Thursday, October 23, 2008

Recent comment posts 4

(see links at right for the related postings)

-------------

Jerry Volkman

23oct08

That’s a very impressive amount and depth of study, Jerry. In order to help clarify the workings of such a complex model for the tutors at the final presentation (and for potential users of such a system) could you make some additional explanatory diagrams:

1) relating the variables to plans and sections (or whatever is most appropriate) so that it can be seen directly where a particular variable will have effect; and

2) relating the variable to each other, and to final and intermediate results, so that the chain of dependencies is clearly visible (of course, the GC graph is supposed to do this, too, but I’m sure you could improve on that method of display)

Thanks.

 

-------------

Bert van Diepen

22oct08

It seems you’ve had a long, hard struggle so far, Bert, but you’re getting close. I’ve gathered that your main strategy is to present differently lit renderings of the same scenes, for a collection of scenes. So when setting up the survey, it would be good to cross-check people’s responses somewhat by, presenting the scenes in different ways. For example, you can show the same scene with two or more kinds of lighting, but then also present two or more different scenes with the same kind of lighting. (Also, the angle of view can have an effect on people’s impressions, so it might be useful to include a few oblique (non-central) views for examining this.)

 

--------------

Rein Roosma

22oct08

Thanks for posting this material, Rein. Can you find some way to reduce the number of calculations and/or simplify your model (for example to a single floor, as we suggested last week) so that you can draw some intelligent conclusions about the system’s behavior and feed those back into your design without having to spend a week just running analyses?

 

--------------

Christian van Gruithuijsen

22oct08

Now you’re starting to get some interesting results, Christian. It would be good to see more clarification of the operations (explosion, booleans, etc.) which generate your forms. Then, your next steps could be both taking those operations deeper, to better articulate the architectural organization (not just a skin – though the two skins do make some progress in that direction) and also changing the rules/operations and/or the inputs to those, to see what effect these have on the outcomes. One example of this is, of course, your stated goal of testing the effect of various street patterns on the building form. But you can also look at what effects result from changing the relationships between the context and the project.

(And when presenting your research next week and preparing your report, please be sure to include images and diagrams which support the claims and explanations you’re writing here.)

 

--------------

Rein Roosma

20oct08

Hi, Rein The work on the lighting analyses you showed last time we met looked good, and it would be great if you could post here a summary of it - as well as things you've been doing since then - for the benefit of the tutors and in preparation for next week's presentation. Thanks, AC

 

--------------

Auke Verbraaken

20oct08

So, Auke, how's it going with modeling additional variants to try improving on the problem of keeping the floors horizontal, which you pointed out in your last presentation? By the way, it would be great if you could post a summary of that presentation, since it was a good example of the feedback process: building models to answer a question, looking at the results, and then doing further exploration to improve the results. Cheers, AC

ps Please turn on display of all comments you've received (see examples of your classmates' blogs 'Recent Comments', etc.) for the benefit of the tutors reviewing and commenting on your work. Thanks!

 

--------------

Arjan Klem

20oct08

It's been good to see the way you were able to realize some of the strategies we discussed for translating your analysis results into 3D models of the facade, Arjan. Have you been able to work out the errors you mentioned when trying to model variations in placement of openings? If not, perhaps you could rely on your earlier experiments with introducing openings into the facade and instead look at the possibility of using different materials in the tension and compression zones, as a way of tapping into this difference for a source of expression. AC

 

---------------

Arnoud Herder

20oct08

Really impressive effort on the physical modeling and actuation challenges, Arnoud!

Are you seeing any feedback from this to your ideas about how to incorporate such a tensegrity mechanism into your architectural designs? AC

 

---------------

17oct08 – mtg w/ Mark Min about NetLogo model

 

---------------

Bert van Diepen

16oct08

Hi, Bert

If you’re still not getting satisfactory results from rendering with colored light, I’d suggest you change your approach to coloring the surfaces of your model(s) via their material settings – as is more commony done – and using neutral-colored (e.g. white) light. In this way you can avoid losing more time to technical problems, and it will also give you more control over the scenes you generate, since the monochrome and two-tone images you presented so far do not seem really usable.

So don’t get hung up by the software choice – do whatever you can to achieve some satisfying and meaningful results in the time remaining. (You can even produce differently colored scenes by hand and put scanned copies of them into your surveys.)

AC

 

--------------------

14oct08 – mtg w/ Christian about GC modeling

 

--------------------

13oct08 class presentations, comments

 

--------------------

e-mail from/to Christian:

... yes, Christian, I think this model is showing something more in the spirit which you seemed to be seeking.

However, here, too, as in the model on which I commented earlier today, the spatial effect(s) of the context upon the site/project are still quite simple. So in order to generate the level of complexity needed to really begin architectural 'space-making' I thought that you could use the 'irregular diagonal' model from earlier in the place of your proposed new building.

One way of doing this, I believe, would be to set all of the height control points of the building footprint to zero (or small, anyway) while leaving the surroundings at various heights, and then placing the 'irregular diagonal' model onto the site. Next you could look at ways of linking its control points to the context, rather than being 'independently' driven. This way, you may be able to generate complex (irregular diagonal) building designs both from a rectangular street pattern and from more complex streets patterns like the one you're showing here. I hope this is clear, but in any case we can discuss it further on Monday.

Have a good weekend!

Andre

ps I'll post these e-mails to your blog, also, so you have all comments collected in one place. I don't see any comments (neither mine, nor anyone else's) appearing on your blog, though. It would be helpful for the instructors/tutors to be able to see those also, not only what you're posting yourself, so please turn on the comment-publishing functionality of your blog, if you are able. (If not, see the blogs of some of your classmates, where links exist for 'Recent comments', 'Latest comments', etc. Thanks!

-----Original Message-----

From: Christian van Gruijthuijsen [mailto: xxxxxxxx]

Sent: Fri 10/10/2008 15:16

To: Chaszar, A.T.

Subject: RE: [cvangruijthu] New comment on "Parametric deconstructivism"

Hi Andre,

Thank you for placing a comment on my post. I also agree that the results as shown in the blog 'parametric deconstructivism' are not really what I want to achieve. I understand from out your comment that it would be the best to bring back the adjustable irregular / diagonal streetpattern and relate this to the new to rise design? At this moment I got a model the same as 'parametric deconstructivism' but with an adjustable streetpattern. By adjusting the streetpattern I create the irregular, diagonal negative volumes which forms the streets within the city-grid. Also the footprint of the new to rise design changes. In the supplement I attached a quickly made render from the model as it is now. Do you think I approach this way the research question more then the model you gave comment on?

Greets,

Christian

---------------------

Christian van Gruijthuijsen    Generative design/rule based design and deconstructivism

http://gruijthuijsen.weblog.tudelft.nl/

10 oct 08

It seems you’re beginning to find ways of relating your project’s context (in this case the buildings surrounding your site) to some of the characteristics of your proposed design, which is the basic aim of your research. It’s not clear, however, whether the relationships and results are those which your really want – expressed for example in your earlier model of a volume subdivided with irregular, diagonal negative volumes, which we suggested on Monday would be a good starting point for the proposed design, rather than for the surrounding context.

Therefore, I’d suggest that in addition to the exploration of modeling other contexts, which you’ve said you’re planning next, you should also try implanting that earlier model into the grid you currently have, in order to see what effect a different set of relationships has within the same context.


Tuesday, October 7, 2008

Recent comment posts 3

Dave Koomen

3oct08

Some clarification so as to try to avoid misunderstandings of the feedback you received after the presentation:

- the list of tasks you’ve posted here for your further work needs to be focused on your design research goals. If you are still interested in providing daylighting via lightpipes for a set of underground spaces – rather than in designing lightpipes – then it’s probably better to first use the type of pipes and the type of space geometry accommodated by the software you’re using, and only after that consider whether the software’s limitations are seriously impacting the quality of your design.

Instead of movable lightpipes, you can consider heliostats for catching light from different directions and multiple pipes over different parts of each space with adjustable lids or louvers for accommodating varying outdoor light intensity (due to changes in weather, time of day and day of year), lenses and/or reflectors for redirecting light within the spaces, and so on. These technical devices don’t need to be modeled in detail at this stage, however, since changing the intensity of available light entering into each pipe can simulate their effects reasonably well (assuming this can be varied individually in the software) except for redistribution of light inside the spaces.

So rather than getting bogged down in trying to study all of these factors in detail, it seems better to consider the relationship between what you’re learning about lightpipes and what you’re aiming to accomplish architecturally.

 ---------------------

 

Marco Visser

3oct08

With reference to the discussion after your presentation, the aim in terms of modeling structural behavior is not to sum moments from adjacent spans but to use the equations for continuous spans rather than individual, simply supported ones (still assuming for the time being that the beam section is constant.) Another direction, also discussed there briefly, is to account for varying crossection not only as it affects load distribution and torsion, but also in the resistance to those loads, which would allow using more of the bridge material as active structure, rather than just building a spine and cladding it.

These two directions can also be combined, but in any case there is still not much feedback to the overall design – aside from the crossection depth being driven by the beam depth – since there are no limitations set. Therefore, rather than trying to further refine the behavior of structural modeling, it would be good at this stage for you to pose some research question(s) which you can address through the model you’ve built, and then decide whether more detailed behavior modeling is needed to get the answers.

------------------

 

Bart van den Broek

3oct08

Whichever way you eventually resolve the scaling problems that you mentioned during the presentation in relation to the software you’re using to model windflow, and even if you are able to test some physical models in an actual, not virtual, wind tunnel, it is very important that you consider more than one wind direction for shapes whose plan form is not radially symmetrical, since you would get misleading results concerning wind effects and performance except in the very unlikely case that the building site has only one strong wind direction.

Also, rather than attempting such an exhaustive survey of shapes, it would be best to prioritize, for example by trying three or four which differ from each other significantly and then exploring further in the directions which look promising.

----------------

 

Arjan Klem

3oct08

It seems that your research is motivated at heart by a search for interesting (-looking) forms underpinned by some sort of non-standard structural logic – which is a valid pursuit as long as engineering considerations are not strongly prioritized. It’s important therefore to take into account what you’ve discovered in your experiments so far: that regular, symmetrical loadings on homogeneous materials result in regular, symmetric structural patterns.

Observing this, you can now consider what sources of irregularity, asymmetry and heterogeneity may be applicable to your design work, especially architectural ones such as program, or others based on environmental performance, for example. (Structural motivations alone won’t lead you in this direction, since strictly speaking – in the absence of other requirements – the best structural solution is for all activities to take palce on firm ground.)

Also, in order to not get bogged down in running long, highly detailed simulations, you need to find ways to simplify the models as far as possible while still capturing the relevant factors in the variations you test.

-----------------

 

Mark Min

3oct08

The direction your research is taking is interesting and contrasts strongly with most of your classmates’ work, in the sense that most of them are pursuing topics which are largely geometric in nature. This presents a challenge in terms of the structure and aims of this course, however, since you have not yet been able to construct any experimental framework within which to test aspects of your proposal, nor compare it to other possible proposals.

In the time that remains, it is very important for you to demonstrate some aspects of the proposed façade system’s behavior, rather than relying only upon an argument constructed solely of words and diagrams. The preferred methods in this course thus far are digital/computational, and since there are not sufficient time and resources for you to build any physical mockups to test, I’d suggest you, too, should try a computer model.

Since the geometric modeling tools (Generative Components, Maya, Rhino, etc.) which the others are using don’t suit this problem, I’d recommend you try something like NetLogo, whose agent-based structure allows simple but useful modeling of fluid and molecular behavior. You can download the current version from the internet and look through the example models included with the program. It would be also good if you could begin modeling yourself – at least by working through the tutorials – within the next few days.


Thursday, September 25, 2008

e-mail from/to Auke

... copy of response to your e-mail:

Hi Auke
What I had in mind in the first place when suggesting GC was being able to model how the mechanisms you're proposing alter their shape, separately from consideration of actual loads and material properties
After that, as a second step, you could either export the geometry from your first (GC) models to a structural analysis program (maybe Diana, since that's what seems to be preferred by Andrew and Eliza) or perhaps some structural analysis could be programmed into the GC model, as for example your classmate Arjan is doing ... though I think that option may be too ambitious for the time available in this course.
In any event, yes, I think GC would be good at this stage for studying movement, since such a model would allow you to drive the deformation of the structure both by displacing nodes (which is essentially what happens when the structure reacts to external loads, passively) or by changing the lengths of elements (as in an active structure.)
Andre
ps I'll post this comment also to your blog, so that you have it all in one place.
pps Axel, let us know if you have any comments on this, please.

-----Original Message----- 
From: Auke Verbraaken [mailto:xxxxxxxxx] 
Sent: Thu 25/09/2008 16:42 
To: Chaszar, A.T. Subject: 
RE: SUA comment: 11386-080901  

Hello Andre,
Thanks for your comment. It was my aim to indeed at least model one variant for Monday. Is GC the most suitable program to model the different variants? Is it easy to exchange a GC model to a program to test the performance, or is it possible to test the performance in GC itself?
Auke Verbraaken

Recent comment posts 2

Auke Verbraaken            Active passive moving buildings/structure
http://averbraaken.weblog.tudelft.nl/

25sep08

Your diagrams give a nice analysis of system types whose performance you could explore.

It would be great if you could model at least one of them in time for the midterm presentation, with the aim of modeling all four (or more if other ideas come to you in the meantime) by the end of the course, allowing a direct comparison of their performances under identical load conditions.

Are you able to start modeling, for example in GC?

Gr, AC

--------------------

Christian van Gruijthuijsen    Generative design/rule based design and deconstructivism

http://gruijthuijsen.weblog.tudelft.nl/

25sep08

Congratulations on this step! It looks like you've developed a good way to generate the city 'grid' - which is similar in this example to recent work from Hadid's office, for instance as shown by Patrick Schumacher at last February's SmartGeometry event.

Of course, this way you don't really obtain the contrast between your building and the rest, which you've said you're aiming for, but you can always set the generating 'street' forms to rectilinear for that. In this case you can perhaps 'flatten' your building site by building a new volume whose footprint shape is the same as the site's (but perhaps offset outward if the surrounding street walls are not vertical), giving it sufficient height (or depth, in the downward sense) and doing a further boolean subtraction with it.

On the other hand, if your intent is to keep the surroundings rectilinear anyway, then perhaps it's best to model that by starting from a grid of points - irregularly spaced, if you prefer - and putting a 2-D array of planes and vertical vectors on that set of points, so you can build simple boxes with independent heights right from the beginning.

Anyway, looking forward to seeing the midterm presentation, AC

 

Tuesday, September 23, 2008

Recent comment posts

Marco VIsser            Bridge geometry responding to Architectural Program
http://1212257ar0960.blogspot.com/

Following up on Axel's comment about bringing the technical analysis results back into the design of the form, two of the (many) approaches you could take are:
1)inverting the deflected shapes so as to arrive at more compression-controlled structural elements;
and/or
2)using the undeflected and deflected shapes to define trusses rather than beams (which have the potential to form a space, not just support the loads/program.)
AC

September 23, 2008 5:40 AM (?)

---------------------------

Christian van Gruijthuijsen    Generative design/rule based design and deconstructivism
http://gruijthuijsen.weblog.tudelft.nl/

23sep08

As I discussed with Axel on Friday, it might be helpful for you to approach your study of Decon by means of comparisons. That is, if you're searching for parametric/associative or other generative rules for defining Deconstructivist forms, you can also look at rules for Constructivist, or Modernist, or other formal languages, and explore how changing those rules might give you the rules you want.

Of course, it's important for the purpose of this course (since the time is so short) that you work with a fairly simple design, just complex enough to be recognizably belonging to the language you're working in, but not requiring too many rules, so that you can easily test variations (both within the rules, and variations of the rules) and reach some conclusions about the question you're researching.

The rules can be expressed both in words and diagrams, to aid you in modeling and/or scripting them.

AC

 

----------------------------

 

Bert van Diepen            Relation of colored lighting and architectural space
http://bertvandiepen.blogspot.com/

... also, the survey can help you get more information about the color effects/interpretations you're reading in the literature - for example, whether a significant number of people agree with those, and how those who may disagree react (or at least describe their reactions.)
This type of feedback may then help you in refining your designs.
AC

September 23, 2008 1:58 PM

----------------------------

 

Dave Koomen            Daylight and Underground buildings
http://davekoomen.blogspot.com/

Since you have these relatively simple relationships between the pipes' height, width (&length) and reflectivity, you might also try making some diagrammatic and spreadsheet-based studies to get a feeling for how the system works. You could use such studies to guide your modeling in Dialux, and/or to quickly compare results of changing a parameter such as W or r. (It's not so easy for H, since if the heights of the pipes and rooms are linked, the effect of the room height on distribution of light also becomes a factor.)

September 23, 2008 1:38 PM

----------------------------

 

Rein Roosma            Generation of facade patterns
http://rroosma.weblog.tudelft.nl/

22sep08

Regarding your next steps, I would suggest that creating a program for the building at this stage may be too time-consuming (and not necessarily consistent, since the program should also have influenced the shape and orientation of the project.) Instead, you might try just defining some general goals for the light to be made available inside (...with energy loss and/or generation addressed later, if there's time.)

Examples would be:

1) uniform lighting of floor surface;

2) more intense light at center of floor;

3) more intense light at edges of floor;

4) more intense at one end, less at other;

Just these simply stated goals (even without specific, numeric illuminance level requirements) can be quite challenging to achieve, especially at differing times of day and days of year.

AC

Stand-Up Architecture - Fall '08

Having been asked to assist, I'll be posting commentary on projects and topics from the course, as appropriate. The feedback pertaining to specific projects can also be found at the students' own blogs. They are collected here for convenience and an overview, as well as offering a possible forum for broader discussion.