Numerical Investigation of Heat Transfer in a Conical, bubbling gas-solid fluidized bed

Abstract

The prediction of heat transfer in gas–solid fluidised beds remains non-trivial, particularly once nonisothermal effects and non-standard geometries are introduced. While the hydrodynamic behaviour of these systems is reasonably well established, the coupling between flow structure and thermal transport is less robustly captured in current modelling approaches. This limitation becomes more apparent in conical geometries, where the superficial gas velocity decreases with height, inherently producing spatial variations in fluidisation regime. In the present configuration, the use of a nozzle-based gas inlet further complicates this behaviour by introducing a localised high-velocity jet near the distributor. This
is likely to alter both bubble formation and local heat transfer rates in a way that is not well represented by correlations originally developed for uniformly distributed gas injection.

In this work, heat transfer in a laboratory-scale conical bubbling fluidised bed is investigated numerically using an Euler–Euler two-fluid framework implemented in ANSYS Fluent. The system consists of porous γ-alumina particles fluidised by nitrogen, with gas introduced through a nozzle at the base of the conical section. The model resolves the coupled momentum and energy balances for both phases,
with interphase heat transfer described using the Gunn correlation in the dense regions of the bed and the Ranz–Marshall correlation in more dilute zones. This combination was selected to reflect the strong variation in local void fraction, although the transition between regimes introduces some uncertainty, particularly in intermediate regions where neither correlation is strictly valid.

Three heating strategies were considered: wall heating, gas preheating, and a combined configuration. The simulations indicate that the mode of heat input has a clear impact on both the rate and spatial distribution of temperature rise. In the gas preheating case, the bed approaches a relatively uniform temperature profile more rapidly, which suggests that interphase convection dominates under these
conditions. In contrast, wall heating produces a more gradual thermal response, with temperature gradients persisting over longer times. This behaviour is consistent with a transport mechanism limited by particle–wall heat exchange and conduction through the dense phase. The combined case exhibits the
fastest overall heating, although the improvement appears to be largely additive rather than indicative of a new dominant mechanism.

The conical geometry introduces a noticeable axial dependence in the thermal behaviour. Near the inlet, where gas velocities are highest, heat transfer appears to be convection-driven, likely due to increased slip velocities and more active bubbling. Further up the bed, as the gas velocity decreases, the system transitions towards a denser regime where conductive effects become more significant. This
spatial variation is reflected in the temperature fields, which show a clear delay in heating in the upper sections. The influence of superficial gas velocity is not straightforward: increasing velocity enhances local heat transfer coefficients, but at the same time reduces the solids fraction, which limits the overall heat capacity of the bed. The resulting effect is therefore non-monotonic, suggesting the presence of an optimal operating range rather than a simple scaling with velocity.

Overall, the simulations capture the expected qualitative trends in heat transfer behaviour, but some limitations remain. In particular, the results are sensitive to the choice of interphase heat transfer correlation and to the local prediction of void fraction. Despite these uncertainties, the model provides a useful framework for interpreting the interaction between hydrodynamics and heat transfer in conical fluidised beds. The findings highlight the importance of geometry and inlet configuration in determining thermal performance, and suggest that conventional design correlations may require modification when applied to such systems.