scope:

INTRODUCTION

Energy and load calculation procedures have relied for some thirty years on the assumption that zone air is thoroughly mixed. The application of a single control volume with a uniform zone air temperature at any point in time is reasonable for typical forced air system configurations where relatively good mixing is a design intent. The "well-stirred" zone model might be inadequate for some system designs or operating modes including:

• Air system off

• Displacement ventilation

• Underfloor air distribution

• Chilled beams

• Natural ventilation

• Mixed-mode ventilation

• Baseboard and convective heating

• Large or tall spaces, such as atria, auditoria, and stairwells

In such situations, the airflow in the indoor spaces cause nonuniform zone air temperatures. Some designs require the nonuniformity of the zone air temperature to improve energy efficiency and/or indoor air quality. Building rating systems need fair and accurate methods of comparing energy efficiency of alternative designs to conventional forced-air systems. Current trends of increased use of underfloor air distribution and natural or hybrid ventilation may create a need for engineers to account for their unique performance characteristics in sizing equipment and estimating energy use. To address these needs, it is important to estimate the impact of nonuniform distribution of indoor air temperature on the building load and energy simulations. It is therefore desirable to combine air modeling with load calculations.

Coupling air and load routines is not new. Detailed zone models of the thermal network type are available with both mass and energy balances and already offer the capabilities envisaged here (Sowell 1991; Walton 1993). These models represent the thermal zone with both surfaces and air nodes in a single network and present a single representation of the thermal zone to an HVAC component simulation. Researchers and engineers have long had the ability to formulate detailed network models of a thermal zone and solve them using a variety of software tools such as SPARK (1997) or IDA (EQUA 2002).

On the other hand, computational fluid dynamics (or CFD) has been used to model building room airflow for nearly 30 years. Chen and van der Kooi (1988) and Negrao (1995) coupled CFD to a building load/energy simulation program and, later, Beausoleil-Morrison (2000) expanded these capabilities. Other coupling work between CFD and a load/energy program includes those from Srebric et al. (2000) and Zhai et al. (2002). As pointed out by Srebric et al. (2000), a direct coupling of CFD with an energy simulation program for hourly simulation of building performance over a year is too demanding computationally.

This investigation tries to systematically build up a framework that allows an easy combination of different air models with load and energy models. Figure 1 diagrams the classification of room air models used in this paper. Such models have been developed for more than 30 years and are plentiful. The goal of the framework is to allow using all such air models with the ASHRAE toolkit (Pedersen et al. 2001). Although the terms nodal and zonal are used interchangeably in the literature, for the purposes of this study a distinction is made between them. The distinction is basically one of how strictly and how resolved the geometry of the control volumes is defined. A "nodal" model treats the building room air as an idealized network of nodes connected with flow paths. A "zonal" model uses a grid of well-defined control volumes. In both cases, energy balances are solved, but zonal models typically have many more fluid balance relations.

Three models were selected as examples for demonstrating and testing the coupling framework. Two of these are nodal models for sidewall displacement ventilation by Mundt (1996) and Rees (1998) or Rees and Haves (2001). The third is a simplified three-dimensional airflow model referred to as the momentum-zonal model (Griffith and Chen 2003). The CFD models are excluded from this study because of their high computing costs at present. However, the framework developed here is CFD-ready. CFD models can be easily plugged in should it become computationally affordable in the near future. Nevertheless, the air models themselves are not presented here in order to focus the discussion on a framework for use with any room air model, which forms the objective of the present investigation. More detailed information regarding different air models can be found in the literature (Rees and Haves 2001; Haghighat et al. 2001; Inard et al. 1996; Griffith and Chen 2003) and have been compared by Chen and Griffith (2002).