Use of Simulated Porosity for Avoidance of Casting Defects
Nutzung simulierter Porosität zur Vermeidung von Gußfehlern
Simulation de la porosité pour éviter les défauts de coulée
Arno Louvo and Matti Sirviö VTT Manufacturing Technology, Finland Presented at World Foundry Conference 1994 in Düsseldorf, Germany
The simulation programs are used to achieve sound, high quality castings. Usually the commercial programs use a material data found in the literature. Consequently the simulation programs can predict temperature distribution or hot spots in the castings.
The materials and their behaviour vary from foundry to foundry. The material database has to be defined with experiments to suit the melt used in the foundry. In order to predict casting defects the thermal properties are defined as a function of the temperature and porosity prediction is simulated based on mass flow above solidus temperature.
In this study it is shown that casting defects in SG iron castings can be prevented by simulation after the correct material properties for eutectic reaction has been found.
Simulationsprogramme werden benutzt, um dichte Gußstücke höchster Qualität herzustellen. Gewöhnlich benutzen kommerzielle Programme Materialdaten aus der Literatur. Konsequenterweise können die Simulationsprogramme Temperaturverteilungen und Wärmezentren vorhersagen. Die Materialien und ihr Verhalten variieren von Gießerei von Gießerei. Die Materialdatenbank muß mittels Experimenten definiert werden, um zur speziellen Schmelze der Gießerei zu passen. Um Gußfehler vorherzusagen werden die thermischen Eigenschaften als temperaturabhängige Funktionen definiert. Porositäten werden aufgrund einer Simulation des Stofftransport oberhalb Solidus vorhersagt.
In dieser Studie wird gezeigt, daß Gußfehler verhindert werden können, wenn in jeder Gießerei die eigene eindeutige eutektische Reaktion simuliert wird.
Des programmes de simulation sont utilisés pour obtenir une coulée sans defauts, de haute qualité. Generalement les programmes, que l'on trouve dans le commerce, utilisent un ensemble de données tirées de la littérature. Par conséquent à l'aide du procédé de simulation on peut prévoir les variations de températures ou de "hot spots" dans la coulée. Les materieux ainsi que leurs propriétés diffèrent de fonderie en fonderie. Ainsi, si l'on veut une masse de coulée convenable, il est nécéssaire de définir les données de base des matériaux à l'aide d'expériences de simulations. Pour prévoir les défauts de coulées; les propriétés thérmiques sont à définir comme fonction entre la température et la porosité simulée dans la coulée de la masse au dessus de la température solidus.
Dans cette étude on a pu démontrer, que les défauts de la coulée ne peuvent étre éviter, qu'uniquement si la réaction eutectique est simulée dans chaque fonderie.
Foundrymen are using the simulation programs to achieve sound, high-quality castings. The main interest for the foundrymen is not merely to see temperature distributions or the hot spots in castings. The programs should also predict the solidification shrinkage and porosity formed during solidification.
Despite the enormous progress in algorithms based on both finite element and control volume or finite volume approaches, most programs still use empirical formulae based on simulation results to predict defects in castings. Very little or no effort is done to predict casting defects directly as a consequence of the physical phenomena modelled by the algorithms in the software /1/.
Moreover, the engineers and casting designers require a user environment that communicates with them rather than some computer jargon /2/. Programs should be provided with good visualization technology to enable the simulation results to be explored easily /1/.
The commercial programs still seem to be too slow for use as everyday design tools in foundries. Most of them run on massive computer systems. Programs are often too tricky to use. In addition, the value of mere macroscopic solidification analysis is quite limited /3/.
One important goal was then to develop a code that could solve 3D problems on personal computers reasonably fast /3/. The aim of this study is to show that the computer program can be used as an efficient tool in foundries for predicting casting defects when the modelled mushy zone is set in close agreement with the melt used in the particular foundry /4/. The program calculates most of the post-processing information during the simulation and stores it in the form of movie files for time-based animation and tomography presentation /3,5/.
2. Demands of Industrial Users
As simulation software becomes accessible to foundries, it is increasingly used to improve castings already in production. Typically a foundry strives to reduce or eliminate the existing defects.
On the other hand it would be even more beneficial to avoid the problems in the casting beforehand /3/. Previously this has been difficult for foundries which have many products and small series. The amount of time that was required for the 3D modeling, the simulation response and the visualization of the results has been far too long. Moreover, temperature distribution alone has not been sufficient to predict and classify casting defects. It is quite obvious that foundrymen would prefer to see the locations and forms of possible defects in their design, not the hot spots themselves.
3. The Simulation Program
The simulation program developed by VTT can be considered as comprising two main parts: a casting problem-specific user interface and the general solver to handle the numerical simulation of heat transfer and phase transformation. It includes a common interface for pre- and post processing of problems and it controls the solution of the numerical simulation. /3,5/
The time-consuming modeling of the casting geometry can be avoided by using the existing models in commercial CAD-programs like AutoCAD, DUCT and I-DEAS. The program has an IGES interface which transfers the geometry to the program. On the other hand, the geometry can be created even with a text editor, for example with Microsoft Word or Word Perfect /5/, because the mesh data is stored in the form of ASCII characters, each of which represents a cubic element, e.g. character '1' for casting, '2' for core and '3' for the mold.
The program solves numerically in three dimensions the general heat transfer equation, including the release of latent heat based on the fraction-of-solid curve. For a casting undergoing solidification, if only the heat conduction is considered, the energy equation for a three-dimensional geometry in a Cartesian coordinates is given by
where r is density, k thermal conductivity, cp is specific heat at constant pressure, L the latent heat of the phase transformation and fs is the solid fraction in the mushy zone. The solution is based on the Control Volume Method and the Enthalpy Method. The calculation mesh is divided into independent areas, making it possible in the future to use parallel runs utilizing the power of computers in the network if needed. The number of elements can be significantly reduced by making the mesh rougher in the areas where the temperature gradients are small during the simulation. The heat transfer equations are solved independently on each calculation area. The solver is based on the Gauss-Seidel method. /3/.
The program includes a material database, where the right materials for the particular problem can be chosen /2/. Thermal properties are given as a function of temperature. The driving force for the mass flow of melt is the change of the density as function of temperature and gravity force. The porosity prediction is simulated based on mass flow above solidus temperature. For example, the expansion (because of graphite formation) and shrinkage (because of the formation of austenite) in SG iron can be simulated. Whereas the driving force for mass flow is based on the physical properties of the material, e.g. density curve, which are read into the program from the material database, the program can be tuned up to handle in correct way the very different "process melts" found in the foundries.
With the project database the user can handle the set of subprojects to facilitate the comparison of the different simulation runs. The program calculates most of the postprocessing information during the simulation and stores it in the form of movie files for time-based animation and tomography presentation of, for example, temperature isotherm pictures. Tomography means in this context the movie file including slice-by-slice information in one axis direction /5/. It was found that the 'movie files' format makes it possible to check the results in an efficient way after the simulation. /5,6/
4. Case Study
The simulation program was installed at Valmet Paper Machinery Inc., Rautpohja Foundry. The material database was defined with experiments to suit the melt used in the foundry. The experiments were carried out with step castings and cube tests coupled with various sizes of risers and riser necks. The tests showed that the prediction of the casting defects is depending on the melt used in the foundry. The difference in simulated porosities between two melts is shown in Fig.1.
Figure 1. Comparison of default (in the material property database of the program) and tuned-up material properties of a SG iron. The Fiq. a (left) and b (right) show the casting (a test cube) and feeder head and neck cut along the symmetry axis. Color coding in the fiqure is as follows: light blue = completely solidified, yellow-to-red = mushy zone, dark blue=microporosity, crey=empty cells. a) Simulation based on default material properties.; b) The SG -iron based on the melt of the foundry expanded less and a large area of porosity can be seen inside the cube.
Running wheels have been a problem casting for the foundry for many years. Various systems for casting design were tried and finally after considerable trial and error the foundry was compelled to use the system described in Fig. 2.
The casting system was redesigned with one riser located outside the rim of the running wheel. Then a 1/8 portion of casting was modelled and simulated. The aim of the simulation was to optimize the feeder head and the feeder neck. The results showed that porosity remains at the feeder neck. Then four test castings were made. Two of those were found perfect, but the other two had small areas of porosity leading from the feeder neck to inside the casting.
The thickness of the feeder neck was increased towards the riser and the system was simulated again. Results showed that the porosity had moved towards the risers. The simulated results were confirmed with successful test castings.
As a consequence of the successful simulation, the production cost was lowered and the lead time was shortened by remarkable amount. The savings would have been much larger, if the program could have been used when the casting system was originally designed with trial - error method in the foundry.
The new casting system is described in Fig. 3.
The simulation program was used to solve the solidification of the casting as a three-dimensional problem. The original enmeshment included 297 000 cubic elements. The program made the mesh rougher in the mould where the temperature gradients are small (not near the metal and mould interface) thus reducing the problem size to 100 000 control volumes. The solution time for the solidification was about 200 minutes on a Macintosh Quadra 840 personal computer. Altogether 45 time steps were calculated. The post-processing including tomographs (slice-by-slice animation files) was done during the calculations.
The simulation results are shown in Fig. 4 and Fig. 5.
The solidification sequence in SG irons, as has been normal so far, is not of primary interest. In order to predict casting defects, the thermal properties have to be set as a function of temperature and have to be more accurate than needed in solidification sequence calculations for predicting casting defects. Also more time steps are needed for calculating the porosity more precisely.
The installation of the solidification programs should be carried out together with experiments on the test castings in each foundry. The materials and behaviour vary from foundry to foundry. Very few foundries have a melt corresponding to the data found in the literature.
(1) CROSS, M., Development of Novel Computational Techniques for Next Generation of Software Tools for Casting Simulation. Modelling Casting and Welding Processes VI, Palm Coast, Florida, March 21 - 26, 1993.
(2) LOUVO, A., PELLIKKA, E., Utilization of Macroscopic Solidification Simulation in Optimizing the Feeding System design of Nodular Cast Iron Castings. Modelling Casting and Welding Processes V, Davos, 1990.
(3) SIRVIÖ, M., A Computer Program for Simulation of Solidification in Three Dimensions Based on the Control Volume Method. Diploma Work, Espoo, December 15, 1992.
(4) LOUVO, A., PELLIKKA, E., ALHAINEN, J., EKLUND, P., Criterion Functions based on alloying and cooling rate for simulating the microstructure and mechanical properties of SG iron castings. 95th Annual Meeting of American Foundrymen's Society. Birmingham, 5-9 May, 1991
(5) LOUVO, A., MARTIKAINEN, H., SAARHELO, J., SIRVIÖ, M., Heat Transfer Code for the Simulation of the Casting Process, Techninal Reports, Technical Research Centre of Finland(VTT) , Internal Report, Metallurgy Laboratory, 1993.
(6) LOUVO, A., SIRVIÖ, M., VILPAS, M., Simulation of Weld Solidification in Submerged Arc Welding. Modelling Casting and Welding Processes VI, Palm Coast, Florida, March 21 - 26, 1993.