During the start-up and shutdown sequence of rocket engines, immense forces are generated that are required to propel the payload into earth's orbit or outer space. The bulk of the generated forces are in the form of thrust. However, forces are also generated that act perpendicularly to the thrust direction. These side-loads are unwanted byproducts of the complex flow features present in the nozzle and as such, research has been conducted in the past to address this issue and discover their origin. Conventionally, research into side-loads has been conducted using subscale nozzles that don't deform as the flow passes through them. Due to the rigidity of these models, the nozzles don't deform to any degree that would cause it to interact with the flow. This fluid-structure interaction, present in full-scale rocket nozzles due to their sheer size, is therefore not taken into account when testing on subscale nozzles. To this end, a rocket nozzle testing facility was designed and built to investigate the fluid-structure interaction present in rocket nozzles using flexible subscale rocket nozzles. Due to the laborious nature of designing and manufacturing the testing facility, no working subscale flexible nozzles could be manufactured within the given time.

To continue building upon this previous work, subscale rocket nozzles were poured using various types of polyurethane. To cast these nozzles, multiple mould parts were manufactured. To guarantee a smooth surface finish for the inner nozzle wall, an inner mould was CNC machined. To enable the casting of multiple nozzles with different wall thicknesses, outer moulds were 3D-printed. After having determined the correct variant of polyurethane, Smooth-On's 'Smooth-Cast 66D', three different nozzles were cast on which tests were conducted. The nominal nozzle was designed with a wall thickness of 2mm, with two additional nozzles being cast with a wall thickness of 1.5mm and 2.5mm.

Having manufactured the nozzles, they were mounted to the test facility. The test facility consists of a stand connected to an air tank capable of containing air at pressures of up to 40 bar. Due to pressure losses throughout the valves, the maximum attainable nozzle pressure ratio, or NPR, amounted to slightly over 30. Various tests were performed during which the NPR was either altered during the test, done by manually opening or closing the control valve contained in the system. Further tests were conducted at a number of constant values of NPR at which excessive vibrations occurred to gain further insight into the flow phenomena present at these critical values of NPR. Lip tracking was done as well as schlieren photography. For the lip tracking, correction fluid was applied to the nozzle lip as well as spray painting the outer wall of the nozzle black. The area surrounding the nozzle was made dark such that the predominant feature photographed would be the circular nozzle lip. Once the images were taken, the nozzle lip was discretised in 180 points and was compared to an image of the nozzle taken when the wind tunnel was closed. The end result is a discretised function of 180 points displaying the lip deflection at each azimuthal location. With the help of the discrete Fourier transform, the dominant eigenmodes could be extracted from the nozzle vibrations in the form of their Fourier coefficients. In addition to lip tracking, schlieren images were taken simultaneously with the aim of visualising the flow downstream of the nozzle exit. The brightness of a set of pixels was tracked and their spectral make-up was determined.

Along the entire scale of NPR, two regions were found in which most of the vibrations occurred. The first region of activity, was purported to be linked to the transition of separation regime from free shock separation, FSS, to restricted shock separation, RSS. The second region of NPR, in which the deflection was often even more excessive than the first, is highly likely due to the presence of a recirculation bubble passing over the nozzle lip. Out of the first three eigenmodes, named the breathing, bending and ovalisation mode respectively, the ovalisation mode accounted for the largest proportion of the deflection by far. Based on these transient tests, further tests were conducted at a constant NPR of NPR = 21.7 and NPR = 27.7. These were chosen as all three nozzles showed large deflections at these NPR. For the nominal 2mm nozzle, two additional values were chosen of NPR = 22.7 and NPR = 27.5 to conduct further tests at. The energy present in each of the first three eigenmodes was determined by calculating the variance of the Fourier coefficients. The energy in the modes during the constant NPR runs were compared to the energy in the modes during the transient tests. These were obtained by determining the energy of an NPR-envelop spanning 1 unit of NPR centered around the NPR at which the constant run was performed. By and large, the energy storage in the first three modes remained similar when comparing transient tests with constant NPR runs. The bending mode showed the largest deviation from this trend, mainly when in RSS. Frequency analyses showed the nozzle to oscillate at rates predominantly within 130Hz-400Hz. Wavelet power spectra showed no clear trend between the NPR and the frequency with which the nozzles vibrated. Strouhal calculations further confirmed that NPR would not have a drastic effect on the unsteadiness frequency of the flow. The testing helped provide insight into the mechanical vibrations of nozzles of various wall thicknesses in different flow conditions. However, due to the absence of quantitative flow visualisation and measured flow unsteadiness frequency, no concrete link could be made to the aeroelastic coupling and the fluid's role in the forcing of the vibrations.