In the present study we develop a complete Computational Fluid Dynamics (CFD) modeling procedure suitable for accurate simulations of industrial boilers fed alternatively with gaseous and liquid fuels. The model is developed and validated by means of data from on-site testing.

Two different boilers are considered: a 6-burner steam generator of a refinery for the model definition, and a 32-burner thermal power plant boiler for validation, fed by flare-gas/natural-gas or Heavy-Fuel-Oil (HFO). Selected reliable experimental data coming from performance testing are used for both the set-up of the CFD simulations as well as to compute the Flue Exit Gas Temperatures (FEGTs) employed as validation criteria. The flare-gas testing on the 6-burner boiler allows us to accurately determine that a membrane wall emissivity around 0.60 is appropriate as a boundary condition for radiation. By a sensitivity analysis, the relevance of wall emissivity with respect to fouling thermal resistance on the overall heat transfer inside the furnace is established. For boilers alternatively operating with gaseous fuel and HFO a value of 2.8 kW/(m²·K) has been found appropriate. HFO testing on the 6-burner boiler provides data to develop the additional soot modeling, crucial to properly catch oil flame emissivity, based on the model of Khan and Greeves. This allows the matching of a mean soot mass fraction of 1%–3% in the flame zone, a range found by in-flame measurements in the literature and confirmed in the present study by overall heat exchange data in the whole furnace coming from HFO boiler testing. Finally, the 32-burner boiler is considered for validation: using the same procedure and parameters, excellent agreement is found between experimental and CFD results, both on a 100% load natural gas and on a 107% load HFO performance testing.

The study demonstrates how field data can be used to validate CFD simulations and confirms that the use of physically meaningful parameters and models exempts from repeated tuning: no change is needed for emissivity and fouling thermal resistance between gaseous and liquid fuel operation, and even a simple soot modeling can be used, just ensuring physically consistent particulate concentration in the oil flame.