





Philadelphia, United States
Core 601 Studio: Education & Housing University of Pennsylvania, Fall 2023
Instructor: Scott Erdy
Building Area: 50,000 sf Individual Work
This project integrates campus and student housing, repurposing the existing 46th Street Station in Philadelphia. In the context of the post-pandemic world, remote learning has become increasingly significant. The key challenge, therefore, is to explore how to balance remote learning, in-person education, public spaces, and student housing. By incorporating emerging mixed-reality technologies, the project introduces a hybrid learning experience that seamlessly blends physical and virtual environments, challenging the traditional educational model.
Student housing is uniquely designed as individual virtual reality cubes. These cubes serve as gateways to diverse VR landscapes, significantly expanding the students’ learning horizons beyond the confines of traditional academic environments. Complementing this, the station’s central public courtyard, interconnected through a series of corridors, is outfitted with study pods that are infused with augmented reality capabilities, adding an AR layer to the educational space. These housing units and study pods are intertwined with the station, offering a seamless, enriching educational experience to both the students residing there and the commuters passing through, ensuring that every transit experience has the potential to become a moment of learning and discovery.
The facade was inspired by the Schuco system consists of a series of lightweight aluminum frames. Adjustable louvers ensure the privacy of residential units. These shape-shifting modules are distorted by sine function to create a curving pattern, echoing the railing design, while the periodic function ensures repetition for easy fabrication.
The section reveals a rich scenario of simultaneous activities occurring across diverse functional settings. The upper residential units depict residential environments, while the lower station showcases public activities. Everything is integrated within the building, intertwining without interference.
A Faraday cage is an enclosure designed to block electromagnetic fields It works by redistributing electric charges on its surface, effectively canceling electric fields inside. Our prototype installation also has a Faraday cage as envelope.
Since we are using energy collection device on the landscape to transfer energy produced by visitors into usable energy for facade moving, energy conservation is crucial. For
this project, the conventional metal frame Faraday cage is combined with electromagnetic shielding glass to prevent signal from escaping. The charging spaces also utilized wireless charging technology to transmit electromagnetic forces to supply the visitors, the feature of Faraday cage also ensure this function is shield and reduce energy loss.
China
Undergraduate Work: Kindergarten
University of Nottingham, Spring 2021
Instructor: Maycon Sedrez, Jing Xie
Building Area: 27,000 sf
Individual Work
Located in the master plan of the garden city in Ningbo, this creatively conceived building is affectionately named “Bubble Wonderland”. This project is a bold venture to design a kindergarten that transcends traditional learning spaces, offering children unique areas to play freely and explore It’s envisioned as a nurturing environment designed to ignite their curiosity and encourage exploration with wonder and enthusiasm.
Drawing inspiration from the simple, captivating world of bubbles, a favorite in children’s play, this concept emphasizes unity and a sense of belonging. The spherical shape of bubbles inspires the design of the activity spaces, which blend organically with standard function rooms, yet present an engaging contrast. Echoing the way bubbles shimmer with vibrant colors in sunlight, the building is envisioned to provide a rich, dynamic space, creating a wonderland-like environment. The programs of this project is designed according to educator Montessori’s theory of learning sensitive periods of children. This space is filled with excitement and imagination, offering children an inspiring place to enjoy, learn, and grow.
Circulation System & Programs for Different Groups
These installations encourage children to explore spaces and learn about privacy by seek and find.
Montessori introduced learning sensitive period theory to describe key learning aspects in children’s different age spectrum Combined with traditional three-class kindergarten mode different programs are needed for each class to fit education targets.
Variation of forms provides kids with corners for gathering and to learn about their relationship.
Cooperation tables provide kids tools to make handwork and observe others.
Little galleries with photos and paintings improve children’s aesthetic abilities.
Multi-level stages encourage children to explore their own identities and relationships.
Reading rooms are mainly used by elder children with better comprehension.
Kids play in a large group to better learn collective rules.
The study spaces are more independent to learn about identity.
The least degree of openness for junior class children to avoid safety issues.
More open spaces create more chances for these children to explore the nature.
Children can play in small groups to learn about relationship.
The most open spaces for the eldest because they have better security sense.
Into the Rabbit Hole Competition - Healing Center Competition work, Summer 2024
ARCHIOL International Competition, Second Place
Building Area: 67,500 sf
Individual Work
This building carries the function of psychological inquiry and treatment for children with mental trauma. It integrates two typical psychotherapy methods: Gestalt therapy and psychodrama therapy to propose a composite methodology including psychological exploration and treatment.
The whole treatment is mainly organized by typical psychodrama therapy proposed by psychologist Moreno. Different forms of theaters hold psychodrama shows in different themes in specific acting phases. In addition, the gestalt labyrinth is used in early-stage for preliminary analysis.
The Gestalt cycle experience helps to excavate the disordered links of the experience process. In this maze, a series of spaces are translated from these links to provide a complete exploring experience to amplify and analyze the deep thoughts of the patient in a spatial narrative way. A similar maze outside is open for the public to experience and enliven the atmosphere. This approach not only aids in the healing process of the patients but also fosters a community understanding of children’s mental health challenges.
The diagram shows the cycle experience to describe the conventional complete experience. Number 1-7 show different obstacles in different steps.
Block stimulus from the inner region.
Take everything from outer for granted.
Introversion
Transference or deny part of self.
4. Retroflection
Repression diverts energy to the self.
Treatment System Structure
Neglect or transfer inner/ outer stimulus.
Focus too much on self and thoughts.
Desensitization
Extroversion
1. Desensitization 5. Deflection 2. Introjection 6. Egotism
Deflection
The whole treatment follows Moreno’s psychodrama theory. The programs and spaces are designed specifically for these steps. Between the warm-up and action steps, a series of maze spaces are introduced for children to explore to help analysis the deep reasons for psychological disorders behind patient’s mind.
Integrate with others thus lose self.
The psychological types by Jung can match obstacles to three dimensions:
Confluence 3. Projection
Introversion
Extroversion
Thinking Feeling
Sensation Intuition
Key Users
Key Elements of Psychodrama & Users Analysis
Psychodrama Elements
The design of the psychodrama performance stage refers to key elements proposed in the psychodrama theory. By analyzing the roles and functions of these users within these elements, the circulation of the psychodrama performance and program design is organized.
Projection
Introjection 4. Retroflection
Gestalt Maze Types
Thinking Feeling
Egotism
Sensation
Intuition
Confluence
The psychological types are transfered into spatial qualities. The spatial qualities are contrary to psychological features to stimulate the patient. Each obstacle is classified into three features to generate the 7 gestalt maze types. The Gestalt maze design is inspired by these features.
THEATER CONSULTANT FLOORS DISTRIBUTION
Rendering Provided by Outsource - For Display Purpose Only
Xinjiang, China
This hotel project celebrates the rich history and diverse culture of Xinjiang. Inspired by traditional folk houses, its design features a “River Valley Courtyard” layout that blends traditional and modern elements, reflecting local lifestyles and Silk Road heritage.
The project emphasizes sustainability with green building concepts and modern technologies. Key areas like treeshaded arrival zones, mountain-view corridors, and cultural pathways connect the landscape with functional spaces, creating a harmonious and eco-friendly destination.
The project’s architectural design is inspired by the folk houses of Yining. With clear outlines and uniform lines, the buildings are neatly planned to echo one another while maintaining unique identities, and key parts such as entrances and porches are highlighted. I contributed to the project’s early stages by researching and analyzing dual-hotel case studies in China and conducting regional feature studies.
I prepared extensive InDesign reports, created diagrams in Illustrator edited and colored hand-drawn designs in Photoshop based on the design leader’s sketches, and assisted with SketchUp modeling. The image below is an example from another project, I modeled this existing building entirely by myself referring to PDF drawings.
I contributed to design of curved spandrels and the entire podiums forms. I designed, modeled and rendered various curtain wall types. I also utilized Grasshopper to randomly divide openable windows and keep the sub-pieces in the same size for easier fabrication.
For the sales office of this project, since it was a renovation of an existing building, I helped model and clean up the entire original building and helped model the exterior facade and draw sections. I was mainly responsible for the interior design, designing and adjusting the organic interior form using Grasshopper and detailing it in response to different functions and utilized Enscape for interior rendering.
Groupmates: Xingyu Xie, Finn He, Chris Chen, Louis Huang
Contributed to initial code optimization and fabrication code design
Grasshopper & Galapagos for prototype
Drawing from the traditional Chinese “wan” pattern symbolizing infinity, this parametric design emerges from lofting two ellipses at different rotation angles, with wind direction analysis via Dragonfly and solar optimization through Galapagos and Ladybug. I utilized OpenNest to design an auto-label, flatten and easy-retrieval code for fabrication.
Wind direction & Shadow Range Analysis Fabrication (Detailed Fabrication Code Omitted)
StudioRAP Workshop
Groupmates: Pouria, Zherui, Riddarth, Toto
CIS 519: Applied Machine Learning
Groupmate: Kailun Wang, Yuner Zhang
Contributed to data collection, model design and report writing
We trained an Ensemble machine learning model for GPS prediction of input photos of a specified area in the Upenn campus. We trained and tested a variety of different models as well as schedulers several times. Finally, four sub-models were trained, and the final values were calculated as the output using the weighted average. The final validation RMSE was 26.57m, better than baseline’s 41m.
Input Images & GPS Sample Initial Models & Schedulers Comparision
Contributed to design, visulization, fabrication
During a week-long workshop organized by StudioRAP, we learned about clay printing using the ABB IRB120 robot The workshop covered the complete workflow from form generation to fabrication in Grasshopper including output code generation and optimization.
Housing Analysis
MUSA 550: Geospatial Data Science with Python
Groupmate: Wenyi Zhang
Contributed to coding & analysis
We conducted a data analysis study and a series of interactive visualizations on affordable housing in New York City, exploring its correlation with the distribution of low-income groups and privately owned public spaces (POPS), and conducted K-Means analysis to explore the combined impact. The final results are assembled in the multi-page Quarto website.
K-Means Analysis Group and Cluster Visulization
Data Collection Heatmap
Final Ensemble Model Design Final Model Performance
Input: 5980 Training Dataset, 100 Validation Dataset Size: [32, 3, 224, 224]
(Latitute/Longitute)
We aggregated the predictions of four sub-models using a distance-weighted method. Each model’s prediction’s weight was determined as the reciprocal of its average prediction distance from the others, making models with closer consensus weight more. The normalized weights were applied to compute a weighted average, reducing the impact of outliers and reflecting the majority consensus. The diagrams show obvious improvement for the bagging model’s performance compared to the individual models.
University of Pennsylvania
Master of Architecture
Email: yuchen4@upenn.edu
Tel: +1-445-225-9891
Address: 3720 Chestnut St, Philadelphia, PA