Conference Agenda

Artificial intelligence (AI) is an emerging technology with recent research utilizing large language models (LLMs) to advance climate change. One of the strengths of LLMs in climate research is providing predictions based on probabilistic modeling. However, the potential of AI in predicting embodied carbon in the early building design phase hasn't been explored. This study investigates AI as a prediction tool in early building LCA using existing data sets. We quantitatively analyzed buildings' carbon emissions in the early building designs by comparing various LLMs' scenarios.

Existing literature indicates that the early conceptual design phase affects LCA more than the late stage of design since there is more flexibility to change. Our research emphasizes the timeliness of LCA in early building design to mitigate uncertainty and provides predictions of different design stages. A methodology was developed to predict the potential embodied carbon reduction depending on when the LCA can be implemented. Determining what information needs to be provided at each stage is the next critical step for our process-driven research. Several important decisions are made during the early design stage, such as size and height, shape, materials, and structural system, and those were defined as time frames.

Our knowledge-intensive process provides embodied carbon predictions at each defined time frame. Moreover, the process can suggest different actions representing embodied carbon reduction strategies at the prediction point. These predictions provide insights into customized strategies for early building projects with accumulated data. Following this process, designers are guided in implementing optimized early building designs. This approach reduces reliance on specialized sustainability consultants and streamlines the building system choices and design process. By addressing both the timing and amount of embodied carbon reduction strategies, this study explores the prediction capabilities of AI during the early LCA process. This information can eventually lead to optimized building designs during the early design process.

Digital technology continues to pervade all aspects of modern life, from the way we socially interact, our defense industries, how we find & cure diseases, how we power and heat our homes, and how we transport ourselves in society. The increasing pervasiveness of digital technology also fuels concerns related to the environmental impacts of semiconductor production and design. This paper explores challenges associated with understanding the total environmental impact of digital technologies across their life cycle. The holistic approach addresses difficulties in understanding for instance: (a) the environmental "footprints" created during the production phase and use of digital technologies; (b) the environmental "handprints" from digital technologies as these technologies could help reduce emissions in other industries; (c) uncertainties and difficulties in performing LCA and handprint analysis for digital technologies; and (d) the inherent difficulties in understanding what the future, and specifically, new innovations may bring.

Climate change poses a formidable challenge to humanity, manifesting through both gradual stresses and extreme disasters that threaten efforts to achieve and foster sustainable development. Among its many impacts, average temperature increases and heatwaves are particularly evident examples of stresses and shocks. These phenomena claim tens of thousands of lives annually, even in developed nations, while placing immense strain on energy systems and infrastructure. Moreover, they can exacerbate and trigger other disasters, such as wildfires, amplifying the need for resilience-focused planning and design. Conventional design and assessment approaches often lack the capacity to address these challenges, particularly at the high temporal and spatial resolutions required for robust analyses. Furthermore, existing temperature records and models often fail to capture the impacts of climate change and extreme heatwaves, limiting their utility for long-term resilience and sustainability assessments.

To address these critical gaps, we have developed a suite of tools for generating synthetic temperature data scenarios. These tools, based on 30 years of monthly data averages and 19 years of daily data averages, enable the creation of high-resolution temperature datasets that are both realistic and flexible. The tools can generate synthetic daily temperature scenarios spanning decades, from present-day conditions to the end of the century, while accounting for gradual temperature increases derived from observed or projected climate change scenarios. Additionally, they incorporate the capacity to simulate heatwaves at chosen rates or randomly, with probabilities and durations increasing over time in alignment with climate change models. This novel integration of stochastic processes ensures realistic variations while aligning with observed data, providing a robust foundation for resilience assessments. The methodology assumes that monthly temperature distributions follow a normal pattern, with summer maximums and winter minimums situated at two standard deviations from the mean, in their respective seasons. Randomly generated monthly averages are compared with reference averages derived from daily temperature records, and daily averages are then adjusted to ensure alignment. This process allows the tools to produce datasets that are representative of real-world conditions while maintaining sufficient variability for meaningful scenario generation. The average temperatures align with real data, and the day-to-day and year-to-year variations remain meaningful and realistic.

This innovative approach enables the assessment of resilience and sustainability across diverse scales, such as pavements, the energy-water nexus, and buildings. Applications include forecasting future demand under various scenarios, testing the flexibility and preparedness of infrastructure, and enhancing disaster planning and operational practices. Moreover, the tools’ potential for further refinement, including higher temporal resolution (e.g., hourly data) and the incorporation of spatial dimensions, enhances their relevance for urban climate studies. This could support investigations into urban heat islands, albedo effects, and the interplay between urban design and infrastructure. By bridging the gap between climate data generation and practical resilience assessments, this work represents a significant advancement in climate adaptation tools. The ability to generate synthetic yet representative temperature data with high temporal resolution addresses critical needs for planning sustainable and resilient systems in the face of climate change.

There are numerous initiatives towards envisioning better cities for the future. Three intertwined initiatives are sustainability, resilience, and smart cities. Sustainability means the ability to sustain; the goal of sustainability is maintaining a state at a certain level. Resilience has been typically defined in technological domains (such as engineering and disaster response) as robustness, rapid recovery and resourcefulness – where a system should withstand the force of a disturbance and, should it fail, return to its original performance levels quickly with readily available resources. And, the smart city is deeply rooted in the usage of information and communication technology in urban policies to solve diverse problems (e.g., energy, parking, safety, etc.). Oftentimes, research on these initiatives will reference one or two of the other initiatives. Therefore, we set out to describe how are sustainability, resilience, and smart city concepts converging (or not). We conducted a scoping literature review to clarify the relationships between sustainability, resilience, and smart cities in urban systems. We followed the process suggested by Pham et al. (2014) and refined our literature pool using the Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA). For each article, we coded the title, author, year, and keywords. This coding was conducted by each of the three authors independently and then validated through consensus. Combining this literature review and coding, we established an integrated framework for sustainability, resilience, and smart cities in urban systems.

Unsurprisingly, the topics of sustainability, resilience, and smart cities show alignments and contradictions. It appears that resilience is a key component of both urban sustainability and smart cities, but there are not agreed upon approaches to this assimilation. Both resilience and smart city concepts are distinctive yet can share similar goals of having sustainable development. We discuss and map these interconnections to help navigate the concepts. The review led to a few interesting discussion points, including the complexity of urban systems and how this complexity can lead to lock-in and create difficulties in pursuing physical or institutional adaptation and transformation within any of these three initiatives, the increasing amount of information generated in interconnected systems and its exchange can bring opportunities and challenges (such as demand on the environment for data storage, AI, etc.), and the role of the citizen within the urban system. There is not a silver bullet for urban planning to address sustainability, resilience, and smart cities. By understanding how these three topics interrelate, decision-makers can avoid over-emphasizing one topic over another in the design process. By understanding how each topic impacts the built environment, they can also strategically place resources based on their local context.

Road-stream crossings (RSCs) are critical infrastructures for stream ecosystems and transportation networks. Many RSCs are aging, in disrepair, and/or undersized, threatening the resilience of freshwater habitats and transportation systems. Given the limited resources of state agencies and municipalities that manage these assets, it is essential to prioritize these RSCs for replacement effectively. However, there is a significant lack of coordination among stakeholders. This research develops a multi-objective optimization (MOO) framework that incorporates the interests of multiple stakeholders and compares its results with conventional scoring and ranking (S&R) methods. We employed the non-dominated sorting genetic algorithm (NSGA-II) to optimize environmental, transportation, and replacement cost objectives, achieving optimal solutions at the watershed scale. To determine the optimal population size, initialization method, and termination criterion, we utilized the modified inverted generational distance (IGD+) performance indicator. Our custom-seeded initialization method demonstrated superior performance and faster convergence compared to other initialization methods. Compared to S&R methods, MOO consistently resulted in higher scores for environmental and transportation objectives across various budget limitations, with increases of at least 19.57% and 37.68%, respectively. The frequent selection of certain RSCs among Pareto optimal solutions highlights the significant impact of their replacements. Analysis of this selection frequency in relation to RSC characteristics revealed that structural condition had the highest correlation (Pearson correlation of 0.60), indicating it as the most significant factor. This systematic approach promotes more comprehensive and effective infrastructure management by aligning multiple objectives and addressing the diverse priorities of stakeholders.