Los Angeles: Natural Ventilation Potential from Present-Day to 2080




Year: 2024
Course: Data Science for Environmentally Responsive Buildings
Instructors: Elence Xinzhu Chen, Sunghwan Lim
Team: Sara Segura, Ryan Otterson


This experiment calculated the natural ventilation potential for a generic room in Los Angeles for both the current and future climate. The room measures 6 x 6 x 4 meters, with two 2-square-meter windows on the east and west facades. Using a polynomial regression model based on 2020 climate data, researchers predicted cooling loads for 2080, enabling two approaches to assess ventilation potential. The first modified window openings hourly for optimal comfort, achieving 4,684 hours of natural ventilation out of 7,122 cooling hours. The second predictive approach increased this to 4,779 hours, a 2% rise.

This experiment underscores the importance of natural ventilation for cooling enclosed spaces as temperatures rise. Cities
like Los Angeles can expect higher future temperatures, with natural ventilation becoming insufficient for cooling even in mild climates. While responsive window operability can provide sufficient cooling for most hours, it cannot fully compensate for the projected temperature increases, leading to higher energy demands. This contradicts climate change mitigation efforts and may overwhelm energy grids already struggling, especially in cities like California. The inability to provide passive cooling could leave vulnerable populations even more at risk. Increased cooling loads will further exacerbate global inequities in both the developed and developing world.


Los Angeles Weather Data


Los Angeles was chosen as the site of study for its Mediterranean climate, reflected in the current weather data. As a city built within a desert, the researchers anticipated its cool nights and mild temperatures to be ideal for natural ventilation. The city experiences hot summers with low relative humidity, and brief periods of high humidity in the winters as a rainy season. Its humidity conditions were also promising for natural ventilation potential, as high relative humidities in summers could have a compounding effect on temperature in other cities. As a coastal city, Los Angeles’ wind speed and clear wind direction were expected to produce positive effects on natural ventilation potential (Figure 1).

After simulating the changed weather conditions of Los Angeles in 2080, a sharp increase in dry bulb temperature was observed year-round, especially in the summer months (Figure 2, 3). Relative humidity remained consistent primarily in the summer months but dropped between 5-10% during the winter months (Figure 4). Solar radiation total kWh increased by 5.1% in 2080. Wind factors remained constant or only slightly changed.




Figure 1: Wind rose for Los Angeles based on existing weather data (2020).
Figure 2: Dry-Bulb Temperatures and Relative Humidity (2020 and 2080 Values)
Figure 3: Dry-Bulb Temperatures (2020 and 2080 Values)
Figure 4: Current and 2080 Relative Humidity in Los Angeles



Calculating NV Potential


This experiment calculated the natural ventilation potential for a generic room in Los Angeles for the city’s current and future climate. The box-shaped room measures 6 x 6 x 4 meters, with two windows on the east and west facades, each measuring 2 squared meters. Natural ventilation availability was calculated using a discharge coefficient of 0.6 and a pre-set list of pressure coefficient values based on wind directions (Figure 5). Wind speeds were directly pulled from weather data. Indoor target temperatures were calculated based on the prevailing mean outdoor temperature for each calendar month, with dry-bulb temperatures provided by the weather data standing in for operative temperatures. These indoor target temperatures and their respective natural ventilation requirements were calculated to achieve optimal thermal comfort, as defined and measured by ASHRAE-55. Window openings were adjusted on an hourly basis to achieve the target temperature, as actual wind speeds fluctuated and wind directions changed air pressure values. The final values for each category are shown in Figure 6.  As shown in Figure 7, the unit was cooled using natural ventilation for 1,711 hours out of the 1,920 total cooling hours of the year, representing more than 89% of the cooling hours. Heating loads and cold discomfort were not calculated at this phase of the experiment.




Figure 5: Pressure coefficients for windows on the west and east facades



Figure 6: Statistical summary of calculations for current weather conditions in Los Angeles
Figure 7: Cooling Hours vs Natural Ventilation Hours for Current Weather



Model Testing & Selection


The researchers tested three machine-learning methods for predicting sensible cooling loads: a linear regression model, a polynomial regression model, and a KNN model. The data from Figure 6 was modified to remove sensible cooling loads and split into training and testing sets (75% to 25%, respectively) for each model. Multiple input variable combinations were tried and selected to achieve the lowest mean squared error in the testing dataset, shown in Table 1. In the linear and polynomial regression models, the variables chosen were dry-bulb temperature, wind speed, and solar radiation for their correlation values and their resulting mean squared errors. The KNN model was limited to using only dry-bulb temperature as its sole input. The polynomial model was ultimately selected for its superior accuracy and reduced mean squared error.



Future Predicted Cooling Loads


Knowing its superior accuracy above the other models, the polynomial regression model was used to predict the sensible cooling loads for the 2080 climate. As anticipated by the increase in dry-bulb temperatures, monthly aggregates of cooling loads jumped significantly between 2020 and 2080 (Figure 8, 9). While the absolute number of cooling hours increased primarily in the summer months, the predictions reveal the starkest relative increase in cooling hours during the late fall and winter months (i.e., November through March). Strikingly, August cooling hours increased by more than 250,000 J/S while December cooling hours increased by 2,157%. 

Having the sensible cooling loads from the polynomial regression allowed for the same calculations to be undertaken as with the 2020 weather data. Mean monthly temperatures, indoor target temperatures, natural ventilation availability, natural ventilation requirements, and respective window openings as area measurements were added. Natural ventilation hours were then calculated from this information (Figure 10, 11).





Figure 8: Monthly Aggregated Hours of Sensible Cooling Loads (> 0 J/S)

Figure 9: Comparison of 2020 and Predicted 2080 Cooling Loads

Figure 11: Monthly NV Hours: Current and 2080 Prediction

Figure 10: Statistical Summary of Calculations for 2080



Future NV Potential: Two Methods


Two methods for cooling were employed. The first method modified window openings to meet the current cooling loads for each hour of the year, adapting to the changing wind speeds, wind directions, and air pressures. Consequently, the unit was cooled using natural ventilation for 4,684 hours out of the 7,122 total cooling hours of the year. The second approach involved pre-cooling the space by anticipating hours in which the natural ventilation availability would be insufficient to satisfy their requirements or cooling loads would be high (i.e., greater than 500 J/S). Its logic is diagrammed in Figure 12. This predictive cooling method allowed for 4,779 hours of natural ventilation rather than 4,684, representing a 2% increase (Figure 13). In both scenarios, natural ventilation was able to compensate for a majority of the additional cooling hours in 2080. However, their respective percentages relative to the total cooling hours each month were lower in 2080 than in 2020 (Figure 11).




Figure 12: Predictive Cooling Algorithm Logic




Figure 13: Natural Ventilation Hours in 2080 with and without Predictive Cooling


Discussion & Conclusion


This experiment speaks to the importance of natural ventilation as a means of cooling enclosed spaces in both the short and long term. Cities like Los Angeles can anticipate much higher temperatures in the years to come. There is already a significant global discrepancy between those that can provide cooling using air conditioning equitably and meet their local energy demands. As shown in the 2080 natural ventilation calculations, this experiment reveals a significant increase in the number of hours in which natural ventilation will not suffice for cooling indoor spaces, even in cities with mild climates. While responsive window operability to maximize thermal comfort can provide sufficient cooling for a majority of the future hours, it cannot compensate proportionally for the increase in temperature predicted for the coming decades. This marks a sharp increase in the energy demands for providing comfortable spaces in the future. Not only does this contradict the global mandate for climate change mitigation, but also it may prove infeasible for energy grids that are already struggling to meet energy demands. Despite its global status as a cosmopolitan city, Los Angeles is already faced with frequent power outages. The inability to provide passive cooling could render already-vulnerable populations significantly more vulnerable in the coming decades. As the results of this experiment illustrate, increased sensible cooling loads can exacerbate global inequities and increase heat vulnerability in both the developed and the developing world.