Implementation Year: 2023

This project focused on optimizing the mechanical supply and exhaust ventilation system for an infectious patient ward. The optimization was achieved by identifying the most effective placement for air supply and exhaust points within the room, while keeping operational parameters like total airflow and temperature constant. The primary evaluation metric was the relative concentration of airborne pathogens in the room volume with an occupant. A genetic algorithm was employed to efficiently navigate the vast number of possible configurations. The study concluded that a specific configuration of ceiling exhaust and wall-mounted supply vents resulted in a significant reduction of pathogen concentration, nearly halving it compared to other effective layouts, thereby improving air quality and potentially reducing infection risk.

The project involved optimizing the mechanical supply and exhaust ventilation system for an infectious ward. Optimization was performed solely by determining the best placement for air supply and exhaust points. System parameters like total airflow rates and supply air temperature were held constant to demonstrate the potential of this optimization approach. The main criterion for evaluating the system’s effectiveness was the relative concentration of pathogens in the room’s volume with a patient present.

The infectious ward was represented as a simplified computational domain for simulation. The room is rectangular, measuring 3.8⨯2.8 m (10.6 m²) with a ceiling height of 2.8 m. Medical equipment is concealed behind a false wall. A standard medical bed (1960⨯800 mm, 550 mm high) is placed across from it, with a mobile cabinet (1000⨯400⨯900 mm) in the corner.

Pathogens in the air have negligible mass and spread through the room via diffusion and advective transport. Pathogen diffusion occurs due to the thermal motion of air molecules, with its rate determined by the concentration gradient and quantified by the diffusion coefficient (for air, D = 10⁻⁵ m²/s). Pathogen advection is their passive movement with air flows, generated by the ventilation system’s supply/exhaust points, patient respiration, and natural convection from temperature differences.

A model of diffusion-convective transport for an incompressible Newtonian fluid in a steady-state formulation was used. As no turbulent flows were expected, the model is described by three equations: continuity, momentum, and energy.

Room air temperature, per regulatory code SP 2.1.3678-20, must be maintained at 20-26°C; thus, supply air temperature was set to 25°C. Ventilation system capacity, per the same code, must be at least 80 m³/h per bed; therefore, total supply and exhaust flow rates were fixed at 80 m³/h for all configurations.

According to WHO data, patient breathing generates a flow of approximately 6 l/min of contaminated air at about 32°C. The human body also emits roughly 100 W of heat.

The ward is located within the building interior without external walls or windows, so heat flow through enclosing structures (walls, floor, ceiling) was disregarded.

A well-established mathematical model was used, requiring no additional validation against experimental data. However, simulation results depend heavily on correct model application. Since low mesh quality and significant residuals are primary causes of erroneous results, these factors were used for verification (checking the correctness of the equation system solution).

5.1. Mesh Convergence

The computational mesh divides the domain into elementary volumes (cells). The model’s equation system is solved for each cell, storing local values for temperature, pressure, velocity, etc. Generally, more cells yield higher detail and solution accuracy. However, increasing mesh density raises computational load. The goal is to find the coarsest mesh providing acceptable results. Mesh methods are sensitive to resolution only up to a threshold, beyond which further refinement doesn’t change the outcome. Finding this threshold is meticulous and time-consuming work.

In this case, the dependency of pathogen concentration in the room’s center on mesh resolution was straightforward to interpret for selecting the optimal mesh. The chosen mesh contained 3.4 million cells with an average size of 5.1 mm. A calculation on this mesh took nearly 9 hours.

5.2. Iterative Convergence

The model’s equation system is solved using an algorithm (the semi-implicit SIMPLE method in this case). For steady-state problems, the solution is approached iteratively. A threshold number of iterations exists, beyond which the result doesn’t change significantly.Δx, mm

This threshold can be found by monitoring residuals and task-significant parameters. Residuals are solution inaccuracies that decrease to a minimum in well-configured models. Stabilization occurred after 2,600 iterations in this case. Monitoring key parameters (pathogen concentration and flow velocity) at a single point (room center) showed stabilization after 4,750 iterations.

Optimization focused solely on the placement of air supply/exhaust points. To explore options, walls and the ceiling were divided into 20⨯20 cm squares, and four 80⨯10 cm rectangles were defined on the door.

Calculating one variant took 8.9±0.4 hours, making a brute-force search of all combinations (2⁹⁷⁷) infeasible. A genetic algorithm was used based on these hypotheses: 1. Exhaled air flow from the patient is primarily vertical. 2. Exhaled air rises due to its lower density (2.3% less dense, with a 7°C temperature difference from room air).

The vertical direction of the contaminated flow suggested placing exhaust points on the ceiling above the bed. An alternative considered placing exhaust devices in the false wall at the head of the bed, operating as a lateral exhaust.

Supply air could be introduced through walls or the ceiling. One wall faces a corridor or airlock, allowing supply only through the doorway. The room is longitudinally symmetric, leading to four basic supply point placement options: doorway, side walls, rear wall, and ceiling.

The total surface area of supply and exhaust grilles differed between options. This does not reduce the reliability of comparing results, as volumetric flow rates were constant in all cases.

Labeling supply and exhaust placement options with letters and numbers yielded 8 basic configurations.

Configuration effectiveness was evaluated based on the proportion and location of contaminated air. This initial assessment used a single criterion: the volumetric share of contaminated air. A normalized pathogen concentration threshold of C* = 1‰ was used.

Based on initial results, only two configurations (B-1 and C-1) were selected for further transformation. Subsequent analysis identified configurations B-1 and B3-1 as most effective at removing pathogens.

Visualization of simulation results led to the selection of configuration B-1 as the final recommendation. This configuration features more pronounced zones of higher air mobility (reaching 0.16 m/s), but these are located significantly below patient level and should not cause discomfort. Configuration B-1 achieved nearly a twofold reduction in pathogen concentration compared to configuration B3-1.