director <elaine.hunt (at) ces.clemson (dot) edu>
Laboratory to Advance Industrial Prototyping
Clemson University 206 Fluor Daniel Bldg.
Clemson, SC 29643-0925
864-656-0321 (voice) 864-656-4435 (fax)
http://rafiki.vr.clemson.edu/credo/persall/persall.html
http://www.vr.clemson.edu/rp/
http://www.vr.clemson.edu/credo/rp.html

Greg Pettengill says fine art applications are the best examples.

Elaine,

I think that the most common occurrence of this is Fine Art.

Best Regards
Greg Pettengill
http://www.indirect.com/www/greg_p/index.html

Andy Christensen says medical modeling is one of the few production applications of fabbers.

Elaine,

The medical modeling industry is probably one of the few represented here that use rp as a production method. Because of the 1-up nature of this business it works extremely well. That's not to say it couldn't be improved with better, faster, cheaper processes...

Regards,
Andy Christensen
Medical Modeling Corporation
17301 W. Colfax Ave., Ste. 300
Golden, CO 80401
303/273-5344
303/277-9472 fax

Marshall Burns says he’s working on articles for Rapid Prototyping Report on this subject.

Elaine,

That's a great question. I don't know if you've seen my new column yet in the RP Report about the future of fabber technologies, but starting in December, I'll be doing a series on this very subject of direct manufacturing in fabbers. I'd be very interested in seeing the responses to your question above, and will mention some of them (if nonconfidential) in my articles.

Best regards,
Marshall Burns
President, Ennex Corporation

Marshall (at) Ennex (dot) com
Los Angeles, USA, (310) 824-8700
www.Ennex.com

[Editor's note: See Burns’ article in the December 1999 issue of Rapid Prototyping Report, entitled UPS Foretells the Fabber Revolution, which talks about a future in which people order products on the Internet and have them made on their own fabbers in their homes and offices. A similar article was published in the January 2000 issue of Zone News, called Fabbers and the Internet—Fulfillment in the 21st Century. This scenario is a good deal further in the future than many RP-ML subscribers are willing to think seriously about.]

Carsten Freyer expresses confidence that fabbers will be used that way one day.

Hi Elaine, hi to all,

yes, why not.

I think this could be one of the strategies for future. Of course, it won't be the solution for all the world's problems, but why not considering about understanding "RP" as "Rapid Production" instead of just "Protoyping". In many cases - and that in future much more than today - the RP/T techniques may open ways we did not had before in terms of time needed or features that will be able to be produced.For many fabricants there seem to be still some walls in their heads that have tried to break down when talking about RP: "OK, that was the prototype - it's fine, but now let's come to the real parts..." Please do not misunderstand me, but in many cases this seems to me the way of thinking. I think it is a question of acceptance and confidence the people have into our RP-technologies. However, that does not disengage the manufactuerers and researchers dealing with RP from having to do their very best to improve the results we have now. When the first guys have been jumping from hills with selfmade wings, nobody did also expect people flying to the moon some day....

Have a nice weekend,
best regards,
Carsten :-)
___________________________________________________
Dipl.-Ing. Carsten Freyer
Fraunhofer Institute of Production Technology IPT
Steinbachstr. 17
D-52074 Aachen
Germany
Tel. +49 (0)241 8904 124
Fax. +49 (0)241 8904 198
MailTo:freyer@ipt.fhg.de

Nick Osborn gives examples of Sanders tooling and masters, production human busts by LOM, and others.

Elaine

In terms of precision waxes, we routinely use the Sanders system to make small numbers for direct casting, or as masters for our Precision Silicone Tooling (PST) process. As Tom Richards and others have mentioned previously, Sanders parts burn out fine and produce good metal castings.

Following on a recent topic, we have built full size busts of human heads in LOM, which were lacquered and left as the final "objet d'art" - see website for small pic. of unfinished version.

2 " high heads work well in Sanders for casting in solid bronze (the weight seems to impress lay-people as much as the technique, which is interesting...)

For moulded products we use Swiftool, so that the customer can choose the material, rather than being limited to the native RP material / structure (layered, microporous, etc.)

On sintered / infiltrated metal parts from DTM, Jim Williams from Paramount had some nice looking parts at the TCT show recently.

Regards
Nick

Nick Osborn
Managing Director
------------------------------------------------
Swift Technologies Ltd.
140 - 144 Station Road, March,
Cambs. PE15 8NH, UK

"Real Parts Real Quick - Because Time Is Money"

Tel: +44 (0) 1354 650 789
Fax: +44 (0) 1354 650 799
www.swiftech.co.uk

Dan Eifert tells of using SLA and Actua to make molds for production wiring harnesses and hand grips.

Hello all!

I have been using the output from our company's 250 and 500 SLA machines as well as the Actua for making potting molds. I am in the field of Electronic Packaging and have been responsible for designing custom wiring harnesses for aerospace applications. Fitting wire bundles with connectors sometimes requires molding the wires off of the connectors at very specific directions or angles. I have used the thermoplastic from a 3D systems Actua to make such molds. It was easy to use and could have been dissolved in hot water (so I'm told) but I didn't want the water in my connector, so I froze the entire assembly and the mold came off in pieces but did not stick to the connector or potting. Subsequently I used a mold release to avoid destroying the molds and got up to 3 shots from the molds. On another occasion I supplied a subcontractor with an SLA mold which they loved. I started with a cover form in Pro-E and added mass to the thin areas. The final mold had metal inserts installed for screwing the pieces together, a small "V grove" at the edges to act as a dam for the viscous potting material and the ability to see into the mold for voids was a definite bonus which avoided much rework and rejection. Our company has also molded a hand grip for use as a "show and tell" for a PC based simulator but the plastic was strong enough and since only 3 pieces were needed they were used as the final product.

Dan Eifert
Electronic Packaging
Lockheed Martin, Missiles and Fire Control - Orlando
E-Mail: dan.eifert (at) lmco (dot) com

Ron Clemons cites hundreds of end-use SLS parts, e.g. utility pole guard, gives four reasons such projects come up.

We've produced hundreds of SLS parts that have been and are being used in final products. They have been used in robotics computers, printers and as templates for assembly in the manufacturing of telecommunications equipment to name a few.

One of the more interesting applications was when a utilities company had us produce reverse engineered parts because they were discontinued by the manufacturer. As I remember, they were guards used to keep snakes and critters from getting into the electrical works that are mounted on telephone poles. (No one likes a chicken snake in their face when they're 20 feet off the ground!) Some of the reasons for using SLS parts as production components have been delays in tooling, low quantities needed, product discontinuance and quick turnaround. We've never received a complaint for the performance of these parts that I can remember. SLS is most appropriate for production applications when the parts are small and needed for function, not cosmetics.

We've been amazed at some of the clever ways our customers have used our SLS parts.Ron
***********************************
Ron Clemons
Harvest Technologies
Where Innovation Takes Shape

Glenn Whiteside echoes Freyer’s excitement, wonders why the vendors aren’t leading the charge on this.

>"When the first guys have been jumping from hills with selfmade wings,
>nobody did also expect people flying to the moon some day...."

Great stuff Carsten. With some more material development effort there should be no reason why we can't move to the next level and have "Rapid Production."

This is the next frontier to conquer and will open up a whole new manufacturing industry based on customization, small lots, just-in-time production, quick-response manufacturing (QRM), production on demand (POD), etc.

RP equipment manufacturers should also be leading the charge on this but I have heard nothing from them on this subject - has anybody heard from them on this?

Regards,
Glenn Whiteside

Robert Welther says many customers use DSPC castings in production, e.g. motorcycle engine barrels, explains benefits.

Many of the functional prototype castings we provide are also the "production parts" for some of our customers. One such customer is racing a 500 GP class motorcycle in the FIM Grand Prix circuit that has aluminum engine barrels cast using the DSPC process. How well will it perform? On the dyno and track, great. I guess the rider will have the real challenge.

For large production runs we usually transition from pure DSPC cast parts to hybrid production tools that we make using this RP process. For hundreds and thousands of castings, conventional tooling is still going to be the most cost effective method. But getting to that production tool stage, making those tools once, and correctly the first time, is the value that SFF (solid freeform fabrication) methods yield. We're seeing an increase in this paradigm shift among engine designers to go to these SFF methods to functionally test their designs before launching production tools. What this will bring about is better products through the flexibility to test multiple iterations, in parallel, in a compressed time frame and ultimately production tooling that is on time, on budget and made correctly the first time.

Robert Welther
Soligen, Inc.

Tom Richards says they often ship 25 to 75 castings made from fabbed masters; he assumes they're used in production.

Elaine: We're a foundry, not an O.E.M. Our customers are O.E.M's. However, we are more and more shipping quantities of 25 to 75 pcs of Direct RP precision castings. I think it's reasonable to assume that they're for production use.

Nuno Reis refers to a publication and a conference that highlighted this subject.

Dear Marshall and List,

You may find a large part of the ONR funded research on direct manufacturing using SFF technologies on the Naval Research Reviews, No. 3, 1998 (the whole volume is entirely dedicated to this subject).

In addition, there was a workshop held in Oxford last year to discuss materials opportunities in layered manufacturing technologies. This was organized by John Halloran (U. of Michigan) during his sabbatical in the UK. I believe John will make his report available if you request it. Interesting key findings included the fabrication of a new class of optimally designed "artificial materials" (using the homogenization techniques introduced by Kikuchi and others) that display a proscribed property tensor, and VPC (Virtual Pattern Casting for single crystal turbine blades) using ceramic stereolithography.

Regards,
Nuno

Tom Richards mentions Molecular Geodesics is also doing fascinating work in this area.

Molecular Geodesics of Boston does fascinating work in this area.

Tom Richards, Metallurgist

Imamura asks Richards for clarification of his casting terminology.

>Elaine: We're a foundry, not an O.E.M. Our customers are O.E.M's. However,
>we are more and more shipping quantities of 25 to 75 pcs of Direct RP
>precision castings. I think it's reasonable to assume that they're for
>production use.

Dear Mr. Richards,

What is the meaning of Direct RP precision casting?
I's means RP master?

Best reagards,

Imamura

Tom Richards explains his casting terminology.

>What is the meaning of Direct RP precision casting?
>I's means RP master?
>Imamura

In Direct RP the RP built pattern is destroyed, as in the lost wax investment casting process. In Indirect RP the RP pattern(s) are used to create a wax injection mold from which a number of prototype castings can be economically produced.