|
|
Resin Transfer Molding (RTM) is an important process for manufacturing polymer composite parts. The RTM process involves injections of liquid resin into a closed mold containing fiber preform and subsequent curing.
Consistent mold filling is essential for producing defect-free RTM parts. However, this is often difficult to achieve due to variations in preform permeability, race tracking, and varying resin viscosity. These variations cause uncertainties that significantly affect the filling patterns and the Last Point to Fill (LPF) locations. One of the most serious defects is dry spot, which is a region not wetted by the resin in the final part. A common practice to eliminate the dry spot is to predict the LPF location and place an exit vent at that location. However, due to the aforementioned uncertainties, the LPF may not coincide with the preset exit vent and a dry spot may form.
The objective of this research is to develop adaptive control strategies for RTM by using numerical modeling to account for the aforementioned process uncertainties. A system model that describes the RTM filling process is developed. The system model is characterized by its non-linear time-varying parameters. Adaptive control techniques are applied to regulate the RTM filling pattern.
Various example cases of RTM are simulated with the control volume based Finite Element method. Under the control of aforementioned adaptive controllers, the LPF positions are regulated to the preset exit vent locations.
Fortran codes were developed and run under Windows.
To topA computational study of the pseudosteady-state 2-D mixed convection within rotating spherical containers is presented. the computations are based on an iterative, finite-volume numerical procedure using primitive dependent variables, whereby the time dependent continuity, momentum and energy equations in the spherical coordinate system are solved. Natural convection effect is modeled via the Boussinesq approximation. For a fixed Prandtl number of 4.62, parametric studies were performed by varying the Rayleigh number in order to cover the laminar regime adequately. For a given Rayleigh number, the ratio of Gr/Re2 was varied between 0.1 and 10. Given a Rayleigh number, the streamline patterns maintain their general shape with a dominant rotating vortex. As the forced convection effect becomes less marked, the streamlines exhibit less pronounced gradients near the surface of the sphere. As the rotational effect become more marked, the extent of deviation from the limiting case of non-rotating spheres becomes more noticed. However, the bottom of the sphere still remains to be the region with enhanced heat transfer. Given a rotational Reynolds number, the streamline patterns are not affected greatly as the natural convection is promoted, however the temperature gradients near the surface are markedly enhanced. It is noticed that as natural convection effects are promoted, the greater portion of the sphere's surface experiences enhanced heat transfer rates. Given Rayleigh number, the contours of the azimuthal velocity exhibit a nearly vertical equally-spaced pattern suggesting that solid-body rotation for high rotational Reynolds numbers. However, as the natural convection effects are enhanced, the contours become more slanted. The variation of the mean Nusselt number with the Reynolds and Rayleigh numbers is also quantified.
Code was written in Fortran, run on CRAY-XMP supercomputer under Unix.
To topConstitutive equations for hot forming process of metals are developed in terms of microstructural evolution characteristics. The Torsion test has been chosen as a mechanical test and the recorded torque-deformation curves are then used for identifying the microstructural as well as rheological parameters in the constitutive and microstructural equations proposed. The Finite Element method together with an inverse method are used to optimize the identification process. The numerical model reproduces faithfully the shape of the torque-deformation curves and a good agreement with experimental data is obtained for a constant strain rate test. The coupling with the microstructural evolution is also investigated. The simulation is conducted with Fortran under Unix.
To topA set of general purpose CFD Fortran code for 3-D fluid flow and heat transfer is developed. The flow model is built on Navier-Stokes equation and k-e turbulence model. The discretization is performed on the Control Volume based Finite Difference method. SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) method is adopted for the calculation.
The code has been applied to several industrial applications. Take two examples from the projects where I am one of the Co-PIs. One is with Shanxi Steel-Making Company. The company imported from Germany a set of 200 × 200 continuous billet caster for steel making. In order to ensure quality production and maximize productivity, more has to be known on the flow pattern in the mold of the caster, which is the primary objective of the project. Computer simulations are conducted on VAX machines to provide detailed knowledge on the velocity field in the mold. A water model is also built for the velocity field measurement with Laser Doppler Anemometry (LDA, DANTEC55X with DIS55L90a counter). Other aspects of the project includes: the effect of Strand ElectroMagnetic Stirrers (SEMS) on the flow in the liquid core and the temperature distribution in the mold of the billet caster.
Another project is with Baotou Iron-Steel and Rare Earth Company. One of steel making processes is conducted in furnaces with submerged gas injections. The hot liquid metal in the vessel is driven by the injected gas to circulate. Strong and well distributed circulation pattern favors the refining reaction inside the vessel. The project is to investigate the effects of arrangement of the submerged gas injection nozzles on the flow pattern.
To top