ICLASS-2006 Aug.27-Sept.1, 2006, Kyoto, Japan 1. INTRODUCTION The emissions and operability of liquid fired gas turbine combustion systems are strongly affected by the fuel preparation process. In particular, the atomization and dispersion of the liquid droplets and subsequent vaporization and mixing of vapor produced with oxidant all play a role in the resulting combustion behavior. Conceptually, these processes each contribute to the overall time scale within which combustion occurs. In general, each of these processes is complex and presents a significant challenge to designers of these systems. In many liquid fired gas turbine applications, the focus is on attaining performance at full power. In these cases, temperatures and pressures generally result in a very high rate of vaporization, to the point where the atomization and vaporization steps may be “neglected”. However, for modern engines that need very high specific heat release, the vaporization step may well be the rate limiting step. As a result, efficient, accurate models that can describe the vaporization process in these applications are needed. Further complicating the problem is that the fuels used in these applications are comprised of hundreds of components. As a result, the vaporization process must account for this “distillation” process when using practical fuels. The simplest models basically indicate that the vaporization rate is directly related to the surface area of the liquid [1]. This is only true, however, for steady state vaporization which is not the case in many combustion applications. Depending on the conditions, the time for the liquid to heat up may approach the total vaporization time which creates a non-linear relation between surface area and vaporization rate. Early models attempted to capture this distillation effect using an “effective vaporization constant” concept [2] which was a major step in sophistication for design applications. More sophisticated models include the “Rapid Mixing Model” which basically assumes that mixing within the liquid occurs instantaneously; therefore the temperature is constant throughout the evaporation. However, rapid diffusion may not be valid for distillate fuels at high pressures [3]. More sophisticated physics are incorporated into so called “Diffusion Limit” (DL) model [4] [5]. This model becomes very inefficient computationally as more and more fuel components are added. As a result, application to practical fuels may be limited. More recent efforts have attempted to capture the multi-component behavior using a “Distillation Curve Model” which uses the mean molar weight of the liquid as a correlating parameter for vaporization rate [6]. In this model, for example, Jet-A can be represented by 3 pure alkanes. Compared to DL model, DC model has the efficiency of single component model with the possibility to calculate practical engine fuels more accurately. The DC model is based on the uniform temperature model and incorporates the distillation curve of practical engine fuels like Jet-A, JP-4 and DF-2 to determine the fuel vapor molar weight. The fractional boiling process is accounted for as a function of only one variable: the actual mean molar weight of the fuel inside the droplet. Thus the multi-component fuel evaporation is Paper ID ICLASS06-184 EXPERIMENTAL VALIDATION OF A DROPLET EVAPORATION MODEL Sosuke Nakamura 1 , Qing Wang 2 , Vince McDonell 3 and Scott Samuelsen 4 1 Visiting Scientist, UCI Combustion Laboratory, University of Calfornia, Irvine, sosuke_nakamura@mhi.co.jp 2 Graduate Student, UCI Combustion Laboratory, University of Calfornia, Irvine, qw@ucicl.uci.edu 3 Associate Director, UCI Combustion Laboratory, University of Calfornia, Irvine, mcdonell@ucicl.uci.edu 4 Director, UCI Combustion Laboratory, University of Calfornia, Irvine, gss@uci.edu ABSTRACT The pollutant emissions generated by liquid-fuel fired gas turbine engines are strongly influenced by the fuel preparation process that includes atomization, evaporation and mixing. In order to accurately predict the fuel preparation process, sufficiently precise models of the key thermophysical processes are crucial. In the present paper, the performance of fuel droplet evaporation models are considered as applied to a spray produced by a practical gas turbine fuel injector under actual conditions. Of particular interest are numerically efficient models including the “Distillation Curve” (DC) model. The DC model can account for the behavior of multi-component fuels like diesel fuel #2. Fractional boiling is described by the molar weight as a single process variable. This way, the fractional distillation process during evaporation of droplets is taken into account. In addition, the thermophysical properties of the fuel are supplied as a function of the molar weight. Real gas effects are also taken into account in order to improve accuracy at elevated pressures. A major advantage of DC model is that algebraic expressions are derived for the multi-component droplet vaporization. Thus, the DC model combines both numerical efficiency and accuracy. In the present study, experiments were conducted using a co-axial plain jet airblast atomizer at engine conditions used in a commercial 30kW gas turbine engine operated on DF-2 (diesel fuel #2). Measurements including visualization, and droplet size and velocity are obtained from which vaporization rates are inferred. These rates are compared to rates predicted using the DC model and effective vaporization models. The predictions compare well with the experimental results, but also illustrate how critical the initial droplet size is to the calculation. The results also show that both modelling approaches provide a reasonable estimate of the vaporization rate for the spray studied in the present work. Keywords: Vaporization, Distillate Fuel, High Pressure, Phase Doppler