1. Magnetic Field Control of the Mould Filling Process of Aluminum Investment Casting
S. Eckert1, V. Galindo1, G. Gerbeth1, W. Witke1,
R. Gerke-Cantow2, H. Nicolai2, U. Steinrücken2
1Department Magnetohydrodynamics, Forschungszentrum Rossendorf
P.O. Box , D Dresden, Germany,
2TITAL Ltd., P.O. Box 1363, D Bestwig, Germany
Sino-German Workshop
on Electromagnetic Processing of Materials
Oct , Shanghai, China

2. Magnetic Field Control of the Mould Filling Process of Aluminum Investment Casting
Overview
Motivation – Aluminum investment casting as example of applied electro-magneto-hydrodynamics
Measuring Techniques in liquid metals
Numerical simulations – “volume of fluid method (VOF)”
Experiments
Conclusions and remarks

3. Aluminum investment casting
Motivation
Aluminum investment casting
casting of complex shapes
casting unit: U-bend with the filling channel
(down sprue) and the mould being the legs
connected by a horizontal channel
the flow is exclusively driven by hydrostatic
pressure
Problem
Too high fluid velocities in the early stage of the pouring process  entrapment of bubbles and impurities  worsening mechanical properties
(first) Solution
Damping of the turbulent flow with a D.C. magnetic field

4. Strategy
Magnetic Field Control of the Mould Filling Process of Aluminum Investment Casting
Numerical simulation
- finite element code FIDAP
Labor experiments with a “cold” liquid metal
- Flow observation, velocity measurements
- Validation of numerical results
Extrapolation to industrial scales
- Numerical simulation
- Magnet system design
Model experiments with GaInSn
Experiments with Aluminum melt under industrial conditions

5. Numerical Simulation (non-dimensional) Governing equations I
momentum conservation: time dependent Navier–Stokes equation with Lorentz force density term
mass conservation:
where is the Reynolds number and is the
electromagnetic interaction parameter is the density, is the dynamic viscosity, is the electric conductivity, B0 is a characteristic magnetic field strength, L is a characteristic length and v0 is a characteristic velocity.

6. Numerical Simulation (non-dimensional) Governing equations II
In the case of a present static magnetic field it is necessary to solve an additional equation for the electric potential
electric charge conservation: taking into account the Ohm‘s
and Kirchhoff‘s law, following equation for the electric potential holds:
boundary conditions: isolating walls, non-slip

7. Material properties – characteristic numbers
Numerical Simulation
Material properties – characteristic numbers
The system is defined through 2 independent non-dimensional characteristic parameters: Re and N
Material properties Al
GaInSn
density ρ
2355
6360
Kg/m³
dynamic viscosity η
1.45
2.164
×10-3 Pa s
electric conductivity σ
3.73
3.2
×106 -1m-1
taking L=0.03 m (channel height), v0=0.5 m/s and B=0.5 T we obtain:
Characteristic parameters Al
GaInSn
Reynolds number Re
24362
44085
interaction parameter N
23.8
7.6

8. Numerical Simulation
Finite element code FIDAP: solution of the Navier-stokes equation for the flow and, in the case of applied static magnetic fields, an additional „species“ equation for the electrical potential Φ
4 nodes (2d) or 8 node (3d) per element, bilinear interpolation
Segregated algorithm for coupled equation system
Volume of fluid (VOF) method for simulation of the filling process – void fraction is a new unknown scalar
2 turbulence models: Prandtl mixing length hypothesis and standard k-ε
Sketch of the volume of fluid (VOF) method
Interface: surface with a given constant value of the void fraction

9. Numerical Simulation Application of a static magnetic field
down sprue
Electromagnetic force density acts as damping force
mould model z y x
pole shoes
Sketch of the casting unit
computational grid

10. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
Attenuation of the maximal velocity value at the beginning of the filling process
t = 0.15 s
B = 0 T
B = 0.5 T
Velocity vector plot on a cut plane in horizontal channel. Vmax=1.25 m/s corresponds to color red

11. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
t = 0.25 s
B = 0 T
B = 0.5 T
Velocity vector plot on a cut plane in horizontal channel. Vmax=1.25 m/s corresponds to color red

12. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
t = 0.5 s
B = 0 T
B = 0.5 T
Velocity vector plot on a cut plane in horizontal channel. Vmax=1.25 m/s corresponds to color red

13. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
t = 0.75 s
B = 0 T
B = 0.5 T
Velocity vector plot on a cut plane in horizontal channel. Vmax=1.25 m/s corresponds to color red

14. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
t = 1 s
B = 0 T
B = 0.5 T
Velocity vector plot on a cut plane in horizontal channel. Vmax=1.25 m/s corresponds to color red

15. 3D-Numerical Simulation with the VOF Method
Application of a static magnetic field
Simulation with the help of the VOF (“volume of fluid”) method; Video 
left: without
right: with magnetic field
video shows the first 2.5 seconds of the filling process
B=0T
B=0.5T

16. Model experiments Perspex model model fluid: GaInSn,
down sprue
model fluid: GaInSn,
liquid at room temperature
D.C. magnetic field:
up to 850 mT
transparent walls
mould model
horizontal channel
magnet pole shoe

17. Measuring techniques Visual Observation
Ultrasound Doppler Velocimetry (UDV): see presentation about measuring techniques
Inductive Flowmeter (IFM)

18. Measuring Techniques Inductive Flowmeter (IFM)
Measurement of the perturbation of the magnetic field by the flow
voltage is proportional to the flow rate
high temporal resolution
can be applied at high temperatures
G: geometry factor

19. Experiments
Velocity measurements obtained at a pouring experiment with
a) InGaSn at room temperature b) AlSi alloy at about 700°C

20. Model experiments: UDV measurements

21. Model Experiments
UDV measurements in the down sprue
B = 0
B = 0.85 T

22. UDV velocity measurement in the model experiment
Model Experiments
UDV velocity measurement in the model experiment
in the down sprue channel
in the horizontal channel
dangerous peak velocity in the early stage removed
velocity fluctuations become smaller

23. Validation of the numerical simulation
Experiments
Validation of the numerical simulation
Comparison of numerical and experimental results regarding the flow rate Q as a function of the magnetic field strength (related to the flow rate obtained at B = 0)

24. Al casting Experiments Casting units evaluation
Visual inspection of the resulting metal surface
UV-light visualization of surface defects
B=0.25 T
B=0.75 T in the first 5 seconds

25. Al casting Experiments Casting units evaluation - Statistics
CNF
6-4
4-4
4-2
3-2
CWF
4-3
4-1
3-3
1-2
FNF
2-2
2-3
FWF
6-2
2-1
1-1
Clear tendency: the d.c. magnetic field always lead to an improved quality of the casting unit with reduced amount of entrapped oxides

26. Conclusions I
Velocity and flow rate measurements are needed for better understanding of the flow phenomena and filling process of the investment casting of Al
Low temperature metallic melt InGaSn has been used for model experiments; key advantage: at this temperature a sufficient number of different measuring techniques are available
Validation of numerical codes using such liquid metal models provides a profound basis for an extrapolation of the numerics to the real scale problem and turned out to be essential for the reliable simulation of the real Al casting process
Main problem in the pouring process: the occurrence of large velocity values at the beginning of the casting processes leads to an accumulated generation of vortices inside the pouring channel 

27. Conclusions II
A high rate of turbulences in the flow is supposed to entail the transported impurities, oxides or gas bubbles from the walls and the free surface into the bulk of the casting patterns. As a result the mechanical properties are deteriorated
The external d.c. magnetic field damps the high flow velocities at the beginning of the pouring process. A significant reduction of the peak velocities, leading to a generation of vortices inside the pouring channel, has been shown by model experiments and numerics, and has been demonstrated in the real Al casting process afterwards
As a important input for the control system, a contact-less flow rate sensor has been developed and successfully applied
The statistics of a multitude of cast units showed a clear tendency of reduced oxide entrapment due to the magnetic field influence

28. Perspective
Next step: linear A.C. traveling field which breaks initially, and pumps at the end  constant flow rate during the whole process
Scheme of the coil system to generate the traveling magnetic field
Induced electromagnetic force density compute with a finite element Maxwell solver OPERA