Huilan Huang

Journal Entry For
Module 8 - Gen Des and ML
ACC Folder Link
Link to Student

Module 8 Questions:

  1. How does defining the design objectives influence the outcomes in a parametric design study using Generative Design in Revit? Provide examples of how different objectives might result in varied design options.
  2. Defining specific design objectives influences the outcomes in a Generative Design study in Revit because it directly contributes to what is being calculated in the algorithm. The objectives serve as guidelines for the study, helping the designer evaluate tradeoffs and narrow down which options best align with the goal.

    Different objectives may result in varied design options. For example, maximizing the slenderness ratio of a skyscraper design may produce building forms that are more tall and slender, neglecting factors such as cost, structural concerns (e.g story drift), and constructibility. Another example is setting the objective to be maximizing daylight and outdoor views. This may produce building forms that have a smaller floor area per level (e.g. if the space is smaller, more people could sit by the window rather than in the middle of the room). Another issue down the line is that this may increase heating/cooling loads, and thus, operating costs. Most designs will have tradeoffs involved, but comparing the different options helps designers decide what is more important (cost vs. aesthetics vs. energy efficiency, etc). Generative Design studies with clear objectives allow designers to quickly explore many options, aiding with the decision-making process.

  3. Discuss the importance of identifying target taxonomies when generating synthetic datasets in architectural design. How can this help in managing large datasets and ensuring diversity and accuracy in your designs?
  4. Identifying target taxonomies is important when generating synthetic datasets in architectural design because it provides structure to the dataset and ensures that the dataset aligns with the goal or problem you are trying to address. Target taxonomies help in managing a large dataset because they make it easier to search for and filter specific results. They also help ensure diversity and accuracy in the designs by helping you identify gaps in the dataset (e.g. so no single category dominates) and promote balance.

  5. What are the potential benefits and challenges of using automated workflows for generating diverse synthetic datasets in parametric design? How can modularity and scalability be achieved in such workflows? Using automated workflows for generating diverse synthetic datasets in parametric design can save time and provide consistency. For example, it can greatly speed up the generation of large amounts of design data. However, one challenge that may arise is managing the complexity of the parameters and understanding how they affect one another. Another issue that may occur is that the automated workflow could reinforce design biases, creating a dataset that lacks diversity. Modularity can be achieved by breaking workflows into smaller components that can be reused or modified without affecting the entire study. Scalability can be achieved by designing the system such that parallel computing tools can be used.
  1. Explore the role of iterative processes in optimizing design options within the generative design framework. Why might it be necessary to tweak and repeat studies, and how can this approach lead to better design outcomes? It may be necessary to iterate through runs to identify design flaws and evaluate tradeoffs. This allows designers to learn from the previous iterations and improve aspects of the study to create better outcomes. This process promotes continuous learning and exploration while helping designers come to an optimal solution.

Questions Related to the Autodesk Class:

  1. Describe the general workflow of creating a generative design study in Revit, as presented in the lecture. What are the key steps involved, and how do they contribute to the generation of optimized design options?
  2. The first key step in creating a generative design study in Revit is to identify the purpose of the study, along with any constraints and goals. Then, you can define the design variables and objectives, turning the variables into inputs and the objectives into outputs. The next step is to choose an appropriate solver, an algorithm that automatically samples the input variables (e.g. Randomize, Cross-Product, Optimize, Like-This). Then, you can begin generating design options and exploring the generated designs. If you are satisfied with the results, you can export them to another tool for further analysis (e.g. Microsoft Excel) or visualization. If not, you can make changes and repeat the process until the results match the objectives of the study.

  3. Given the properties of a good synthetic dataset outlined in the class, such as being large, labeled, parametric, expandable, diverse, and balanced, how would you apply these principles to create a dataset for a specific building morphology study using Dynamo? Outline your approach.
  4. With the 6 properties listed above in mind, I would begin my study by thinking about the goals/questions I have in mind, then creating key parameters that can be adjusted (e.g. base width, height, or twist angle of a building). Next, I would make sure my inputs and outputs are clearly labeled so that the data can be easily categorized and analyzed if it is exported. To make the dataset large, balanced, and diverse, I would try to ensure that a range of parameters are included so that a wider variety of designs are generated. The “cross-product solver” tool may be helpful for this purpose. Also, I would try to build my study such that more parameters could be added on later and not cause issues to the existing study. This helps to create expandable and large datasets. After running the study and receiving the desired results, I would make sure to export the data to another tool, such as Microsoft Excel, for more convenient storage, analysis, and sharing.

  5. Identify and discuss the four different solvers mentioned in the lecture that can be used to generate building masses in the Generative Design tool. How do these solvers impact the sample space and variety of design options produced? The four solvers mentioned in the lecture that can be used to generate building masses in the Dynamo Generative Design tool are: the Randomize solver, Cross-product solver, Like-this solver, and Optimize solver. Details for each of these can be found below.
    • Randomize solver: Assigns random values to each of the inputs in the study. Generates a diverse range of design options.
    • Cross-product solver: Useful for generating combinatorial datasets as it provides outputs for all possible combinations of parameters across input categories.
    • Like-this solver: Applies small variations to the current design; Useful in sensitivity analysis datasets.
    • Optimize solver: Iterates over multiple designs to try to find one that is “optimal”, based on quantifiable objectives for the dataset. Useful when you are trying to find the “best” design solution given multiple inputs (possibly with constraints) and a clear objective.
    • Each of these solvers is useful in their own way and can affect the variety of design options produced. For example, the Like-this solver and Optimize solver may be more beneficial if you are trying to refine a design you like. The outputs/design options produced may not be drastically different from one another. However, the Randomize and Cross-product solvers may generate a more diverse range of options, which can be useful for ideation and earlier stages of design. Ultimately, choosing which solver to use depends on your study objective and what you want the resulting design options to look like.