This thesis is concerned with advancing the modelling of indoor air flow and internal surface convection within dynamic whole-building simulation. The path taken is the conflation of computational fluid dynamics (CFD) techniques with dynamic whole-building simulation, with an accurate treatment of the co-dependencies between these modelling domains.

Two flow responsive modelling techniques were devised and implemented within the ESP-r simulation program to achieve the research objectives. The adaptive convection algorithm enhances ESP-r's thermal simulation domain by dynamically controlling the simulation of internal surface convection. Empirical methods were extracted from the literature and a new method for characterizing mixed flow convective regimes was created to provide the algorithm with a basis of 28 convection coefficient correlations. Collectively these methods can calculate convection coefficients for most flows of practical interest. Working with this suite of correlations, the algorithm assigns appropriate equations to each internal surface and adapts the selection in response to the room's evolving flow regime.

The adaptive conflation controller manages all interactions between the thermal and CFD modelling domains. The controller incorporates the latest turbulence modelling advancements applicable for room air flow simulation and possesses a suite of handshaking and thermal boundary condition treatments. The job of this adaptive conflation controller is to monitor the evolving thermal and air flow conditions in the room and dynamically select an appropriate combination of modelling approaches for the prevailing conditions. The two control schemes implemented to demonstrate the controller make use of a double-pass modelling approach. Each time-step that the thermal domain handshakes with CFD, the adaptive conflation controller performs an investigative simulation to approximate the room's flow and temperature field. Using these estimates, the controller calculates dimensionless groupings to determine the nature of the flow (forced, buoyant, mixed, fully turbulent, weakly turbulent) adjacent to each internal surface. This information is used to select suitable boundary condition treatments for each surface. A second CFD simulation is then performed using the refined modelling approach to more accurately resolve the room's air flow and temperature distribution, and to predict surface convection. In order to protect the thermal domain, a two-stage screening process is used to assess (and where necessary reject) the CFD-predicted surface convection estimates.

These adaptive modelling techniques advance the modelling of indoor air flow and internal surface convection within whole-building simulation.