Surrogate Model Design and Implementation for Earthquakes; Analysis of Urban Mass Timber Construction in the United States and the Push for Sustainability

The general topic that couples the STS research paper and capstone technical paper together is earthquakes. The capstone technical paper describes research completed on artificial intelligence neural networks in an attempt to predict earthquake occurrences. The STS research paper’s connection to earthquakes is a bit less obvious. For this paper, the topic is about discovering the extent to which social pressure for sustainability has influenced US construction companies to experiment with mass timber construction methods. The connection here can be found in the concept of timber construction. Timber materials are known to have more elastic properties that help circumvent structural damage by seismic energy. Although this concept of earthquake mitigation is not fully explored within the STS research paper, it is mentioned here and in the prospectus to justify its connection to the earthquake prediction in the capstone technical paper. Performing research on this topic is important because earthquakes cause catastrophic physical and economic damage. Despite the thousands of years that humans have spent investigating and addressing earthquakes, there is still so much that we have yet to accomplish. Therefore, to prevent further loss of life and economic damage it is essential to develop technologies that can protect society.

The title of the capstone technical paper is “Surrogate Model Design and Implementation for Earthquakes”. The paper details my work with a research team at UVA which designed and implemented surrogate models. A surrogate model is a type of artificial intelligence model which is typically used when simulating large data scenarios, such as earthquakes. As the name implies they are surrogate, or substitute, models which are an approximation of what a true model for the scenario would behave like. However, they are created beforehand on other data since forming a new model based on thousands of variables would require too much time and computational power. During the research period, I was the main developer of the model and other backend logic for our system. I experimented with various model architectures and with the assistance of my teammates we presented our work to stakeholders. Over the course of our research, we eventually discovered that a generative surrogate model would be much more reliable and easy for users to use. In response to this, we left behind a rough design blueprint for future researchers for our organization to develop.

The STS research paper’s title is “Analysis of Urban Mass Timber Construction in the United States and the Push for Sustainability”. This paper focuses on the sociotechnical system of urban construction in the United States along with the sustainability of mass timber construction. Mass timber construction is a newer methodology where special types of timber are used to replace steel and concrete within larger buildings. Depending on the technique used, some forms of mass timber can replicate the same structural stability as steel with only a fraction of the environmental impact. Concrete and steel production industries are some of the world’s leading carbon emitters and a switch to mass timber within the construction industry would have profound impacts on this. Studies have shown that the general public is becoming increasingly interested in sustainable housing options, which generates more pressure for construction companies to deliver on. The STS research paper conducts two literature reviews, one to analyze the sustainability of mass timber construction and another on social pressure influencing companies. The overall conclusion of the paper was that in most cases mass timber is a sustainable practice, and sustainability is a major consideration for companies. However, until enough research is completed and there are no doubts left about the technology, it is very unlikely that many companies will adopt it. Although, some groups have begun experimenting with it on larger scale projects to inspire confidence among their peers.

The research accomplished in these papers is by no means anything groundbreaking, but I would argue that it is another small contribution to our collective pool of knowledge. It is the small additions to this pool that lay the foundation for others to come and make their own contributions. For my technical paper, my teammates and I left behind detailed directions to help guide future researchers along. Specifically, they should investigate the creation of a generative AI surrogate model and implement something called a variational autoencoder for earthquake data. The STS research did not leave behind any specific plans for others, but it did create a starting point for people who are interested in the topic. The reason that two literature reviews were used was due to the lack of existing research on the topics together, so I believe that using this paper and those referencing within it would be an excellent aid. Overall, the work completed was worthwhile but there is still much more to be completed in the future.

School of Engineering and Applied Science

Bachelor of Science in Computer Science

Technical Advisor: Rosanne Vrugtman

STS Advisor: Kent Wayland