Random Delete

Many of you are now getting into the evaluation part, and you might be finding that the analysis is quite slow sometimes. If you have 20,000 windows, and you try the analyse Sky View Factor, you will probably get the dreaded 'Memory Allocation' error...

So you need to reduce the number of surfaces. If you have a lot of surfaces, then you can actually just select a random sample to analyse. As long as the selection is random, the results will still be representative of all the surfaces. So in the case of 20,000 windows, you can analyse just 10% or 20%.

To help you do this, I have created a Random Delete node:

• http://files.patrick.janssen.name/2010/random_delete.zip

Scenic View Factor

While helping Ying Ying this morning, I created another evaluation node, called Scenic View Factor. This node will allow you to input a set of points that identify scenic views. The inputs are as follows:

• Input 1: The faces you want to analyze
• Input 2: The obstructions that might block your view
• Input 3: The scenic view points.

 The node will then work out the number of scenic points that you can see from each face that is analyzed, and display the result as a colour and a percentage. So 100% means that you can see all the scenic points from that face, will 0% means that you can see none of them.

• http://files.patrick.janssen.name/2010/scenic_view_factor.zip

Triangular grids

Some of you need grids that are triangulated with equilateral triangles. Here is an otl to do it:

• http://files.patrick.janssen.name/2010/tri_grid.zip

Bad designs are good

During tutorials yesterday, we had a number of discussions about the importance of keeping your low performance designs.

You need to present all your variants, even if they are very low performance. You will have created developmental and evaluation processes. You will of course be hoping that the developmental process produces design variants which, when evaluated, have good or at least reasonable performance. But in the end, this may not be the case.... the performance may be terrible! But this does not matter - do not try and hide these designs. We are interested in the success of the process, not the success of the designs. And a process that highlights that your designs are weak is a good process... You can then try and change your developmental process to produce  better designs.

Crit next monday

This is a reminder that as stated in the schedule, we will have a crit next Monday. For this crit, the most important thing is to have a Pareto graph... we want to see a set of variants evaluated according to some criteria and plotted onto the Pareto graph for comparison.

Other things that you should also have are:

• developmental diagrams
• evaluation diagrams
• a map of your design process which shows what you have been doing in the last few weeks

ps I have recovered from the flu, so tutorials as usual today

Passive Zone Proportion (updated)

I have updated the Passive Zone Proportion node so that it now works much faster. Instead of 'brickering' polygons, I now use an offset, which is a lot faster. The inputs are also a bit simpler - no need to input the walls any more. Just the floor polygons will do. The node is in the same place as before:

• http://files.patrick.janssen.name/2010/pzp.zip

For those who still want access to the old node, it is here:

• http://files.patrick.janssen.name/2010/pzp_old.zip

Evaluation nodes

In order to help people started with their evaluation processes, I have been creating various evaluation nodes that that I have been uploading on the blog. So far, I have created the following nodes:

• Unobstructed Vision Area node for evaluating horizontal views 
• Passive Zone Proportion node for evaluating % passive floor area. This relates daylight levels inside the space.
• Sky View Factor node for evaluating % visible sky. This relates to levels of daylight and views (feeling of enclosure).
• Sun View Factor node for evaluating % of visible sun. This relates to solar radiation.

There are various other things that can be measured. I may add a few more nodes. There are also more advanced simulations using EnergyPlus and Radiance, but it is better to start with the simpler types of evaluations.

They key for you is to start thinking about how you might use such nodes in you evaluation. Also, try and figure out the process at a small scale - then, once you know what you are doing, scale it up. These evaluations can be slow, so don't just put thousands of polygons into one of these nodes and then wonder why it does not work. You will probably not be able to evaluate the whole 1 km2 in one go - you will need to break it down.

A number of people have asked about how to insert these evaluation nodes and the otl files. When I create a node, I typically create one or more otl files, and one hipnc file. The otl files contain the actual node, and the hipnc file is just an example of using the node. If you want to use one of these nodes in you file, you need to install the otls. First copy the otl's to the folder where you are working, and then in Houdini menu go to File > Install Digital Asset Library

Iterative Design Process

So far, most of you have been focusing on your developmental process that will be used to generate design variants. You will then need a way of comparing your variants. So at this stage, it is important to start working on your evaluation process. It is important that you identify at least two criteria for evaluation, so that you can plot your design variants on your pareto graphs.

At the end of semester, we will be focusing on the your iterative design process, rather than the final design. It is therefore vital that you have more than one iteration - preferable more than ten iterations. To do this, you need to have both a developmental process and an evaluation process.

Since the focus will be on the process, it is important not to get caught up with trying to solve everything. At the moment, many of you are trying first to solve everything before starting on your iterations - if you do this you will never get there. You need to start iterating as soon as possible - even if the iterations are based on a very simple developmental and evaluation processes. You can then gradually add in more complexity as you go, and only where it is really necessary. It is better to have a simple design idea where the iterative design process can be applied successfully, than a complex idea with only one or two iterations.

Sun View Factor

It has been very quite - not many requests. So in the meantime, I worked on another otl for evaluation. I have created a node called "Sun View factor". It is similar to the Sky View Factor node:

• For the Sky View Facto, it calculates the % of the sky dome that is visible from the centre of any face.
• For the Sun View Factor, it calculates the % of the sun dome that is visible from the centre of any face.

The sun dome is the area of sky where the sun may be at some point in the year. It is a slice of the sky dome. So , if you get a Sun View Factor of 100%, it means that the sun is always visible, 100% of the year.

The other thing I have added to the node, is the option to apply Lambert's cosine law. This will give a result that is more closely related to the amount of solar radiation hitting the face. Basically, when this law is applied, the calculation takes into account the fact that the sun hitting a face straight on is more intense than the sun is hitting a face obliquely. If you apply the Cosine Law, then the values you get will always be lower than without the law - often about half. So you will never get 100% - even for you roof.

Here is the node:

• http://files.patrick.janssen.name/2010/sun_view_factor.zip

If you right click on the node, and select Help, you will find some help for inputs and parameters.

The wave building

I have done an example of a building in a wave form - a number of you have had some questions about this.  I think it is also a useful example to look at, even if you are not doing wavy buildings.

You can change various parameters - one subnet generated the wave curve, and another subnet then offsets the curve. This offset was harder than it might first seem, since you have to try and avoid self intersecting forms. The example also displays the total floor area - which can be very useful.

• http://files.patrick.janssen.name/2010/wave.hipnc

Flu

I have been a bit unwell today, so I will work from home. I will  conduct tutorials etc from home all of tomorrow. If you have technical queries, then send me the file via email. If you want to have discussions, then we can phone or skype.

Passive Zone Proportion

I have created a Houdini node to calculate the Passive Zone Proportion (PZP), as described in the book "Energy and Environment in Architecture: A Technical Design Guide", by Nick Baker and Koen Steemers.

The PZP compares the floor area a certain distance from the facade to the total floor area. So for example, if you set the range to be 6 meters, then it will calculate the total % floor area that is 6 meters from the facade. This area of the floor is called the passive zone, since it has the potential to be passively lit and ventilated. 

Usually, the depth passive zone is assumed to be twice the floor to ceiling height. So if your space if 4 meters hight, the passive zone is the floor area 8 meters from the facade.

The node is a bit more flexible, and it allows you to enter multiple ranges. For example, if you enter "3, 6, 9", then it will calculate 4 values: 1) the percentage floor area between 0 and 3 meters from the facade, 2) the percentage between 3 and 6 meters, 3) the percentage between 6 and 9 meters, and 4) the percentage above 9 meters.

• http://files.patrick.janssen.name/2010/pzp.zip

Update

On Monday we came to the following overall breakdown of the 5 million square meters:

• Assume that there are 100,000 people working and living on our site. There will be some overlap between these groups - i.e. some people will be both working and living there. But there will also be some people who live there but work elsewhere, and some people who work there but live elsewhere. But I guess that it would be safe to assume that most of the 100,000 people who are living there will be working either within our site or close by somewhere in the CBD. This may affect how you approach the issue of parking. 
• For office, we can assume 20m2 per person. So for 100,000 people, we get 2 million square meters of office. You can assume that office space is fully air-conditioned. A key challenge will be daylight - how to make sure that offices get reasonable daylight.
• For living, we will need to break down the remaining 3 million square meters into 1) residential, and 2) other activities (shopping, recreation, etc). 
• In order to get some idea about the amount of space required for the other activities, we need to know the number of households. The current average for Singapore is 3.5 people per household, so that is 28,571 households. Lets round it up to 30,000 (since the overall trend is that households are getting smaller). Based on 30,000 households we can use Belinda's paper to calculate that we need about 0.2 million square meters. You need to think where to put this space - and how it relates to residential and office spaces.
• This leaves 2.8 million square meters for residential.  So if we assume 30,000 households, then we get 93 m2 per household. Note that this includes common circulation and any other common spaces in addition to the size of the actual flat. I am not sure what the percentage is - but lets say it around 14% - then we get the average floor area of the flat to be about 80 m2 for 3.5 people, which is about 23 m2 per person. This sounds reasonable. Note that this does not mean that all flats should be 80 m2 - there may be a mixture if different sizes. You can assume that the residential will be naturally ventilated, with AC being used only when it gets too hot.

Singapore stats

For statistics in Singapore, there is a good government site with some useful data here:

• http://www.singstat.gov.sg/stats/keyind.html