GTP-20-1719 Experimental investigations of superheated and supercritical injections of liquid fuels

Single- and multi-component liquid fuels are injected in a jet-in-coﬂow conﬁguration at elevated temperatures and pressures with both a custom plain oriﬁce nozzle and a commercial pressure-swirl atomizer. The transitions in spray morphology from mechanical breakup to superheated/supercritical regimes are characterized qualitatively by laser shadowgraphy and evaluated based on quantitative measures of superheat. Although fuel preheating exhibits no discernible effect in the mechanical breakup regime, dramatic jet-to-plume transition as well as build up of fuel vapor in the spray chamber are observed with increasing level of superheat. The difference between two different atomizers in terms of spray behavior diminishes at high levels of superheat, suggesting the predominant role of thermal effect on spray morphology in superheated/supercritical regimes. For a mutli-component fuel such as Jet A-1, the transition into a fully ﬂashing spray occurs at temperatures lower than expected values, which are calculated by treating Jet A-1 as a single-component fuel. Additionally, pressure drop is shown as a sensitive indicator for the departure from mechanical breakup and the onset of thermal effect on the spray. Comparisons between measured and estimated pressure drop also reveal the differences in susceptibility to thermal effects between the plain oriﬁce and the pressure-swirl atomizers. leveraged. plays in spray morphology in this regime. Detailed comparisons of measured and estimated ∆ P F show a different trend with respect to superheat compared to the case with the plain oriﬁce nozzle, demonstrating the necessity of investigating atomizers with complex geometries under superheated and supercritical conditions. These results also demonstrate that fuel spray undergoes notable changes in morphology even at fairly low superheat levels, which could affect the combustion process especially in cases where the fuel


INTRODUCTION
The commercial aviation sector is expected to continue its robust growth with doubled passenger numbers in the next two decades [1]. In order to meet the ambitious goals outlined in the Flightpath 2050 by the Advisory Council for Aeronautic Research in Europe (ACARE) to curtail pollutant emission [2], disruptive combustion technologies are needed to ensure both economic and environmental sustainability.
As the operating temperature and pressure of aero engines continue to be raised for improved overall efficiency and reduced CO 2 emission, it becomes increasingly likely that liquid fuels will be injected into the combustion chamber under superheated or supercritical conditions. Superheated (or flashing) atomization occurs when a liquid is discharged into a gaseous ambient at a pressure lower than its saturation pressure. As the fuel temperature and pressure prior to injection approach its supercritical limits, the liquid fuel behaves more and more like its gaseous counterpart. Under these conditions, fuel sprays can attain finer droplets, wider opening angles [3] and smaller penetration depths [4] comparing to injections at normal conditions [5,6,7]. These advantages, when fully leveraged, could enable novel strategies for achieving compact flames and low pollutant emission [8,9] that are not accessible through conventional means of combustor modification. Although many applications of superheated/supercritical injections can be found in wide-ranging industries, relatively few work exists with regard to gas turbine combustion that focuses on multi-component fuels and atomizers with complex geometries.
Under the framework of the EU2020 Soot Processes and Radiation in Aeronautical Innovative Combustion (SOPRANO) project, this work aims at providing experimental data for the improvement of numerical models to account for thermal effects on fuel injection, which may not be negligible in certain combustion systems that do not incorporate active cooling on fuel delivery line and/or nozzles. A major effort of this work is to establish a database to allow quantitative and ambient). Fuel sprays from both a custom plain orifice nozzle and a commercial pressureswirl nozzle are characterized with laser shadowgraphy. The energy barrier for bubble nucleation demonstrated in Ref. [6] is adopted here alongside classical measures of superheat to compare and analyze the results.

Modified HiPOT facility
In order to examine non-reacting injection of liquid fuels under elevated temperature and pressure conditions, the High Pressure Test (HiPOT) facility at DLR was modified to incorporate several newly constructed modules, as schematically shown in Fig. 1a (with a list of major components given in the caption). With the standard configuration (grayed out parts), HiPOT is capable of ©2021 by ASME. This manuscript version is made available under the CC-BY 4.0 license http://creativecommons.org/licenses/by/4.0/ achieving up to 30 bar inside the optical pressure vessel (No.2) and up to 700 K of main air temperature at a maximum flow rate of 400 g/s. In this work, a fuel preheater with a maximum capacity of 3.6 kW (No.3) was introduced to allow raising the liquid fuel temperature (T F ) up to 650 K at a maximum design flow rate ofṁ F =2 g/s. After injection, the liquid fuel was subsequently consumed inside an afterburner (see No.7 in Fig. 1b), which was designed following the FLOX concept [10] with a ring of 12 nozzles and separate fuel (methane) and air plena. The afterburner was ignited by a gas pilot (ZMI, Kromschroeder) that was mounted on the afterburner combustion chamber (No.1) and operated with non-premixed methane and air, generating a pilot flame of the length of 2 to 10 mm. The stability of the afterburner was monitored via a small window on the combustion chamber.
Fuel injection was implemented in a jet-in-coflow configuration inside the optical pressure vessel (No.2), as illustrated in Fig. 1b. A liquid fuel injector with a replaceable nozzle was situated at the center of a square housing (No.5), which was furnished with a sintered plate that homogenizes the main air and creates a stable coflow. Fuel spray was injected into an optical spray chamber (No.6) with a clear aperture of about 80 mm by 120 mm on four sides, before being throttled by a converging nozzle into the afterburner (No.7). A custom-designed plain orifice atomizer was installed at the tip of the fuel injector for most of the experiments presented in this paper. It had a nozzle diameter of d 0 =200 µm with a length/diameter ratio of l 0 /d 0 =10. In addition, sprays generated by a commercial pressure-swirl atomizer (Delavan, FN=1.1), often used in aerospace applications, under selected operating conditions are presented as comparisons. Fuel temperature (T F ) prior to injection was monitored via a thermocouple (T 1 ) 55 mm upstream of the nozzle exit (closest point possible due to limited accessibility). Air coflow temperature was monitored by another thermocouple (T 2 ) at the exit plane of the sinter plate. Fuel pressure (P F ) could not be measured directly in the injector due to high fuel temperature and was determined instead upstream of the fuel preheater. Tests were conducted to ensure negligible pressure loss between the measurement location and close to the nozzle exit. Additionally, readings from other sensors built into HiPOT such as chamber pressure, flow rates and material temperatures were logged every second during the measurement for cross referencing purposes.

Operating conditions
A total of five single-and multi-component fuels were examined during this work, including cyclohexane, iso-octane, n-nonane, Jet A-1 and heating oil extra light (HEL). Their critical temperatures (T c ) and pressures (P c ) were listed in Table 1. Fuel flow rates were maintained either at 0.5 g/s or 1 g/s, resulting in an exit velocity of up to approximately 100 m/s (depending on the fuel type and preheating temperature). Three different chamber pressures P ∞ were used for each fuel: 1.5, 3 and 6 bars. Due to safety concerns, fuel temperature inside the preheater was kept lower than the corresponding saturation temperature at a given fuel pressure. The hot coflow was then utilized as a second tier fuel preheater inside the main air plenum (No.4 in Fig. 1a) to further increase T F to be closer or larger than T c . The coflow temperature (measured at T 2 , see velocity was maintained at 2-4 m/s (depending on the pressure and temperature), such that it was significantly slower than the fuel sprays and was not expected to influence the spray behaviors.
The afterburner was operated at λ ∼1.3 and a thermal power of up to 180 kW (at 6 bars). The small amount of liquid fuel was not seen to influence the stability of the afterburner throughout the measurements.

RESULTS & DISCUSSIONS
In the following sections, only results from cyclohexane and Jet A-1 are shown, since the same trends were observed in other tested fuels. In order to compare the fuel sprays at various chamber pressures P ∞ , the classical definition of the level of superheat ∆ T [11]: was used to classify the cases. T F measured at T 1 (see Fig. 1b)) was taken as the injection temperature T inj . The saturation temperature at the back pressure T sat (P ∞ ) was inferred from fuel properties taken from [12,13] at measured chamber pressure P ∞ . Single-shot shadowgraphs of cyclohexane injected at P ∞ =1.5 bar (blue dashed line on the P-T chart),ṁ F =1 g/s and various initial fuel temperature T F and pressure P F (red dots on the P-T chart), using a plain orifice nozzle. Each shot is 10 mm by 25.5 mm in physical scale.  Figure 2 shows the nine cases at P ∞ =1.5 bar. As T F was increased, higher P F was needed to sustain the same fixed mass flow rate. As can be seen, the chosen cases span a wide range of ∆ T : from no superheat in Cases 1 and 2 (i.e., ∆ T ≤0) up to about 170 K in Case 9, which still lies under T c at this P ∞ . Before T F crosses T sat , the fuel spray appears as a jet with a negligibly small spreading angle, as expected with this type of atomizer at conditions controlled by mechanical breakup. Preheating within this regime has no discernible effects on the spray. As soon as T F crosses T sat , i.e., injection in the superheated regime, the spray appears to contain a downward skewed main jet with fine particles seeming to "peel off" from it, as shown in Case 3.

Superheated and supercritical injection of cyclohexane with a plain orifice nozzle
This downward skew is likely the result of gravity, as the jet-in-coflow configuration is horizontally laid out and the fuel is injected from left to right in this view. As T F inches higher, the main jet diminishes while more fine particles could be seen building up alongside and above the main jet.
The effective spray angle is therefore significantly broadened from Case 3 to Case 4. Further preheating leads to the formation of a plume of dense particles from Case 5 to Case 6, resembling the fully flashing sprays reported in the literature [6]. From this point on, additional superheat results in the shrinkage of the plume, i.e., a reduction of fuel penetration into the spray chamber.
Fuel vapor can also be identified as bright and small structures trailing the plume. Figure 3 shows the nine cases with matching levels of superheat at P ∞ =3 bar. As can be seen, increasing ∆ T causes the fuel spray to go through a fairly similar jet-to-plume transition observed at P ∞ =1.5 bar in Figure 2. Although Cases 1 and 2 (mechanical breakup) do not seem to differ much from those at lower chamber pressure, the plume appears to attain a noticeably smaller spread angle. In addition, Case 9 with a ∆ T close to 170 K is already in the supercritical regime at this P ∞ . Instead of abrupt changes in spray behaviors as seen from mechanical breakup to superheated injection (Cases 1&2 to , supercritical injection appears to continue the trend of a shrinking plume of particles and increased presence of fuel vapor in the field of view. These general trends hold for P ∞ =6 bar, as displayed in Fig. 4. Comparing to the lower pressure cases, the plume formed in superheated conditions (Cases 3-6) appears much narrower with respect to its spread angle. Besides, it also seems that the transition from a "diverging" jet form to the plume form (Cases 3-5) occurs much swifter than at lower pressures. The transient behavior of fine particles "peeling off" from the main jet is not present at this chamber pressure. Moreover, when compared to Figs 2 and 3, at comparable levels of superheat throughout the superheated and supercritical regimes, the penetration depth of the fuel jet (plume) reduces starkly with increased P ∞ . Particularly at Case 9 at P ∞ =6 bar, the plume retreats into a small spherical shape attached to the nozzle exit while the majority of the liquid fuel appears to be vaporized.
In order to provide insights into the observed evolution of spray form from Cases 1 to 9 at various chamber pressures, additional measures of superheat, R p and χ, proposed in Ref. [6] were derived for the nine cases using their corresponding operating conditions. These two parameters are defined as: with P sat (T inj ) representing the saturation pressure at the injection temperature, χ the energy barrier to nucleation and Θ the dimensionless surface tension following Ref. [14]. It is demonstrated that the onset of fully flashing can be correlated solely to the condition of χ=O(1) and is independent of the Weber number. Moreover, the threshold R p needed for the onset decreases with increasing back pressure P ∞ . Figure 5a summarizes the measured liquid fuel and air coflow temperatures T F (colored symbols) and T A (colored dashed lines) for Cases 1-9 at the three pressures shown in Figs 2 to 4. As described in the experimental section, they were measured using thermocouples at T 1 and T 2 in the injector housing (see Fig. 1b). Since the cases were selected based on the similar levels of superheat ∆ T , the absolute T F increases with increasing back pressure (due to the increase in T sat (P ∞ ) in Eq.1). Since the coflow was also used to preheat the liquid fuel, T A was higher than T F . The difference was likely smaller at closer to the nozzle exit since T 1 was located about 55 mm upstream of T 2 . Notice also that the temperature difference becomes smaller at higher pressures due to a generally slower coflow velocity at similar air mass flow rates (a certain minimum amount was necessary for the operation of the air preheater) and hence a better heat exchange between the coflow and the fuel line. where the transition from jet to plume becomes noticeably abrupt at P ∞ =6 bar, suggesting that bubble nucleation is likely the dominant process for controlling spray morphology responding to thermal effects as found in Ref. [6]. Note that the observed reduction in spray spreading angle with increasing pressure is also consistent with the observations in Ref. [6].   of Jet A-1 provided in Ref. [12], which was used to treat Jet A-1 as a single-component fuel in numerical simulations [15]. As can be seen, despite having matching levels of superheat, the spray appears to enter the Category II described above at Case 2, before superheated condition is achieved, such that Cases 2-4 for Jet A-1 seem to resemble Cases 3-5 for cyclohexane (see

Estimation of pressure drop
The influence of superheat on fuel injection was also inspected by comparing the measured pressure drop ∆P F across the nozzle at various T F to estimated values based on monitored operating conditions and the discharge coefficient of a plain-orifice nozzle [17]: where A 0 and ρ F are the cross sectional area of the nozzle and fuel density, respectively. For the plain orifice nozzle, the discharge coefficient C D was calculated based on the empirical formula from Ref. [18]: ©2021 by ASME. This manuscript version is made available under the CC-BY 4.0 license http://creativecommons.org/licenses/by/4.0/ Fig. 10. Single-shot shadowgraphs of cyclohexane injected at P ∞ =6 bar (blue dashed line on the P-T chart),ṁ F =1 g/s and various initial fuel temperature T F and pressure P F (red dots on the P-T chart), using a pressure-swirl nozzle. Each shot is 10 mm by 25.5 mm in physical scale.
where C Dmax = 0.827 − 0.0085l 0 /d 0 . In order to minimize the role of cavitation, only the cases withṁ F =0.5 g/s were considered. The results are plotted in Fig. 9, with the cases included in the shadowgraph collages in Figs. 6 and 7 highlighted with larger symbols (matching those used in Fig.8). The corresponding T sat at P ∞ =6 bar is also indicated in the plot for reference (blue vertical dashed line). Note that the constant 0.827 in C Dmax was slightly inflated to 0.9 for Jet A-1 to better match the measured pressure drop at low temperature conditions.
As can be seen for cyclohexane in Fig. 9a, measured pressure drop only starts to deviate from mechanical breakup across T sat (i.e., in Category II defined in Fig. 5). This suggests that measured ∆P F could also serve as a sensitive indicator for monitoring the effect of superheat and the departure from a classical regime dominated by mechanical breakup. This is particularly the case for Jet A-1 shown in Fig. 9b, where the measured ∆P F outpaces the theoretical value about 70 K lower than the theoretical T sat (considering Jet A-1 as a single-component fuel). Notably, Case 2 is seen here to no longer behave as a mechanical breakup, consistent with the observations from the shadowgraph shown in Fig. 7.

Thermal effect on a pressure-swirl nozzle
Since the ultimate goal of this work is to examine the impact of superheat on fuel injection in aero engine combustors, laser shadowgraphy was also applied to fuel injections facilitated by a ©2021 by ASME. This manuscript version is made available under the CC-BY 4.0 license http://creativecommons.org/licenses/by/4.0/ Fig. 11. Comparisons of measured and calculated pressure drop over a pressure-swirl nozzle for cyclohexane at P ∞ =6 bar, m F =1 g/s and various initial fuel temperature T F . The cases included in Fig. 10 are highlighted with larger symbols. T sat is indicated by the blue vertical dash line.
Delavan pressure-swirl nozzle (described in the experimental section above).
In a similar manner to those shadowgraphs carried out with the plain orifice nozzle presented above, shadowgraphs with the Delavan nozzles were selected based on the operating conditions to match the levels of superheat for injection of cyclohexane at P ∞ =6 bar,ṁ F =1 g/s (see Fig. 4).
The results are shown in Fig. 10. Since the transition from Case 3 to 4 appeared too abrupt, an additional case at ∆ T =44 was added in the collage. Case 8 was then removed since it exhibited similar form as Cases 7 and 9. As can be seen, at no superheat (Cases 1 and 2), the fuel spray enters the chamber with a spreading angle of approximately 90°(expected of this nozzle). As in the case of the plain orifice nozzle, preheating below T sat does not seem to have any impact on the spray morphology. However, as soon as the superheated regime is entered, the spray angle starts to shrink drastically (between Cases 3 and 4). At the same time, the penetration depth of the spray reduces with increasing amount of fuel vapor present in the chamber. From Case 4 onward, it is almost indistinguishable from the behavior of the sprays generated from the plain orifice nozzle shown in Fig. 4: the plume of fine particles retreats towards the nozzle with increasing superheat. The results seem to suggest that, once superheat becomes the predominant mechanism in controlling the spray morphology, the differences in mechanical breakup mechanisms (e.g., plain orifice vs. pressure-swirl) become insignificant. This observation may prove crucial when designing injection systems for future aero engines where the advantages of superheated and supercritical injections were to be leveraged.
©2021 by ASME. This manuscript version is made available under the CC-BY 4.0 license http://creativecommons.org/licenses/by/4.0/ As pointed out above, the transition from a cone spray to a plume in the case of the Delavan nozzle occurs later than the jet-to-plume transition with the plain orifice nozzle, which is made obvious by comparing Fig. 10 with Fig. 4, specifically from Case 3 to Case 4. In order to gain more insights into this difference, pressure drop ∆P F over the Delavan nozzle was also estimated based on the manufacturer specifications (d 0 =0.35 mm and C D =0.45) and Eq. 4 and compared with the measured values, as plotted in Fig. 11. The cases used for the shadowgraph collage in Fig. 10 are highlighted with larger symbols and labeled correspondingly. It appears that up until Case 3, ∆P F behaves as expected without any thermal effect. From then on, the measured ∆P F starts to deviate strongly from the estimated values, which corroborates well with the observations in

CONCLUSIONS
In this work, a jet-in-coflow configuration was implemented at the High Pressure Optical Test Examinations conducted with the pressure-swirl atomizer demonstrates that as T F enters the superheated regime, the spray contracts dramatically from a standard cone shape to form a plume of fine particles, which shrinks with increasing ∆ T . Up from a certain high level of superheat, the spray behavior becomes nearly indistinguishable from that observed with the plain orifice nozzle, suggesting the predominant role thermal effect plays in spray morphology in this regime. Detailed comparisons of measured and estimated ∆P F show a different trend with respect to superheat compared to the case with the plain orifice nozzle, demonstrating the necessity of investigating atomizers with complex geometries under superheated and supercritical conditions. These results also demonstrate that fuel spray undergoes notable changes in morphology even at fairly low superheat levels, which could affect the combustion process especially in cases where the fuel ©2021 by ASME. This manuscript version is made available under the CC-BY 4.0 license http://creativecommons.org/licenses/by/4.0/ injection system is not actively cooled. The comprehensive datasets obtained in this work could be used to derive spray boundary conditions for constructing numerical models that account for thermal effects on fuel injection.