EXPERIMENTAL AND NUMERICAL INVESTIGATION OF CONTAMINANT CONTROL IN INTENSIVE CARE UNIT: A CASE STUDY OF RAIPUR, INDIA

Proper ventilation is an important strategy in the practice of infection control. Hospitals are complex atmospheres that require ventilation for thermal comfort of patients as well as control of harmful pathogens infection emissions. The present study is performed on a hospital at Raipur (21.2514° N, 81.6296° E), India, to analyze the avoidance of airborne infections from the mouth of patient to protect the doctor and other patients in the intensive care unit (ICU) using Computational Fluid Dynamics (CFD) software FLUENT. Incense smoke is used to for capturing velocity field. Twenty seven (27) cases of simulation were executed using different air change per hour (ACH) (6, 9 & 12) and different inlet and outlet positions talking into account the constant inlet temperature (20 °C). The wall temperatures were taken out from ISHRAE handbook for Raipur region. The velocity vector and capturing the flow field were also performed experimentally. All three turbulence model (Standard, RNG & Realizable) predictions have shown to be in good agreement with the experimental data. It can be effectively employed to validate the extensively used k-ε model which was commonly used for ICU.


INTRODUCTION
The dispersion of air borne disease can be impeded if infected patients are segregated from those who are not infected. The effectiveness of proper ventilation and contaminant control analysis has not been performed well till date. The quality of air in hospital environments is of specific concern as patients may serve as a source of pathogenic bacteria to staff and hospital people. Bacteria, virus and other air borne microorganisms which are harmful to living beings are easily spread through the air we inhale from the surroundings. A hospital is an institution for treatment of various diseases caused by many organisms. Even at an institution for treatment, diseases are still spread through improper handling and maintenance of infected area or beings. One such method of infection transfer in hospitals is the spread of the pathogen through air. Computational fluid dynamics (CFD) is used as an effective tool for studying various real time problems which are expensive and difficult to solve experimentally. An important application includes the use of commercial CFD software in solving fluid flow patterns with variable parameters such as boundary conditions and ventilation strategies. CFD using Eulerian and Lagangian methods have the ability to predict the steady state particle concentration distribution. Various case studies using CFD for fluid flow in ventilations of multiple/single bed hospitals were performed. Particulate dispersion in hospitals equipped with heating, ventilation and air conditioning (HVAC) and variable air volume (VAV) systems were also considered. The effect of the location of the infected patients in transmitting air borne diseases with hospital using CFD is also presented. [1-10].
The effect of ventilation, flow differential and passage on the air flow patterns across an isolated room was studied and the generation of greater flow in the hinged door than the sliding door through baseline smoke patterns was studied. Tracer gas with various ventilation rates of 6 and 12 air change per hour (ACH) had a less impact on the exchange of air across the hinged door. They have concluded that the sliding door model in isolated rooms of a hospital have better effectiveness in reducing air volume exchange during opening of door [11]. The cross-infection risks of air borne diseases are reported to have been reduced through higher rate of ventilation in hospitals. It was reported that natural ventilation has more effectiveness than mechanical ventilation in-terms of higher ventilation rate and energy efficiency through opening both doors and windows in a ward, with highest ventilation rate obtained in 69.0 ACH [12]. Another study reported the spread of severe acute respiratory syndrome (SARS) and H1N1 virus through coughing and speaking. Though the study was conducted with use of particle image velocimetry (PIV), the results have yield that the average initial coughing velocity was 15.3 m/s (male) and 10.6 m/s (female). The average initial speaking velocity was 4.07 m/s and 2.31 m/s for male and female respectively [13]. Tracer gas is used effectively in studying the fluid flow and air borne pathogen transmission in a six-bedded isolated room through walking motion of health care workers (HCW). RNG k-ɛ model is used to validate the results. Conditions like HCW walking with arms & legs swinging are taken as mixing process. The study have resulted that flow disturbance after stopping the motion of HCW (5.4 s to 25.4 s ) have larger impact in spreading the gases, thereby requiring more than 30-60 s for the flow to recover in its initial position even after the motion of HCW is stopped. 6 ACH is reported to have more spreading of gases than 12.9 ACH [14].
To accurately define boundary conditions like contaminants source, distribution and mixing with supply air in CFD, series of transient distribution of contaminants is developed through transient accessibility indices and analytical expressions. Experimental analysis is performed to validate the accuracy of the analytical expressions [15]. The contaminant control in a minienvironment through particle concentration measurement using CFD is performed through which particle concentrations and transfer route of contaminants were measured. They have proposed the concept of improving the yield and energy consumption of the system [16]. The control of air flow through rearrangement of air return by regulation from the ward cubicle using CFD on a general ward ventilation design was performed by researcher in a study. It assures the removal of microbes effectively at lower cost, minimizing cross infection [17]. The location of outlet in a single ceiling air diffuser plays an important role in the desired occupied zones. Adoption of variable inclination angles at inlet is recommended for improving air distribution profiles on applications like medical clean room [18]. The importance of error evaluation for solving air flow & heat transfer around human health body problems using CFD were discussed [19]. An additional study also revealed that the respiratory events such as breathing, sneezing, talking and coughing were the main source of transfer of contaminants.
The study uses a model of equations using various parameters like rate of flow, area of mouth orifice opening etc, and proposed that the model be used for describing source of contaminant transfer due to talking and breathing [20]. SIMPLE and SIMPLEC algorithms for steady, laminar and incompressible flow under Boussineq's approximation for various positions of inlet and outlet in a room heated with warm air is used with variable Grashof and Reynolds number. It was concluded that locating the outlet at a higher position than inlet led to enhanced temperature distribution [21]. The air flow patterns of an operating room (OR) during opening and foot traffic are studied. Even though OR have slightly higher pressure than other adjacent rooms, a small volume of air still enters during a cycle of door opening and closing even without any person entering the room. The study have revealed higher volume of air enters the OR if the person enters the OR [22].
The evaporation and condensation of expiration droplets and their size (coughing and speaking) have been found to have negligible impact on usual droplet size from human beings (13.5 µm from coughing and 16.0 µm from speaking) for average expiration velocity of 11.7 m/s and 3.9 m/s for coughing and speaking respectively [23]. Contaminant distribution in an office environment of 6.6 m (L) x 3.7 m (W) x 2.6 m (H) dimension with air conditioning and mechanical ventilation was also studied. Tracer gas (SF6) is used for simulation of contaminants on a model room and CFD was used for validating the results. The study revealed that the pattern of contaminant dispersion depends greatly on the velocity flow field. The layout of various objects like furniture also influences the pattern of air flow and contaminant. Other study suggests that CFD can be used effectively in predicting the spatial distribution of bio-aerosol in indoor environments like hospitals. The study was conducted at three different layoutsempty, single bed and two bed rooms. Deposition of the particles have no correlation with relative surface concentration and source distance but partition among the patients proved to be effective in reducing cross contamination among patients. [24,25]. Other areas in using CFD for numerical study of fluid flow and particle trajectory includes one bed hospital, particle motion in ICU, air pollution in hospital modelling, ventilation of hospital rooms, deposition of particle and air flow in IC Engines, nano-fluid study, fluid flow and pressure distribution in bearings etc. [31][32][33][34][35][36][37][38][39][40][41][42][43][44]. While CFD is an effective tool for numerical study, other numerical study involving ANN may be conducted for enhancing the results, as studies reported earlier have confirmed the same [45][46][47][48][49][50][51][52][53][54].
The above authors have performed various experiments in different regions using the respective environment conditions. The authors in this paper have performed real time study through the application of both experimental and numerical methods on the environment conditions of the hospital in Raipur (21.2514° N, 81.6296° E, 298m altitude above sea level),Chhattisgarh, India taking into consideration the wall temperature from ISHRAE handbook (2007) for Raipur [26].

THEORY PROBLEM STATEMENT
In the present study, numerical simulation of ICU room for single-patient has been carried out. Total nine (9) different positions of inlet and outlet for three (3) different Reynolds numbers have been considered. The overall dimension of ICU room has been considered as 4 m (width) × 3 m (Height) × 6 m (length). The room consists of a patient (source of contaminant), doctor, bed, light, stool and oxygen cylinder. In all the twenty seven (27)

Computational fluid dynamics
The CFD models are used to predict the air velocity, turbulence level, particle tracking, pressure and temperature of the computational models. In the present work, the k-ɛ model (Standard, RNG & Realizable) is used for all numerical simulation process. The corresponding equations are given in equation (1 -6) in which equation (1) ,+ + ' -+ ( + + The following are the assumptions made during the numerical modeling and simulations. Heat and mass transfer between air and contaminated particles trajectory are neglected. The walls, floor and ceiling of the room are considered to be well insulated. No particle rebounds on solid surfaces, such as walls, floors and ceilings have been considered.

Grid independence test & Y + value
FLUENT software is used to perform the steady state numerical solutions. Tetrahedral elements with an unstructured mesh have been used to discretize the computational domain of ICU. Different grid sizes (5.5 x 10 6 and 10 x 10 6 cells) were used for grid independence test (GIT). The maximum gradients of transport parameters such as air (outlet, inlet), near humans (patient and doctors) were chosen for mesh refinement to obtain flow and heat transfer characteristics. The computational domain in the inner, outer and center zone has been kept in the range of 1≤ Y + ≤ 7. The study uses various cell sizes (0.035 m, 0.05 m and 0.08 m)

Particle trajectory
The following equation (7) is used for calculation of the discrete phase particle trajectory.
where ? ( $ − @ ) is drag force per unit particle mass, is gravitational force per unit particle mass (m/s 2 ), Fx is additional force exerted on the particles. Some forces are negligible as they were generally small. The FD is defined by using the Stokes law (equation (8). During the computation, all contaminated particles are assumed to be of spherical solid shape, heat and mass transfer between air and contaminated particles are neglected and no particle rebounds on solid surfaces (walls, floors and ceilings). The Coefficient of drag (CD) from equation (8) was calculated [27] which regulates the value of CD for the particle (spherical) over a extensive range of Reynolds number. The particles have no control on the gas [28]. The above assumptions were validated through checking during the post processing of the results [4].

(11)
Where Z is the ratio of particle mass flow rate to gas mass flow rate, σ is standard deviation of the particles fluctuating velocity [5] The particle deposition effects were neglected since the deposition velocity and coefficient of loss deposition were low [29].

RESULTS AND DISCUSSION
The result is presented in both numerical model and experimentation work. The contaminant control in the various twenty-seven different models was presented herein the numerical solutions.

Contaminant particle movement
The movement of contaminants in a hospital is complex and difficult to predict. Different types of contaminants react different flow paths at different velocities. Every contaminant has a unique set chemical properties and physical properties. shows the movement of mass less contaminated particle coming from the mouth of patient when velocities are 0.5 m/s, 0.75m/s and 1.0 m/s respectively. The contaminated particle starts from the mouth of patient, moves through tortuous path and leaves through the outlet without affecting the doctor or patient. The total time taken by the mass less contaminated particle to leave the ICU is tabulated in the table 5. Path of mass less particle is more/less tortuous and varying from case to case. In case VI-(c), from figure 7, the contaminated particle is not leaving the room and it is settling in the ICU. Thus, it is affecting the health of the patient and other people.

Experimental work and flow visualization
In order to visualize the flow field by smoke it is necessary to generate smoke. The use of smoke in wind tunnels began towards the end of the nineteenth century and has subsequently been developed into an important research tool. The principle of flow visualization in the present work is based on the injection of smoke in the flow field and taking pictures of the flow pattern. The smoke particles follow the path of airflow. The smoke being visible, the flow pattern is visible to the naked eye and can be photographed. The dispersion of smoke particles by the airflow is captured by a still camera. Visualizing the flow in the acrylic glass is complicated, because there are many different reflections that interrupts the image and makes it nearly impossible to recognize the flow. Smoke visualization is the most widely used method for natural convection flow visualization.
Experimental measurements have been carried out under steady-state conditions by stabilizing the room (thermal and fluid conditions) for more than 3 hours before recording air velocity, air temperature and visualizing smoke concentration. The function of the illumination system is to provide a uniform source of high intensity light which is dispersed by smoke particles so as to take their photographs. An interesting features (as shown in fig.11) have been observed when the smoke particle produce from incandescent stick enters in the room along with air smoke, it clearly shows the path of main stream two re-circulating cell have been clearly captured one near the ceiling and one near the floor.

Validation
The geometrical structure of the model room for the purpose of this investigation has width, depth and height of the room as 1.2 m, 0.6 m and 0.8 m respectively. In order to ensure the required unidirectional airflow at the inlet of the model room, a duct is fabricated. Air is allowed to enter the model room through one inlet vent through duct and leave through the other opposite outlet vent. For velocity measurement during experimentation using a thermal anemometer, a slot of 100 cm x 3 cm is created at the top and bottom of the model to fit the strip of same dimension with holes of 6 mm at every 15 cm. During measurement of velocity using thermal anemometer as shown in fig. 12, by inserting sensor probe through a hole, other holes are closed by a tape to ensure no undesirable air leakage. A calibrated anemometer was taken for conducting the experiments with an uncertainty of ±0.001. The readings have been taken after five (5) consecutive trails. The simulated air velocities along the two horizontal inlet and outlet lines (line 1 & line 2) in fig. 13 have been validated against the measured results. Line1 passes through the center of inlet and extends straight (parallel to the floor) to the opposite wall. Line 2 passes through the center of outlet and extends straight to opposite wall. The simulated air velocities along the two horizontal inlet and outlet center line (line 1 & line 2) using k-ɛ turbulence models (standard, RNG and Realization) have been validated against the measured results. Good agreement has been found between predictions using standard k-ɛ turbulence model and experimental measurements as shown in fig. 14 (a) and (b).

CONCLUSION
In the present work, the authors have taken into account twenty-seven (27) conditions of ICU for a government hospital at Raipur (21.2514° N, 81.6296° E, 298m altitude above sea level), C.G., India region. The work is performed both numerically and experimentally to validate the results. In a case [Case-VI: (c)], it was observed by the researchers that the contaminant particle got settled in the ICU itself and thus they are highly hazardous to the health of the patient, doctor/nurse and other people on the ICU.
• The path of the main stream, as seen by smoke visualization technique, matches with the computed results.
• The CFD predictions using the standard k-ɛ turbulence model have resulted in good correlations with the air velocity measurements as well as smoke visualization. • The ventilation in case-VI was found to take more time in leaving the contaminant from the ICU, as compared to other cases. • Cases II, IV and VIII were found to be optimal for the contaminant to leave from the ICU.
In this study, it can observe that the nanoparticles play a significant role in the UV-Vis wavelength ranges where a large amount of the solar energy is dissipated. The extinction coefficient of the media is enhanced and affects the radiation transfer performance. Particle size and number have important effects on the thermal radiation transfer, increasing particle concentration leads to enhance the radiative properties.