Gas-side fouling of heat exchanger surfaces

Abstract

Fouling is an unpleasant and unavoidable occurrence in heat exchangers that decreases the performance of the heat exchanger. Fouling reduces the overall efficiency of a system and can lead to equipment failure. Understanding the mechanisms of fouling deposition and removal of fouling material is important to minimize the overall fouling in a system.

The following work consist of a presentation of mechanisms for particle transport, particle surface interaction and particle resuspension. Further, numerical model for particle transport in turbulent channel flow is presented. The flow and temperature conditions are set to resemble a heat exchanger environment to consider the effect of turbulence and temperature on the transport and deposition of particles. Furthermore, the effect of gravity has been evaluated based on different orientations of the flow. The commercial flow simulation program Fluent ANSYS has been used for simulation of the main flow and the particulate flow. The particle flow was simulated by DPM model of Fluent ANSYS based on an initially solved flow and temperature field. The DPM model was extended through user-defined-function (UDF) macros to account for Cunningham correction factor and to include the effect of Brownian diffusion on the particles. The deposition of particles was evaluated based on transport mechanisms to the wall. At the wall-cells the deposition was stored in user-defined-memory-location (UDML) that stored the number of deposited particles in the cell values. Different attachment (rebound) models were considered through implementation in MATLAB. Based on the models one attachment (rebound) model was implemented into Fluent ANSYS through UDF macros to account for particle rebound behavior in the flow.

The resuspension is considered as a separate simulation in MATLAB and can be viewed as a post-process simulation. In this regard the quasi-static Gaussian RnR model proposed originally by Reeks and Hall(2001) and the non-Gaussian RnR model proposed by Zhang et. al. (2013a) have been implemented in MATLAB. In addition, resuspension is considered through moment balance approach with a Monte Carlo simulation. Both RnR and Monte Carlo resuspension model considered surface roughness effects as they are important for resuspension. The models used during this thesis are appropriate for the initial stage of deposition and resuspension, before multilayer effects become relevant. The combined effect of deposition or attachment (rebound) with resuspension was evaluated based on real particle size distribution in the flow.

The simulation results, based on turbulent channel flow, show that RSM model may successfully be used to describe all deposition region. The k- ω SST model on the other hand gave inconsistent results for the lower part of the zone between the diffusion and inertia controlled region, and the diffusion controlled region. This inconsistency was associated with the assumption of isotropic turbulence of k-ω SST model. When the effect of thermophoresis was included the models gave more consistent results as the dominant transport mechanism was associated with temperature gradient. Introducing rebound effects showed a decrease in deposition for the biggest particle, but with an uneven effect considering gravitational orientation. The deposition decreased drastically for the biggest particles when gravity was against the flow direction. The combined effect of deposition and resuspension with a particle size distribution in the flow further showed the difference in deposition and deposit layer for gravitational orientation against and with the flow. Although it remains to be examined further, the present simulations shows that gravity against the flow direction results in least deposition for the simulated turbulent channel flow. The presented numerical tools and approach of simulation incorporates the desired features of particulate fouling such as rebound of particle at sufficient velocities, removal of particles by hydrodynamic forces, the effect of particle motion after rebound and the effect of particle-substrate material properties. Combining the presented tools with capabilities of Fluent ANSYS and MATLAB one can evaluate the initial stage of fouling for other geometrical cases, operation conditions and gravitational orientations.