Characterization of Laser Beam Welding
of Metal Matrix Composites

K. Mukherjee and S. Kudapa Sponsored by: REF

Introduction

In most fusion techniques, the lack of control on the heat input into the job, generates large fusion- and heat affected- zones (HAZ). Usage of filler material leads to dilution of secondary reinforcement in the weldment. Autogenuous collinear dual laser beam welding (CD-LBW) appears to be a promising candidate for fusion joining of MMCs. Bead-on-plate laser welds were produced on 6061 Al-10 and 20vol% Al2O3(p) composite. The samples were extruded and heat-treated to T6 condition. A 2.5kW CO2 laser was used for the experiments. Fig. 1 illustrates the experimental set-up. The advantages of CD-LBW is to avoid the presence of solid phase at the root of the weldment which leads to micro-fissuring. It also results in less HAZ width, through penetration, absence of porosity, less visible distortion, simultaneous preheating, lower solidification rates etc.

Figure 1. Schematic illustration of the CD-LBW set-up

Results and Discussion

In Fig. 2, a transverse section of the weld produced by CD-LBW is shown. The 2.5kW laser power was split into two beams of 60% (1500W-side A) and 40% (1000W-side B) of the power. The top half of Fig. 2 corresponds to beam A and the bottom half to beam B. On side B of the weldment, porosity of diameter 30 to 100 m can be seen and side A exhibits significant loss of Al2O3 particles in the fusion and HAZ. Under these parameters, the sub-surface porosity being driven from side A did not have enough time to escape out from side B, since the processing speed was high (2.1 cm/sec.) and only a small depth of the bottom face was in molten state, which solidified before the porosity was completely eliminated.

Figure 2. Optical micrograph of transverse section of the polished weldment at 2.1 cm/sec of 2.4 mm thick 20vol% composite

The absence of porosity and micro-fissures, makes the CD-LBW process more amenable for full-penetration welding at high process speeds. The advantage of high process speed is more than compensated by the sizable loss of Al2O3(p) in the fusion zone of A. A slight reduction in the traverse speed (1.7cm/sec) resulted in less particle depletion. With further decrease in speed to 1.3cm/sec, there is not only a significant drop in the depletion zone width, but also the fusion zone on side B did not retain any porosity. The power on side B was efficiently utilized for forming molten pool and in heating the particles.

Figure 3. Optical micrographs of (LEFT) the cross-section of weld bead, exhibiting the profile of both side A and B weld beads and (RIGHT) HAZ of the bead formed on side A

This efficiency in laser power absorption on side B is due to the larger heat input from side A of the weldment. Alumina particles have lower thermal conductivity than the base matrix- at high traverse speed, there is not sufficient time for the thermal energy to be transfered to the particles by conduction. The absorbed energy is utilized in melting the matrix, thereby forming large zones of precipitates. At high temperatures, the molten pool is less viscous. The increased fluidity of the alloy contributes to alumina particulate agglomeration. From these observations the optimal process speed was found to be 1.3 cm/sec for a 2.4mm thick sample reinforced with 20vol% Al2O3(p). Optical-microstructures of the transverse section of the weldment at these optimal conditions is shown in Fig. 3. For this case, the focal point of the side A was about 2.0mm inside the sample and the focal point of the side B was about 0.5mm inside the sample.

Advanced Composites Conference and Exposition
Research Project Awards

The advanced Composites Conference and Exposition was held November 6-9, 1995 at the Hyatt Regency, in Dearborn, MI. Over twenty MSU College of Engineering students presented poster papers and several presented podium papers on advanced composite materials. In addition, several MSU faculty and students received awards for their research projects and presentations as noted below.

Most Valuable NSF Center Research Project
"Electromagnetic Processing of Polymers and Composites"
under the direction of Dr. Martin Hawley

Best Academic Paper, Advanced Composites Conference and Exposition
"Some Approaches to Improve Efficiencies of Bolted Joints in Composites"
by Dr. Gary Cloud

Best MMPI Poster for Technical Committee II, Composites Processing,(tie)
"Microwave Processing of Composites Using Variable Frequency Method: Preliminary Results"
by Yunchang Qui and Dr. Martin Hawley
"Heating Characterization of a Microwave Part-Shaped Cavity"
by Trent Shidaker and Dr. Martin Hawley

Best MMPI Poster for Technical Committee III, Recycling and Waste Management
"High Purity Plastic Streams from Automotive Shredded Residue Using Hydrocyclone Classifiers"
by David Carlson and Dr. Charles Petty

Environmental Scanning Electron Microscopy

Environmental Scanning Electron Microscopy (ESEM) is a new innovation in scanning microscopes specifically designed to study wet, oil bearing, or insulating materials. Polymers, biological cells, plants, soil bacteria, concrete, wood, asphalt and liquid suspensions are readily imaged in the ESEM without specimen preparation of application of a conductive coating Samples may be examined in water vapor or in other gasses at near atmospheric pressures due to th unique vacuum system of the ESEM.

Energy Dispersive X-ray Spectroscopy (EDS) is a powerful technique for quantitative micro-analysis. The ESEM is wquipped with an Link ISIS system that has a resolution of 130eV. The Si(Li) detector utilizes an Atmospheric Thin Window that permits detection of elements ranging from Carbon (Z=6) to Bismuth (Z=83) at low vacuum conditions. X-ray mapping allows for imaging of the elemental distribution in samples.

Special stages are available for in situ investigations for almost any application. Hot and cryogenic stages permit examination of samples over a wide temperature range of -80 to 1000 C. A tensile stage enables in situ tension, compression, or bending experiments to be conducted at temperatures from -50 C to 900 C. Samples can be tested in different environmental conditions while applying loads up to 1000 lbs. A microinjection system allows direct application of liquids onto samples.

New Developments in Computational Models
for
Laminated Composite Structures

R.C. Averill and Y.C. Yip
Department of Materials Science and Mechanics
Michigan State University
East Lansing, MI 48824-1226

Introduction

Fiber-reinforced composites were initially employed in structures in such a way as to take advantage of their excellent inplane stiffness and strength properties. These early composite structural components were thin, and were designed to withstand primarily inplane tension loads. More recent applications of composite materials in automobile front aprons, bridge frames, submarine hulls, and primary load-carrying members of aircraft involve the use of thick-section laminates and/or sandwich construction that may contain over one hundred layers of fiber-reinforced material. Furthermore, such structures are typically designed to withstand complex mechanical loading states in the presence of harsh environmental conditions. Because the importance of laminated composites in aerospace, automotive, civil, and marine structures continues to increase, the development of accurate structural theories and efficient computational models for analyzing these structures is currently a critical area of research. Especially important is the development of accurate, efficient, and convenient models that can be used for static, dynamic, and progressive failure analysis of thick or thin laminated structures.

Brief Description of the Current Models

A new technical theory and associated finite element model have been developed for analysis of thick or thin laminated composite and sandwich panels. The theory employs the sublaminate concept, in which each computational layer (or sublaminate) contains several, even many, physical layers. Within each sublaminate, an accurate approximation of the displacement field is employed that accounts for discrete-layer effects without increasing the number of degrees-of-freedom (DOF) as the number of layers is increased. This is accomplished by satisfying analytically the continuity of transverse shear stresses at layer interfaces as well as the shear tractions conditions at the top and bottom surfaces of each sublaminate. Because the resulting through-thickness variation of inplane displacements takes the form of a piecewise nearly linear function, the theory is often called a zig-zag theory [1]. The operative DOF in the theory are displacements and stresses at both the top and bottom surfaces of each sublaminate, which facilitates the satisfaction of continuity conditions between sublaminates [2].

The accuracy and efficiency of the present theory is thus adaptable, depending on the number of sublaminates chosen to model a given laminate. For most global structural analyses, only one sublaminate is needed through the thickness of the entire laminate to attain the desired accuracy of overall structural response measures (e.g., deflections, vibration frequencies, and buckling loads) and bending stresses. More sublaminates may be used to increase the accuracy of overall structural response predictions or to capture local effects such as interlaminar stresses. Depending upon the ply stacking sequence and the type of global and/or local response measures that are sought, the optimal number of sublaminate approximations required for accurate analysis of thick multilayered composite laminates or sandwich panels will often be greater than one but far less than the number of layers in the laminate. The adaptable nature of the current theory can thus be used to great computational advantage.

Based on the above theory, a finite element model has been developed using state-of-the-art element technologies, including both anisoparametric interpolation and consistent strain field concepts. While the finite element geometry could take the form of a four-noded plate element, it is more advantageous to cast the element in the form of an eight-noded brick, as shown in Figure 1. This topology allows the laminate thickness to be conveniently subdivided and modelled by multiple finite elements (representing sublaminates). It is thus possible to increase the accuracy of the finite element model, as needed, to capture through-the-thickness gradients and transverse (interlaminar) stresses. It is also possible to simulate delaminations using the redundant node concept.

Figure 1. Topology of an 8-noded plate finite element based on a zig-zag sublaminate theory.

Both the theory and the finite element model have been developed in two forms. The first form, called here LZZ3, is based on a high-order zig-zag theory. The finite element model LZZ3 contains seven DOF per node: u, v, w, Qx, Qy, Jx, Jy. The first five DOF are the usual ones used in plate/shell finite elements. The remaining two DOF, Jx, Jy, are the transverse shear tractions (or interlaminar shear stresses, as the case may be). If only one sublaminate (element) is used through the thickness of a laminate, then these transverse shear DOF can be eliminated at the element level, leaving only the traditional five DOF. The second form of the model, called here LZZ1, is based on a first-order zig-zag theory. The finite element model LZZ1 contains the usual five engineering DOF u, v, w, Qx, Qy at each node.

For symmetrically laminated composite and sandwich panels, LZZ1 is only slightly less accurate than LZZ3. When the lamination sequence is unsymmetric about the midsurface of the panel, LZZ3 can be significantly more accurate than LZZ1, depending upon both the lamination sequence and the span-to-thickness ratio of the panel.

Sample Application

Numerical results are presented for bending of a simply-supported laminated square panel subjected to a double-sinusoidally varying transverse load (see Figure 2). The laminate is very thick, having a span-to-thickness ratio of four. Numerical examples such as this are an excellent test of a model s ability to capture local effects. Physically, this example is similar in many ways to local models of the region directly beneath an impact load.

Figure 2. Schematic of loading and boundary conditions for example problems
Note: Load is distributed over entire top surface

Numerical results are obtained using a quarter-model with a discretization of 8x8 elements in the plane of the panel and either one or three elements through the thickness. Only results using LZZ3 elements are shown. Comparisons are made between predictions of the current model and an exact elasticity solution [3].

The example utilizes a lamination scheme that is similar to that present in the U.S. Army TACOM s Composite Armored Vehicle (CAV). The laminate consists of 55 layers and is nearly two inches thick. The laminate can be divided into three sections, as described below:

1. Inner Shell: four plies of S-2 Glass/Phenolic Fabric and 37 plies of S-2 Glass/8553-40 Epoxy Tows.
2. Armor Core: one layer of EPDM rubber and one layer of ceramic tile with inserts.
3. Outer Shell: twelve plies of S-2 Glass/8553-40 Epoxy Fabric.

Two levels of thickness discretization were used. The first model employed a single element through the thickness of the entire panel, and the other used three elements through the thickness one for the inner shell, one for the armor core, and one for the outer shell.

Results of the analyses are shown in Figures 3-5, where predictions by the two LZZ3 models of the through-thickness variations of inplane displacement ux, inplane stress Fx, and transverse shear stress Jxz are compared to the variations predicted by three-dimensional elasticity. Even for this extremely thick laminate, it can be seen that good predictions of ux and Fx are obtained using only one element (sublaminate) through the thickness. When the thickness is discretized using three elements, the model predictions of ux and Fx are nearly indistinguishable from the elasticity solutions. However, three elements through the thickness are needed to capture the variation of Jxz in this example.

Figure 3. (ABOVE) Axial displacement versus normalized thickness coordinate in a simple-supported square plate(CAV) subjected to sinusoidal load

Figure 4. (ABOVE) Axial stress versus normalized thickness coordinate in a simple-supported square plate(CAV) subjected to sinusoidal load

Figure 5. (ABOVE) Transverse shear stress versus normalized thickness coordinate in a simple-supported square plate(CAV) subjected to sinusoidal load

Conclusions

The current models show excellent promise for efficient and accurate analysis of thick and thin laminated composites and sandwich panels. The current models are much more efficient and potentially more accurate than 3D continuum-based models for analysis of laminated composites and sandwich panels. They are also computationally competitive with traditional plate elements.

References

1. Averill, R.C., Static and Dynamic Response of Moderately Thick Laminated Beams with Damage, Composites Engineering, Vol. 4, p. 381-395, 1994.
2. Averill, R.C. and Yip, Y.C., An Efficient Thick Beam Theory and Finite Element Model with Zig-Zag Sublaminate Approximations, to appear in AIAA Journal, 1996.
3. Burton, W.S. and Noor, A.K., Three-Dimensional Solutions for Thermomechanical Stresses in Sandwich Panels and Shells, ASCE Journal of Engineering Mechanics, Vol. 120, pp. 2044-2071, 1994.

Upcoming Conferences and Seminars ... March - October96

American Society for Composites - 11th Technical Conference on Composite Materials
Location: Omni Hotel at CNN Center, Atlanta, GA
Date: October 7-9, 1996
Chairperson: Prof. W. Steven Johnson, Georgia Inst. Tech.
Phone: 404-894-3013

SME International Conference on Education in Manufacturing - Manufacturing Education for the 21st Century
Location: San Diego, CA
Date: March 13-15, 1996
*Abstract was due: April 28, 1995
*Paper was due: August 31, 1995
Phone: 313-271-1500 Ext. 506
Fax: 313-271-2861

MACM 96 Marine Composites Conference - Sixth International Conference on Marine Applications of Composite Materials
Location: Cocoa Beach, FL
Date: March 19-21, 1996
*Abstract was due: September 15, 1995
*Paper due: January 12, 1995
Phone: 407-951-9464
Fax: 407-728-9071

Materials Research Society (MRS) - 1996 Spring Meeting Symposium on Microwave Processing of Materials
Location: San Francisco, CA
Date: April 8-12, 1996
*Abstract was due: May 15, 1995
*Paper was due: November 1, 1995 Phone: 412-367-3004
Fax: 412-367-4373

SAMPE - 41st International Symposium & Exhibition Location: Anaheim, CA
Date: March 25-28, 1996
Phone: 818-0616 Ext. 610

NSF Center Project Review Session
Location: MSU Management Center, Troy, MI
Date: March 27, 1996
Phone: 517-353-3969
Fax: 517-353-9854

NSF Center - MMPI Symposium
Location: E. Lansing, MI
Date: May 22, 1996
Phone: 517-353-3969
Fax: 517-353-9854

Composite Materials and Structures Center
College of Engineering
Michigan State University
East Lansing, MI 48824-1326

Please send any comment or request to

CompositeWEB page