Experimental Identification of Local Mold Pouring Times and Its Using

Experimental identification of local mold pouring times and its using
for validation of filling modelisation for complicated Al-Si die castings

Zenon IGNASZAK

PoznanUniversity of Technology, POZNAN, Poland

Stanisław ZIĘTKIEWICZ, Andreas RIEBANDT Jerzy MIELCARSKI ,Jacek PIASECKI, Piotr BARTKOWSKI

Volkswagen Foundry, POZNAN, Poland

Abstract. Confidence to the forecasting results is conditioned by the state of validation of the models used for purposes of the simulation system. The validation includes as well the time of the mold cavity filling period. In the paper the work including methodology and results of an excerpt from a broad research refering to experimental validation processes of the simulation software codes, is described. The chosen excerpt of the researches refers to measuring partial times of filling particular sections of the mold, that were obtained in the result of a closure of an electric circuit by the liquid metal in the system of contacts located in definite locations inside the mold. Thanks to the test we can check to what degree the former tests and the conclusions drawn there from the given castings have a universal character, or what more accurate statements or even more important modifications of particular pre-processing parameters during preparations to the simulation process should be necessary. The result of the simulation has been considered as satisfactory as compared to real pouring process

Introduction

The requirements related to safety, protection of the environment, and reduction of manufacturing and exploitation costs of complicated technological products induce a new dimension of quality of the modern technologies, particularly the ones related to the automotive industry. Many components of the devices used in the industry, inclusive of complicated motor heads, are manufactured by casting. The heads are produced, among others, of Al-Si-Cu alloys in metal molds (dies). The casting technology is subject to optimization in order to attain the required quality, by application of professional simulation systems. Such an approach makes a standard of technological offices of the foundries. Availability of multi-processor computers and development of numerical techniques enable enlarging product engineering modifications and achieving their effective optimization in a virtual dimension that may be transferred to actual realization. The optimization quality criteria are based on the expectations with regard to map of local casting properties, related to the structure parameters and their compactness. Gradient of the properties being the optimization criterion depends on the casting destination and the conditions of its operation. Confidence to the forecasting results is conditioned by the state of validation of the models used for purposes of the simulation system. The validation includes as well the time of the mold cavity filling period. Unfortunately, commonly known difficulties related to obtaining sufficiently accurate empirical data often preclude full validation of the system.

It should be mentioned that virtualization of the process should allow avoiding the errors that may be eliminated on the most easy and simple manner just at this stage of the procedure. Omitted errors generate, in practice, considerable financial losses once the product is put into production. Therefore, the need of validation of the casting quality forecast system becomes evident. Such a simulation tool allows the designer to decide on possible corrections and production engineering (process) optimization. The present paper represents the methods and measurement results of local (partial) values of mold pouring filling time under actual industrial conditions in order to apply the experimental results for purposes of checking correctness of modeling the mould filling process.

Short characteristics of the filling model

The process of mold cavity filling is modeled based on the mass conservation equation.,

- where - are mass flow rate through the j-th face of the cell (control volume)

The full continuity equation may be formulated as follows:

,

where - are the axes of the Cartesian system

ρ – density of the fluid

u1, u2, u3 – Cartesian components of velocity vector

m – mass

For the incompressible liquids when :

- is typical simplified equation of mass conservation.

The Navier-Stokes equation may be balanced as follows: the accumulation term + convection term = the term related to pressure force+ gravity force term + the main diffusion term. Hence:

where:

- the term of additional diffusion force

- additional forces (not divergence free)

- the term of special external forces

- absolute viscosity or - kinematic viscosity

A simplified form of the equation makes a basis for the model used for simulation of the mold cavity filling process and waspresented in [7]. The problem that should be solved in the model consists in consideration of the relationship between viscosity of the liquid alloy and the temperature. The coupling with the temperature field below the transformation temperature (liquidus) requires enlarging of the notion of the Newtonian liquid to the liquid-crystals suspension. The simulation systems, e.g. Magmasoft or NovaFlow&Solid linearize dependence of the pseudo-viscosity below liquidus (c.f. Fig. 1) without defining the so-called castability limit. This affects forecasting of the cast faults, i.e. a so-called misrun, consisting in solidification of the flowing metal before entire filling of the mold cavity. Even without so drastic stoppage of the flow a so-called cold lap may occur at the contact surface of two metal fluxes, the fronts of which contain inclusions and oxides. This causes leakage in these locations of the cast. Mechanisms of the phenomena are not provided by the model and, therefore, forecasting of such a condition may be only indirect. Hence, the optimization stage of the mold gating system should be improved by simulation of its filling process. Taking into account the above mentioned simplification of the model with regard to theoretical and practical knowledge on actual course of the phenomena, the experimental validation becomes “condicio sine qua non” for the expected success.

Fig. 1. Viscosity as function of temperature in database of Magmasoft system [1]

(without distinguishing of real and substitutional viscosity range).

The essence of the experimental validation

It should be noticed that, according to the proceeding principles, the validation tests should be referenced to the given casting, particularly the casting of highly composed structure (configuration and variable wall thickness). Such phenomena as physical-chemical condition of liquid metal, varying metal friction against the mold walls, secondary oxidation of the metal stream and surface, turbulences and agglomeration of inclusions, dynamics of temperature field and its effect on real or apparent castability, should be repeatable. All these items evade the control during simulation of the process, according to the casting (mold cavity) design. Comparison of similar validation tests [2] carried out in recent years for another die casting (also a cylinder head) and other cast alloy confirm the above. This confirms the need of execution of validation tests for every new casting with a view to check to what degree the former search and conclusions drawn for some definite castings are of universal character and whether further more precise statement for the pre-processing parameters are necessary.

The need for the research results from the publication [5] based on the experience of the NDCS Company ((Nemak Dynamic Casting System) Nemak is the largest independent manufacturer of cast cylinder heads worldwide with three development sites (Monterrey/Mexico, Windsor/Canada and Wemigerode/Germany). In collaboration with the two America based sites, a new dynamic casting method was developed at Wernigerode- building on many years of experience in the production of high performance cylinder heads. The new casting technique is particularly suitable for the production of heavy loaded Diesel cylinder heads. In this project Nemac Europe also continued the close relationship with Ottovon-GuerickeUniversity of Magdeburg, a relationship established over few years ago by Rautenbach-Guss Wernigerode GmbH [5].

Measurement of partial times of die filling with a liquid alloy. Methodical requirements and their realization

Information on partial times of filling of particular casting cross sections is a parameter the recording of which is relatively the easiest. The method consisting in closure of an electric circuit by the liquid metal in the system of contacts located in definite locations inside the mold is known and widely used since years [cf. for large casting – Z. Ignaszak in FC], The zero-one (0-1) response obtained in case of possible turbulences (stream retreat) may be incomplete when the contact points are located near to or directly on the mold surface. The assumptions formulated for a new method of intended research include some conditions:

• the time required for the liquid metal to attain an a-priori level is to be measured with contact probes in the middle of the wall thickness;
• the test casting meeting such a condition should be only designed for destructive and non-destructive testings (is considered as a non-sold casting):
• in the contact points the temperature probes (thermocouples) should be located as well, distinguished by possible low time constant (possible to be determined);
• consideration of the effect of radiation from the free metal surface on the temperature measured by the probe, still before the direct contact;
• taking into account the number of the measurement points inside the cavity of the cylinder head die and considering the need to carry out the measures within the manufacturing cycle (without disturbing the cycle of die temperature variations) the set for assembling the contact and temperature probes should ensure measuring availability within 30 seconds (since removal of the previous casting).

A special tooling for accurate location of the probe inside the mold cavity had to be designed and executed.In order to check actual location of the probes the test castings should be examined with X-ray or subjected to destructive tests (by milling). The located measurement points are introduced to a 3D model representing a CAD geometry of the casting-die system. The mold pouring process is then simulated with the simulation system, thus answering the question of partial times of particular filling stages (in the points defined based on the experiment). As a basic simulation tool is used the Magmasoft system implemented in VW Foundry Poznań.

Initial conception of the test consisted in making grooves in the core surface (Fig. 2), where copper wires protected with a special matter of thermal and electric insulation properties should be placed. The ends of the wires connected with appropriate measurement system serve then as the contact probes. Conception of the measurement consists in recording the 0-1 electric signals generated by closing the circuit by the liquid alloy in the measurement points.

Fig. 2. Core of motor-head water jacket made of Croning sand. Part of core with probe placed in especially groove (first idea) [1].

Nevertheless, after the preliminary tests the above conception has not been accepted. Therefore, a new device for accurate probe location inside the mold cavity has been designed and effected. A special support with a centring sleeves has been designed for this purpose. It is located on the core and, thanks to such a solution, the probes (probes) are precisely located in selected measurement points (Fig. 3).

The course and results of the tests

Preparation to tests consisted in positioning of the support in the main cylinder head core (Fig. 3) and assembling of the probes connected to the data acquisition computer logger. The points have been chosen based on analysis of the needs resulting from identification of the flow of local metal streams inside the die cavity (4 and 5).

Fig. 3. Core model with support steel for probe clamping (by CAD Proengineer system) [1].

Fig. 4. Scheduled positions of contacts and thermoelements in motor head casting (black points), top view [1].

Fig. 5. Scheduled positions of contacts and termoelements in motor head casting (black point), sectional view [1].

Among the contact-temperature probes tested at this stage (Fig. 6) the jacket thermocouples connected into a bundle have been chosen. Their jacket made of heat resisting steel allowed for using it as an electric contact in the independent circuits signaling the filling level. Figure 7 shows the stage after removing the casting from the die.

Fig. 6. Conceptions of the contact-temperature probes tested during the investigation: a – combination of the K-type thermocouples with a weld and no weld ones (the third metal principle); b – a bundle of K-type thermocouples (all with welds); c – a bundle of commercial jacket K-type thermocouples (the jacket used as an electric contact); d – view of a bundle (the case b) assembled in the core [4].

Fig. 7. View of motor head casting with core after shaking-out from die, with equipment to signal contact and temperature monitoring during pouring (right- detail of measurement bundle type c-fig.6, con- contact, Th- thermoelement plug) [4].

The tests have been carried out for two filling types: into a stationary die (without a tilt) and into a preliminarily tilted die (the TILT method, Fig. 8).

Fig. 8. The TILT casting method and its realization during the tests – the poured die is tilted (the die tooling visible) [4] and [5].

Aspects of the contact-temperature probes have been recorded by means of the data acquisition computer logger. The data transferred to Excel after reformatting of the files enable their further transformation, processing, and presentation in the form of diagrams.

The simulation tests have been carried out too (with the use of Magmasoft system), with consideration of the thermophysical data available in the database. In order to transfer the virtual points corresponding to actual locations of the probe terminals in the CAD geometry and, afterwards, in the discretized casting volume, every real measuring casting has been dissected, and coordinates of these points determined, according to previous plans (Fig. 9).

Fig. 9. Cross section of the cylinder head casting in the location of the probes bundle (the points determine virtual locations of the thermocouples) [1]. The picture shows actual locations of the probes after section milling procedure.

Diagrams made based on experimental and simulation tests are shown below. They include a comparison of the times of the metal level attainment to the probes.

Figure 10 shows the T=f(t) patterns and the 0-1 signals defining the times of attainment particular points while pouring the mold under actual conditions and simulations.

Fig. 10. Temperature and probe signal contact as function of time [6].

Table 1. Specification of the metal level attainment time to the probes located in the measurement points in the experiment and the simulation [6].

No and place / Real experiment / Simulation experiment (Magmasoft)
rib / [s] / [s]
1(down) / 7,5 / 6
2(middle) / 12 / 8,8
3(upper-feeder) / 17,5 / 11,8
injector
4(down) / 3 / 1,8
5(middle) / 6 / 4,8
6(upper-feeder) / 21,6 / 16
rib
7(down) / 10 / 7,9
8(middle) / 13 / 10,5
9(upper-feeder) / 17,5 / 13,5

The data specified in Table 1 enable stating that the simulated pouring process undergoes more quickly than the actual one, on the average by 3 seconds. Comparison of particular measurement points allows to state that the least difference occurred in the point the nearest to a so-called zero-plate (the die bottom), amounting on the average below 2 seconds, while in the riser head direction the results were as follows: from 3 until 5 seconds in the riser head. Such a difference in the results might be caused by differentiation of the discretization mesh in particular parts of the casting. The cast riser mesh has been generated with the Magmasoft module named “standard”, while the mesh parameters were, on the average, 5-fold (linearly) larger than in the other part of the cylinder head casting, that has been discretized with the help of the “advanced 2” module, in order to improve accuracy of the simulation calculation related to particularly responsible parts of the casting.

The analysis enabled stating that the discrepancy of local pouring times (speeding up) may be probably caused by “pseudo-pressure” virtual pouring of the die under simulation(Fig. 11) consisting in fictitious pressure occurring after filling the virtual pouring cup. Under real conditions the metal should flow outward the pouring cup. Therefore, the geometry of the gating system has been improved by appropriate enlarging the pouring cup with a view to avoid the phenomenon while simulating the process.

Fig. 11. Filled pouring cup in the second mold pouring stage (pseudo pressure mold pouring, Magmasoft system) [1].

The result after the correction is shown below (Fig. 12). It may be stated that the simulated pouring process has been shortened on the average by 2 seconds. Consideration of the measurement error amounting to 0.5 s, the effect of the mesh size on the simulation results, and an error of real probe location, allows to state that the simulation result is satisfactory and comparable to the real conditions.

Fig. 12.Temperature and probe signal contact as function of time after change of geometrymodel [6].

Summary