The world's energy demand by 2042 is estimated at about 18 billion tons of oil equivalent with 80% fulfilled by the combustion of fossil fuel. Combustion is predicted
to be the most important way of generating energy to cater for these energy needs. The need to address energy sustainability (fuel depletion) and environmental
pollution (emission) has led to an increased interest in energy efficiency improvement. Combustion technologies with higher thermal e�ciency and biogas (renewable) fuels are possible long-term solutions. The Moderate or Intense
Low oxygen Dilution (MILD) combustion technology can play a significant role to produce higher thermal effi�ciency and reduce emissions. MILD combustion has achieved great success, however it needs further fundamental study due to
the current limited research on open-end furnaces.
In this study, an open-end furnace with an enclosure wall was used to capture the exhaust gas and utilised it as Exhaust Gas Recirculation (EGR). The EGR recirculates a portion of exhaust gas back to the combustion chamber to dilute the oxygen before the oxidant is mixed with the fuel and increases the reactant temperature. This setup is an open-end furnace because it allows a portion of
the exhaust gas to flow out and be utilised as external EGR. In the case of a closed furnace, the exhaust gas is recirculated internally to dilute the reactants
and increase their temperature. The wall thickness for the open-end furnace is very thin compared to the closed furnace which normally has a thick wall. The
development and operating cost for the open-end furnace is cheaper than the closed furnace but the external EGR structure is an additional installation for
the open-end furnace. The main objectives of this thesis are to conduct numerical modelling using Computational Fluid Dynamics (CFD) to design and develop the
furnace, to develop and fabricate the new open-end furnace combustion chamber and to optimise the furnace performance to achieve MILD combustion.
The numerical modelling for the MILD combustion using CFD has been extensively conducted on different burners and different scale furnaces. In this study, the three-dimensional CFD model was utilised to develop and optimise the furnace. The Reynolds-Averaged Navier-Stokes (RANS) equations were solved using realisable k–ε turbulent models which has been shown by others to be a reasonably accurate model to predict the combustion temperatures and combustion
products. In this research, the results of the simulation slightly over-predict the flame temperatures by about 2 % compared to the experimental results. The non-premixed, partially premixed and premixed combustion with chemical equilibrium and non-adiabatic energy treatment for the thermo-chemical database were used. The Discrete Ordinates
(DO) model was used to solve the radiative transfer equation for a finite number of discrete solid
angles in the Cartesian co- ordinate system. The Weighted Sum of Gray Gas Model (WSGGM) was used for the absorption coefficient. The method of meshing was tetrahedrons patch conforming with advanced sizing function of proximity and curvature. The active assembly, fine span angle centre
and fine relevance centre setting were used for all grids. All the governing equations were solved using the second order upwind discretisation scheme for higher accuracy of the calculation. The pressure-velocity coupling scheme with a least-squares cell-based gradient was used and presto was used as the pressure discretisation scheme. The early stage of the combustion chamber development started with a CFD simulation of a basic enclosed wall and top open combustion chamber. After analysing the results, two external EGR pipes were added. Then two more EGR pipes
were added to make a total of four EGR pipes. The final model was improved with regard to the pipe size and the top part of the chamber design to ensure it collected an appropriate amount of the exhaust gas. This model was later used for the experimental and numerical modelling.
The laboratory scale combustion chamber (total volume of 0.33 m3 and thermal ntensity of 18.8 kW/m3 atm) was developed and fabricated. In this study, the air and fuel supply system was also developed especially for this furnace operation. The parameters for the
experimental study were the fuel compositions and the equivalence ratio. The experimental data was
collected with 42 thermocouples (R-type and K-type), a gas analyser, an oxygen sensor and pressure transducers. The EGR flow rates were determined based on the pressures and temperatures in the EGR pipes. A National Instruments compact data acquisition system was used with analogue to digital converter modules and controlled by Labview software. Experimental tests were conducted where the
secondary air supply was heated and non-heated while the reactants were varied to produce non-premixed, partially premixed and premixed flames. Methane fuel ignited more quickly than biogas due to its higher calorific value and lower self-ignition temperatures. The combination of heated
secondary air with partially premixed fuel was the quickest to ignite. The combustion chamber, EGR and exhaust temperatures and the exhaust gas species were experimentally studied for the various equivalence ratios. The flame temperature for the biogas is lower than methane due to the lower calorific value of the fuel. Both methane and biogas flames produced very low NOx (2 ppm) for all
flow rates whereas for carbon monoxide, biogas produced almost zero once the flame became steady.
The numerical modelling for the partially-premixed methane and biogas using the same geometry and conditions as the experimental work is conducted and the flame temperatures are 1,499 K and 1,368 K respectively. These can be compared
to the experimental flame temperatures for methane and biogas which are 1,483 K and 1,358 K respectively. The numerical modelling over-predicts by 1.13 % and
0.73 % respectively, which is good agreement. The exhaust gas species (CO2, H2O, O2 and NOx) were analysed at the exhaust pipe and the downstream of the EGR pipes. Both methane and biogas flames are lean. The combustion is completed with zero unburned hydrocarbons detected and excess oxygen was recorded for both the exhaust pipe and the downstream of the EGR pipes. At the downstream end
of the EGR pipes, the excess oxygen is much higher due to the fresh supply of secondary air diluting the exhaust gas as expected. The NOx emissions for methane and biogas are very low (< 3 ppm) for both locations. The numerical sensitivity test was conducted to study the effect of the
chamber wall temperature boundary condition on the flame temperature. The results show that the flame temperature is very sensitive to the combustion chamber wall boundary conditions. It concluded that biogas has advantages over methane due to lower peak temperature making the
combustion chamber and burner last longer and be more economical to operate. More numerical modelling was conducted for the experimental furnace geometry; when appropriate oxygen dilution (3–13 %) and the preheated oxidant were applied, the model could achieve the MILD regime. The limitation of the developed experimental furnace was discussed where it required a minimum of
0.1 MW/m3 atm thermal density to achieve and sustain
Further simulations were conducted using the same furnace geometry except a luff-body air-fuel nozzle to successfully achieve MILD combustion. The oxygen mole fraction is
diluted between 3 % and 13 % and the oxidant supply is preheated to achieve the overall reactant temperature above the fuel self-ignition temperature. The average and maximum chamber temperatures were almost identical at
3 % inlet oxygen mole fraction. The chamber temperature uniformity ratio was ≤ 20 % when the oxygen mole fractions were ≤ 13 %. The conventional flame was produced (the
uniformity ratio of the chamber’s temperature was > 20 %) when the oxygen mole fraction reached > 14%.
A new open-end furnace was designed, fabricated and tested experimentally as ell as numerically. The open-end furnace with the enclosed chamber numerically operated in the
MILD combustion regime for both the industrial burner and the bluff-body burner. These results can be utilised by the heating industry and also for further studies by combustion researchers. It is recommended that future work to extend this study be carried out experimentally and numerically. The further studies can be undertaken using different fuel compositions, a new gas burner, higher
secondary air preheating temperatures and a modified combustion chamber.