self-designed and built architecture
Michel Bauwens
physical design co. seems very significant,
Could you possibly comment on it for our blog?
Many thanks!
Michel
http://mass-customization.blogs.com/mass_customization_open_i/2009/03/interview-the-next-generation-of-architectural-design-daniel-smithwick-from-physical-design-co-on-a-.html
The central idea behind Physical Design Co. is to provide consumers with easy-to-use online tools that engage them in the design and manufacturing process and enables them to become the producers of their own architectural-scale designs. Our web platform also allows consumers to utilize local manufacturing via our distributed fabrication network which not only reduces carbon emissions, but it also strengthens local economies. Essentially, we’re re-thinking how our built environment is designed and constructed – with the Physical Design Co, online users, whether they live in rural China, or they are busy professionals interested in design, they can now play an active and participatory role in the built world around them.
--
Working at http://en.wikipedia.org/wiki/Dhurakij_Pundit_University - http://www.dpu.ac.th/dpuic/info/Research.html - http://www.asianforesightinstitute.org/index.php/eng/The-AFI
Volunteering at the P2P Foundation:
http://p2pfoundation.net - http://blog.p2pfoundation.net - http://p2pfoundation.ning.com
Monitor updates at http://del.icio.us/mbauwens
The work of the P2P Foundation is supported by SHIFTN, http://www.shiftn.com/
Bryan Bishop
<michel...@gmail.com> wrote:
> physical design co. seems very significant,
"Daniel wants to offer another alternative: You design your dream in
SketchUp, the free CAD software by Google, and his company will
translate your uploaded design in a custom kit of interlocking CNC-cut
parts that you can then easily assemble after delivery. His promise:
"With Physical Design Co Web Platform anyone can design, remotely
manage production, and assemble their own full-scale inhabitable
creations!""
"Second, we have developed a patent-pending technology which
automatically translates the user’s design into a unique kit of
interlocking, easy-to-assemble parts. For example, let’s say you
wanted to design a backyard shed. Instead of having to digitally
model all of the individual parts, consider how they all attached
together, worry about the structural integrity and verify that it is
indeed possible to put it all together, with the Physical Design Co.,
all you have to do is model the shape of your design. Our technology
automatically and digitally translates the design shape into a
kit-of-parts that can then be CNC fabricated and subsequently
interlock together without the need for nails, screws or any
additional hardware., "
http://www.physicaldesignco.com/
http://www.physicaldesignco.com/blog/
I wonder what algorithm they are using to split up surfaces and other
structures. Basically, the way that this would work is that users
upload the 3D CAD files, and the surfaces are split up into workable
units, and then each intersection between part is automatically made;
I suspect that what they are doing is using a limited number of
interconnection types from a specific pool of possibilities that they
have already done math for across the combinatorial possibilities. So,
anyway, given a 3D design file, you cycle through all of the faces
(surfaces), edges and then convert this into a set of connections
between the individual parts. But they do claim that they allow a
unique kit of interlocking parts- so perhaps the interlocking parts
are unique to each individual kit? That makes replacement parts
slightly harder to come by, so I don't know whether or not they are
really doing that. Anyway, for any two intersecting surfaces, you can
do a simple imaginary extrusion and then take half of that, and that
can be a 'hole' or depression made for the other part to fit there,
but are there any other type of connections (hooked connections maybe)
that might be appropriate here? My intention is to figure out the
types of interconnections and go off and implement my own (free)
algorithm for this.
Does anyone have any thoughts re: structural integrity calculations?
Part mating is geometrically simple, but how do you know what the
structure is going to be rated for in terms of maximum load and so on?
For instance, have it build a skyscraper.
Anyway, this is all quite humorous because I slipped by the Physical
Design Co. display at the Austin Maker Faire without a second thought-
I think I got a splinter or something, but otherwise avoided it. Oops.
Bryan Bishop
> I wonder what algorithm they are using to split up surfaces and other
> structures. Basically, the way that this would work is that users
Well, I was completely wrong. Some further investigation tells me that
this is probably something about shape grammars, which are a specific
subset of generative grammars, much like the graph grammars that are
used in the Automated Design Lab. In shape grammars, the left-hand
side of the grammar rules look for points, lines, planes, solids, and
then the right-hand side is some sort of substitution rule. Whether or
not parts fit together is still an issue of course- I somehow doubt
that it's merely an issue of checking for complementary shape
structure (which would be something closely related to an isomorphic
graph, I think).
Introduction to shape grammars
http://portal.acm.org/citation.cfm?id=1401132.1401182
pdf: http://portal.acm.org/ft_gateway.cfm?id=1401182&type=pdf&coll=GUIDE&dl=GUIDE&CFID=27720622&CFTOKEN=73214644
Abstract: "The theory of shape grammars defines a formalism to address
the ambiguity that quantitative and symbolic computations mostly help
us rule out in creative processes. The theory was first launched by
Stiny and Gips in 1972 and has evolved into a groundbreaking
pragmatist philosophy of shape and design since. The course, composed
of a 2 hour lecture and an optional one-day workshop for 10-12
participants, introduces the fundamentals of the theory and optionally
a venue for attendees to put these to practice in a hands-on workshop.
The lecture will focus on giving some basic knowledge of shapes, shape
algebras, and shape rules in order to explain how shape grammars
translate visual and spatial thinking into design computation.
Multiple examples of generative designs produced using shape grammars
will be presented. The workshop consists of one exercise where
participants will explore spatial relations between a number of
shapes, leading to the production of a series of designs to be built
by hand, out of a prescribed material such as wooden blocks or paper."
On page 73 of the pdf, there's a diagram by Sotirios Kotsopoulos
showing an inversed tree capturing the construction details of some
part/device thing that, as far as I can tell, seems to be mostly
random (the seed design, I mean- which is not a big deal, it's just an
example). The original part is classified into three major components,
which are then each individually split up into their own separate
parts, and then when the smallest prismatic parts are found, the
intersection of two parts is identified. In this instance, the design
can be decomposed into a list of interconnections between individual
parts at the smallest level, and then you could imagine walking back
up the branches of the tree to recover the instructions that can be
used to re-assemble the overall item. (shape/part decompisition- i.e.,
breaking it down into a set of flat faces, much like origami crease
patterns, or STL for the representation of triangular faces)
My searching around for Stiny's papers on parametric shape grammars
brought me over to the "descomp" project, a 'comprehensive' design
computation library licensed under the GPL:
descomp
http://code.google.com/p/descomp/
Interestingly enough, descomp is a project headed by 'thedawnthomas',
who I was duking it out with on the brlcad-devel mailing list a few
days ago about graph grammars for constraints in parametric design,
specifically for BRLCAD. So that's neat.
Anyway, back to the physical grammar that the Digital Design Group has
been keeping under wraps-
http://ddf.mit.edu/projects/PHYS_GRAMM/index.html
"Also a method to shape objects based on structural loading and
mechanical strategies. Calculation and generation of these objects
with features (such as attachments) occurs before the slicing
algorithms found in layered manufacturing software. This paper
introduces a production system as a schema that specifies information
for materializing a design built of units. This system is defined as a
physical design grammar used to transform a design model (surface
model) into sets of scalable descriptions for production by layered
manufacturing. These descriptions can be non-uniform models of the
type shown in; these object-based descriptions differ in geometric
description to satisfy most physical design constraints. In theory,
the resulting geometry is scale-less; it can be used to build a small
model or the full-scale building. The maximum size of each object
represented as b in is limited by the machine size. The artifact
itself, though, is limited in size by the number of parts, not the
maximum length of b,as illustrated in the photos."
So, they are including attachments/interconnections/ports within the
right-hand side of their (parametric?) shape grammar rules. I'm not
entirely sure if parametric shape grammars are parametric in the sense
of maintaining physical units and variables, instead it's probably
more about maintaining the relative size of different parts of an
overall image or something. This could ultimately be used to take an
SVG image made of straight lines and preserve the relative sizes of
different aspects, then go on to generate a 3D model and a set of
instructions for assembling that design, or something- which also
happens to be the idea behind the algebraic origami graph grammars,
but the problem there is that if you start folding paper into the next
Sears tower, you're going to fall through the floor once or twice
(maybe). Also, not only are the attachments on the right-hand side of
the shape grammar rules, but this just might be a system maintained by
friction between components. See this:
http://ddf.mit.edu/projects/CABIN/index.html
"The Instant Cabin is a structure fabricated of one material (plywood)
assembled with muscle and a rubber mallet. Nails, screws or glue are
not needed for assemble and support. The assembly of studs and
sheathing are sustained by friction alone. .. The computer is used to
convert a starting shape into an assortment of specially shaped
components complete with a number. The designer builds a computer
model in 3D, subdivides the object into components then flattens the
objects to a horizontal position in CAD before sending each component
to a computer controlled router. Joinery between components is so
precise that all parts stay together by friction alone."
People to keep track of re: the instant cabin:
* Nic Rader
* Marcel Botha
* Margaret Nelson
* Diana Nee
* Victoria Wang
* Shu Wang
I'd like to see a mathematical proof that the thing stays in place
with friction alone- how does any type of joint or fitting work, for
that matter? In the most simple of case of a vice grip, there is a
compressive force that keeps the object from sliding in an x/y
direction, but what about one of the z directions? Is an object being
held in a vice grip also simply friction? Another issue that comes to
mind is that if you are going to be loading things into the "instant
cabin", there must be a maximum load that each joint can support; as
you walk around a house with floor boards, you hear creaking and
cracking because there are optimal positions over the surface at which
to be that would distribute tension most evenly (and reduce creaking
when tiptoeing downstairs for some water at night, or something). Some
FEA simulation is in order- I wonder to what extent you would see
increased tension or stress near the joints being held entirely by
friction, and how much mass we could stack and when it would break.
It's kind of important information. Other than that, though, they are
claiming that the precision of the CNC tools allows the dimensions of
the fabbed parts to be so precise that friction is enough to maintain
it.
Short paper on "the instant cabin":
http://ddf.mit.edu/projects/CABIN/cabin_mit_2005.pdf
In the paper, Lawrence Sass talks about a 'computable construction
description' and the plight of hand driven markings on wood for
fitting parts together in typical construction projects. The component
attachment methodology is 'tab and slot' included in the manufacture
of each component. Apparently this is called "integral attachment" in
the plastics industry. [13] Later the paper goes on to mention that
"joineries are built based on the relationship of component geometry,
each joint is provided a mating attachment from the list of attachment
types discovered in early assembly testing." So which is it- is it a
constant pool of possible 'tab and slot' joint designs, or is it based
off of the system that is being built?
[13] Genc S, Messler R, Gabriele G, “A systematic approach to integral
snap-fit attachment design”, Research in Engineering Design, Vol. 10,
No. 2, pp. 84–93, 1998.
http://heybryan.org/books/papers/A%20systematic%20approach%20to%20integral%20snap-fit%20attachment%20design.pdf
Abstract: "Traditional integral snap-fit attachment design focuses
almost exclusively on the individual locking features, such as
cantilever hooks, bayonet-fingers, compressive hooks and others. The
positioning and orientation of other significant features on parts,
such as those that facilitate or enhance engagement and eliminate
unwanted degrees of freedom left by locking features, i.e. locating
features and enhancements, are not considered. This paper builds on
relatively new methodologies and guidelines for arranging all
attachment features on plastic parts comprising snap-fit assembly.
Classification of features into categories of locking features,
locating features and enhancements of these is used as the basis for
discussion. A systematic approach to attachment design is presented."
One thing that I've been wanting to work on for a while now is a
simple 2D (and later 3D) canvas where a line can be drawn (and later a
surface) that would define contact between two complementary parts,
and then information on whether or not the parts would fit together,
whether or not they would stay put, etc., when subjected to various
external forces, etc. For deformable materials with snapping joints,
it's hard to figure out how to do that without FEA and lots of PDEs.
From this system you will have a 'silhouette' (like a circular inlet
for a lego brick) that could then be used to quickly scan through a
database of possible shapes, especially if the shape of the silhouette
is reduced to a set of geometric primitives (i.e., you could set
resolution such that a parabola looks like it might fit a triangular V
silhouette in a search process, and then increase the resolution when
you have a handful of possible results that you want to test against,
if you want to be speedy about it). Friction might ultimately be
cheating, you know. Plywood can come with a lot of friction, even
after rubbing sand paper over it for hours, so this might be a
domain-specific deal. Although this plastics interconnect paper
suggests otherwise- elastic deformation and deflection during assembly
(physical mating). In deflection, the parts undergo some deformation
but then recover shape (and "snap") after passing through a lock
structure or some such.
So, it occurs to me that the integral attachment method of connecting
parts together is kind of a bit of a hack. In old computer systems,
there were certain unused bits in memory, which were used in times of
desperation when you really needed extremely awesome performance from
your assembly code (or something). And in the case of integral
attachments, there is a set of forces with certain vectors (particular
directions) that can be used to mate or attach two parts together.
During the operation of the overall device, it's "hoped" that the
motions, forces and vibrations resulting from operation of the overall
system do not result in the dislodging of the parts. But, this isn't a
very good assumption to make about the DOFs (unconstrained
directions). Anybody who has fallen off a chair because the screws or
bolts apparently left their respective homes, know how terrible it is
to make those assumptions about "unused degrees of motion" or
something, where you have specific obscure motions for inserting
objects, because there's still a likelihood that those motions will be
executed in the operation of the lifetime of the part. There's
probably a specific way to come up with a connection shape that is a
special case and shows how it won't be coming apart "in a million
years", but I haven't found a method yet to search for that type of
attachment shape or cross section.
For mathematical models of feature-based attachments, seek the
following references from that paper:
Mobay [3], Lewis et al. [14] [15], Knapp et al. [16][17]
[11] uses a matrix of possible allowable combinations of parts (though
I don't know why- you could just do a geometrical analysis of whether
or not the surfaces and edges conflict with each other, in particular
by aligning a set of edges in one part to a set of edges in the other
part, such that two edges might be touching, or something, which would
be an intense way of calculating the different possibilities of how
one part fits in with another part, if it is at all capable of doing
so- other quick metrics include surface area calculation, etc.).
Wood frame grammar
http://web.mit.edu/ddfg/papers/06_lsass_caad_future_2005.pdf
"friction fit connections"
"dog bones" to help connect "panels to panels" and "studs to studs"
strength is in the corners, this method does not need special
fasteners at connections between panels
Holy crap. What on earth? They wrote VBA scripts for this and called
it CAD? Ugh. So many things are wrong with that.
"The fourth script checks for a proper tolerance connection between
each part assuring a solid friction fit connection. The technique used
to friction fit wood panels follows past
techniques on self assembled building components (Sass 2004)."
"The second paper was a shape grammar based computer program used to
generate sheet metal panels for the manufacture of lightweight parts
(A. Soman et al. 2002). Rules for the grammar are based on the
material and assembly properties of sheet metal and CAD CAM
operations. The grammar is built of rules for notching, bending and
punching sheet to fabricate stereo encasings. Last and most recent is
a paper (A Kilian 2003) presenting a computer program that generates a
puzzle connection between two flat or curved CAD surfaces. The program
calculates the relationship between surfaces then generates new
geometry built of semi circular extensions and subtractions. Each
extension fits into the subtraction of the opposing sheet. The program
generates zipper joints of varying scales based on the level of
curvature between the two surfaces."
Sass, L. 2004: Design for Self Assembly of Building Components using
Rapid Prototyping, Architecture in the Network Society [22nd eCAADe
Conference Proceedings] Copenhagen (Denmark) 15-18 September 2004
pdf: http://ddf.mit.edu/papers/04_lsass_ecaade_2004.pdf
Soman A, S Padhye, M Campbell, 2003: Toward an automated approach to
the design of sheet metal components, Artificial Intelligence for
Engineering Design, Analysis and Manufacturing
(.. hm, don't I work for that guy?)
Kilian, A. 2003, Fabrication of partially double-curved surfaces out
of flat sheet materials through a 3d puzzle approach, In “ACADIA 2003:
Connecting Crossroads of Digital Discourse,” Muncie Indiana, Pages
74-81
pdf: http://www.designexplorer.net/newscreens/proto/Axel_Kilian_acadia2003.pdf
http://designexplorer.net/
http://www.axelkilian.com
Other papers that are probably awesome:
* Kilian, A. Block, P., Schmitt, P., Snavely, J., “Developing a
language for actuated structures”, Adapatables conference Eindhoven.
(2006)
* Kilian, A., Schmitt, P., Künzler, P., Joachim, M., Garcia, E. J.,
Mitchell, W. J., “Designing an actuated vehicle, the H-Series”, Game
Set Match II, TU Delft. (2006)
* Kilian, A., “Design Exploration with circular dependencies - a chair
experiment”, Caadria. (2006)
* Kilian, A. “Design innovation through constraint modeling”, IJAC. (2006)
* Kilian, A., “Design innovation through constraint modeling”, ecaade,
Lisbon. (2005)
* Kilian, A., Ochsendorf, J., IASS Journal (2005)
* Kilian, A., “Hanging chain models and Fabrication”, Acadia 2004,
Waterloo. (2004)
* Kilian, A., “Fabrication of partially double-curved surfaces out of
flat sheet material through a 3d puzzle approach, Acadia 2003,
Indianapolis. (2003)
* Brygg Ullmer, Elizabeth Kim, Axel Kilian, Steve Gray, Hiroshi Ishii,
Strata/ICC: Physical models as computational interfaces, CHI 2001.
(2001)
* Testa, P., O’Reilly, U.M., Kangas, M. Kilian, A. “MOSS:
Morphogenetic Surface Structure. A Software Tool for Design
Exploration.” Proceedings of Greenwich 2000: Digital Creativity
Symposium.
Looks like Axel beat me to it when it comes to the shape optimization
for silhouetted joints and connections:
http://www.axelkilian.com/newscreens/nmm/final/finaljoint/jointmatrix.html
Anyway, the reason I went off looking into Axel was because of the
paper with the '3d puzzle' in the title. In that paper, he mentions
that he's using pressure fit joints using highly precise CNC cuts. Due
to the spatial curvature of the shapes he wished to CNC, you can't
just have a ridiculously tight fit at some random location in three
dimensions, because then the force that you need to apply to the tight
fit to get it to stick might dislodge other parts, or just be
generally impossibly hard to get manual hand access to. Sequential
assembly of individual parts requires the parts to be large enough to
be assembled one at a time, and held by humans and so on. But make it
too large, and the curvature of the part might make the fit too loose
and cause the pressure fit joint to fail. Hm. A standard dimensionless
fitting description language .. wonder how that works?
"2.3 Changes in standardization: With the generation of geometry
through scripts that allow for parameter-based variations, the notion
of standardization shifts from dimension based descriptions to a
topologic system of relations. It is no longer necessary to fix a
certain dimension at the interface between components, but the
relationship of the two parts defines the shared dimension. The
standard shifts to the description of the relationship. It is
important to know how elements relate to each other, but it is not
necessary to know their absolute dimensions. Parametric modeling has
set the stage for the expression of elements as a set of relations
that have variable dimensions. A non-dimensional based standard relies
heavily on fabrication techniques that allow for the production of
varying geometries based on a system, as in the example of the joint
shown here."
"2.9 Shift of the design focus from geometry toward building systems expressed
through generative or parametric systems
Increasingly, the introduction of digital-based modeling will shift
the focus of design
away from fixed geometry towards parametric or generative constructs capable of
producing a variety of geometric output. The important difference is
that parametric and
generative models can be set up to contain fabrication constraints and
material properties
that allow for the exploration of form within these constraints."
"A surface is modeled with the help of a custom-written quadratic
spline surface modeler
in Autocad using an Autolisp script. The surface is defined through
eight slopes, two each
at the surface corner points. The slopes define the incoming and
outgoing slope of the
surface edges of the spline surface. The spline surface is adjustable
in its grid cell
resolution through a custom command. The grid cell resolution
determines how close the
surface approximates the numerical spline surface and is the basis for
the generation of
the puzzle joint detail as well."
"Once the desired form and resolution is achieved one can use another
custom command
to develop the strips in one of the UV directions of the surface as
strips onto the ground
plane. The process uses a crude triangulation technique, but since the
topology of the
surface elements to be flattened is known and each strip has only one
neighboring strip
with one shared edge, this approximation is not a problem. The
flattening technique also
ensures that there are no overlaps between the parts when they are
flattened. Once the
parts are flattened, the joint detail is generated based on the
resolution of the surface strip.
The variation of the edge line spacing is based on the overall grid
resolution of the initial
surface."
"Once the cutting line layout is generated, the testing begins to
fine-tune the balance
between cutting tolerance, material thickness, material stiffness, and
the friction of the
faces of the cut. This balance influences how well the joint will
pressure fit. Since the
pieces are curved, the assembly can only be done one joint at a time. The offset
introduced through the curvature is much larger then in a planar
configuration. The joint
has to have enough play to be assembled. But if the tolerance is too
great, the joint will
not stay connected through the material friction. The actual joint
geometry was born out
of the quest for a geometry that would allow for fastener free surface
connections, which
would be able to transfer forces within the surface, both in tension
and in compression."
"Since the puzzle joint does span over the edge-line of the
developable surface strip, it is
geometrically not a clean component. It is attached to one strip with
single curvature, but
through the joint properties it is forced to lie within the surface of
the neighboring strip,
which does not share the same curvature. It therefore takes on a
partial double curvature.
This actually helps the pieces to approximate the numerically double
curved surface the
strip approximation is based on. But the process is very uncontrolled
and hard to predict.
In many cases it fails, and the bending force pulling the puzzle node
out of its position is
greater then the friction force keeping it in place. However, the
process also has positive
effects. The joints are pre-stressed which increases the friction in
the joints through the
increased forces acting on the joint seam."
"a certain correspondence between surface properties and strip
properties. But it does not
take into account the difference in curvature between neighboring strips. A more
promising approach would be the use of a tiling scheme that analyses
the local curvature
condition and orients the seams in the optimal orientation to the
slope of the surface. A
study was done by the author using Genetic Algorithms for the
description of finite state
automata that would walk across the surface reacting to the local
curvature. It is a
relatively complicated interconnected model that does not allow for
very reliable results
due to the number of parameters that can be adjusted. The main
challenge is to define a
reliable fitness function for selection of the FSA for the strip
production that does allow
for new patterns to emerge but does not limit the selection too closely."
I am still not entirely sure if the quest for a geometry to transfer
tension and force over the surface (instead of through fasteners) was
ever found (even in 'tab and slot' friction joints), and whether or
not that attachment scheme is being used in the instant log cabin, or
in the Physical Design Co. kit generator scripts. Sass says it's just
tab and slot, but then where are the calculations and numbers on
whether or not friction is enough in pressure fit joints? The paper on
'integral attachments' from the plastics researchers didn't really say
much, certainly not doing much in terms of calculating anything. And
then there's the further question of whether or not the kits from
Physical Design Co. are using custom joint geometries for shapes in
different architectures (uploaded by users)- which would mean that
there's some mathematical method of evaluating the performance of
attachment geometry?
This might be a good photograph of a tab-and-slot friction joint:
http://www.ucsart.com/files/u1/frame1.jpg
http://www.ucsart.com/files/u1/frame2.jpg
http://www.ucsart.com/files/u1/frame3.jpg
http://www.ucsart.com/files/u1/frame4.jpg
http://www.ucsart.com/files/u1/frame5.jpg
http://www.ucsart.com/files/u1/frame6.jpg
http://www.ucsart.com/files/u1/frame7.jpg
http://www.ucsart.com/files/u1/frame8.jpg
http://www.ucsart.com/files/u1/frame9.jpg
http://www.ucsart.com/files/u1/frame10.jpg
http://www.ucsart.com/files/u1/frame11.jpg
http://www.ucsart.com/files/u1/frame12.jpg
http://www.ucsart.com/files/u1/frame13.jpg
http://www.ucsart.com/files/u1/frame14.jpg
http://www.ucsart.com/files/u1/frame15.jpg
from: http://www.ucsart.com/how-to-assemble-stretcher
Still need to figure out maximum load calculations. I guess we can
just throw designs into FEA software packages and run simulations, but
some rule of thumb would be amazingly useful that would relate the
size of tabs, slots, coefficients of friction, direction of some
supplied force or something, and whether or not it would break, and
whether or not simulating this in individual units or sections would
be appropriate so that we can just know the general behavior (i.e.,
where the stress will be) overall in modular components. Then I might
feel less silly when I go off to write a tab-and-slot design generator
and parametric shape grammar to one-up Physical Design Co. (in a
friendly way!)
Okay, that's it for now.
ben lipkowitz
> Still need to figure out maximum load calculations. I guess we can just
> throw designs into FEA software packages and run simulations, but some
> rule of thumb would be amazingly useful that would relate the size of
> tabs, slots, coefficients of friction, direction of some supplied force
> or something, and whether or not it would break
for an interference fit, the holding force is easy to calculate.
you need to know:
the amount of interference (distance)
the material thickness
the material stiffness (elastic modulus)
the coefficient of friction
now your normal force is equal to the distance times modulus, and friction
force is normal force times coefficient of friction.
or if you like letters; f = E * d * t * c
for plywood, elastic modulus is about 10GPa and friction is about 0.5
this is a first order approximation which ignores buckling. buckling can
be sidestepped by using small tab/slot size relative to thickness.
Bryan Bishop
> for an interference fit, the holding force is easy to calculate.
Thank you. For the record, the other day when we talked, we were
mentioning a few other aspects of these projects. For instance, if
there somehow is a way to make min-a-max or tensegrity or other
alternative assembly methodologies more simply 'fit' together like
legos, or as easy to assemble as tab-slot, then I can go ahead and
implement shape grammars and other design software that could do
Buckminster Fuller architecture under the hood for everything that
physicaldesignco is claiming to be able to do. Eric, I suspect you
might be particularly interested in this, and it's also interesting
that I mentioned this in my email re: the Austin Fab Lab discussion.
http://groups.google.com/group/openmanufacturing/msg/7bb83ef18ae60f28?&q=%22Hello+world%22
"It's entirely possible that I haven't understood T-slot in general, or
the possibilities that it would allow. Let's say that I could come up
with a way to automatically generate assembly instructions for certain
T-slot designs, i.e., "put this part here, put that part there, etc."-
then for all of these myriad different designs and tools that
structurally use T-slots, would it be useful to have a design
generator? For instance, would it be useful to let the user specify an
input mass and a requirement for force stability (or something), and
then calculate whether or not a certain design supports that? or maybe
it's more useful for motion stability. Again, this is probably just
because of my misunderstanding of what's going on, and if I had a
better understanding, some software could quickly be whipped up that
I'm sure you'd like. I just need some specifications or a better
understanding of the design methodology that comes from reasons to
think about the T-slot in the first place (etc.)."
Another thing that I remember now is to somehow force fenn to write up
a ranty email about his background in stewart platforms and hexapods
(oops, I meant to say "hexayurt" in my last email re: disasters),
which we haven't seen much from him on that subject, so it might be
about time because of these convergent threads of thought.