Reason: Access restricted by the author. A copy can be requested for private research and study by contacting your institution's library service. This copy cannot be republished
Squeeze flow during assembly of novel joints in composite aircraft components
thesis
posted on 2017-01-10, 05:51authored byBurka, Patryk
An adhesive bonding process for composite spar-to-skin structures which can be found in various aircraft components is proposed. In this process, referred to as the Insertion Squeeze Flow (ISF) bonding process, a spar is inserted into a substructure which is integrated into the composite skin. The cross-sectional shape of the substructure is similar to the Greek letter π (Pi), the roof of the π being attached to the skin, and this substructure is referred to as a Pi-slot. Before the insertion process is started, adhesive is placed into the Pi-slot bottom and due to the insertion of the spar distributes into the gaps, or flow channels, between the spar and the Pi-slot.
The adhesives that can be used for the conduction of the ISF processes were analysed in order to develop an adhesive material model that can be used to represent the adhesive in computational analysis. The adhesives Hysol EA 9395 and Hysol EA 9396 were selected to be used for the ISF bonding process. A mixing ratio by weight of 70 – 30 EA 9395 to EA 9396 was determined to have the lowest acceptable viscosity. The upper viscosity limit was determined as the viscosity of EA 9395, which is the more viscous of the two adhesives. Rheological tests showed that all studied adhesives are non-Newtonian, shear thinning fluids. Furthermore, their time dependence appeared to be small and their elasticity negligible. Constitutive material models (a Power law model and a five parameter rational model) were derived based on shear viscosity versus shear strain rate results.
In order to develop a two-dimensional (2D) numerical model for ISF using computational fluid dynamics (CFD) software, a simplified ISF process was studied first. A Newtonian fluid was specified as the fluid to be displaced by the insertion process and numerical predictions were compared to the solutions of a derived analytical model for the same problem setup, showing good agreement. To simulate the actual ISF bonding process, the material models developed for the adhesives were implemented into this numerical model. The agreement between experimental data and numerical predictions was good.
ISF bonding processes conducted at constant insertion speed were studied numerically applying the developed numerical 2D model. Insertion forces and pressures acting along the Pi-slot walls were predicted and discussed for various insertion speeds, adhesive viscosities, flow channel widths and insertion plate head designs. The main findings were a linear relationship between the insertion force and the insertion speed as well as a linear relationship between the maximum pressure along the Pi-slot walls and the insertion speed. The pressure was found to distribute approximately linearly along the Pi-slot wall, with a maximum reached at the root of the Pi-slot wall. The ratio between the insertion force and the maximum pressure was found to be independent of the insertion speed and the adhesive viscosity. The established understanding of forces and pressures during ISF supports the development of an ISF bonding process in terms of component design and in terms of bonding facility design.
The effect of lateral misalignment was studied numerically in order to ensure complete adhesive distribution during ISF. A dimensionless parameter ξ was defined referring the wide to the narrow flow channel width and its effect on the adhesive distribution evaluated. A second dimensionless parameter ψ was introduced which defines the ratio between the flow front in the narrow and the flow front in the wide flow channel. One main finding of this evaluation was that these two dimensionless parameters were found to be linearly related with each other. Furthermore, it was found that this relationship was not affected by the insertion speed, adhesive viscosity, initially applied adhesive volume and scarcely affected by the insertion plate width variation. It was, however, affected by the shape of the insertion plate head, with the rectangular head shape found to be the one most difficult to fill. Procedures were proposed to ensure entire filling of the flow channels, consequently leading to a desired Pi-joint quality, for this rectangular head shape.
Finally, the developed 2D numerical model was extended in regard to four aspects: the consideration of the insertion control (at constant insertion speed or constant insertion force), the consideration of a slight variation of the ISF process (ISF with adhesive pre-application), the involvement of a fluid-structure-interaction (FSI) and finally the consideration of ISF modelled three-dimensionally (3D). An ISF process conducted at constant insertion force control was implemented into the numerical model and predictions showed that relationships derived from constant insertion speed simulations were also valid for constant force insertions. The effect of a FSI on the adhesive showed a negative effect on the adhesive distribution compared to rigid Pi-slot walls, and two suggestions were proposed to eliminate this effect in practice. Finally, three-dimensional (3D) simulations were conducted to study the effect of a longitudinal misalignment. In the considered range the adhesive flow was scarcely affected by this misalignment.
The detailed understanding of the adhesive flow during ISF is supportive for the design of an adhesive bonding process that can be used to join spar-to-skin structures as found in aircraft components. The outcomes of the presented research work can be used as a guide to ensure the joint quality of these spar-to-skin structures.