Skip to content Skip to navigation

Slum home

Motivation and Objectives

Respiratory diseases are a leading cause of child death globally, killing approximately 1.3 million children per year. Poor indoor air quality is a major cause of these infections, and there are indications that improving ventilation could reduce respiratory illnesses. The objective of this project is to develop and validate a computational framework for predicting ventilation rates in a variety of low-income household layouts. The framework will provide essential information for analyzing results of an initial randomized control trial to evaluate the impact of ventilation interventions in homes in Bangladesh, and it will support the formulation of global ventilation recommendations. Eventually, we envision that encouraging parents to secure housing with adequate ventilation will become a standard child health recommendation, just like breast-feeding or vaccinating a child.

Research progress
We have developed an integral model with uncertainty quantification (UQ) to predict the indoor air temperature and air change per hour (ACH) in a representative slum home, and we performed field measurements of these quantities in a representative home in Dhaka. The test home was rented by the local collaborating team in Dhaka, and had a variety of openings to test different ventilation configurations.

We considered 4 different combinations of openings; for each configuration we measured time series of the temperature at 15 locations in the home, and we measured ventilation rates at 3 or 4 different time instances. The integral model solves for the time evolution of the volume-averaged indoor air temperature. The UQ framework accounts for uncertainties in weather conditions and in parameters used to model physics and material properties. These uncertainties are propagated using Monte Carlo sampling. In addition, we quantify uncertainty due to observed variability in the natural ventilation flow pattern by using two different models, representing either cross- or single-sided ventilation, driven by both buoyancy and wind.Validation is performed by comparing the temperature and ACH predictions to measurements in the target house. 

Figure 1 shows a comparison of the predicted mean temperature and 95% confidence interval for the configuration with a window and a skylight. The plot indicates that the model predicts a mean volume-averaged temperature that is within 3 degrees of the measured volume-averaged temperature during the day, and within 2 degrees at night. The uncertainty in the measurements and predictions can explain most of this discrepancy. 

Figure 1: Left: measured temperatures at three different heights and five different locations in the target home; center: comparison of spatially averaged measurement to model prediction, assuming cross-ventilation; right: comparison of spatially averaged measurement to model prediction, assuming single-sided ventilation.


Figure 2 plots the predicted vs modeled ACH in the different configurations. The predictions depend strongly on the choice for the ventilation model (cross- vs single sided). The comparison to the measurements indicates that a small vent located far away from the skylight (far left) might not significantly affect the ventilation, i.e. ventilation is primarily driven by single-sided ventilation through the skylight. When increasing the size of the vent and locating it closer to the skylight (center left), cross-ventilation might become more likely to occur. In this case, the area of the smaller opening becomes a limiting factor, and the effective ACH might decrease compared to the skylight only solution. A similar observation holds for the window/vent configuration (center right). Finally, the model significantly overpredicts the ACH for two large openings (far right). Understanding this overprediction is a focus of ongoing CFD analysis of the flow pattern in the homes; it could potentially be explained by an opposing effect of wind and buoyancy. The CFD analysis will also investigate the reason for the higher variability between the different experiments as compared to the variability between the different predictions. A potential explanation is that the ACH is sensitive to physical phenomena that are currently not incorporated in the model, such as the effect of the wind direction.

Figure 2. Comparison of night-time ACH measurements and predictions for the different configurations tested.



This research is funded by an EVP grant from the Woods Institute for the Environment.