(Georgia Institute of Technology, 2020-04-25)
Chang, Soowon
Due to the increasing population, cities are requiring more energy. Among urban elements, buildings account for about 40% of energy demands and 30% of carbon dioxide emissions globally. To address the increase of energy demands and environmental responsibility, existing buildings should be transformed into highly energy efficient forms.
This research explores how to support decisions that affect performance-driven smart and resilient urban systems focusing on building renovations. The research scope covers the redevelopment of existing built forms at multiple scales. Since urban objects influence urban patterns at other scales, both individual and collective performances of buildings at larger scales should be evaluated to support proper redevelopment decisions. In addition, the transformation of existing buildings will encounter different problems and challenges at different scales in urban areas. On an individual building level, the selection of different envelope options can project the future architectural environment of buildings. On a block level, the performance will be changed along with combinations of building typologies such as land use, height, floor area, etc., and therefore changes to building typologies should be managed collectively to improve the performance. When PV are applied in buildings and hourly electricity demands are recognized, the dynamic energy flows on a community level will become complex to manage.
In this respect, this research is devised to identify and address redevelopment problems at different scales: individual buildings, block, and community. On the individual building level, this research studies how to support decision-making when optimizing the selection of building envelopes by using a Genetic Algorithm (GA). Based on the findings from optimizing at each scale, an interdependence of building parameters and multiple performance is observed. Therefore, decision frameworks across multiple scales are extrapolated to support community-driven and building-driven decisions. On the block level, this research explores how existing building typologies influence multiple performance indicators in a collective manner to support reconfiguring decisions using a Bayesian Multilevel Modeling. On the community level, this study addresses how the community can optimize block boundaries for resiliently managing the energy demand and supply of groups of buildings by using K-nearest neighbors (KNN) and community clustering algorithms.
This research will contribute to making appropriate decisions for investment, regulations, or guidelines when renovating physical building assets at different scales in urban areas. The research findings will consolidate theoretical understandings about the relationships between building design and construction parameters considering multiple performance indicators at multiple scales in urban areas. Since many cities are at the tipping point trying to become more resilient, increasingly focusing on sustainability, economic feasibility, and human well-being, a better understanding of the impact of built forms at multiple scales will support urban development decisions for the future smart and connected communities.
(Georgia Institute of Technology, 2020-04-25)
Hu, Yuqing
Effective coordination across multi-disciplines is crucial to make sure that the locations of building components meet physical and functional constraints. Building information modeling (BIM) has been increasingly applied for coordination and one of its most widely used applications is automatic clash detection. The realistic visualization function of BIM helps reduce ambiguity and expedites clash detection. However, many project participants criticize automatic clash detection, as many detected clashes are irrelevant with no significant impact on design or construction work, thereby decreasing the precision of clash results and the benefits of BIM. In addition, clash detection consists of discovering problems, but it does not entail solving these clashes. Even though some studies discussed automatic clash detection, they rarely discussed the dependence relationships between building components. However, a building is an inseparable whole, and the dependent relationships among building components propagate the impact of clashes. Relocating one object to correct one clash may result in other objects violating spatial constraints, which may directly cause new clashes or indirectly cause them through relocating other components. Therefore, figuring out the dependency among clash objects with peripheral building components is useful to optimizing clash solutions by avoiding change propagation.
Algorithms are designed to automatically capture dependency relations from models to construct a component dependency network. The network is used as an input to distinguish irrelevant clashes for improving clash detection quality by analyzing the relations between clash components and the relations between clash components with their nearby components. The feasibility to harness the clash component network and graph theory are also explored to generate the clash component change list for minimizing clash change impact from a holistic perspective. In addition, this study demonstrates how to use BIM information to refine clash management, and specifically focus on designing a hybrid clash correction sequence to minimize potential iterative adjustments.
The contributions of this study exist at three levels. The most straightforward contribution is that this research proposed a method to improve clash detection quality as well as to provide decision support for clash resolution, which can help project teams to focus on important clashes and improve design coordination efficiency. In addition, this research proposes a new perspective to view clashes, switching the clash management focus and inspiring researchers to focus on finding global optimal solutions for all clashes other than a single clash. The third level is that even though this research focuses on clash management, the optimization algorithms based on graph theory can be used in other interdependent systems to improve design and construction performance.