Stage 1: Create Two New Evaluator Nodes
For this stage, a continuation of the Evaluator Node from Module 5, Stage 2 is used. The other metrics for evaluation are all added into a single evaluator node to more easily implement in the testing logic (does not require linking input parameters again).
Evaluator Node 1: Material Cost
This evaluator uses the structure’s surface area and gross floor area to estimate the cost of materials required. It is split into 3 outputs, including structural material cost, architectural finishing cost, and the total cost of the two. The desired material for structural and architectural can be chosen. The structural material include steel framing and reinforced concrete framing. The architectural finish can be a glass facade or metal panels. Within the custom node, sample costs per square foot are used to get the final material cost of the whole structure.
Evaluator Node 2:
The evaluator here completes a solar analysis on the structure in Dubai for a singular day. It specifically outputs the cumulative total insolation on the entire structure. This is used to calculate the total insolation in the year by multiplying by 365. The start and end time can also be updated at the beginning of the analysis, but would cost more in the analysis, so this method is a quicker average. This part of the node had struggles running, which I discussed with the TA about. Hence, the table for these outputs are empty for now.
Stage 2 Summary Table:
Do the new evaluation metrics that you’ve designed capture the meaningful differences between the building form alternatives?
The material cost, or any kind of construction cost metric, will likely see a logical difference between the building form, where larger section profiles will automatically indicated larger costs. Depending on the developer’s preference for the shape of the structure, they can look at the building profile alternatives and see how small dimensional changes will change the total structure cost.
The solar analysis will provide a more interesting difference in the building form. Since it is mostly dependent on the surface shape of the structure, changes in the profile should show more drastic differences to the final insolation values. For example, because my structure is a lofted profile, if I make one profile smaller than the one above it, the surface may seem more shaded areas and change the insolation value.
What other metrics would be useful to compute to help understand and make the case for which alternatives are truly better than others?
Other economic metrics will likely be useful, as the price and labor required for the project is usually limited. Even if a developer wanted certain aspects on their building, they still have a budget, so showing an alternative with a cheaper construction cost will likely be more appealing than others.
More structural metrics could also be used, such as the performance of the building, especially during natural disasters. It may be important to the developer to have a resilient building that can continue to function after floods or heavy wind. Human comfort is also very important, especially for tall structures. Therefore, structural framing that prevents large drift/sway might be necessary.
Since the developer is interested in the cumulative insolation of the structure, other forms of understanding sunlight on the building could be helpful. Specifically, modeling the structure in the real estate that it will actually be in with existing structures around it will likely change the insolation throughout the year to make it more accurate. Similarly, this relates to the building’s line of view. Since this is such a large structure in a city such as Dubai, there will be many featured objects around the area that would be valuable if in the view of people in the structure. Measuring directiveness of nearby buildings and objects of different heights will make the building more desirable and rentable.
The overall cost of the structure will likely be the guiding metric for deciding the structure’s building profile. If only small changes are made to the structure inputs, costs will likely stay very similar. Therefore, the developer will likely choose alternatives that will push up the insolation value overall.
For 3 or More Units: Develop a Single-Objective Optimization Scheme
What overall strategy do you feel best captures the relationship between the evaluation ?
In this stage, I added another metric which is the rentable floor space revenue. This uses the gross area and lets the user choose their desired rent per square footage when running the model. This gives another metric for the single-objective optimization scheme to use for its final score calculation.
Single-Objective Optimization scheme
The metrics used to calculate this optimization is the total material cost, rent revenue, and then insolation values. The goal of the developer is to minimize the surface area of the structure, which can be shown in the total material cost which uses the surface area to estimate the architectural facade costs. The goal is to minimize the material cost, while maximizing the rent revenue and insolation values.
Because of the running speed of the insolation value, the calculation is done by looking for the smaller material costs that have more variation area of the structure (too see more varied alternatives), and then the insolation values are calculated separately (with the Stage 3 file which takes individual instances).
What overall strategy do you feel best captures the relationship between the evaluation metrics?
I think the single-objective optimization is a decent, easy scheme to calculate the best design, and checking which metrics align and don’t align are helpful for this. These specific measures I chose are easy to quantify together, but additional metrics will likely make this more difficult. The overall weighing system is helpful for having more control over what metric is more important. However, additional metrics may not have as clear of a relationship with each other to easily decide if they are aligned or are tradeoffs. Adding more insolation analysis, especially with shading from surrounding buildings, may require much more different profile changes to be more optimized. Being able to do multiple objectives next week will likely show a even more detailed alternative result.
Table Summary
The weights chosen for each of the three main metrics can be seen below. The values are normalized to the max value of each metric, and then taking the ratio of each alternative to the maximum. For the Insolation Values and the Rent Revenue, the ratio is kept just as is because the higher the final value, the better. The Material cost is normalized by taking the ratio mentioned above, but using 1-ratio instead so that the larger value material cost is penalized. The weights are chosen to weight the insolation value the most because it is a key part of the developer’s interests. The total cost and the rent revenue are somewhat related to the same values in the building calculation, so should be weighted less to not double count. These calculations are done directly in Excel.
What propelled the recommended alternative to the top of the list?
The recommended alternative is chosen by the single-objective scheme which calculated a weighted score as described in the table above. Because the weight of the insolation values are higher than the other two metrics, that was what governed. The rent revenue was also helpful being at a decently high value due to the large gross area in this alternative. The material cost is weighted less, but that specific design is also cheaper. Overall, this scheme better shows that a larger structure is not necessarily better in all metrics.
Are there important nuances or tradeoffs that got lost is the single evaluation?
The choice of the weights for the single evaluation are subjective to the user. I chose the weights based on what I estimated would change with the insolation values, for example, and how it is less dependent on the calculated surface and floor areas. However, other weights can be analyzed. Additionally, the normalization used for the scheme rewards the largest/smallest of the series of data more, which may mean that certain metrics should not actually get such a high score. Multi-objective evaluation could produce a more balanced weight of multiple metrics.
For 4 Units: Visualize the Recommended Alternative
Here, I took the best alternative dimensions and applied it to one instance of the structure. I used the solar insolation values to decide the opening amount of each of the quad panels. This is done by mapping the insolation values to a minimum of 0.05 to 0.5. Panels with large openings will have smaller insolation values and vice versa. This can be a good visual for desired window/shading, where areas of the structure with more sun may want more cover-up for the people inside. Oppositely, the smaller panels will show that a larger area of solar panels, for example, are desired since it sees more sun wattage.