| 摘要: | 本研究針對汽油主要成份異辛烷(iC8H18)於近似燃氣輪機條件，量測其球狀火焰之層流和紊流燃燒速度(SL和ST)。實驗條件為當量比(equivalence ratio)1和1.2，其有效Lewis數分別為Le ≈ 1.43 > 1和Le ≈ 0.93 < 1、初始溫度T = 373 K(異辛烷沸騰溫度約99°C)、初始壓力p = 0.5~3 atm，方均根紊流擾動速度u′ = 0 ~ 4.2 m/s。實驗在已建立之高溫高壓雙腔體三維十字型燃燒設備執行，此設備設置一對反向旋轉風扇及空孔板於大水平圓柱之兩端，它可在三維十字型燃燒室中心區域產生近似等向性紊流場。實驗前需將異辛烷燃氣預先汽化，再以分壓比方式將所設定值之汽化異辛烷燃氣分壓與空氣注入十字型燃燒室並混合，以高速紋影法紀錄中央引燃往外傳播球狀火焰，以獲得其火焰半徑之時序資料R(t)，來估算其燃燒速度。實驗結果主要含四個部分：(1)整合本實驗室先前不同溫度之量測，含三個不同溫度(T = 358K、373K、423K)，來探討溫度效應對於SL和ST之影響，在、p atm下，SL ~ (T/T0)1.3，而ST ¬~ (T/T0)0.07當u′ =1.4 m/s、ST¬ ~ (T/T0)0.64當u′=2.8 m/s和ST ¬~ (T/T0)1.07當u′ = 4.2 m/s，其中 T0 = 298K，顯示增加溫度，SL和ST值均會增加。當u′ =1.4 m/s時，溫度對ST值之增加效應較不明顯，但此效應隨著u′值之增加而顯著地增加。(2)有關壓力效應對於SL和ST之影響，在、p =atm下，SL ~ (p/p0)-0.26，而ST¬ ~ (p/p0)0.15當u′ = 1.4 m/s、ST¬ ~ (p/p0)0.22當u′ = 2.8 m/s和ST¬ ~ (p/p0)0.17當u′ = 4.2 m/s，其中 p0 = 1atm，顯示SL值會隨壓力增加而下降，但ST值會隨壓力增加而上升。後者是因為壓力上升會造成運動黏滯係數(Kinematic viscosity, v)下降，使得紊流雷諾數(ReT,flow = u′ LI /v)上升，這是ST上升的主因，而非傳統上認為壓力上升會使紊焰越不穩定所主導，因為當ReT,flow固定時，ST就像SL一樣，均會隨p上升而下降，其中LI為紊流積分長度尺度。(3)我們發現Le < 1紊流火焰( = 1.2, Le ≈ 0.93)，其ST值明顯高於Le > 1紊焰(= 1.0, Le ≈ 1.43)，這是在 SL和u′ = 1.4m/s為近乎相等的情況下所得的結果，其原因為Le < 1紊流火焰會受到熱擴散不穩定性之影響而增加ST值。(4)最後，本研究分別探討四個有關S¬T之關係式(未考量Le數)：(1) ST,c=0.5/u′ = A(Da)B，其Damköhler number Da = (LI/u′)(SL/L)、下標c ̅為平均傳遞變數、L為層流火焰厚度、A和B為實驗係數；(2) Kobayashi et al.所提出之 ST,c=0.5/SL = A[(u′/SL)(p/p0)]B；(3) Chaudhuri et al.所提出之[(1/SLb)(d/dt)] = A(ReT,flame)B，其中ReT,flame= (u′/SL)(/L)、SLb為未經密度校正之生成物層流燃燒速度；(4) Bradley et al.所提出之ST/u′ = A K B，其Karlovitz number K = 0.157(u′/SL)2(ReT,flow)-0.5。將本實驗所獲得之Le > 1與Le < 1之ST值，代入前述四個ST之關系式，可以發現Le < 1之ST值均會大於Le > 1之ST值，故需作Le之修正，分別為：(1) ST,c=0.5/u′=0.084(DaLe-1)0.5；(2)ST,c=0.5/SL = 2.97[(u′/SL)(p/p0)Le-1]0.32；(3) ST,c=0.5/SL = 0.24(ReT,flameLe-1)0.5；(4)ST,c=0.5/u′ = 0.32(KLe)-0.45，將原本分散之不同Le、T、p、u′值之異辛烷ST資料，經Le數修正後，所有資料均可藕合成一曲線，顯示球狀火焰具有相似性，且四個關係式加入Le之考量後，均可成為一般通式。此研究結果，對高溫高壓預混紊流燃燒相關之應用，如車輛及航空引擎之燃燒研究有所助益。;This study measures high-pressure, high-temperature turbulent burning velocities (ST) of iso-octane/air mixtures (iC8H18 is a major component of gasoline surrogate) relevant to gas turbine conditions. Two equivalence ratios (= 1.0, 1.2) with different Lewis numbers (Le ≈ 1.43, 0.93) are used, and the ranges of experimental conditions are: The initial temperature T = 373K, the initial pressure p = 0.5~3 atm, and the r.m.s. turbulent fluctuation velocity u′ = 0~4.2 m/s. Experiments are conducted in a high-pressure, high-temperature, double-chamber, cruciform fan-stirred premixed turbulent explosion facility, capable of generating a near-isotropic turbulence for conducting combustion experiments at constant T, p, and u′ conditions. Before a run, the iso-octane is injected into a pre-vacuumed heated vessel to make sure that it is fully vaporized. Then we inject the pre-vaporized iso-octane into the cruciform fan-stirred burner by using the partial pressure method, and the pre-vaporized iso-octane and air are well-mixed. A run begins by centrally-igniting the combustible mixture. The radii of spherical expanding flames as a function of time are recorded by high-speed Schilieren imaging to estimate the corresponding burning velocities. The experimental results are mainly consisted of four parts: (1) Together with our previous data, the effect of T (358K, 373K, 423K) on SL and ST is discussed, which show that when =1.0 at 1 atm, values of SL and ST increase with increasing T, i.e. SL ~ (T/T0)1.3, ST¬ ~ (T/T0)0.07 at u′ = 1.4 m/s, ST ¬~ (T/T0)0.64 at u′ = 2.8 m/s, ST¬ ~ (T/T0)1.07 at u′ = 4.2 m/s, where T0 = 298K. At modest u′ = 1.4 m/s, the incremental effect of T is rather weak, but such effect becomes stronger and stronger as u′ increases. (2) Results show that when =1.0 at 1atm, SL ~ (p/p0)-0.26but ST¬ ~ (p/p0)0.15 when u′ = 1.4 m/s, ST¬~ (p/p0)0.22 when u′ = 2.8 m/s, ST¬ ~ (p/p0)0.17 when u′ = 4.2 m/s, where p0=1atm. The latter is because when p increases, the kinematic viscosity (v) decreases resulting in a decrease of turbulent flow Reynolds number ReT,flow = u′LI /v, where LI is the turbulent integral length scale. This ReT,flow decrease is the major reason causing the decrease of ST, not the commonly-held view that ST increases with increasing p due to the enhancement of wrinkling flame instability by the decrease of the flame thickness. When ReT,flow is kept constant, ST decreases with increasing p in the same manner as SL, showing the global response of laminar and turbulent flame speeds to pressure. (3) It is found that the effect of Le plays an important role on ST, because Le < 1 flame ( = 1.2, Le ≈ 0.93) have higher values of ST than that of Le > 1 flames ( = 1, Le ≈ 1.43), both at the same SL ≈ 0.4 m/s and u′ = 1.4 m/s. This is because Le < 1 flames experience additional thermodiffusive instability. (4) Finally, this paper applies four correlations to analyze the present ST data: (1) ST,c=0.5/u′ = A(Da)B, where the subscript c is the mean progress variable, Damköhler number Da = (LI/u′)(SL/L), where L is laminar flame thickness, and A and B are experimental coefficients; (2)A correlation of [(1/SLb)(d/dt)] = A(ReT,flame)B proposed by Chaudhuri et al., where SLb is the burned laminar burning velocity, and the turbulent flame Reynolds number ReT,flame = (u′/SL)(/L); (3)A correlation of ST,c=0.5/SL = A[(u′/SL)(p/p0)]B proposed by Kobayashi et al.; (4) A correlation of ST/u′=AKB proposed by Bradley et al., where the Karlovitz number K=0.157(u′/SL)2(ReT,flow)-0.5. In the above four correlations, Le < 1 flame have higher values of ST than Le > 1 flames, suggesting a need for Le modification. The Le modifications we propose are as follows: (1) ST,c=0.5/u′ = 0.084(DaLe-1)0.5; (2) ST,c=0.5/SL = 0.24(ReT,flameLe-1)0.5; (3) ST,c=0.5/SL = 2.97[(u′/SL)(p/p0)Le-1]0.32; (4) ST,c=0.5/u′ = 0.32(KLe)-0.45; all scattering ST data are collapsed onto single curves showing self-similar propagation turbulent of spherical flames regardless of different Le, T, p, u′. These four general correlations with the consideration of Le are important for high-pressure, high-temperature turbulent premixed combustion relevant to auto and aviation engines. |
