Natural convection from heated room surfaces

Abstract

Current convective heat transfer coefficients (CHTC's) for internal room surfaces have, in most cases, been based upon data for small, free-edge heated plates. An extensive survey of CHTC data has shown that a very wide variation exists in CHTC values for vertical and horizontal surfaces. For example, a CHTC value in the range 1–6 W m⁻² K⁻¹ has been obtained for walls. Both building thermal and CFD models require accurate CHTC's for calculations of the thermal conditions and the air movement in a room. However, most such models use convective coefficients obtained for free-edge heated plates. This paper presents convective heat transfer coefficients for the heated surfaces of an environmental chamber and a small box measured under controlled conditions. Using uniformally heated plates attached to an internal surface of the chamber or the box and by accurately measuring the surface and air temperatures, the CHTC's were deduced after allowing for conduction and radiation losses from the plates. Data is presented for a heated wall, a floor and a ceiling for natural convection.

Introduction

Surface convective heat transfer coefficients (CHTC's) are an integral part of any building thermal model. However, the thermal models which are more commonly used do not all use the same values of CHTC's. Widely used CHTC's are those found in the ASHRAE Fundamentals Handbook [1], CIBSE Guide [2]and Alamdari and Hammond [3]. However, these coefficients have all been based upon experiments using small free-edge heated plates. The applicability of these coefficients to room surfaces is somewhat debatable, as the air movement over a surface in a room is different from that over a small free edge heated plate.

The differences in CHTC's have been highlighted in a report by ETSU [4]. The report shows that a reasonable level of agreement is found between many well known sources for isolated surfaces (free edge heated plates). However, it also shows that the CHTC's for walls, windows and floors can vary between about 1 and 6 W m⁻² K⁻¹, and for ceilings between about 0.1 and 1.2 W m⁻² K⁻¹.

The likely parameters, which could influence the values of CHTC's for the surfaces of real buildings in natural convection, have been identified to be:

• the shape of the enclosure

• the surface temperature distribution

• the presence of forced air movement, caused by draughts, people, fans and other devices such as radiators, underfloor heating, solar gains, etc.

• the roughness of the surface.

The data that was reviewed in this report included mainly expressions for free edge heated plates, except that of Min et al. [5]which was an early paper that dealt with heated surfaces in an enclosure. Clearly a variation of 1 to 6 W m⁻² K⁻¹ would have a significant influence on the heat flux calculated using a thermal model. In a study of thermal models by Lomas [6]in which four CHTC values were used within the thermal models ESP and HTB2, it was found that the estimated annual heating energy demand varied by around 27% depending on the value of CHTC used. However, when the two thermal models tested were implemented with the same value of CHTC they produced an estimated annual heating energy demand within 1% of each other. This clearly shows the impact the CHTC value has on the energy analysis for a building.

There has been a number of papers that have dealt with the convective heat transfer in rooms. The main problems with the experiments detailed in many of these papers were the accuracy of the instruments used for measurement and control of temperature in the test cell. A review of the literature for heat transfer data in enclosures by Hatton and Awbi [7]shows that some of the early work into surface CHTC's in enclosures have used small boxes, in some cases filled with water, Bauman et al. [8]. Such a set up does not depict a real situation and results from such a study are unlikely to be applicable to buildings.

Work reported by Bohn et al. [9]involved heating whole surfaces of a small 305 mm cubic box. The walls, considered to be isothermal, were heated or cooled using water that was pumped through milled channels in aluminium plates. However, only one size test cell was used in the study and the effect of size of heated surface on the CHTC was not investigated.

The work of Min et. al. [5]was the first to investigate convective heat transfer in a full-size enclosure. In their experiments, the authors used three differently sized chambers so that the size effect could be considered. The measurements also took into account the radiation exchange recorded by a radiometer. However, it is not clear from their work the number of temperature sensors (thermocouples) that were used to measure the surface temperature.

Work into full-size enclosures has increased since the 1980s. Khalifa and Marshall [10]carried out experiments in a full-size test cell which consisted of two enclosures, the larger one represented a living area with interior dimensions 2.95 m×2.35 m×2.08 m. The other enclosure was a cold zone, which was used to control the temperature on the exterior surface of one of the vertical walls. All the walls and roof were constructed from 50 mm thick isocyanurate board covered with aluminium foil on both sides. Twenty-one thermistors were used to measure the temperature on the wall that divided the two enclosures. Seven thermistors were used to measure the air temperature 60 mm from the interior surface of the wall (the edge of the thermal boundary layer). In their study, Khalifa and Marshall neglected the radiation effects. Error analysis has shown that the total uncertainty was 21% for a wall to air temperature difference of 1 K. However, details of the analysis method used are not given in the paper.

It was concluded by Khalifa and Marshall that the CHTC on the interior surfaces of the vertical walls of the cell were found to be a factor of 1.7 higher than values which were in use at the time in building thermal models. The local CHTC on a vertical wall of a full-size enclosure was found to deviate by approximately ±10% from the average for the whole wall.

Later work by Delaforce et al. [11]with a test cell (2.034 m×2.034 m×2.334 m) constructed from brick and polystyrene used a fan heater to heat the room. The convective heat transfer coefficients were calculated for the surfaces of the test cell. The errors due to radiation were ignored as it was claimed that small temperature differences were considered. Single CHTC values were found for a wall, a floor and a ceiling which were 1.6, 4.8 and 0.5 W m⁻² K⁻¹, respectively.

It is clear from the above discussion that precise experimental techniques and equipment are required to obtain accurate data for CHTC's in a room. This paper presents a precise and thorough study of the convective heat transfer from the heated surfaces of a room. The most accurate temperature sensors and measuring techniques available were used in the measurement of the CHTC's from the heated surfaces of two chambers. This allowed the authors to investigate the scale effect on the results. The emmmissivity of the room surfaces have been measured using an infra-red thermal imaging camera and the loss by thermal radiation was calculated using the measured emmissivity. The thermal radiation exchange was then allowed for in the calculations of CHTC.

To visualise the air flow patterns and temperature variations in the room, a number of CFD tests have been carried out. In addition, smoke tests were used for flow visualisation and the investigation of the transition point for the boundary layer over a heated wall. Finally, the results from this work are compared with equations commonly used in building thermal models for calculating the CHTC.