Now showing 1 - 5 of 132
- ItemFurling Fern: A Sculptural Design From Construction To Installation(2022) Walker-Andrews, TristanThis project follows the production of a large scale public sculpture from initial design to fabrication and installation. With the goal of sustainability, this project primarily utilized recycled steel, salvaged from past engineering projects at Swarthmore. The steel square tubes were cut and welded together to generate a large sculpture with an environmental inspiration. The stress in the sculpture was analyzed with a conservative model and found to be less than the yield strength of the steel with a large factor of safety, greater than 18, under self weight and 80mph wind load. The moments generated under this loading were then used for the design of a base to support the sculpture. The installation of the sculpture was proposed to the Swarthmore College Art Committee and denied.
- ItemLife Cycle Analysis of Mass Timber Products Use of Crude Oil and Possible End of Life Reduction(2022) Moeller, Stefen; O’Donnell, FionaMass timber products promise a new form of construction over traditional building materials like reinforced concrete and steel with a variety of advantages including reduced construction time, the benefits of greater carbon sequestration, and low global warming potential. While there have been many life cycle analyses (LCA’s) investigating these different facets of mass timber, little has been done to measure the potential material consumption, especially beyond just the wood itself. Of particular concern is the potential of these mass timber products to consume oil through the use of plastic resins and as such, this study evaluates the potential consumption crude oil due to the adoption of CLT manufacturing in mid- and high-rise commercial construction. The scope focuses on the United States over the next 75 years specifically, constructing a python-based model to use the total amount of installed building stock per year to get an approximation of crude oil consumption through a variety of conversion factors determined from literature review. Additionally, the model embraces some of the LCA framework, measuring how the oil may be consumed further in the use phase and how the use end-of-life options might help alleviate some crude oil demand. Ultimately, the model determined that the range of total crude oil consumption in an aggressive adoption scenario to be 2-3 billion lbs, well below 0.1% of our current annual consumption, suggesting the direct material demand for oil in CLT construction is likely not problematic for mid to high commercial construction.
- ItemPilot-Scale Plant Wall for Singer Hall Commons(2022) Atkinson, Lucy; Bronkema, Bethany; Tate, Spencer; Everbach, CarrFor this engineering design project, a pilot-scale, self-watering plant wall with an interactive audio element was designed and constructed for the purpose of enhancing Singer Hall Commons. An Arduino based system was used with a capacitive moisture sensor and a relay-triggered electrical outlet operating an inexpensive water pump. Physically, the plant wall structure was constructed out of found materials from the Swarthmore College machine shop, including aluminum framing, Unistrut material, and small wheels. Previously used plastic trays were located at a nearby store and repurposed as plant trays for this project. Finally, for the audio element, nature-based sounds that correlated to collected signals of light levels and distance sensing were indexed in MATLAB to create a varied listening experience.This project was successful in achieving its goals at the pilot-scale and could feasibly be extended to occupy a larger space in Singer Hall Commons.
- ItemIntegrated LiDAR Based Cycler Safety System(2022) Malcolm, Patterson; Neureiter, Luke; Cheever, ErikThis project aimed to produce a working prototype of an onboard rear detection safety system for cyclists. The project goals were outlined as iterative with the main priority being an operational rear-facing detection system using a LIDAR sensor and the following priorities were Bluetooth integration, an operational mounting system, and GPS tracking. After background research of best practices and feasibility, an initial system was set up using an Ardunio mega integrated with the LiDAR sensor. The final product is a working prototype with detection capabilities, bike mounts, Bluetooth, and GPS tracking.
- ItemUsing Support Vector Classifiers to Determine Cancerous Breast Histopathology Images(2022) Webb, Owen; Piovoso, Michael; Moser, AllanWith breast cancer being a leading type of cancer in women, any help that can be given to support early detection should be applied. Therefore, the goal of this project is to use machine learning to aid a pathologist in the early detection of cancer in tissue samples. The machine learning model researched here is Support Vector Machines (SVM). These are used to classify breast cancer histopathology images as cancerous or benign. Using support vector machines from sklearn and raw pixel data from Cruz-Roa et al , we tested the performance of a single Support Vector Classifier and a Bagging classifier with eight support vector machines as its base models. On 22% of the dataset, a single SVM performed at 71.0%. On 22% of the dataset, the 8 base model bagging classifier performed at an accuracy of 76.1%. This is not as accurate as the CNN tested in previous work, roughly 84%. We can conclude that there is unlikely a use case for SVM as a predictor of breast cancer.