produce a continuous stream of salt (NaCl) particles. Three types of experiments were conducted to test a range of strategies.

• Stationary experiments
  • Retrofit experiments: Evaluated the impact of high-efficiency cabin filters, air exchange rates, installation of barriers, and stand-alone air cleaners.
  • Redesign experiments: Assessed the effectiveness of a parallel flow ventilation system (as opposed to a conventional horizontal, or longitudinal, flow ventilation system), particle arrival time, and particulate matter (PM) removal rate.
• On-road experiments: Measured air exchange per hour and PM removal rates.
• Computational models. Computational fluid dynamics (CFD) methods and models were used to simulate airflow in a transit bus to evaluate the effects of conditions such as ventilation configurations, barriers, and thermal plumes. Computational models are often used to test theoretical and real-world scenarios using data and computer modeling techniques. This approach allowed a range of scenarios to be tested and evaluated.
• Closed box modeling. An Excel-based airborne virus transmission model was used to show how a simple technique could be used to predict the impact of variables on virus transmission in an enclosed space. This model does not consider flow distribution, unlike CFD models, but it is exceptionally easy to use and can predict the relative impact of changes. A multivariable regression was created to analyze the influences of selected input variables for a transit bus, light rail unit, and an airliner. The model uses input variables such as occupancy, breathing rates, mask usage, ventilation rates, temperature and pressure conditions, and space volume, and produces information related to the probability of infection. Using the examples, the model can be applied to a range of different cases.

Key Findings

EFFECT OF CABIN FILTRATION IMPROVEMENTS

Researchers found the use of new minimum efficiency reporting value (MERV) 13 cabin filters to be at least 20% more efficient than the existing in-use MERV 13 cabin filters. This is because the static charge that comes with the filter diminishes with frequent and prolonged use.

CABIN ACH RATE DURING STATIONARY TESTS

Air changes per hour (ACH) measures the number of times per hour (hr^-1) that the air inside a room or enclosed space is replaced with new air from outside. This is an important parameter that characterizes the ventilation condition of a confined space, such as a room in a building or a vehicle cabin (e.g., a bus). Assuming the outside air contains a lower concentration or zero airborne virus particles, a higher exchange rate (i.e., ACH value) can result in lower concentrations of in-cabin airborne particles (or viruses) and lower infection rates.

There is a lack of data on ACH rates in bus cabins. To address this, the study measured ACH rates under various conditions on buses. ACH rates were obtained by releasing carbon dioxide (CO₂) gas into the bus and then measuring how fast the CO₂ gas was removed from the air by the ventilation system (i.e., the rate of decay), and by mixing the air inside a bus with fresh outside air. CO₂ was used because it is non-toxic at low concentrations, cost-effective, and easily measured by low-cost sensors. Additionally, because CO₂ is exhaled by humans, it is naturally present inside bus cabins when passengers are onboard. ACH rates were collected in bus cabins with bus doors open and closed.

• Bus cabin doors and windows closed. Fresh outside air enters buses through the ventilation system and worn-down seals (e.g., door gaps) due to the air pressure difference experienced when driving. The average ACH rates collected for this study ranged from 0.67 ± 0.03 hr^-1 to 2.5 ± 0.2 hr^-1 in a stationary bus when the air conditioning (AC) system was set to high and all doors and windows were closed. This means the air inside the bus was entirely replaced up to 2.5 times within 1 hour. The bus cabin volume and age of door seals may affect the ACH rates.
• Bus cabin doors open. This condition was tested because it is a part of the normal operation of public transit buses. It was found to have a significant impact on ACH rates. Although not tested in this project, opening windows would likely have a similar effect.

  Transit buses typically have a front door and a back door. Traditionally, passengers are asked to enter through the front door to make payment and exit through the back door. Sometimes passengers may exit through either door. During the COVID-19 pandemic, many transit agencies modified their door-opening practices by allowing fare-free services so that passengers were not required to use the front door for payment and could enter and exit through both doors. Both scenarios were tested.

  • One door open. Opening one of the bus doors produced similar results with no great difference between opening the front or rear door. ACH rates were comparable at 12.8 ± 0.6 hr^-1 and 12.2 ± 0.8 hr^-1 respectively. This ACH rate was 8 times higher than the ACH rate when the doors were closed and the AC system was set to high.
  • Both doors open. Opening both bus doors (front and back) was found to make a significant difference in air exchange rates during the stationary tests. The ACH rate was 36 ± 7 hr^-1. This is 3 times higher than the ACH rate collected under the one-door-opened condition.

ON-ROAD (STOP-AND-GO) EXPERIMENTS

On-road testing was conducted in an attempt to replicate the driving pattern of a public transit bus by driving laps around an urban city block with a perimeter of 1.1 miles and road speed limits of 35 mph to 45 mph. Stop-and-go means that the bus doors were open for approximately 20 seconds at the end of each lap to simulate air exchange conditions during passenger loading and unloading at a routine bus stop. The ACH rate for the on-road stop- and-go test was 27.1 ± 0.9 hr^-1. The intermittent opening and closing of doors to load and unload passengers at bus stops facilitated good ventilation. The ACH rate without doors opening for the on-road test was 8 hr^-1. This is five times more air exchange compared to the stationary testing condition. The dynamic pressure difference created by the motion of the bus helped increase ventilation rates.

PM REMOVAL RATE BY eACH RATE

In this project, NaCl particles were used to mimic the behavior of airborne COVID-19 viruses. PM sensors were used to quantify the concentration change of the airborne NaCl particles. As such, particle (or PM) removal rate should be interpreted as airborne virus removal rate and particle (or PM) arrival time should be interpreted as airborne virus arrival time.

Particle removal rates are expressed by an equivalent air changes per hour (eACH) rate. The higher the eACH value is, the lower the risk of virus infection in the cabin. The eACH is calculated similar to the ACH except the PM concentration, from the aerosol generator, is used in place of CO₂. The eACH rate gives a measure of how fast particle concentrations are reduced within the bus through the filtration system, dilution from outdoor air, and deposition to the walls and surroundings. The aerosol generator was placed at the front, middle, and back of the bus to allow particle concentrations to stabilize before it was turned off. Once the aerosol generator was off, the decay rate could be determined.

The following experiments were conducted to examine and measure PM removal rates.

• AC and MERV 13 filters. Having the bus AC system on in the conventional ventilation system improved the eACH rate from 4.6 ± 0.4 hr^-1 (AC off) to a rate of 9.5 ± 0.2 hr^-1. In other words, turning on the AC and ventilation systems improved the removal of airborne viruses by a factor of 2. This shows the effect of the AC system and in-use MERV 13 filter in removing particles (or airborne viruses).
• Aerosol source location. Changing the location of the aerosol source from the front to the middle of the bus improved the eACH to a value of 10.9 ± 0.2 hr^-1. This indicates that airborne particles from a coughing passenger sitting in the middle of the bus can be removed more quickly than those from a coughing passenger sitting in a front seat.

Therefore, if an individual is coughing, it is recommended that the individual sit in the back of the cabin.

• Leaky cabin air system. Although the eACH in the leakier cabin air system in Bus A was comparable at 9.2 ± 0.2 hr^-1, there was a trade-off between virus removal through dilution and exposure to air pollutants that came in from outside air.
• Parallel flow ventilation system. The largest eACH rates were observed to be at least 4.6 times larger in buses using the parallel flow ventilation system compared to the conventional bus ventilation system (stationary and on-road).

PM REMOVAL eACH COMPARISON: ONBOARD AIR CLEANERS AND ON-ROAD TESTING

Some of the methods for reducing airborne contagion considered in this research included strategies to address emergency situations. These strategies would not be continued under normal operating conditions because they have some undesirable characteristics, such as obstructing passenger movement or reducing seating capacity on a bus. The use of onboard air cleaners was tested and compared under stationary and on-road conditions.

• High efficiency particulate air (HEPA) purifiers in stationary experiments. Two stand-alone Honeywell HEPA purifiers were installed in Bus A to conduct stationary tests of airborne contagion mitigation. The HEPA purifiers were used together with the bus air conditioning system to test the effect of onboard stand-alone air cleaners. The air purifiers were placed in the middle of the aisle and were spaced evenly apart. HEPA purifiers proved to be a very effective remedy to remove particles in the air as eACH rates increased significantly to values of 23.0 ± 0.4 hr^-1 and 24.7 ± 0.8 hr^-1 with the aerosol source (e.g., source of contagion such a coughing passenger) at the front and back respectively. This is 2.5 times the eACH compared to the AC system without stand-alone HEPA purifiers.
• Air exchange in a moving vehicle. The on-road eACH rates were examined and found to be higher overall than those found in the stationary tests with values of 12.46 ± 0.09 hr^-1 and 12.76 ± 0.09 hr^-1 from the front aerosol source and middle aerosol source respectively. This is indicative of dilution effects from the outside air penetrating into the bus due to air pressure differences while the bus was in motion. If the HEPA purifiers were installed, particle removal was at its highest when the bus was in motion. The eACH rate was observed to be at 31.8 ± 0.4 hr^-1. The HEPA purifiers improved eACH (i.e., virus removal rate) by 2.5 times under on-road test conditions.

PM ARRIVAL TIME

In this study, particle matter arrival time is defined as the time (in minutes) after the aerosol generator is powered on, the PM concentration reaches a value equal to ½ the maximum recorded concentration. This method provides a measure of when the aerosol (which is a surrogate for airborne virus) reaches each sensor throughout the cabin. This data was normalized by using a maximum concentration from all sensors when the aerosol generator was at the front, middle, and back of the bus. With the regular HVAC system, particle arrival time depends on the aerosol source location.

• Front of vehicle contamination. When the aerosol generator was located at the front of the cabin, particle arrival times to the sensor were 2 to 3 minutes for a short bus and 4 to 5 minutes for a longer bus.
• Middle of vehicle contamination. When the source of contamination was at a middle seat, there was only a 1-minute difference between arrival times to the front and back of the cabin inside the short bus and a 3-minute difference for the longer bus. However, the aerosol cloud can immediately reach the center of the cabin aisle, and while the contaminated air makes its way toward the back filter, arrival times can then be much faster. Therefore, this variable should be investigated further.

EFFECT OF BARRIERS

Experiments in the stationary bus examined how to reduce the spread of airborne contagions by using plastic barriers to impact airflow. The barriers varied in height between 2 feet 10 inches and 3 feet 4 inches, depending on their position within the bus and their proximity to the ceiling handlebars. General lengths were measured at 33.5 inches with a depth of 1 inch. A total of 10 barriers were installed on each forward-facing seat to create partitions throughout the cabin of Bus A.

• Plastic barriers with a conventional ventilation system. When barriers were installed in conjunction with the conventional ventilation system, the aerosol reached sensors at faster times than when no barriers were installed. However, the concentrations were lower with barriers installed, indicating some particles were being removed through collisions with barrier obstructions.
• Plastic barriers with a parallel flow ventilation system. Barriers in the parallel flow ventilation system showed the same behavior with faster arrival times. However, the concentrations were lower overall when the aerosol source was at the front, middle, and back.

CFD MODELING RESULTS

Flow simulation was used to analyze the airflow within a transit bus to examine the overall effectiveness of the parallel flow ventilation system under various testing configurations. These conditions include the following:

• The addition of plastic barriers behind each row of bus seats.
• The added effect of thermal plume given off by a body. (The overall influence of thermal plume on airflow throughout the bus was studied to determine if this factor can drastically change the overall flow patterns within the bus.)
• The differences in flow direction:
  • by having air supplied through the upper vents and removed through the lower system, referred to as “downflow,” or
  • by having air supplied through the lower vents and removed through the upper ventilation system, referred to as “upflow.”

The simulation results and analysis indicate that the implementation of a parallel flow ventilation system in either upflow or downflow generally shows a reduction in airflow and a narrower distribution of particles across the length of the bus compared to the traditional ventilation system. The addition of plastic barriers showed an improved effect on the airflow and particle distribution throughout the bus when the parallel flow ventilation system was in its downflow configuration, but there were questionable results in the upflow configuration. Finally, the overall effect of thermal plume on airflow can cause either erratic flow or improve the overall efficiency within the bus and needs to be evaluated on a case-by-case basis.

MODELING PARALLEL FLOW VENTILATION SYSTEM IN ALTERNATIVE MODES OF TRANSPORTATION

While many public transportation modes incorporate HVAC systems, not all existing designs may be optimal. Consequently, in addition to buses, this research examined subway and tram vehicles using a CFD model to assess whether (1) introducing plastic barriers between rows of seats or (2) introducing a hybrid ventilation system could reduce the transmission of contagion on these modes. The specific objective of the modified ventilation systems was to reduce the travel distance of particles throughout the vehicle by confining airborne particles to the area in which they were emitted. Leveraging CFD simulations to model additional public transportation modes allows for a better understanding of the efficacy of

incorporating barriers and modified ventilation systems to curb the transmission rate of diseases on public transportation vehicles.

The research concluded that trams and subways align with models and features seen in buses. Both have ventilation systems that run along the ceiling of the vehicle.

• Subway model. Simulations showed that the parallel flow ventilation system with and without barriers was neither better nor worse than the traditional HVAC system. The changes to the subway system showed little effect, due to the pre-existing parallel flow ventilation system.
• Tram model. Installing just the barriers or the parallel flow ventilation system showed no significant improvement compared to the conventional HVAC system in a tram. The addition of barriers and a parallel flow ventilation system showed improvement in reducing the movement of particles, almost completely restricting them to the rows they were released from.

CLOSED BOX MODELING

The research team used the COVID-19 Aerosol Transmission Estimator tool (https://tinyurl.com/covid-estimator) developed by J. L. Jimenez and Z. Peng to model the influences of different factors in the spread of viruses in public spaces. Use of the model helped the team understand the relative impacts of various factors on the spread of viral contagions in a bus, light rail vehicle, and airplane. The model takes input variables such as occupancy, breathing rates, mask usage, ventilation rates, temperature and pressure conditions, and space volume, and returns information related to the probability of infection. It relies on multivariable regression to produce a summary of the influence of each input variable for three types of public transportation vehicles.

This closed box model estimates the concentration of infectious material per unit volume of the space based on a set of input prevailing conditions. It uses the well-known Wells-Riley infection model to calculate infection probability. One limitation of the model is that it assumes a uniform concentration of contagion and does not consider flow distributions highlighted in the bus experiments and modeling work conducted in this research.

The rationale for highlighting this model is as follows:

• It provides straightforward insight into parameters that affect airborne virus transmission in any space.
• It is free.
• It is relatively easy to use.
• It provides immediate results.
• It allows for easy investigation of “what-if” scenarios to aid practitioners in decision-making.
• It accounts for prevailing conditions of interest such as the size of a space and the number of people in it.
• It is readily adaptable to other spaces that other practitioners find of interest.

Reducing the model to a single equation made it easy to identify key factors to focus on and provided a clear direction for minimizing the aerosol transmission risk. Additionally, a detailed example of changing transit bus airflow and filtration was analyzed to illustrate an approach and metrics that can be used when making investment decisions to improve public health. Beyond demonstrating the utility of the tool, this example offered specific recommendations for transit buses.