Rapid Prototyping 1997
Matthias Böttger + Peter Fennell
Introduction
Rapid prototyping is simply the production of 3D CAD models that you can handle and use.
Simple to say but for those of us, who are only now beginning to get using the 3D capacity of CAD in our design, and are still excited by printing a perspective, the idea of ‘printing’ a physical three dimensional model would seem pure science fiction. It is, but it is also real. This report investigates the very tangible collection of processes already achieving extraordinary commercial viability under the collective sound byte: ‘rapid prototyping’.
Reasonable well established in the sphere of product design, it seems to have been born unnoticed by architects; notwithstanding which, those who hear of it immediately become excited by the potential. The end products can be highly finished and incorporate fully-enclosed internal volumes with extremely slender walls if required; yet they can be robust to the point of being employed as structural test parts. The range of materials succumbing of these various processes is equally tantalising: paper laminates bearing a striking resemblance to wood; nylons, polycarbonates and glass-reinforced plastics - pure white, transparent or otherwise; and perhaps most astonishing of all, metal alloys (steel, tungsten carbide et al).
Legend has it that Sir Christopher Wren used to carve large scale mock-ups in timber for his masons to work from. Such detail is rare in architectural design today; masons are few and short of work and such labour of the architect unheard of. Sir Chris’s process is certainly not the model of current architectural practice; the process has been blown into altogether different waters and so have forms. However, RP and its unforeseen offspring may have ramifications to send architecture much further adrift from those days … or perhaps return the ship through 180°.
RP could render mason and modelmaker alike redundant and yet turn back the tide of mass-produced uniformity. Usher in an age where form can be realised unfettered by manufacturing constraints then design horizons become frighteningly infinite. The architect holding the Zeitgeist of new technologies as God would be morally obliged to explore wonderful and difficult virtual 3D form (because he can) to be prototyped on his desktop and cast with ease from the prototype - he would have to become a sculptor like Wren only with pixels instead of wood and RP instead of masons.
The brief history of RP
The phrase was first coined in the mid-80´s for the original stereolithography; the relatively crude process developed at Texas Institute by Charles Hull. That first process could not create hollow trapped volumes nor handle delicate projections or walls more slender than ±2 mm. The capital investment was colossal and models were further limited in size to 150 mm3. Today all those constraints are tumbling and Moore’s law for computer technology would probably apply: that processing power doubles every 18 month and cost for the previous generation halves.
Today, resolution can be refined to as little as .003 mm; a stand alone machine can be acquired for a mere £49,100; build envelopes are up to 1 m3 and a search on the Internet for ‘rapid prototyping’ will uncover 18.000 hits for the exact phrase. Imminent improvements include unlimited size, extended range of materials and, of course, the daily operation of Moore’s law.
Rapid prototyping is leaving its infancy; when it reaches puberty it will be called rapid manufacturing; and when it comes of age it will be casually known as desktop manufacturing.
This is, providing we still use desks.
State of the art …
The purpose and employment of RP are, though extraordinary, rather simple; the theoretic physics which underpins them is quit the opposite. We are indebted to the indefatigable work of institutes of technology, primarily in the USA, for this practical application. Partial absorption of actinic laser photons by photopolymers is one such theoretical field with enough syllables to discourage further enquiry. Suffice it to say that this stuff is state of the art.
Back in the real world, it is the way in which RP forges links between well-developed existing technologies - which were previously unconnected clusters - that makes it so significant. Some of these could be considered state-of-the-art in their own right; the following examples participate variously before, during and after the prototyping itself.
The 3D digital scanning process, Computer Tomography (CT scan), is used to digitise (create a computer model of) existing objects - from sculpted wooden products to the skulls of head injury patients; in the latter instance, the model is manufactured to tolerances of +/- 1.2 mm and can be put up to plan bone repositioning and correction.
3D CAD models in AutoCAD, standard Drawing Exchange Format (DXF) etc. are converted into Standard Template Library (STL) pattern files. This represents the typical (and most logical) procedure to directly drive the manufacturing process.
Having produced a prototype, various casting techniques employ it in the same way that they would conventionally employ a sculpted wood, clay or wax template to create the 3D negative cast which, in its turn, is used for mass manufacture.
Pratt & Whitney’s Rapid Prototyping Casting Laboratory, in East Hartford, Conn. on investment casting: “Last year, we made 2000 RP castings at a cost between $600,000 to $700,000. Using previous methods, the cost of those castings would have been about $7 million. That’s an order of magnitude reduction. In addition we’ve realised time savings of 70 to 90 percent.“
Alternatively, the prototype is itself a negative, further reducing the stages of production by creating a pattern for moulding or soft tooling.
Porsche have replaced the sand-casting process altogether by rapid prototyping stressable metal components for immediate incorporation in test engines and estimate cost reduction from $75,000 to $120,000 each time.
Imminent architectural significance
RP is currently serving product and engineering design where the tolerances are demanding. Architectural elements are frequently larger than available envelope sizes; further, they rarely demand such tight tolerances - despite the tendency of high-tech designers towards highly engineered details. Consequently, RP has not made much of an impact - for parts that are small enough, the price is higher than cruder alternatives which nevertheless meet the required tolerances. However, prices are coming down fast. The SLA250, which cost £250,000 in 1993, cost £130,000 in 1996, which though not quite rivalling the computer processor revolution, is nevertheless indicative of the typical downward trend of new technology - which alternative older technologies will not mirror. So, if for instance RP is not commercially viable for prototyping a custom Planar Glazing then it will be very soon. As envelopes increase, prototyping 1:1 elements will become very real.
Another application within reach is scale modelmaking - for which large architectural firms are accustomed to paying thousands of pounds. In house, 3 model makers on £8 per hour would cost about £1000 a week - equivalent over 1 year to the price of Modelmaker 2 which is already more accurate than the SLA system. It is not that modelmakers need to be superseded, but if the trend continues, in 2 to 3 years there will be considerable cost savings to be made. The principle advantage to the designer would be instant feedback on his 3D CAD designs. Modelmaking would become available in more instances and with more immediacy.
In 1993, entry level 3D printing systems were forecast for 1994 to cost $20,000. We were not able to discover whether this came true but it is comparable with the cost of the first Canon Colour Laser Copiers appearing in architectural offices. It would not be unreasonable to predict similar case histories leading to ubiquitous desktop prototyping.
Large format colour printers in offices tend to encourage a certain profligacy. One partner complained of his staff re-printing three times A1 colour plots for the sake of spelling mistakes - each of which would cost £30-100 from a bureau. Likewise one could foresee architectural offices awash with discarded plastic models - but this detracts little from the appeal and utility of desktop prototyping/manufacturing.