f Experimental and numerical studies on convective heat transfer in a neonatal

incubator

Y.-H. Kim I C.-H. K w o n I S.-C. Yoo 2

1Department of Biomedical Engineering, College of Health Science, Yonsei University, South Korea Research Institute for Medical Instrumentation & Rehabilitation Engineering, Yonsei University,

South Korea 2Department of Mechanical Engineering, I-lalla University, South Korea

\

Abstract--Thermo-neutral i ty is one of the major environmental factors affecting a premature or low-birth-weight neonate inside an incubator. Severe temperature differences inside an incubator lead to neonate heat loss, hypothermia and apnoea, which are closely related to air f low and air velocity. In the study, f low visualisations, hot-wire velocity measurements and computational f luid dynamics simulate the air f low inside a neonatal incubator. An anatomical ly correct neonate model is designed using a three-dimensional laser scanner system and a rapid prototyping machine. Flow visualisations demonstrate that large-scale rotating air f low is produced inside the chamber, and a number of small, stationary eddies are found in regions between the air inlet and the neonate. Hot-wire measurements show that air velocities along the long inlets are not uniform. Computational f luid dynamics show relatively uniform temperatures of about 34°C on the neonate's anterior aspect and the highest temperature of 36.1 °C at the right armpit and the crotch. Flow fields from air f low visualisations, hot-wire measurements and computa- t ional f luid dynamics are very similar, both qualitat ively and quantitatively. The smal l eddies produced between the neonate and the mattress could interfere with convec- tive and evaporative heat transfers from the neonate. Therefore it is important to el iminate eddies around the neonate in future designs of neonatal incubators.

Keywords--Neonatal incubator, Flow visualisations, Hot-wire anemometry, Rotating vortices, Turbulence

Med. Biol. Eng. Comput., 2001, 39, 114-121

J

1 In t roduct ion

A PRE-TERM or low-birth-weight neonate should be nursed at thermo-neutrality inside the neonatal incubator. When exposed to cold, a normal baby tries to conserve heat by a variety of means: peripheral vasoconstriction, increased basal metabolism with increased oxygen consumption, increased voluntary muscular activity, involuntary muscular activity, or shivering, and non-shivering thermogenesis (LEVENE et al., 2000). However, a pre-term neonate's limited ability to flex limbs and tnmk increases the exposed surface area for heat loss. Furthermore, limited shivering ability and deficiency of sub- cutaneous fat tissue makes the newborn prone to heat loss (LEVENE e t al., 2000).

A reduction of 22% in the mortality rate was found when neonates were nursed in incubators that had carefully controlled air temperature (PERLSTEIN et al., 1976). Careful considerations of temperature, airflow and humidity are necessary for effective control of a neonatal incubator. Air velocity influences both

Correspondence should be addressed to Dr Y.-H. Kim; emaih yhkim@dragon, yonseL ac.kr Paper received 8 May 2001 and in final form 22 August 2001 MBEC online number: 20023627

© IFMBE: 2002

convective and evaporative heat loss from the neonate to the environment. LIBERT et al. (1997) reported that the effect of air velocity should be analysed, with its simultaneous effects on convective and evaporative heat transfer. Using numerical analysis, YAMAGUCHI et al. (1996) reported very uneven temperature distribution inside an incubator in a transitional unsteady state. They noted that the convectional mixing in the incubator was mostly dependent on large-scale fluid motion.

Few studies have been performed on airflow, because it is highly complicated and varies greatly with different air circulation systems. BELL (1983) reported that velocity magni- tudes in all tested incubators were about 0.05 m s -1, which was beyond the sensitivity limit of a hot-wire anemometer. A good discussion of the computational method for obtaining the flow characteristics inside an incubator was given by HASEGAWA et al. (1993). The authors showed that the presence of large eddies invalidated the assumption of uniform airflow and produced small vortices that could interfere with convective heat transfer on the surface of the neonate. However, the turbulent airflow characteristics in the incubator were not considered in their studies. Many studies have been performed to visualise major airflow patterns in many different fields other than inside a commercial incubator.

Therefore the objective of this study was to develop and demonstrate an experimental measurement technique to allow direct observation of the salient airflow features in the incubator

114 Medical & Biological Engineering & Computing 2002, Vol. 40

and to compare the results from computational fluid dynamics with an anatomically correct neonate model.

2 M e t h o d s

2.1 Incubator and neonate model

Fig. 1 shows the incubator chamber* with aneonate model for the present study. The chamber had a long inlet at the right side of the neonate and an exit of similar size and shape at the left side of the neonate, so that the major airflow in the incubator chamber would be of a transverse directional nature. A baby manikin was scanned in widths of 5 mm using a three-dimensional laser scanning system t. The scanned model was enlarged to the actual dimension of the neonate, and then the data were trans- mitted to a rapid prototype machine ~. The neonate model was made using a three-dimensional printing technique with micro- stone in a short time. The neonate model was 3.2 kg in weight, and 43 cm in height, with a slightly knee-flexed position, its surface area was 0.155 m 2, which relates to TELLIEZ et al. (1997).

2.2 Flow visualisations

The flow visualisation system used in this study is shown in Fig 2. it consisted of a floodlight, a digital video camera, a slit-

- - 7 - - - - - ~ ~ _ 7 6 0 m m - _ _ - - ~

~ .~_ - - _ - - - - _ t o p _ w a i F - - - - - - ~ - ~ - ~

left wa l l rear wa l l ,,,

- t ~ z . . . . . . - - _ /" ~ Z y - - - - - - _ ,,

/ [ ~ ' f r on t wal l - r igh t wa l l -~U,x~ _ ~

. in let ' . ~ , • " - . . . . \ . ~ 2 ~ f f . : N . _~"~ o u t l e ~ ,

~ ~ - . . . . ~ = _ _ - ~ / 4 7 0 m m - ~ : ~ - ~ - 2 2 - ~ : - _ _ - - = ~ - ~ - _ , - ' ~

_ _ za_~ ~ _ ~ . ~ 38 152 152 152 152

m m 114

4 2 5 m m

Fig. 1 Incubator chamber with neonatal model for present study

2 3

/1\

Fig. 2 Apparatus o f f low visualisation system." (1) incubator," (2) smoke generator," (3) digital video camera," (4) light souree with slit system," (5) mirror

* Ohmeda Medical Care Plus 2000, USA

t Surveyor DS-2016, Laser Design, USA

; Z-402, Z-corporation, USA

Medical & Biological Engineering & Computing 2002, Vol. 40

light system and a smoke generator. The projection lamp was turned on only during actual experiments to eliminate the heating that could cause buoyancy effects in airflow. A 2 kW projection lamp was directed toward the incubator chamber by a set of slit lights. This system provided an angular control with coplanar-orthogonal adjustments. A digital video camera** was used to film airflow images on a PDV-124ME tape.

in the early stage of this investigation, a series of preliminary experiments were conducted using different types of seeding particle, in the first attempt, phenolic micro-balloon particles (5-40 pm in diameter) were tested. The micro-balloons were discrete tracers and provided information about local velocities and the general features of large-scale airflow structures within the incubator. However, our visual observations indicated that an alternative seeding process was required to improve the clarity of the flow patterns, owing to the complex nature of the airflow field in the incubator. Therefore, in this study, a solid- particle generator was modified so that both the particle concen- tration and the airflow rate could be changed to provide appropriate seeding conditions.

Accordingly, a series of experiments were conducted with different sizes of particle (0.5-3 pm in diameter). Motion pictures demonstrated a relative improvement in the level of contrast. However, both the particle concentration and the flow rate were not sufficient to reflect the details of airflow features. This was owing to the fact that the particle concentration in entrained air was not sufficient, in accordance with the previous result, it was evident that a larger concentration of particles would be required to highlight the airflow characteristics of interest. Motion pictures using a smoke generator with special mineral oil showed a significant improvement and a consistency in visual contrast. The smoke generated by this method was clean, dense and pure white in colour. A gas pressure control valve was externally mounted at the front of the smoke generator, so that the gas pressure could be adjusted quickly to suit the particular requirements for which the smoke was being used. The particle size of the smoke varied according to the pressure setting, from 0.5 pm to about 4.5 pm in diameter.

Parts of the incubator were removed and modified to create an optical access through the incubator chamber and an air inlet for the airflow visualisation study. Owing to the presence of seams, inspection doors and rubber connectors with clamps on the clear plastic doors, airflow visualisation would be very limited. To avoid the visibility problem, the incubator chamber was modi- fied using clear transparency film. The transparency film was glued with a clear adhesive to attach it to the chamber around the doors, creating an identical image of the inside of the chamber. The neonate model fabricated by the rapid prototyping process was placed on top of the mattress in the incubator, in a supine position. The ambient temperature in the laboratory was 15.9 °C. The mean airflow rate and the mean Reynolds number based on

3 1 the hydraulic diameter at the inlet were 0.013 m s - and 2084, respectively.

Seeding materials were then introduced at the surface of the filter element. A small slit was cut into the filter to mount 3.8 cm long sections of 0.95 cm thin-wall copper tubing. The tubing was pressed to form a thin oval shape (1.27 cm long by 0.5 cm wide) that was inserted into the filter and pushed through the slit until the tubing edge was flush with the filter surface. The tube was then glued and sealed to the filter to prevent air from bypassing the filter and was tested at various locations on the filter surface until the seed was fed into the area of interest.

A hole of 0.87 cm in diameter was drilled so that a tube could be inserted through the incubator housing and connected to the copper tubes in the filter. The usefulness of these airflow

** Sony DSR-2000A

115

visualisations depends on the degree to which the seed can be introduced without affecting the actual flow. To avoid airflow disturbances while the incubator was seeded, smoke velocity was matched to local airflow velocity. Therefore the smoke velocity from the tubes was approximately the same as that of the air passing through the filter. Observation of airflow pattern changes when the seeding system was over-pressurised or under-pressurised led us to believe that the seeds we used were introduced in an isokinetic manner. Preliminary testing with a stagnation pressure probe inserted through the cover of the incubator showed very small pressure variations between measurements made at the seeding tube outlets and nearby filter surfaces.

A floodlight was used as a light source for the airflow visualisation process. Once the floodlight beam was on, the slit system was adjusted to form a sheet of light through the area of interest. Video stock was then loaded, and the digital video camera was positioned and focused on the sheet of light in the incubator, which was filled with incense smoke to scatter light for focusing. The smoke generator was then connected to the intake system of the incubator. The air circulation fan was turned on, and the laboratory lights were turned offto reduce unwanted light intensity on the camera lens. Smoke was then introduced into the incubator through the intake system, it was necessary to take images at high speed to obtain a satisfactory level of unstable airflow features.

2.3 Hot-wire measurements

A thermo-electric hot-wire anemometer tt was used to measure low air velocities and the degree of turbulence in the incubator, its measuring range was from 0.01 m s -1 to 5 m s -1 with an accuracy of -4-0.1% at a temperature between 0 and 40 °C. Automatic temperature compensation, barometric pres- sure correction for local altitude and simple zero-adjustment guaranteed a high-precision measurement.

Hot-wire measurements along the contour of the neonate, 0.2cm distant from the neonate's body, were employed to measure local airflow velocities and fluctuations. Three velocity components were measured in this study: one component parallel to the axis of the neonate's body (U-component); one component perpendicular to the axis of the neonate's body and parallel to the mattress (V-component); and one component on the measurement plane perpendicular to the mattress (W- component). As a single hot-wire probe was used, it was impossible to distinguish the flow direction. Therefore only absolute velocities are presented in this paper.

Turbulence can be characterised by the local fluctuations in the flow field. Turbulent motion is randomly time-dependent, strongly non-linear and generated in the presence of mean airflow velocity gradients. Within the random variation of quantities, statistically distinct average values could be defined. Turbulent motion could be expressed as a time- averaged part and a fluctuating part

u = fi + u' (1)

The time mean fi of a turbulent function u(x, y, z, t) was defined by

lit u dt (2) U=-T0

where T is an averaging period taken to be longer than any significant period of the fluctuations.

The turbulent intensity was calculated from the fluctuating part and defined by

1i; 7 = T u'2at (3)

The degree of turbulence was defined by

Uz2

T (4)

The turbulent kinetic energy per unit mass was defined by

½ (u,2 + ~,2 + w,2) (5)

2.4 Computational f lu id dynamics

A three-dimensional steady-state convective heat transfer in the incubator with a neonate model was computed using pressure-based finite-volume software, CFD-ACE ~*-. SIMPLEC (VAN DOORMAL and RAITHBY, 1984) was used for pressure corrections, and k and e for the k-e turbulence model were assumed to be 0.5 and 64.3.

The geometry of the neonate was obtained from the scanned data of a baby manikin with a three-dimensional laser scanning system. Fig. 3 shows the grid distribution of the computational model. The density and the viscosity of the air were assumed to be 1.165kgm -3 and 1.86 x 10-Skgm -1 s. The mean flow rate at the inlet was 0.013 m 3 s - 1. The inlet velocities obtained from hot- wire measurements were used as an inlet boundary condition. Mean velocity and mean temperature at the inlet were 1.3 m s -1 and 35 °C, respectively. Thus the mean Reynolds number based on the hydraulic diameter at the inlet was approximately 2084. Governing equations for the three-dimensional steady-state incompressible flow in tensor form are given by

Oui - 0 (6) qox i

Ouju.._____Zi - l o p 3___~ v OUi _ Oxj p Ox i + Oxj Oxj gi f i (T - To) (7)

3uipCpT 3 K O T axj -axj axi +0 (8)

where p, v, fi and Cp are the density, kinematic viscosity,volu- metric thermal expansion coefficient and specific heat of the air. ui is the velocity vector, T is the temperature, and %. is the

2 viscous stress tensor. A constant heat flux of 0.54 W m - at the neonate's body surface was calculated by

q = hA(T~ - Ta) (9)

where q, A, Ts and T, are the rate of heat loss by convection, the surface area, the surface temperature and the ambient

Fig. 3 Grid distributions for computational model

tt ThermoAir3, Schiltknecht, Germany

116

;; CFD Research Corporation, USA

Medical & Biological Engineering & Computing 2002, Vol. 40

temperature. Convection coefficient h was obtained from JOHNSON (1991). The temperature of the incubator wall was 27 °C.

3 Results

Airflow visualisation provided qualitative information on the total flow field at the plane where the light beam was placed. On the other hand, hot-wire measurements allowed quantitative descriptions of the airflow field. The combination of these two experimental methods made it possible to understand airflow patterns inside the incubator both qualitatively and quantita- tively.

Fig. 4 illustrates visualised flow fields in the incubator model in the plane perpendicular to the axis of the neonate's body. Motion pictures of smoke particles have demonstrated that flow in the incubator was turbulent, unsteady and three-dimensional. Near the inlet, irregular airflow was observed to have a turbulent and unsteady nature. Flow was separated at the wall near the inlet, which resulted from the interaction between the air stream and vortex layers springing from the corners of the incubator. Disturbances from the corners of the incubator chamber and the sharp angle of entry of the incubator could affect the airflow instability. A number of small-scale eddies were consecutively produced in areas between the air inlet and the neonate model, where the local shear velocity was high. However, these small eddies were found to be stationary at those locations.

Computational fluid dynamics provided more detailed infor- mation on the airflow fields in the incubator. Fig. 5 shows the velocity vectors on the neonate's transverse planes at the fore- head, armpit, abdomen and thigh. Airflow from the inlet formed a large -rotating vortex near the middle o f the incubator chamber and partly moved towards the exit; this was similar to the airflow visualisation results. A counter-rotating vortex near the fight side of the neonate's armpit was clearly observed and was continuously found on the transverse planes between the nose and the abdomen. In addition, small vortices were also found on

both armpits and the limbs. A fairly rapid flow was observed on the right anterior side of the neonate with a large rotating vortex. The maximum U-velocity was found at the chin area, the maximum V-velocity was found around the abdomen, and the maximum W-velocity was found near the left ear.

Fig. 6 shows the computed velocity distributions along 5 mm distant longitudinal lines near the neonate. In general, V-velocity was dominant along the neonate's longitudinal direction. At the longitudinal line near the neonate's left arm (line a), the negative V-velocity was large in magnitude, but other components (U,/q/) were relatively small. At the longitudinal line near the left anterior surface of the neonate (line b), V-velocities in the face and the thigh areas were relatively small in comparison with those in the chest and abdomen areas, whereas W-velocities were large over the entire line. At the longitudinal line near the neonate's anterior surface (line c), the magnitude of V-velocities decreased at the forehead, mouth and abdomen regions, where locally convex body contours were formed. W-velocities increased near the chest area. At the longitudinal line near the right anterior surface of the neonate (line d), V-velocities were small, because air flow was blocked by the body, but they were large in the areas between the head and the left wall, and between the right foot and the right wall. At the longitudinal line near the neonate' s right arm (line e), velocity distributions were similar to those in line d, and U-velocity increased slightly away from the head region, with a magnitude of 0.2 m s - 1. As shown in line a and line b, negative U- velocity near the head and positive U-velocity near the feet were observed. However, positive U-velocity near the head and negative U-velocity near the feet were observed in line d and line e. This was owing to the flow entrainment to free space.

Fig. 7 shows computed skin temperature distributions of the neonate in the incubator chamber. Relatively uniform tempera- rares of about 34 °C were observed on the neonate's anterior aspect. Convective heat transfer increased where air velocity was large. Temperatures at the right armpit and the crotch were 36.1 °C, and the temperature at the left surface of the neck was 35.3 °C. Temperatures at the medial aspects of both thighs were high, where weak stationary vortices were found. Fig. 8 shows

a b

c d

Fig. 4 Flow visualisation pictures in neonate's transverse plane at forehead. (a) t=2 .5 s; (b) t= l l.O s; (c 9 t=15.0 s; (d) t=26.5 s

Medical & Biological Engineering & Computing 2002, Vol. 40 117

Fig. 5

°C T 36'0 ! 35'8 1 ~ ~,-

/ - " ~ ~ . / ~:.-.-.-.-.-.-.-.-.~~_ ,.-~ ......

35.2 4 4\, 35.0~ 34.8~ 34.6~ 34'4 i 34'2 t a~

\

a b

. . . . ,-:.i i'' 4" "x .,,,'I' i

'~1

c d

Velocity vectors" on transverse planes o f neonate at various locations: (a) forehea& (b) armpit," (c) abdomen; (d) thigh

temperature distributions at various locations along the neonate's longitudinal directions. Temperatures at the head region were slightly lower than those at low extremities. A locally increased temperature was found at the chest area, where airflow velocity decreased.

As shown in Fig. 9, inlet velocities measured by the hot-wire anemometer were not uniform, but varied along the long inlets. The inlet velocity was only 0.22 m s -1 at the locations closest to the neonate's head, 0 .75ms -1 at the middle areas of the incubator, but relatively constant at other locations, within the value of 1.5-1.6ms -1.

Fig. 10 shows absolute velocities in three different compo- nents, 0.2 cm distant from the neonate's body. V-velocities were large, especially at the forehead and foot areas, but they decreased in the neck and chest areas, in general, results from the hot-wire velocity measurements and computational fluid dynamics were similar to each other.

Fig. 11 shows the degree of turbulence and the turbulent kinetic energy, as defined in (4) and (5). in general, the degree of turbulence was higher near the nose, neck, chest and abdomen. in particular, the U-component (longitudinal fluctuation) was larger than the other two components (V, W), except for near the nose and the chest. Turbulent kinetic energy was relatively large near the nose and the chest. The mean airflow velocities near the neck and the abdomen were quite small, which resulted in a small amount of turbulent kinetic energy.

4 Discussion

Neonates exchange heat with their surroundings via four channels: conduction, radiation, convection, and evaporation. HEY (1969) proposed a well-known empirical expression of the

118

heat loss from neonates, but the expression did not take into account convective heat transfer between the neonate and its surroundings. Conduction has been considered to play a minor role in body heat exchanges (STOTHERS and WARNER, 1984).

Convective heat transfer has been assumed to be negligible in some previous studies (HAY and MOUNT, 1972; HEY, 1966; BELL and RIOS, 1983). On the other hand, other investigators (FARANOFF et al., 1972; STOTHERS, 1980; THOMPSON et al., 1984) have been aware of the importance of convective heat transfer, but their studies were not based on accurate 3D unsteady fluid mechanical findings. A lack of air velocity control can mask an increase in body heat loss that unbalances body heat storage, imposing a thermal stress, and can exaggerate the problem of fluid balance encountered in pre-term neonates (OKKEN et al., 1982; LIBERT et al., 1997). Therefore convective heat transfer should be considered as one of the most important parameters in flow in an incubator.

Even though temperature distributions in the incubator chamber were not measured in the present study, results from airflow visualisation studies and hot-wire velocity measure- ments were relevant to those from the computational fluid dynamics. Therefore temperature distributions in the incubator chamber, including the neonate's skin temperature, can be regarded as reasonable. Results from computational fluid mechanics showed that a warmer temperature was maintained on the neonate' s right aspect, especially near the neck, the armpit and the crotch, where air was relatively stationary. Heat losses were larger at the forehead region, but smaller in the extremities; these results were similar to those of a previous study (SARMAN et al., 1992).

Owing to the fact that airflow in an incubator is not directional but turbulent, accurate measurements of air velocities are difficult. Another problem relates to the fact that the air velocity

Medical & Biological Engineering & Computing 2002, Vol. 40

Fig. 6

o4 t 0.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.2 0.1

E 0 0.1 0.2 0.3 0.4 0.5

0.4 0.3 0.2

' - 0.1 E

~ ~ 2 "~

q?. l

0.2 0.3 0.4

M:).5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . q:).6 ~ I H I c I A I T I

0.3 0.~ 0.1

E ~).1 '~ 4 .~ _o

4.~

0.z

7~-"-'% ~

I H I c I A I T I i

0.50 0.42~).33 0.25 0.16 0.08 0.01 0.09 0.18 0.26 x~zo-ord ina te , rn

e

E

_8

E ~q

o

0.4 l "J 0.3 ~ ~ ,/\

0 2 . . . . . . . . . . . . . . , ? - : = : : : - . . . . . . . . . . j . . . . . . . . 0.1 .. . . . . . . . . . . . : . . . . . . . . . . . . . . . . . . . . . 2: ...... • :" '--'-2. ......

0 -"--~:~,'-/- - - ..... "'- ~ " " ' : . . . . . . "--": ...... --":":<- 2"- --"--"-"'--- 'I 0.1

M:).5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . q?6 ~ I H I c I A I T I

b O.2 ,. Ol . . . . . . . . . . . . . .___ !

0 . . . . . . . -x---: ~ - ' ~ : . , ~ _ - - - ' - = ~."-4 . . . . . . . -i

M 3 ' I U : I - - ~ ~ 0.30.40.2

~:).5 ,, ~).6 i I H I C I A I T I ~ i

~).50 ~).42 0.33~).25 0.16 0.08 0.01 0.09 0.18 0.26 x~zo-ord ina te , rn

d .,..'\ \,..

/"" "\ \' "i . j ,,

. " 7htvie 1

U,x

Computed velocity distributions along longitudinal directions of neonate. (---) U; ( . . . . ) ~" (--) W," (~) model," H= head and neck," C=chest; A=abdomen; T=thigh and foot. (y, z)=(a) (O.11m, O.O17m), (b) (O.15m, O.070m), (c) (0.22m, O.110m), (d) (0.30m, O.070m), (e) (0.34m, O.O17m)

Fig. 7

°C T

36.0 I

35"8 I

• ~ ~ ; 3 4 . 8 ~

3 4 . 6 4

34"4 i

34"2 1

Computed skin temperature distributions of neonate m incubator chamber

is often below the sensitivity limit o f relevant measuring apparatuses. Earlier investigators could not estimate convective heat transfer on a quantitative basis, it is generally recognised

that no experimental work has previously been performed to visualise flow structure inside a neonatal incubator with an anatomically correct neonate model. The present study has successfully visualised a complicated airflow in a neonate incubator.

Air entrained into the chamber on the inlet side appeared to control airflow structures in the incubator chamber. Airflow separation at the comers of the incubator created a secondary vortex that could have an impact on major parameters to control the performance of a neonatal incubator. HASEGAWA et al. (1993) showed that smaller eddies were not stationary and were produced continuously, being transmitted by convection along with the largest eddies. However, our results showed that small stationary vortices produced between the neonate and the mattress could interfere with convective and evaporative heat transfers on the surface of the neonate. The incubator chamber design itself could significantly affect the flow structure. Therefore it is important to eliminate the formation of small vortices around the neonate in the design of the incubator chamber by considering fluid dynamics aspects.

Based on the results from computational fluid dynamics, it was found that fluid particles moved in very irregular paths, which could be relevant to the turbulent characteristics. This could cause exchanges of momentum and energy from one portion of fluid to another and would be important in convective heat transfer between the neonate and the environment.

Medical & Biological Engineering & Computing 2002, Vol. 40 119

Fig. 8

o)

&

35.1

35.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34.9

34.534"634"734"8 ~ i _ i_ i_ i_ i_ ! i_ i_ i_ i_ i_ i_ i_ i_ ! i_ i_ i_ ! i_ i_ i_ i_ i_ i_ i_ i_ ! ~ 34.4 34.3 34.2 I H I C I A I Y ] ' l H I c I A I T I '

b

,¢

&

?

o)

35.1

35.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34.9

34.734.8

34.6

34.5

34.4

~4.~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34.2 1 H I C I A I T I '

c

35.1

35.0

34.9

34.8

34.7

34.6

34.5

34.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34.3

34.2 I H I C I A I T I ' -0 .5 -0 .4 -0 .3 -0 .2 -0.1 0 0.1 0.2

x -co-ord ina te , m e

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I H I ? I A I T I ' -0 .5 -0 .4 -0 .3 -0 .2 -0.1 0 0.1 0.2

x -co-ord ina te , m

d

W,z

U,x

/ ' \ \-,

'\ \' "i

, J, '\ .j/ ", , , / r ight v iew

d b

Temperature distributions at various locations along neonate's longitudinal directions. (--) T; (~) model," H = head and neck," C = chest," A=abdomen; T=thigh and foot. (y, z)=(a) (O.11m, O.O17m), (b) (O.15m, O.070m), (c 9 (0.22m, O.110m), (d) (0.30m, O.070m), (e) (0.34m, O.O17m)

2.0

,- 1.6

E

"~ 1.2 _o

0.8 _= o ..a 0.4

I H I o I A I T I -0 .50 -0 .42 -0 .33 -0 .25 -0 .16 0.08 0.01 0.09 0.18

x -co-ord ina te , m

Fig. 9 Inlet velocities measured by hot-wire anemometer

0.26

reported that air fluctuation could induce sudden warming, which causes apnoea. However, from a fluid dynamics point of view, some turbulent fluctuations could be efficient in convective heat transfer between the neonate and the environment.

in the present study, the neonate model inside the incubator remained stationary in its supine position, in practice, the neonate could move continuously and wear light clothing or a blanket. Also, the present study was limited to a specific incubator and cannot be extrapolated to all marketed devices. However, flow visualisations, hot-wire velocity measurements

0.40

0.35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0.30 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E zX .~ 0.25 _o • A

0.20 i ,

0.15 - ' ~ ~. "~. A

-~ 0.10 A []

0 . 0 5 . . . . . . . I . . . . . . . t l . . . . . ~ . . . . . ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,~ . . . . . . . ,, [] • g ~ [] © © ©

-0 .30 -0 .25 -0 .20 -0 .15 -0 .10 -0 .05 0 0 . 0 5 0 . 1 0 0.15 x -co-ord ina te , m

fo rehead neck chest th igh foot nose a b d o m e n

Fig. 10 Measured velocities" along contour of neonate 0.2 cm from body, compared with results from computational fluid dynamics" (CFD). (o) experiment U; (M experiment ~" (,,) experiment W," (o) CFD U; (A) CFD ~" (m) CFD W

and computational fluid dynamics were performed on a commer- cial neonatal incubator. Even though further studies with different incubators are required to provide their fluid dynamic efficiencies, the present study should be helpful in the future design of neonatal incubators.

120 M e d i c a l & Bio logica l E n g i n e e r i n g & C o m p u t i n g 2002 , Vo l . 40

o~

=o o)

"6

103

O) .0_

O) "5

100

90

80

70

60

50 40

30

20

10

O

• ©

&

• ik

0 lk 0.30 0.25 0.20 43.15 43.10 43.05 0 0.05 0.10 0.15

~co-ordinate, rn

lttl t forehead neck chest thigh foot

nose abdomen

0.09

0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0.00

I

0.30 43.25 43.20 0.15 0.10 0.05 0 0.05 0.10 0.15 x co-ordinate, rn

forehead neck chest thigh foot nose

Fig. 11 Turbulent characteristics along contour o f neonate 0.2 cm from neonatal model. (a) Degree of turbulence." (o) dZ/u; (A) v~Z /w (,,) w~Z /w. (b) Turbulent ldnetic energy

5 Conclusions

Both experimental and computational studies were performed to determine convective heat transfer in an incubator with an anatomically correct neonate model. Flow visualisations and the computational fluid dynamics demonstrated large-scale rotating flows inside the chamber and a number o f small-scale eddies in areas between the inlet and the neonate model. Hot-wire measurements showed that air velocities along the long inlets were not uniform. These two experimental methods made it possible to understand flow patterns inside the incubator, both qualitatively and quantitatively. Computational fluid dynamics showed locally increased temperature distributions around the armpit and the crotch, where convective heat transfer decreased. i t is important to eliminate the formation o f eddies around the neonate, and the neonatal incubator should be designed in view of aerodynamic aspects for a stable airflow pattern.

Acknowledgments" This work was partially supported by the Research Institute for Medical Instrumentation & Rehabilitation Engineering. We are also thankful to Joong-Wae Medical Co., Ltd, for providing an incubator for the present study.

References

BELL, E. E (1983): 'Infant incubators and radiant warmers', Early Hum. Dew., 8, pp. 351-375

BELL, E. E, and RIOS, G. R. (1983): 'A double-walled incubator alters the partition of body heat loss of premature infants', Pediam Res., 17, pp. 135-140

FARANOFF, A. A., WALD, M., GRUBER, H. S., and KRAUS, M. H. (1972): 'Insensible water loss in low birth weight infants', Pedia- trics', 50, pp. 236-245

HASEGAWA, T., HORIO, H., OKINO, H., TAYLOR, T. W., and YAMA- GUCHI, T. (1993): 'Three-dimensional visualization of air flow in infant incubators using computational fluid mechanics', Biomed. Instrum. Technol., pp. 311-317

HEY, E. N. (1966): 'Temperature control in incubators', Lancet, 2, p. 202

HEY, E. N. (1969): 'The relation between environmental temperature and oxygen consumption in the newborn baby', J. Physiol., 200, pp. 589-603

HEY, E. N., and KATZ, G. (1970): 'The optimum thermal environment for naked babies', Arch. Dis'. Child., 45, pp. 328-334

HEY, E. N., and MOUNT, L. E. (1972): 'Heat losses from babies in incubators', Areh. Dis'. Child., 42, pp. 75-84

JOHNSON, A. T. (1991): 'Biomechanics and exercise physiology' (Wiley & Sons), pp. 361-368

LEVENE, M. I., TUDEHOPE, D. I., and THEARLE, M. J. (2000): 'Essentials of neonatal medicine, 3rd edn' (Blackwell Science, 2000), pp. 89-90

LIBERT, J. P., BACH, V, and FARGES, G. (1997): 'Neutral temperature range in incubators: performance of equipment in current use and new developments', Crit. Rew. Biomed. Eng., 25, pp. 287-370

OKKEN, A., BLIJHAM, C., FRANZ, W., and BONN, E. (1982): 'Effects of forced convection of heated air on insensible water loss and heat loss in preterm infants in incubators', J. Pediatr, 101, pp. 108-112

PERLSTEIN, P. H., EDWARDS, N. K., ATHERTON, H. D., and SUTHER- LAND, J. M. (1976): 'Computer-assisted newborn intensive care', Pediatrics', 57, pp. 494-502

PERLSTEIN, E H., EDWARDS, N. K., and SUTHERLAND, J. M. (1970): 'Apnea in premature infants and incubator-air-temperature changes', N. Engl. J Ned., 282, pp. 461-466

SARMAN, I., BOLIN, A., HOLMER, I., and TUNELL, R. (1992): 'Assess- ment of thermal conditions in neonatal care: use of a manikin of premature baby size', Am. J. Perinatol., 9, pp. 239-246

STOTHERS, J. U. (1980): 'The effect of forced convection on neonatal heat loss', J. Physiol., 305, p. 77

STOTHERS, J. K., and WARNER, R. M. (1984): 'Thermal balance and sleep state in the newborn', Early Hum. Dew., 9, pp. 313-322

TELLIEZ, E, BACH, V., DELANAND, S., BOUFERRACHE, B., KRIM, G., and LIBERT, J. R (1997): 'Skin derivative control of thermal environment in a closed incubator', Ned. Biol. Eng. Comput., 35, pp. 521-527

THOMPSON, M. H., STOTHERS, J. K., and MCLELLAN, N. J. (1984): 'Weight and water loss in the neonate in natural and forced convection', Areh. Disease Childhood, 59, pp. 951-956

VAN DOORMAL, J. R, and RAITHBY, G. D. (1984): 'Enhancements of the SIMPLE method for predicting incompressible fluid flows', Numerical Heat Transfer, 7, pp. 147-163

YAMAGUCHI, T., TAYLOR, T. W., OKrNO, H., HORIO, H., and HASE- GAWA, T. (1996): 'Computational fluid mechanical study of the convective heat transfer in a closed space simulating an infant incubator', Frontiers" Ned. Biol. Eng., 7, pp. 129-141

Author's biography

YOUNGHO KIM received his MS and PhD in Biomedical Engineering from the University of Iowa, Iowa City, Iowa, USA, in 1989 and 1991, respectively. As a Postdoctoral Fellow, he worked in the Cardiovascular Fluid Mechanics Laboratory at Georgia Institute of Technology, Atlanta, Geor- gia, USA until 1994. His is currently an Assistant Professor at the Department of Biomedical Engi- neering, Yonsei University, Wonju, Kaaagwon,

South Korea. His research interests include cardiovascular biomecha- nics, muscular-skeletal biomechanics, prosthetic and orthotic biome- chanics, and other areas of rehabilitation engineering.

Medical & Biological Engineering & Computing 2002, Vol. 40 121