Space air-conditioning of mechanically-ventilated rooms : computation of flow and heat transfer

Abstract

Computational studies of two- and three-dimensional, turbulent recirculating flows within mechanically-ventilated enclosures are reported. Two principal cases are examined: (i) two-dimensional offset jets: and (ii) three-dimensional flow induced in rooms by supply jets emanating from low or high side-wall registers. The calculations were undertaken using iterative finite-domain proceedures which solve the conservation equations for mass, momentum and enthalpy, together with additional transport equations for the turbulent kinetic energy and its dissipation rate . The effect of buoyancy waS. explicitly accounted for when modelling these equations, in order that they could be employed to simulate buoyant flow in ventilated rooms. Computations of the mean velocity, temperature and convective heat transfer distribution are reported, and compared with experimental data where available. A modified version of the two-dimensional elliptic code of Pun and Spalding (1977) was employed to simulate the offset jet case. These involve the discharge of a turbulent jet parallel to a flat surface and eventually attaching to it. The investigations covered a wide range of offset ratio (3.5-32.4). and the computed flow properties are compared with measurements from several sources. These comparisons show good agreement downstream of the reattachment point, while some discrepancies are evident upstream from this location. The differences therefore occur mainly in the recirculating flow region, and are believed to arise from shortcoming in the starting profiles, the turbulance model and the treatment of the near-wall flow. A three-dimensional elliptic finite-domain code was developed to simulate the complex, jet-induced flow within rectangular enclosures. The code was verified using both laminar and turbulent flow test cases on simpler geometries. Comparisons with the measurements and predictions reported by previous researchers were employed for this purpose. Subsequentlyr the ventilated room simulations were undertaken using three different ventilation arrangements with thermal conditions corresponding to isothermall non-buoyant (constant property) and buoyancy"affected flows. The computations were again compared with experimental and numerical predictions of previous researchers. This comparison displayed generally good agreement with these sources. A study of the flow and convective heat exchange within a warm-air heated rom, for which buoyancy effects are significant# is also reported in a bound paper (Alamdari, Hammonda nd Mohammad, 1986) for three different heat loads. Its aim to assess the balance between accuracy and economy provided by the present higher-level method compared with the intermediate-level convection model of Alamdari and Hammond (1982) when used to supply building thermal simulation programs with accurate convection heat transfer data. The computed results of both models were compared, and indicate that the intermediate-level is a valuable alternative source that can satisfy the needs of building thermal modellers. It provides resonable accuracy at a very modest cost in computing terms.