Search Documents

By Manfred K. Walther

"Tests were conducted on instrumented plates cast in the Abex Experimental Foundry to compare recorded solidification histories with results computed for corresponding models simulated with the C.A.S.T. program. The first plate was cast of low alloy steel and the second of carbon steel with a graphite chill at one end. Both were cast in molds of bonded silica sand. Temperature histories were recorded in the castings, molds and chill. Exceptionally good agreement was achieved between computed and experimental data. Other investigators will find the results useful as a benchmark against which they can compare the results of their own calculations.IntroductionSince the early 1970's our foundry engineers have used both two- and three-dimensional versions of the G.A.S.T. program to gain knowledge of the solidification process [1,2,3]. Study of the computed data gave insight as to how the rigging and risering of specific castings might be changed or adjusted to improve quality and yield. New schemes based on such analyses were then recommended for implementation on the foundry floor. The degree of success of the new practice was subsequently evaluated after the part had been poured, shaken out and examined. Similar conditions were noted in the sectioned castings whenever porosity and voids had been predicted by the computations. Appropriate modifications, when found successful in the model, were applied to the process. This led to a reduction, if not complete elimination, of the defects. Such results gave credence to the use of these programs as an aid in the production of better castings.Earlier investigations on simple shapes revealed excellent correlation between output derived with the G.A.S.T. program and exact solutions. Results also compared favorably with data calculated using other codes as reported in the literature [4] and summarized in the Appendix. However, since the computer program had not been verified by direct experiment, it was decided to run instrumented tests to check the theoretical predictions with actual measurements."

Jan 1, 1986

By J. Szekely, S. Rhee

"A mathematical formulation has been presented to describe heat and fluid flow phenomena in CVD operations. In the statement of the problem, which included consideration of plasma enhanced CVD systems, emphasis was placed on identifying the parameters which play a key role in determining the rate and the spatial uniformity of the deposition process. This was done both through the appropriate dimensionless groups and by presenting an extensive set of computed results. The main conclusions that may be drawn from this work are that the deposition rate will be inherently non-uniform for atmospheric CVD systems; this may be somewhat ameliorated by inclining the susceptor or changing the vessel contours. Operation at greatly redeced pressures and for PECVD systems will give a much more spatially uniform deposition rate.1. INTRODUCTIONIn recent years, there has been a growing interest in chemical vapor deposition techniques for the preparation of thin films for semiconductor, photovoltaic applications and for coatings in general, as evidenced by numerous journal articles and symposium proceedings devoted to this topic (1-6). An interesting variant of this technique is to perform the deposition in a plasma or glow discharge at low pressures because this technology obviates the need for the thermal decomposition of the precursor at the substrate surface and permits operation at greatly reduced temperatures (7-10).By far, the largest portion of the reported work dealing with both these aspects of the technology has been concerned with the characterization of the deposit, quite properly so. However, in more recent times it has been realized that the transport aspects of these processes are also deserving of attention because the understanding of heat flow, fluid flow and diffusional phenomena may provide the key to the solution of problems associated with obtaining uniform deposits and the proper scaling of laboratory measurements in order to design industrial CVD systems."

Jan 1, 1986

By Erwin Flender, Preben N. Hansen, Peter R. Sahm

"Until now FDM-Techniques have been widely used for Modeling Solidification Processes in Castings. Often a rigid, regular type of enmeshment (cubes, , --) have been used. This has the advantage of low CPU-costs, but the disadvantage of a zig-zag presentation of the interface between the casting and the mold. The last point is the reason why FEM-Techniques, giving the possibility of a true presentation of the shape in the enmeshment, are increasingly used. The disadvantage is the high CPU-costs. In this paper, a technique is presented, which uses the FDM-Mesh with the zig-zag presentation of shape and ADI-Algorithms to calculate temperatures and gradients. The new idea is that the calculated parameters are not presented related to the mesh in which they are calculated, but displayed using the true shape of the casting. Such an advantage of both low CPU-costs and shape correct presentation of the calculated results are achieved. IntroductionThe main reason for using a regular type of enmeshment is that it gives the possibility, via ADI-Algorithms, of obtaining a very fast FDM-Technique. The technique presented by P.R. Sahm and P. N. Hansen (1984), and also used by E. Flender (1985), is numerically so stable that average time steps, at least 200 times larger than the limit given by the explicit FDM-Technique, can be applied. To obtain reasonably good results, average time steps in the range of 20 to 50 times the explicit limit seems to be applicable. It should be pointed out,however, that FDM-Techniques can be used on arbitrary nets, the disadvantages being the same as for the FEM-Techniques. This has, among others, been discussed by P.R. Sahm and P.N. Hansen (1984)."

Jan 1, 1986

By George E. Cook, Peter C. Levick, Francis M. Wells

"Relationships are established between the peak current of a pulsed, rectangular current waveform and the pulse current duty cycle under conditions of constant arc power. By appropriate choice of these interrelated parameters, it is shown that the arc pressure may be varied over a wide range even though the arc power is held constant. The methodology is suggested as a means of countering the effect of gravity in 5-G welding, while maintaining constant heat input to the weld. Combined with appropriate penetration sensors, the methodology is additionally suggested as a means of controlling penetration. IntroductionBoth low frequency pulsation (typically 0.1 - 20 Hz) and high frequency pulsation (> 100 Hz) are often used with GTA welding. Benefits derived from pulsing include improved weld quality and better process control.Advantages cited for low frequency pulsation include claims that the total heat input to the weld is reduced, the weld bead size is reduced, weld penetration is increased, grain size is reduced, porosity is reduced, weld distortion is reduced, thinner materials can be welded successfully, and the tolerance of the welding process to material and production variations is improved (1-16). With respect to the latter, pulsing aids control of weld bead penetration by increasing the tolerance to component dimensions and joint fit-up. It also reduces the sensitivity to disparities in heat sink, and it helps to reduce distortion in thin section material or distortion attributable to poor clamping. In certain materials it can reduce the sensitivity to surface oxides and to cast to cast compositional variations (16). Low frequency pulsation is also useful in welding thick materials mated by a prepared groove. In these applications the low frequency pulsation is used to obtain proper sidewall fusion as the tungsten electrode is weaved from one side of the groove to the other. In this case the weld current pulsation is synchronized with the cross seam weave pattern, such that the high current pulses occur at each sidewall. This type of low frequency pulsation has proven particularly useful in welding with narrow width groove designs (17-21). Finally, Ecer (22) has shown that the pulsed current arc is more stable against magnetic disturbances than an equivalent constant current arc."

Jan 1, 1986

By M. J. Bagshaw, J. D. Hunt, R. M. Jordan

"Continuous casting processes, and the roll casting process in particular, offer considerable potential to the Aluminium industry, both meatallurgically and economically.The three major limiting process problems that occur during roll casting are Centre Line Segregation', 'Sticking' where metal adheres to the rolls during casting, and 'Heat Line' formation. 'Heat Lines' are localised semi-continuous longitudinal defects, usually a few cms in width, where the cast metal leaves the roll bite partially molten. These defects occur at relatively low casting speeds when the mean slab exit temperature remains several hundred degrees below the solidus temperature. The objective of the work described here is to develop a better understanding of the roll casting process in general, and in particular, to understand the phenomenon of 'Heat Line' formation.A steady state two dimensional finite difference model has been developed to describe the roll casting process, incorporating a variable heat transfer coefficient through the roll bite. In the model the difference equations are formulated using a control volume method and latent heat is incorporated according to the Scheil equation. A Gauss-Seidel type technique is used to obtain the solution. Because the rate of heat evolution is non-linear the temperature in each box is determined by using a Newton-Raphson method. The solution for a box can be multi valued and therefore lead to instability particularly near the vicinity of the front. A stability 'criterion is presented which shows that to obtain a stable solution it is better to have long narrow boxes rather than boxes with small dimensions in both directions.The model explains why heat lines form in terms of the variable slab/roll heat transfer coefficient.Experimental measurements of slab exit temperatures are also presented. These were obtained on a Hunter twin roll caster for the binary AI-Cu system, at two compositions over as wide a range· of casting speeds as possible. The experimental results are compared with the predicted ones."

Jan 1, 1986

By S. C. Flood, J. D. Hunt

"The columnar-equiaxed transition (CET) has been studied numerically under unsteady-state conditions using mathematical models that: (i) include a non-equilibrium undercooling at the columnar front;and(ii) describe the simultaneous evolution of the columnar and equiaxed zones.The results are discussed and compared with those produced by a steady-state analysis.IntroductionThe majority of the previous work on the columnar-equiaxed transition (CET) has concentrated on the production of equiaxed nuclei, regarding this as being the critical step in the formation of an equiaxed zone. Recently, the growth of equiaxed grains has been considered and this is not unreasonable because, for an equiaxed zone to be present, the grains must grow as well as nucleate. Since the growth rate of the grains will be determined by the thermal environment at and ahead of the columnar tips, it was decided to model the liberation and flow of heat during solidification to discover whether the various trends in the columnar range might be explained by their effect on the temperature profile throughout the casting. The models presented here for the solidification of a casting are the first to include the latent heat liberated ahead of the columnar front. Burden and Hunt (1) consider its effect only qualitatively, and although Lipton et al (2) adopt a criterion for the CET which is based on the thermal interaction of the columnar and equiaxed grains, they do not actually include the latent heat in their calculations.It can be argued that the latent heat liberated by the equiaxed grains in the bulk will retard the columnar dendrites in a casting: once the equiaxed grains are sufficiently well-established, the latent heat that they evolve could slow down the columnar growth to such an extent that the equi axed would dominate. Alternatively, the latent heat produced by the equiaxed grains might suppress further equiaxed solidification by raising the bulk temperature. The nature of the thermal interaction between columnar and equiaxed growth is not obvious, and because it is dependent on the material parameters and the dimensions of the system, it cannot be considered effectively in a qualitative way but has to be studied quantitatively: hence the need for the mathematical models discussed in the following pages. The results of the models are discussed in the light of a steady state analysis."

Jan 1, 1986

By P. Wieser, D. Feke

"Priming of filters with small pores is influenced by the surface tension of the liquid metal and may be complicated by freezing • Freezing followed by melting has been observed experimentally. This process has been modeled on the basis of one dimensional heat flow by radiation to the filter and by conduction in the metal. Also considered are super-heat of the metal and the initial filter temperature. The results of the model provide useful guidelines for the application of filters. IntroductionFiltration has been recognized as a valuable method of refining by removing suspended phases from liquid metals or alloys. Woven fiberglass type fabrics and wire screens were used initially in the aluminum industry. With the introduction of the ceramic filters that exhibit improved inertness and refractoryness, filtration has been introduced to higher melting alloys including cast irons, superalloys, stainless steels, as well as low and medium carbon low alloy steels. A considerable literature is delveloping rapidly on the progress in applications and the advances in understanding the mechanisms and effectiveness of filtration processes (1-8).With the use of filtration for different alloy systems for the production of cast to shape parts a bewildering variety of gating system modifications has been developed (Figure 1) to accommodate the installation of filters in the mold. The modifications have two principal purposes: 1) the runner expansion is intended to compensate for the resistance to flow due to the filter and that of the captured inclusions, and 2) the metal reservoir upstream of the filter, provides a temporary by-pass which permits metal to flow across the filter and heat it until priming can be accomplished. priming of.filters by low and medium carbon steel is readily accomplished but, compared to such alloys as cast iron or 19 Cr-9Ni stainless steel, requires greater precautions to avoid failure to prime by excessive freezing on or inside the filter pores. This problem is further examined below:"

Jan 1, 1986

By A. F. Giamei, F. Landis, M. O. Duffy, R. J. Mador

"The directional solidification of complex advanced gas turbine airfoils must be carried out without the nucleation of undesirable grains and with minimal shrinkage. Such problems are typically eliminated by using a number of mold design and process iterations. A more efficient approach is envisioned where a solidification model is generated and utilized to accelerate process development. Towards this end, the directional solidification (DS) of simple geometrical shapes in a crystal grower apparatus has been simulated with MARC finite element nonlinear thermal analyses.The finite element models of cylindrical (2D) and rectangular (3D) cross-section bars in the DS environment are presented. Temperature dependent properties, the impact of latent heat of fusion and a time-dependent radiation boundary condition constitute the more salient features of the MARC models. Solidification zone thermal behavior is predicted during the withdrawal transient for the specific case of nickel based superalloys grown in ceramic shell molds. Comparisons of analytical results with experiments demonstrate notable prediction accuracy. INTRODUCTIONSingle crystal and uni-directional crystal turbine blades require tightly controlled conditions if metallurgically and mechanically acceptable parts are to be produced. Gated blade geometry and process parameters must be precisely determined if incorrect grain structure, orientation and shrinkage are to be avoided. Since purely experimental development programs designed to evaluate these parameters can be excessively costly and time-consuming, an effort combining engineering analysis and verification by experiment has been initiated at Pratt & Whitney. Comparison of analytical predictions and measured data for geometries simpler than a turbine blade, namely cylindrical and rectangular bars, comprised the initial stage of this effort. Subsequent tasks involve modeling geometries of increasing complexity and ultimately the simulation of the directional solidification of the turbine blade, shown schematically in Figure 1. This paper summarizes the results of the initial stage of the effort."

Jan 1, 1986

By P. H. Chang, J. E. Gould

"Solidification structures and austenite decomposition in some spot welded O.05%C - O.3%Mn steels have been studied with the aid of a numerical thermal model. The model is a one dimensional finite difference formulation, and accounts for melting, temperature dependent thermal properties, effective heat transfer in the liquid, and contact resistances. The model was used to predict solidification conditions during resolidification of the nugget. These solidification conditions were correlated with the primary dendrite spacing using available primary dendrite spacings modelling. The model was also used to predict cooling rates in the temperature regime for austenite decomposition. These cooling rates were used to interpret the underlying microstructure.INTRODUCTIONThe metallurgical transformations which occur during the spot welding of low carbon steels have traditionally been characterized only in the most general of fashions. This is most probably due to the similarity of spot welded microstructures to other welded microstructures, and that the mechanical integrity of spot welds has not been a concern. In addition, little information about the particular thermal cycles experienced during spot welding has been available. The introduction of HSLA (microalloyed) sheet steels has, however, raised new concerns about the integrity of spot welds. These steels have on occasion shown irregular weldability, as well as brittleness in the weld nugget on destructive testing l - 4 .In response to these concerns, considerable work has been done attempting to characterize thermal cycles in spot welds 5-12. This has largely taken the form of thermal modelling, with the end purpose generally to estimate the size of the nugget formed. This work uses the thermal model developed at Inland Steel 13 ,14 to try and understand, in a more fundamental way, microstructural evolution in spot welds. Two facets of this microstructural evolution are examined in detail:"

Jan 1, 1986

By Carlos G. Levi

"The principles involved in the rapid solidification of supercooled metal droplets and existing numerical models are reviewed. The limitations of these models for studying alloy solidification are discussed in terms of the effects of alloying elements on the relevant heat transfer, thermodynamic and kinetic parameters. Analysis of the enthalpy-temperature diagram reveals how the extent of rapid solidification is hindered by the addition of solute. A solidification model incorporating thermodynamic and kinetic relationships into a Newtonian heat transfer formulation is used to analyze the thermal history and concurrent solute redistribution during recalescence. IntroductionIt is well acknowledged that microstructural evolution during solidification is intimately associated with the thermal history of the liquid-solid interface. Structural parameters like chemical homogeneity, morphology and size scale of the phases can be related to the interfacial temperature and velocity. These in turn change by the interaction of the interface with the thermal and solutal fields in its neighboring environment.The thermal history of powder solidification can be conveniently described on the enthalpy-temperature diagram for a pure metal shown in Figure 1 and discussed in detail elsewhere [1). A system can conceivably follow any path of decreasing enthalpy in the supercooled region of Figure 1 (~ < 1), starting at the nucleation temperature (oN < 0) on the liquid line, and ending at some point on the solid line. A vertical path represents an isothermal transformation, of which a limiting case is solidification without significant barriers to nucleation and growth, with the interface at the equilibrium temperature TM (path one in Figure 1). A horizontal path represents isenthalpic (or adiabatic) solidification whereby the system could form an amount of solid proportional to the initial supercooling of the liquid without further heat loss to the environment. A liquid that can completely transform to solid under adiabatic conditions is said to be ""hypercooled""."

Jan 1, 1986

By B. Farouk, D. Apelian

"A numerical model has been developed to predict the temperature history of particles injected in a low pressure d.c. plasma jet. The temperature and velocity fields of the plasma jet are predicted as a free jet by solving the parabolized compressible Navier-Stokes equations using a spatial marching scheme. Particle trojectories and heat transfer characteristics are calculated using the predicted plasma jet temperature and velocity fields. Correction factors have been introduced to take into account non-continuum effects encountered in the low pressure environment. The plasma jet profiles as well as the particle/plasma interactions under different jet pressure ratios (from underexpanded to overexpanded) have been investigated.I. IntroductionIn the plasma spraying process a d.c. plasma jet is used as a heat source to melt and accelerate powder particles which subsequently impinge and solidify on a given substrate. During low pressure plasma deposition (LPPD) prooess, the plasma jet is expanded into a low pressure environment (40-80 Corr) which results in high plasma jet velocities with Mach numbers ranging from 2 to 3. Previous efforts to predict heat transfer and fluid flow in plasmas have made use of incompressible flow formulation employing turbulence models [1-3] for atmospheric plasma spraying. However, these theoretical calculations have been limited to the subsonic flow regime; hence, the compressibility effects and viscous heat dissipation were neglected. In the LPPD process, the plasma jet is operated at the supersonic flow regime, therefore, the compressibility effects and viscous heat dissipation are no longer negligible. This is the main motivation of this work.In addition, production of high quality dense deposits requires that a large fraction of the injected particles be in a molten state when they impact the substrate. An additional motivation for this work is that it is important to understand the plasma/particle heat and momentum transfer since these phenomena dictate both the particle thermal history and particle trajectory.Early computational works in supersonic nozzle field flow [4,5,6] consisted of patching methods that divided the flow field into an invisoid free stream and a viscous boundary and. mixing layers. Each was analyzed independently and coupled through appropriate boundary conditions. Later schemes [7,8] involved solving time-dependent, compressible Navier-Stokes equations over the entire computational domain. These schemes show a great promise in predicting flows with complex structure, and where the viscous effects became prevalent."

Jan 1, 1986

By W. -S. Hwang, R. A. Stoehr, P. Ingerslev

"The flow of molten metal entering complex two-dimensional foundry molds has been modeled using the Simplified Marker and Cell (SMAC) and Solution Algorithm (SOLA) techniques. The molds were for a series of four and three-spoke wheel designs. The computer calculations were compared to physical experiments in which molten cast iron was poured into C02-bonded sand molds. The metal was photographed with a high speed motion picture camera focused through a Pyrex glass window on one face of the mold. To get good agreement between the calculations and the experiments, it was necessary to accurately represent the velocity through the ingate. This was calculated by application of the Bernoulli equation to flow through the feed system. Some difficulties were experienced with mathematical instability, expecially in the three-spoke design with a narrow ingate. This was overcome by using 100% upwind differencing, and over relaxation parameter near 1.0 (i.e. no over relaxation), and variable mesh spacing.IntroductionModeling the flow of molten metal entering molds has the potential for making significant contributions to the design of feed systems. It will supply information to the computer-aided-design systems of the future that will enable the designer to optimize the placement of gates, dimensioning of passages, and designation of casting speeds. Such design criteria have generally been based on application of energy balance (Bernoulli equation) principles to flow in the sprues, runners, and gates. Indeed, this is the easiest way to determine the proper dimensions to give a suitable filling time for a casting of a given size. A number of computer programs, including one by some of the present authors, have integrated such calculations into CAD systems. l Modeling the flow of metal inside the mold cavity requires other techniques that will handle free-surface transient flow. Stoehr and Hwang reported on the application of the Marker-and-Cell (MAC) technique to mold filling problems at the previous conference in this series at Hennicker, N.H. in 1983. 2 Although considerable progress has been made since that time in improving the accuracy, stability, and convenience of these techniques, the shapes to which they were applied remained limited. Also, physical verification of the results of the calculations was limited to water models.The work to be reported in this paper was undertaken with two objectives: (1) to apply these techniques to more complex and realistic shapes, and (2) to verify the results by comparison with flow patterns observed in molten metal poured into actual sand molds."

Jan 1, 1986

By Z. Sun, H. A. Davies

"Current understanding of the mechanisms by which glassy and crystalline thin ribbons are formed in high speed continuous casting processes is considered. A computer model of the combined heat and momentum transfer during melt spinning has been developed. The influence of roll velocity and of heat transfer coefficient (between the melt and the roll) on the temperature and velocity distributions within the melt puddle is analysed and the thickness of the extracted ribbon computed. The relative importance of heat and momentum transfer under different conditions is assessed and differences between the casting of crystalline and amorphous strip highlighted.IntroductionHigh speed continuous casting of thin metallic ribbon and strip is currently of considerable interest both to metal processers and to those studying process mechanisms. The practical interest arises both from the potential economic benefits to be gained from a shorter and more energy-efficient production route and from the ability to obtain extended and uniform thin strip having a metastable microcrystalline or amorphous structure with enhanced or unique properties. The latter arise from the rapid solidification associated with the cooling of very thin sections of melt brought into intimate thermal contact with one or two rapidly moving heat sinks.The principal processes for casting thin strip include free jet, chill-block melt-spinning (GBMS) (1), melt drag (MD) (2), planar flow casting (PFG) (3) and twin-roll casting (TRG) (4). For the first three of these, all essentially variants of melt-spinning, melt flows through a nozzle onto the circumferential surface of a rapidly rotating roll, whereas for TRG the melt passes into the nip between two contra-rotating rolls. We shall be concerned in this paper with the single-roll melt spinning processes only. For these, a puddle is established where the melt contacts the roll surface and the length of this puddle l is a function of the process conditions, notably the circumferential velocity of the roll Us and the volumetric flow rate of the melt Q. The contact time between any point on the roll surface and this puddle, or residence time eR(= l /Us) determines the thickness of ribbon t extracted from the puddle. This thickness, in turn, partly governs the cooling rate T and the amount of undercooling in the melt and thus the microstructure of the solidified ribbon."

Jan 1, 1986

By M. E. Karabin, R. J. O'Malley

"A mathematical model of the roll casting process has been developed which predicts macro-variables, such as roll separating force, roll torque, and strip exit speed for a range of roll speeds. Continuum variables, such as interface pressure, interface shear stress, freeze-front shape, and internal temperature distribution are also predicted.The model consists of two independent segments: solidification and rolling. The solidification model is two-dimensional (thickness and rolling direction) and is capable of handling alloy solidification. The solidification model is employed until solidification through the thickness is complete, then the rolling solution is activated.The rolling model is a one-dimensional (rolling direction) mechanical solution based on the von Karman approach, coupled with a two-dimensional thermal solution. The flow stress is expressed as a function of the average through-thickness temperature of the strip.Experiments were performed using a laboratory roll caster to test and refine the model. Experimental strip exit temperatures were used as input to the model. Experimental and predicted roll torque and strip exit speea were forced into agreement at a single reference roll speed to obtain values for the coefficient of friction and the interfacial heat transfer coefficient. Using these values, good agreement was obtained between predicted and experimental torque and strip exit speed over the entire range of roll speeds examined."

Jan 1, 1986

By Douglas J. Harvey, Chi-Mou Ni

"This paper presents an analysis technique for the solidification of metal castings. This method, based on a modified finite difference scheme with an irregular mesh, is capable of dealing with the complex geometries of castings. The nonlinearities in material properties of specific heat and thermal conductivity, arising from the phase change during solidification, are accounted for in the step-by-step calculations. Two solidification problems are solved with this analysis technique and the calculated results compare satisfactorily with experimental data. IntroductionCasting process design engineers are often faced with the difficult problem of designing castings to be produced porosity-free. Since the solidification of metal is affected by many interacting variables such as the thermal conductivity, density, and specific heat of both the molten metal and the mold, it is easily seen why this is a difficult task.Typically, casting tooling design is an intuitive process which builds on past successful methods. After processing equipment is constructed, trial castings are made and examined for dimensions and integrity. Changes in the tooling are generally required to improve casting soundness or to reduce cycle time. In the permanent mold process, tooling is complex and difficult to alter after initial construction. The water-cooled steel molds are part of the complete automated system. Associated holding furnaces, pouring ladles, casting conveyers, and utility connections all make other than minor adjustments a costly and time consuming task. An analytical method for using the principles of heat transfer for predicting solidification patterns would be of great value in the permanent mold process by minimizing costly changes."

Jan 1, 1986

By P. S. Dunn, A. D. Rollett, H. D. Lewis

"The results of preliminary work on the modeling and measurement of the heat transfer coefficients of metal/mold interfaces is reported. The system investigated is the casting of uranium in graphite molds. The motivation for the work is primarily to improve the accuracy of process modeling of prototype mold designs at the Los Alamos Foundry. The evolution in design of a suitable mold for unidirectional solidification is described, illustrating the value of simulating mold designs prior to use. Experiment indicated a heat transfer coefficient of 2 kW/m /K both with and without superheat. It was possible to distinguish between solidification due to the mold and that due to radiative heat loss. This permitted an experimental estimate of the emissivity, ~ = 0.2, of the solidified metal.IntroductionA feature that distinguishes casting processes from welding processes is the presence of an interface with a low heat transfer coefficient between the mold and the casting. Especially when chill casting, the interface delays the start of solidification and increases freezing times significantly. By comparison welds can be assumed to have a very high heat transfer coefficient. The heat transfer coefficient controls such phenomena as folds and laps on the surface of castings which can be crucial to the acceptability of a casting. The time delay for the start of solidification influences the occurrence of cold shut defects. The standard mold material for uranium casting is high quality graphite which is an excellent refractory when used in vacuum and coated with a refractory wash. Graphite is an effective chill material and has a thermal diffusivity approximately twice that of uranium. Experience has shown that practicable superheat without mold pre-heat is not sufficient for avoiding cold shut defects when casting in graphite molds.TheoryIt is possible to calculate the position of the freeze front s, as a function of time, t, the superheat and interface heat transfer coefficient, h, [1]. However, the value of h cannot be obtained explicitly. The Virtual Adjunct Method (VAM) [2] offers a method for extending the standard equation for an insulating mold where s is proportional to the square root of time. By considering the effect of the interface to be equivalent to finite thicknesses of mold and metal, the following quadratic equation is obtained,"

Jan 1, 1986

By Y. H. Wang, S. Kou

"Three-dimensional weld pool convection was computer simulated using dimensionless variables. The electromagnetic force, the buoyancy force and the surface tension gradient on the pool surface were treated through the use of the dimensionless electromagnetic force, the Grashof number and the Marangoni number. Other typical dimensionless numbers such as the Prandtl number and the Peclet number were also used. The calculated and experimentally observed fusion boundaries were compared. The convection patterns were shown and the effect of the Marangoni number was demonstrated. IntroductionSeveral weld pool convection models have been developed by Oreper et al. (1,2), Chan et al. (3) and Kou et al. (4). These models describe two dimensional convection in stationary weld pools in a semi-infinite workpiece. Since the vast majority of arc welding practices involve a moving heat source and weld pool, and the workpieces are finite in thickness rather than semi -infinite, a 3-D convection model for a moving heat source and a workpiece of finite thickness is expected to be much more useful.Weld pool convection can strongly affect the structure and properties of the resultant welds. Variations in weld characteristics which are likely to occur from changes in weld pool convection are: weld penetration, macrosegregation, gas porosity, solidification structure, undercutting and surface smoothness. Apparently, the quantitative understanding of convection and heat flow in weld pools is of considerable practical interest. Mathematical models describing convection and heat flow in weld pools are essential, not only because of the quantitative understanding they provide, but also because of the difficulties associated with the experimental measurements of weld pool convection. These difficulties are mainly due to the fact that only surface flow, rather than the overall convection pattern in the weld pool, can be observed. In fact, due to arc glow even the observation of surface flow can be rather inconvenient."

Jan 1, 1986

By David E. Hardt

"One of the critical variables to be controlled during arc welding is the size or geometry of the weld pool. The measurement of pool geometry has become an increasingly important problem with the advance of welding automation and control, and this paper addresses that problem. The basic concept of pool size estimation via measurement of pool dynamics is presented. The theory underlying this concept is discussed and is followed by a review of the current work in this area. It is concluded that pool dynamics can be characterized by natural frequencies of oscillation, for stationary welds, and that these can be measured by appropriate signal processing performed on the arc voltage. However, it appears that only full penetration welds show a strong correlation between weld size and natural frequency. Also, the extension of these results to the realistic case of moving welds has yet to be considered in detail. Nonetheless, this method of in-process measurement for real-time weld process control is still quite attractive where applicable since it can be accomplished simply by modifying the arc current schedule and by signal processing of the arc voltage.INTRODUCTUIONThe study of welding process control has been dominated by research into methods of in processing sensing of the weld pool geometry. While it has been shown feasible to measure the topside pool geometry using video methods [1,2], the penetration or backside bead width have remained elusive quantities. This is particularly true if a topside sensor location is required.The concept of using weld pool motion as a pool geometry sensing method was proposed by Hardt et al. [3] and later demonstrated by Hardt and Zacksenhouse [5] and Richardson and Renwick [6]. The concept is based on the fluid dynamics of a pool constrained by a solid container and by significant surface tension forces (i.e. a small shallow pool, not a ""pond"" or ""lake""). Such a pool, if subjected to external forces, will exhibit a surface motion is a function of that force, the properties of the fluid, the surface tension, and the shape of the container. Thus, if this motion can be excited and measured, and the relationship to the pool geometry determined, a means of sensing pool shape will exist."

Jan 1, 1986

By John T. Berry, Robert D. Pehlke, Michael J. Beffel

"Pattern making and metal casting organizations are currently adopting CAD/CAM systems and implementing this technology in design and production. Al though significant progress has been reported, these systems generally are based on two or two and one half dimensional analyses. Very few are based on solid geometric modelers.The present paper emphasizes the need for solid geometric modeling of the casting and rigging system and linkage with a solidification simulator. The nature of some of the technological challenges which have slowed this development are discussed, including accurate description of complex surfaces and their intersections and the bridging of computer encoded information with NC technologies and with FEM-based simulators for stress and thermal analyses. A recently prepared modular-structured FEM simulation code implemented on a micro computer is described.Construction of a solid geometric model of the casting and rigging and validation of the combination for optimum characteristics of soundness, quality and yield by solidification simulation are I inked as key elements in the foundry of the future. Data base support for this CAD/CAM system is rapidly being developed throughout the world.INTRODUCTIONThe simulation of shaped casting solidification using the digital computer, goes back to at least the early nineteen-sixties when Forsund (1) in Denmark applied the technique to the problems of casting surface quality.Prior to that time, the work of Paschkis and co-workers using the analog computer ranges back to the nineteen-forties, when work sponsored by the AFS Heat Transfer Committee at Columbia University revealed many hitherto unknown or only partly anticipated aspects of casting solidification (2).Three reviews of the various historical aspects of simulation should be referenced. Firstly, the AFS monograph by one of the present writers (RDP) and his colleagues - describing background and later work, again sponsored by the AFS Heat Transfer Committee at the University of Michigan (3). Secondly, a review prepared by Durham and Berry (4) which discusses broad aspects of casting solidification simulation. Thirdly, a review by Erickson (5) which contains an excellent bibliography to 1980. In addition to these papers, the publications lists of the CADCAST project* and the proceedings of the two Engineering Foundation Conferences of Casting and Welding Modeling, published by ASME, are also most notable (6) •"

Jan 1, 1986

By R. Hamar, G. Lamanthe

"To realize complex shaped pieces, like turbine blades, superalloys are cast with the lost-wax process. The end properties of these parts depend upon their microstructure as well as upon the occurence of casting defects such as micropores and hot tears. Such features are governed by the local solidification conditions which may be describe by the calculation of thermal transfers between alloy and mould. It needs the evaluation of numerous parameters which are fitted from experimental measurements in real castings. In further simulations, any parameter may be changed, in order to determine its influence. Such a single modification is quite impossible to realize in real castings, which shows that numerical modeling is a very good way to study casting processes.This modeling technique of casting is applied here to the real study of a typical low pressure turbine blade of a civil engine.IntroductionAircraft engine manufacturers use superalloys to produce turbine blades. They have to obtain a given microstructure and to avoid solidification defects to give good mechanical properties to theirs castings. For example, microporosity is induced by a large shrinkage that cannot be completely fed with the remaining liquid alloy .The optimization of casting processes needs a good Knowledge of the solidification process, that must be controlled at every place in the casting, and at every time of its freezing. Material and heat transfers that occur during the solidification, can be estimated from rather complex calculations. To model these physical phenomena, the evaluation of many parameters must be made. In this aim, some casting experiments are fitted with numerous thermal captors. The identification of experimental and computed thermograms let us estimate these parameters.A long-range scientific cooperation between SNECMA and INPG has enabled the realization of such studies on this topic. From all of them, this paper describes an experimental and computed investigation which has been led on a typical low pressure turbine blade of a civil engine."

Jan 1, 1986