Flamelet Model And Large-eddy Simulations For Supercritical Hydrothermal Combustion

Supercritical hydrothermal combustion utilizes the special solubility of the supercritical water,in which the organic fuel mixes with the oxidizer in a single homogeneous phase with no inter-facial resistances.In that single-phase combustion,the combustion time scales are short-ened and the reaction efficiency is improved.At the same time,due to the poor solubility of ionic salts in the supercritical water,they precipitate rapidly and can be handled as the furnace slag.Thus,supercritical hydrothermal combustion is a very promising combustion technology with high efficiency and no pollution emission.However,the numerical investigation of hy-drothermal combustion is rather rare,especially for the relevant turbulent combustion models in industrial purpose.Under these circumstances,this paper develops a numerical platform which is applicable to investigate the hydrothermal combustion.An extended flamelet model is proposed and evaluated under the framework of laminar and large-eddy simulation.First,a numerical platform based on the open-source CFD software Open FOAM has been developed to conduct numerical simulation of supercritical hydrothermal combustion.To pre-cisely predict the mixture properties under supercritical condition,this numerical platform in-corporates comprehensive real-fluid models,including real-fluid equation of state,thermody-namic properties and transport properties.These real-fluid models are validated through com-parisons with the canonical substance property database and the experimental data from a su-percritical jet.Second,the real-fluid effects on the supercritical hydrothermal combustion are systemly investigated.To quantify the effects of each part of the real-fluid models,descriptions of the real-fluid EOS,thermodynamic properties and transport properties are progressively incorpo-rated into the numerical simulations.It is found that,for diffusion hydrothermal flames,the replacement of the equation of state is the most important correction with the flame location be-ing incorrectly predicted by the ideal EOS.For premixed hydrothermal flames,modifications from real-fluid models have little impact on the flame temperature and species distributions,whereas the laminar flame speed is significantly influenced,the error of which can be up to35%if the ideal-gas model is used.Based on the comprehensive real-fluid models,the Lewis number of each species for supercritical hydrothermal combustion is given for the first time.It is found that good agreement is achieved between the results from the non-unity Lewis number assumption and detailed transport approach under different evaluation cases.Third,the real-fluid flamelet/progress variable（FPV）model is formulated to treat the non-ideal effects in hydrothermal flames and then the model performance is comprehensively eval-uated through comparisons with the detailed chemistry approach.It is found that the impact of real-fluid models is quite similar in both the detailed chemistry（DC）and flamelet simulations.The real-fluid model predicts an abrupt change in density and constant pressure heat capacity in the fuel side in both the DC and FPV approaches.The ideal-gas model results in a higher the maximum flame temperature and higher mass fractions of CO and H₂.Then the performance of the real-fluid FPV model is examined in the following via the a priori and a posteriori tests under atmospheric and supercritical conditions with different strain rates.It is found thermal properties and species mass fractions predicted by the FPV model are in overall good agreement with detailed chemistry simulations,which confirms the accuracy of the real-fluid FPV model.At last,the real-fluid flamelet/progress variable model is extended to the framework of large-eddy simulation.To account for the heat loss effects,the FPV model is extended to be non-adiabatic with the heat release damping（HRD）approach.Combined with a presumed prob-ability density function closure,the model is then evaluated in laboratory-scale hydrothermal flames in the context of large-eddy simulation.Results show that the prediction results from the FPV model is better than those from the EDM and EDC models,since it correctly predicts the flame lift-off height as well as the maximum temperature position.The use of the non-adiabatic FPV approach improve considerably the predictions of the flame temperature by comparing with the experimental data.Besides,in order to reduce the memory requirement and speed up the computation of extraction process,this work utilizes artificial neural networks（ANN）to build the library for the flamelet/progress variable（FPV）model and develops the FPV-ANN approach.Results show that the use of ANN can produce a significant reduction in computer memory consumption during the large-scale parallel simulation.Also,it is found that the FPV-ANN approach is 30%faster than the classical FPV approach,which confirms that FPV-ANN approach has better computational performance.