Wind tunnel experiments were conducted at incoming Mach numbers ([ ]) of 2.01 and 2.81 to investigate the interactions between shock waves and weak, moderate, and strong streamwise vortices in a concave channel representing the inward-turning inlet/isolator flow of hypersonic engines. Synchronized high-speed schlieren visualizations and high-frequency wall static pressure measurements were collected to evaluate the steady-state and dynamic characteristics of the global and local flowfields. The interference of streamwise vortices bifurcated the oblique shock wave; the stronger the former, the more severe the deformation of the latter with increasing intensity during downstream development. When [ ], moderate and strong streamwise vortices caused the Mach stem in the shock train leading edge (STLE) to bend upstream; when [ ], moderate and strong streamwise vortices caused the top- or bottom-wall shock waves in the STLE to bend and bifurcate into multiple shock waves. An increase in streamwise vortex intensity weakened the self-excited oscillations of the shock train and enhanced the pressure fluctuations caused by STLE oscillations. The dominant frequency of the interacting flowfield oscillation was 20–40 Hz, and the dominant wave structure in the shock train oscillated rigidly. These results can inform hypersonic engine inlet development.
The flame development prediction of a scramjet combustor forecasts the combustion state and provides valuable information for active flow control. Experiments were performed on a hydrogen-fueled scramjet at different equivalence ratios in a ground pulse combustion wind tunnel with a Mach-2.5 incoming flow. Five image datasets of the flame evolution process were constructed at different predicted periods. The memory fusion cascade network (MFCN) was developed to predict flame images after a certain span using flame image sequences of the previous periods. A complete evaluation system was constructed to compare and analyze the performances of MFCN, Kongs, and ResNet16 models in multi- and long-span conditions. Experimental results show that MFCN achieves a maximum increase of 46.16% of the peak signal-to-noise ratio index, 69.14% of the structural correlation coefficient index, and 5.72% of the correlation coefficient index in the independent test set. Moreover, the volume of the model only reaches the KB level, which has the characteristics of being lightweight. MFCN outperforms other methods in terms of the prediction accuracy and maintains stable prediction results during multi- and long-span tasks.
The prediction of unstart boundaries of hypersonic inlets is significant to the engine control. To correct the previous conclusions based on the one-dimensional and inviscid flow assumption, the present paper numerically investigates the unstart characteristics of two-dimensional contraction ducts in viscous flow that vary with multiple controlling variables. Based on the newly defined dimensionless cowl length and effective internal contraction ratio, the similarity law of unstart is discovered. The universal empirical equation of unstart boundary for different cowl lengths, cowl lip heights, entrance boundary-layer thicknesses, unit Reynolds numbers, and Mach numbers is proposed, which can simplify the design of two-dimensional hypersonic inlets. With varying the dimensionless cowl length, three different types of short-cowl, transitional, and long-cowl unstarts are identified, owing to the different evolution mechanisms of separation region. The unstart displays a separation region growing under the streamwise pressure difference enhanced by the shock system or flow choke. In addition, the wall pressure criteria for unstart are correlated to the characteristic separation scale at unstart critical state and the dimensionless cowl length, which facilitates the unstart detection or early warning. It is observed that the Kantrowitz limit is close to the unstart boundaries for very short cowl and the isentropic limit is close to the unstart boundaries for very long cowl.
Understanding the hypersonic inlet starting characteristics is the prerequisite for avoiding the abnormal unstart state. To make the work close to the actual situation, an experimental study was performed on a scramjet model at a simulated freestream Mach number of 6.0 with pressure and thrust measurements. The inlet working status is determined by the heat release of the injected ethylene with reciprocating variations. The results show that the critical equivalence ratio of the restart state is lower than that of the unstart state, which means that the combustion weakens the inlet restart capability and raises the unstart/restart hysteresis phenomena. Specifically, two novel unstart/restart hysteresis phenomena are found: one may come from the dual-solution characteristics of the shock–combustion interaction and the other may come from the historical effect of reverse flow. Compared to the former type, the latter type requires greater downstream heat release and generates a larger hysteresis loop. In addition, the engine thrust characteristics of the whole unstart and restart processes are analyzed. The thrust increment in the shock–combustion interaction type exhibits nearly linearly. However, the thrust increment meets abrupt changes and strong oscillations in the reverse flow type, accompanied by the reverse flow's formation and disappearance, making the engine more difficult to restart.
Shock-train transitions in simplified curved isolators are carefully studied by simulation. The results show the shock-train behavior is subject to the complex pressure field created by the duct deflection, eventually presenting five modes during a backpressure-varying process. Of them, the most special one is the abrupt shock-train leap. It appears as the leading shocks interact with an adverse pressure gradient and follows a different path after a reversal of the direction the backpressure takes, which causes a shock-train hysteresis. If the curvature increases, the leap phenomenon, together with the related hysteresis, grows in number and intensity. Analysis indicates the background pressure gradients stem from the inherent left-running expansion waves and right-running compression waves. They control alternately the near-wall flow state, provoking the cyclic changes in the pressure gradient sign. Unlike the former, the latter can enhance separation through a positive feedback mechanism, rendering the shock train highly sensitive to backpressure. This is why the leap occurs. Comparing with the previously reported shock-induced leap indicates that there is a marked similarity in their behaviors, suggesting the irrelevance of the occurrence of a leap to the category of incident waves. Nevertheless, a delay in the onset usually follows a compression-wave-typed leap, which reflects that there is a triggering threshold for an incident wave. Given the fact that no local separation is provoked by the compression waves, it is speculated that the threshold should lie below the criterion for causing a separation, as opposed to the impression from the previous research.