Modeling Continually Improves Advanced Materials for Automobiles By BHASKAR PATHAM, PH. D., KIRAN B. DESHPANDE, PH. D., AND SAMPATH K. VANIMISETTI Automakers such as General Motors are playing a key role in creating the science and technology that will help shape the global future in terms of personal transport. As GM focuses on developing automotive solutions that are sustainable in terms of energy and environment, reducing weight of the vehicle and electrification of the propulsion system have emerged as key priorities [1]. These solutions are driven by continual exploration of new materials and manufacturing processes — for example, multi-material solutions and composites for lightweighting and robust battery materials for electrification. Developing a fundamental understanding of these new material systems and the associated manufacturing processes therefore becomes very important and forms a key effort at the GM Global R&D. At the Material Characterization and Modeling Group, which is a part of the India Science Lab of the GM Global R&D, we are integral to this effort, with the mission to develop validated computational models applied to various products and sub-systems performance and the associated manufacturing processes. In this article, using three simple examples, we illustrate how computational tools help us in developing a fundamental understanding of materials and their processing, and in improving material and process performance. Process-Property Interrelationships in Thermosets and Thermoset-Matrix Composites Thermoset-matrix composites hold great potential as automotive materials because of their lower density (compared to metallic alloys such as steel), high specific strength, and good energy dissipation characteristics. With the right combination of suitable fibrous reinforcement and processing, thermosetmatrix composites have the potential for replacing ferrous materials in structural applications, thereby contributing to automotive lightweighting. In addition, thermoset resins, both unreinforced as well as reinforced, find wide applications in automotive component joining as adhesives. Processing of thermoset resins and thermoset-matrix composites involves “cure,” which converts them from their initial viscous fluid (or soft solid) state, through progressive crosslinking, into a stiffer solid. The curing process also typically involves application of heat, pressure (from a mold or in an autoclave), and constraints (globally, from a mold, or the components of a joining assembly, and at the local scales, from the reinforcing fibers). The versatility of thermoset chemistries allows the achievement, upon complete cure, of properties for the base resin that can range from rubbery to high stiffness for semi-structural applications (equivalent to engineering thermoplastics). “COMSOL allows manipulation and redefinition of existing variables without resorting to complex user subroutines.” Processing of thermoset-composite parts and assembly with thermoset adhesives come with their own set of challenges. For example, while steels behave quite well during forming operations and show excellent retention of shape after being removed from a mold or press, this isn’t easily achieved in the case of thermoset composites. This is because cure-induced residual stresses can result in springback of thermoset components after they are released from the mold or in shape-distortion effects in assemblies. Cure-induced residual stresses are driven by the evolution of the modulus of the crosslinking resin, combined with constraints that arise from the relative mismatch between the thermally or chemical-shrinkage induced deformations between the thermoset resin and the mold/components of the joining assembly/ reinforcing fibers. Process engineers may resort to a variety of strategies to minimize shapedistortion effects in thermoset systems: they can redesign the mold, modify the temperature-ramp during the curecycle, or modify the fibers and resins that make up the composite. Of course, in the absence of a fundamental understanding of the various mechanisms that contribute to the development of residual stresses in a particular system, this would involve experimental trial and error, which would be a very expensive proposition. Simulations of residual-stress development in thermosets not only allow the systematic probing of various underlying mechanisms that may cause shape distortions, but also present an inexpensive alternative to experimental trial and error to arrive at optimal design and process modifications so as to mitigate the deleterious consequences of residual stresses. In order to be able to use the estimates obtained from these simulations to formulate guidelines for process and design modifications, it is imperative to account for the cure-, temperature-, and time-dependence of the resin properties with utmost accuracy. As an example, the choice of material model — an elastic material model or a viscoelastic material model — can have significant impact on the predictions of residual stresses. Elastic models are simpler to set up, are computationally efficient, and give reasonable results in many scenarios. Viscoelastic models, on the other hand, capture the material properties much more faithfully and have higher accuracy, but they are not always easy to set up and are more expensive both in terms of computational memory and time requirements. Several prior studies have employed a viscoelastic model for thermosets, which has been implemented in user codes or as user subroutines in commercial FEM packages. While offering comprehensive models, these studies resorted to either elastic or viscoelastic models alone and did not make a quantitative comparison of the two approaches. Figure 1. Residual stresses in an elastic material are only governed by the instantaneous states of temperature and degree of cure, while the stresses in a viscoelastic material are strongly governed by the thermal history experienced by the resin. Our goal in this simple demonstration — involving the mold-constrained cure and subsequent springback behavior of a thick asymmetric 90°-elbow section of an unreinforced thermoset resin — was to develop equivalent elastic and viscoelastic models that can clearly pinpoint the differences between the two with regard to the time-, cure- and temperature- dependent evolution of stresses (For details, please see Ref [2]). It should be noted that cure of thick sections can result in significant exothermic heat release, resulting in large spatial gradients in temperature and degree of cure. In Figure 1, the spatial distribution of stresses in the thick elbow section is shown at the end of the mold-constrained cure cycle, when the degree of cure and temperature are spatially uniform. It is clear from Figure 1 that the instantaneous stresses in a linear-elastic material are only governed by the instantaneous states of temperature and degree of cure, while the stresses in a viscoelastic material are strongly governed by the thermal history experienced by the resin. As seen in Figure 1, the spatial gradients in temperature and cure in thick sections (enhanced by the exothermic heat of reaction) can result in significant spatial variation of viscoelastic residual stresses even after equilibration of the temperature fields and achievement of uniform cure (as evident in a pronounced gradient in stresses along the diagonal of the elbow section). These subtle details, captured by the viscoelastic model, also result in significantly different predictions of the springback behavior. As seen in Figure 2, the elastic material model does predict springback, but does not capture the shape-distortion details as clearly as the viscoelastic models. Thus, we conclude that the choice of the material model — elastic or viscoelastic — is equally, or more significant than the mold-part-interaction details in governing springback predictions. Figure 2. Viscoelastic model (right) captures the subtle influences of temperature and cure transients on the shape distortion effects. We have thus been able to develop a model that accurately captures the evolution of residual stresses throughout the manufacturing process and determines their effect on the final shape of the composite part. For this application, COMSOL offers several advantages, as it allows manipulation and redefinition of existing variables without resorting to user subroutines and thus accommodates more complicated expressions for the evolution of the modulus apart from the basic viscoelastic model setup. The coupling between heat-transfer and diffusion analysis (to account for the exothermic heat of reaction) was readily available as a part of the standard variables offered in COMSOL. The coupling between diffusion and structural mechanics (to account for shrinkage strains), while not readily available as a part of standard variables, could be established in a systematic fashion without resorting to a complex user-defined routine because all the variables employed for the analysis are transparent to the user for easy modification. Figure 3. Corrosion of magnesium joined with steel when immersed in a salt solution. Galvanic Corrosion in Multi-Material Assemblies As noted earlier, a key thrust in the automotive industry is reducing vehicle mass. This has led to the exploration of multi-material solutions in the automotive structure, body, as well as powertrain. Magnesium is the lightest structural material, being 4 times lighter than steel and 1.5 times lighter than aluminum. However, the use of magnesium is quite limited today primarily due to its poor corrosion resistance. The corrosion behavior of magnesium, joined with steel and immersed in an electrolyte solution, is shown in Figure 3, in which it is possible to see that a considerable amount of material has been dissolved in the electrolyte from the magnesium’s surface. Figure 4. Typical 3D results from COMSOL Multiphysics simulations of a nodular battery electrode particle, idealized as a sphere. It shows the subdomain contours of (a) the normalized lithium concentration and (b) normalized strain energy density during intercalation of lithium into the host material while the battery is being charged or discharged. The black outline is indicative of the initial shape and size of the particle. Deformation is scaled for clarity. In order to quantify the dissolution rate (corrosion rate) of magnesium joined with steel, we developed a model using COMSOL Multiphysics. The polarization behavior of magnesium and steel, obtained individually from lab-scale polarization experiments, was used for the boundary conditions. Because the magnesium surface is continuously dissolving in the electrolyte, the galvanic corrosion becomes a moving boundary problem and was implemented using the Moving Mesh application mode. This numerical approach helps in the understanding of the corrosion mechanism, in the selection of materials based on galvanic-corrosion severity, and also in providing design guidelines accounting for cathode-to-anode area ratio. Mechanics of Advanced Battery Materials In recent years, the electrification of vehicles has become an important strategic initiative. R&D activity in advanced electrochemical storage devices aims to offer a potentially robust and clean alternative to fossil fuels. Lithium-ion battery chemistry is the obvious choice for automotive energy storage due to its high gravimetric energy capacity and ability to deliver the power density necessary for driving the vehicular powertrain. Lithium-ion batteries consist of two electrodes into which the lithium intercalates along with an electrolyte in which the lithium ion associates with the anions. Depending on whether the battery is charging or discharging, the lithium intercalates into or de-intercalates from either the positive or the negative electrode. Most electrode materials suffer from volume expansion due to lithium intercalation, which can be as high as 300% for some candidate materials. In addition, phase transformation in the lithiated compounds leads to formation of incoherent phases in the electrode particle that not only lead to misfit strains but also affect further lithiation. The transient in lithium diffusion, along with above aspects, leads to development of large mechanical stresses in the electrode particles. Cyclic intercalation and de-intercalation may cause cracking or accumulation of damage over time in the electrode particle. It is widely reported that this mechanical degradation of the battery electrode material is associated with capacity fading due to loss of active lithium in the formation of SEI (Solid Electrolyte Interphase) on the damaged surface or electronic isolation. The study of diffusioninduced stresses in electrode particles due to lithium intercalation forms a part of our group’s fundamental investigation to understand the impact of electrode design parameters on the performance of the battery. To this end, we created a 3D model in COMSOL by taking advantage of its multiphysics and customization capabilities. The aspects related to diffusion from the Chemical Engineering Module were combined with 3D stress analysis from the Structural Mechanics Module. We developed a special solution scheme to couple the lithium concentration with mechanical strain and subsequently used that to estimate quasi-static equilibrium stresses in the particle. We then used results from the 3D finite element simulations (Figure 4) to understand the effect of microstructural aspects on the possibility of mechanical degradation of the battery materials. Future improvements to the model aim to incorporate a more realistic description of the host material by addressing phase transformation due to lithium intercalation, anisotropy and polycrystallinity. Eventually, the study aims to offer clues to mitigate detrimental stresses and improve battery durability. ACKNOWLEDGMENTS The authors would like to acknowledge B. G. Prakash, A. M.Kumar, R. Narayanrao, H. G. Kia, and M. W. Verbrugge for their constant support and encouragement, and K. Manjunath for assistance with experiments. REFERENCES 1.Taub, A.I., Krajewski, P.E., Luo, A.A., and Owens, J.N., “Yesterday, today and tomorrow: The evolution of technology for materials processing over the last 50 years: The automotive example,” Journal of Metals, 59 (2), 48 (2007) 2.Patham, B., “COMSOL® Implementation of a viscoelastic model with cure-temperature-time superposition for predicting cure stresses and springback in a thermoset resin,” Proceedings of COMSOL Conference, 2009, Bangalore (2009) About the Authors Bhaskar Patham is a researcher at the General Motors R&D, India Science Lab, Bangalore. His research interests are in the areas of experimental and analytical rheology, complex fluids, polymer and composites processing, and processing-structure-property interrelationships in multi-phase polymeric systems and polymeric matrix composites. He holds a Ph.D. in Chemical Engineering and Materials Science from Michigan State University. Prior to joining GM R&D, he was a research engineer with the General Electric Global Research Center at Bangalore. (From left to right) Bhaskar Patham, Kiran B. Deshpande, and Sampath K. Vanimisetti. Kiran B. Deshpande is a researcher at the General Motors R&D, India Science Lab, Bangalore, where he is involved in modeling of galvanic corrosion and its mitigation strategies. He holds a Ph.D. in Chemical Engineering from University of Sheffield. Prior to joining GM R&D, he was a Knowledge Transfer Program associate with University of Sheffield and MHT Technology Ltd., UK. He has contributed a chapter in the book titled Multiphysics Modeling with Finite Element Methods by Prof. WBJ Zimmerman. Sampath K. Vanimisetti is a researcher at the General Motors R&D, India Science Lab, Bangalore. His field of expertise is in the area of computational mechanics of materials with specific emphasis on damage and fatigue. At GM R&D he is employing modeling techniques in COMSOL to understand diffusion-induced stresses responsible for fragmentation in battery electrode materials. Proir to GM R&D, Sampath was a research engineer at GE Global Research Center. He holds a M.Sc.(Engg.) degree in Mechanical Engineering from the Indian Institute of Science, Bengaluru (India). |
分享到
豆瓣网
开心网
人人网
QQ书签
Google
2336个朋友已经阅读过这篇文章
用户评论
没有找到数据. , |