Fuel injection and mixing into air play a crucial role in the operation of hypersonic airbreathing propulsion systems, particularly scramjet engines featuring upstream fuel injection. This study applies an advanced design methodology combining computational fluid dynamics and evolutionary algorithms assisted by surrogate modeling to a multi-objective optimization for fuel injection in a Mach 5.7 crossflow after the initial compression in a scramjet intake operating at Mach 7.6. Optimization is performed for elliptical injector configurations defined by four design parameters (i.e., the injection angle, spanwise spacing, aspect ratio, and radius of the injector), simultaneously aiming to maximize three objectives, that is, fuel/air mixing, total pressure saving, and fuel penetration into the crossflow. Statistical methods based on global sensitivity analysis are employed to assess the optimization results in conjunction with surrogate models to identify key design factors with respect to the three design objectives and additional performance measures. Major effects of the injection angle and aspect ratio have been observed on all considered design criteria. The spanwise spacing has been found to have considerable influence on the total pressure recovery, fuel penetration, and lateral spread when the injection pressure is adjusted to maintain a constant fuel/air equivalence ratio. Low-angle fuel injection through a highly elliptic orifice with wide spanwise spacing demonstrated the most comprehensive advantages in overall aspects.
NomenclatureA j = injector area on floor, m 2 A n j = injector area normal to jet, m 2 AR = injector aspect ratio a j = semimajor axis of injector, m c = mass fraction c s = stoichiometric mass fraction c H 2 = hydrogen mass fraction c O 2 = oxygen mass fraction D = effective injector diameter, m f = vector of objective functions H = vertical height of computational domain, m h f = fuel penetration height (normalized) J = jet-to-freestream momentum flux ratio L = streamwise length of computational domain, m k = number of clustering L 0 = distance to upstream end of injector, m _ m = mass flow rate, kg∕s N = size of population pool p = static pressure, Pa p eb = effective backpressure (normalized) p j = jet static pressure (normalized) p 02 = pitot pressure behind normal shock, Pa p 0 = total pressure, Pa r j = effective injector radius, m S i = first-order sensitivity index S T i = total-effect sensitivity index U = freestream velocity, m∕s u = velocity, m∕s W = injector spacing (spanwise domain width), m w f = fuel lateral spread (normalized) X i = input vector for decision variable x i X = input matrices comprising vectors X i x = streamwise coordinate, m x = vector of decision variables Y = input vector for output parameter y y = spanwise coordinate, m z = vertical coordinate, m α j = injection angle, deg Γ= streamwise circulation (normalized) Δp 0 = total pressure loss (normalized) Δt = time step, s Δx = spatial step, m Δμ χ = mesh sensitivity error of parameter χ η m = mixing efficie...