Skip to main content

ARCH 653 Project 2

ARCH 653 Project 2

1. Objective

This project aims to perform daylighting analyses for Lotte World Tower spaces. The change in parameters of the building facade can result in different daylighting effects. LEED version 4 defines that sDA (spatial daylight autonomy) represents how much of space receives sufficient daylight and it should be over 75. So, this project changes the width of curtain panel windows and analyzes the following daylight factors.


2. Facade Design

The curtain panel chosen for this project is shown in the below figure. 
It has shading walls on each side. The width of the shading walls can be changed parametrically. The middle part is composed of windows. The panel was then placed on the facade of the model.



3. Dynamo - Daylight Analysis

The project used the Honeybee tool to analyze daylighting. For the analysis, Revit 2019 and Dynamo version 1.3.3 or lower are required since the tool hasn't been updated yet for Revit 2020 and Dynamo 2.0.

The tool also doesn't recognize curtain panels as walls or windows and regards it as empty space. Therefore, this project needed to change the method for analyzing daylight to consider the curtain panels.

The conceptual framework for the alternative method is shown below.
The project used machine learning to train daylighting values based on various window sizes.
The trained machine learning then predicted daylighting values based on curtain panel window sizes.


To perform daylighting analysis with the tool, the curtain panels were replaced with simple windows. 


Original Curtain Panels

Replaced Windows for Machine Learning Training



The replaced windows had the same size of window that the curtain panel has.

The below figure shows the geometry for training. 




These are the dynamo node to perform daylighting analysis.



An example of daylighting analysis result is shown below. The color represents how many hours of space receives sufficient daylighting. (Red color means a spot receives sufficient daylighting over 50% of annual occupied hours)



 The daylighting analysis results with different locations and different window sizes were stored in the excel sheet. The different locations will further allow machine learning to predict daylighting at different locations of the building.

4. Dynamo - Machine Learning

 Dynamo loads the excel file created in the previous step. The data are then normalized for Neural Networks training.



The normalized data are then loaded into Neural Networks.



5. Dynamo - Final Outcome

Finally, the machine learning model can predict daylighting values based on different curtain panel window sizes and building locations.


 6. Video Presentation






Comments

Post a Comment

Popular posts from this blog

ARCH 653 Project 1

Lotte World Tower Model Development 1. Building Information • Location: Seoul, South Korea • Description: - 123-floor - 555-meter skyscraper - The 6th tallest building in the world - Opened on April 11, 2017 - Served for hotel, residence, and offices 2. Parametric Mass Design First, multiple reference planes were created to make a truss parametrically work. Based on the planes, reference lines were created to shape the basic form of the truss body. Along the reference line, a round shape form was created. The sizes (width & length) of the top and bottom rectangular were set to be half of the size (width & length) of the middle octagon. An adaptive family consisted of two adaptive points and one reference line was loaded into the body truss and created the remaining part of the truss. An angle between two yellow lines was set to be parametrically changed. The first trial for making the outer walls by pickin
Post a Comment
Read more

ARCH 655 Project 1

Lotte World Tower Model Development 1. Building Information • Location: Seoul, South Korea • Description: - 123-floor - 555-meter skyscraper - The 6th tallest building in the world - Opened on April 11, 2017 - Served for offices, hotels, and residence 2. Mass Design The building has been divided into two parts: bottom and top. NURBS was created in Grasshopper with degrees of three and based on the points in the figure. The NURBS was stacked up to form the bottom part of the model. Later, only the top and bottom of NURBS were used to make a loft. The remaining parts were regarded as useless ones and removed from the model. To make the top part of the model, the reference curve was created based on 'scale non-uniform factor' nodes. The NURBS created in the previous step was resized to have 40% of the original size using this 'scale non-uniform factor' node. Lofts were used four times to form north
Post a Comment
Read more

ARCH 655 Project 2

ARCH 655 Project 2 1. Project Objective This project aims to find the optimal window to wall ratio and the number of facade shadings that meets LEED version 4 daylighting criteria. The target model is Lotte World Tower created in Project 1. 2. Geometry Decision for Daylighting Analysis I decided to analyze daylighting for a single floor to reduce the computational cost for the simulation. The indoor area of the 10th floor was chosen for daylighting analysis.  Windows are (as shown below figure) curtain panel styles and built on the surface of the outside wall. These are the outside walls and facade shadings for daylighting analysis. 3. Parametric Modeling The first step is to create floors.  Once the floors, walls, and shadings are ready, I linked them to the Honeybee zone identifier. And add windows to walls and set transmittance factor to windows. I used 60% transmittance for windows. Then, I created grids on the 10th floor f
Post a Comment
Read more