In metallurgical processes, such as the spray-forming of metal powders, or of near net-shape products, it is often necessary to control the flow of a jet of liquid metal through a small diameter nozzle, from a crucible in which the alloy is melted. In many cases the jet will be of the order of 5mm diameter. Control by conventional means is difficult, because of abrasion and clogging of the nozzle, instability of the jet, and contamination of the liquid metal stream by material picked up from the nozzle wall. It is advantageous to control the diameter of the jet stream, not only by so doing to control the flow rate, but also to keep the jet stream clear of the nozzle wall and so prevent contamination of the liquid metal and wear of the nozzle. Moreover, the crucible in which the metal is melted is often a water-cooled"cold crucible"in which the liquid metal is in contact not with the wall of the crucible, but only with a"skull"of solidified metal which lines the crucible wall, thus avoiding contamination. It is therefore advantageous to be able to define separation point of the jet accurately, so that it separates from the skull and does not touch the wall of the nozzle at all. If this can be achieved it can then be possible to dispense with the nozzle completely.

Various methods are known for the non-intrusive flow control of liquid metals, of which the simplest is the electromagnetic pinch valve. This comprises a solenoid surrounding the nozzle or pouring tube. A high frequency current through the solenoid generates an axial magnetic field which exerts a radial force on the liquid metal and so"squeezes"the stream to a smaller diameter, throttling its flow and also avoiding contact between the jet and the

nozzle wall. In order for the valve to work it is necessary for the operating frequency to be sufficiently high that the magnetic"skin depth"5 in the liquid metal is small compared with the jet diameter. The"magnetic pressure"tending to constrict the jet is then proportional to the square of the magnetic field strength at the surface of the jet. This is independent of the solenoid diameter if the solenoid diameter is much larger than the jet diameter. However, if the solenoid diameter is only a little larger than the jet diameter, the magnetic field at the jet surface is inversely proportional to the gap between the solenoid and the jet surface. It is thus advantageous to use a solenoid of as small a diameter as possible, since this will create a greater force and use less power. In practice, however, the construction of a very small diameter solenoid presents difficulties because of the need to cool the windings, which are generally constructed from copper tube.

It is also known to control flow by inserting a segmented, hollow copper cylinder with parallel inner and outer walls, between the solenoid and the jet, so allowing a larger diameter solenoid to be used. The principle is identical to that of the well-known"cold crucible". Each segment of the cylinder is electrically insulted from its neighbours, so that eddy currents induced on the outer surface are forced to recirculate within each segment. In this way, the current induced on the outer surface is transferred to the inner surface. The radius of the inner surface of the cylinder can be made as small as may be required, and the current flowing in this inner surface acts to compress the jet just as if the jet were surrounded by a solenoid of the same diameter as the inner surface of the cylinder.

This method has the disadvantage that it provides an essentially uniform magnetic field whose intensity reduces in a gradual manner at the inlet to the nozzle. Because of this, the point of separation of the jet stream from the

nozzle wall is ill-defined and may result in unpredictable control of flow, as well as allowing contamination through contact of the jet stream with the nozzle before separation. The method also has the disadvantage that the current density on the inside surface of the cylinder is essentially the same as the current density on the outside surface of the cylinder. If the current density on some portion of the inside surface could be substantially increased compared with the current density in the outside surface, the local field strength would also be substantially increased, so enabling the liquid metal jet to be more easily and more accurately controlled.

In one aspect the invention provides a valve for controlling the flow of an electrically conductive fluid, the valve comprising: a segmented cylinder formed from an electrically conductive material; and an electro-magnetic force generating solenoid surrounding the outer periphery of the segmented cylinder, wherein the inner wall of the cylinder tapers outwardly from an apex toward each end of the cylinder.

The invention provides a non-intrusive control valve for liquid metal, whose geometry is such that an intense magnetic field is concentrated at a well-defined axial location, so determining a well-defined separation point for the liquid metal from the nozzle wall, as well as allowing a greater degree of flow control for a given power input.

There may be at least four segments.

Although the liquid could be any electrically- conductive liquid-an ionic aqueous liquid, for example, or mercury-the primary purpose of the invention is to control the flow of jets of liquids which are molten metals, and in particular non-ferrous metals such as titanium, nickel alloys or aluminium in metallurgical processes such as the spray-forming of metal powders or of near net-shape products.

One example of the present invention will now be described with reference to the accompanying drawings, in which: Figure 1 is a side-cross sectional view of a valve according to the present invention; Figure 2 is a plan-cross sectional view of the valve of figure 1; Figure 3 is a schematic view of a segment employed in the valve of figures 1 and 2; Figure 4 is a plan and side-cross sectional view of a valve according to the present invention employing a cooling mechanism; and Figure 5 is a plan and side-cross sectional view of a further example of the present invention.

Referring to figure 1, a valve 6 according to the present invention is positioned, in use, below a crucible 1 containing an electrically conductive 2 liquid. The crucible 1 has a nozzle 4 through which the liquid 2 passes in use, forming a jet 3. As will be described below, the jet 3 is controlled by the valve 6.

The valve 6 is surrounded by a solenoid 5 which, in use, introduces a magnetic field into segments 8 which are arranged to define a cylinder. Each segment 8 has a hollow core 10 and an inner apex 9 which defines a knife-edge 9 forming part of the inner cylindrical core 7.

Figure 2 shows a plan view of the valve 6 in which the arrangement of the solenoid 5 and segments 8 can be seen more clearly.

The degree of flow control achieved naturally depends on the strength of the magnetic field generated by the valve 6 acting on the liquid jet 3. The averaged magnetic pressure is p is given by p = (Bo2/48) [1- (Bl/Bo) 2] where y=4n x 10-7, B1 is the peak axial magnetic field on the axis of the jet and Bo is the peak axial field at the surface of the jet. For r/5 z 2.25, B1 approximates to

zero. In most applications, a value for B1 between 0.1 and 0.5 Tesla is likely to be sufficient.

In use the current induced by the solenoid 5 in the outer surface of each segment 8 is forced to flow circumferentially around the knife-edge 9 created by the intersection of the upper and lower inner conical faces of segments 8, and in so doing the current density in this region is increased substantially compared with the current density on the outer surface of each segment 8. Moreover, the geometry of the knife-edge 9 itself results in a concentration of the magnetic field in this region, whatever the strength of the currents flowing in it.

It will be understood that as well as the intense magnetic field in the region of the knife-edge 9, there will also be a lower, although still substantial field extending some way along the lower cone axis, and that this will have the effect of stabilising the jet 3, since any movement of the jet 3 towards the wall of the lower cone will result in increased magnetic pressure on that side which will tend to return the jet to its central position.

Although in principle the upper and lower cone angles of the inner wall might take almost any value, it is preferred that the included angle at the cone intersection, i. e. the angle at the apex of the triangular cross-section of each segment, should lie in the range 85° to 130°, and that the upper cone angle should lie in the range 80° to 120°.

In this example the apparatus of the invention includes means (not shown) for creating an axial field around the liquid jet 3 whose frequency is such that the corresponding magnetic skin depth in the electrically- conducting liquid is less that the radius of the jet.

Although any frequency higher than this might be employed, the preferred frequency is such that r/5 is about 2.25, where r is the radius of the jet of conductive liquid and 5= (2/# =4 X 10-7, s iS the electrical conductivity of the liquid, and X is the angular frequency of the

magnetic field. This is because the choice of this frequency minimises the energy lost as heat to the valve 6.

Apart from the obvious economic benefit, this optimisation is usually essential on account of the high power densities involved. For titanium jets for example, of the order of r=5mm, this implies an operating frequency of around 100kHz.

The valve 6 is divided into segments so that the electrical currents induced in the outer circumferential wall cannot flow completely around this outer circumference but are forced to flow through the knife-edge 9 whose position is defined by the intersection of the upper and lower conical inner walls of the segments 8. The segments 8 need not be separated completely from each other provided that they are partially separated in such a way that the length of the current path around the outer circumference of the cylinder is substantially greater than the length of each current path around the knife-edge 9 (see figure 5).

The advantage of not separating the segments 8 completely from each other lies in the possibility of a monolithic construction for the valve 6, which avoids the need for assembly of individual segments with insulating spacers.

It will be understood that the high electrical current densities in the valve 6 will cause substantial heating of the valve material, and that the heat generated is most conveniently removed by water-cooling. Figure 4 shows an example cooling arrangement. In this figure components corresponding to these shown in the earlier figures are numbered identically. Since the valve segments are hollow, it is convenient to water-cool them by allowing water to flow through the hollow segments themselves. It is preferred that the inlet and outlet ports for the water- cooling be electrically insulated from the valve segments, since diversionary current paths would otherwise be created which would reduce the current flowing around the knife- edge defined by the intersection of the conical inner walls of the segments.

Segments of a valve apparatus of the invention are each pierced by a cylindrical aperture 15 which is fitted with a cooling water inlet 6 and a cooling water outlet 17.

The water inlet and outlet of each segment are insulated electrically from the body of the valve 6 by insulating plugs 18 and 19. The valve body is again positioned within a solenoid, which is not shown in this diagram.

The example valve shown in Figure 5 comprises a monolithic hollow copper cylinder with conical inner walls substantially as in Figure 1. Corresponding components are numbered identically. The valve body is positioned within a solenoid which is not shown in the diagram. The cylinder is partially divided into six segments by radial saw-cuts 21 to 26. Saw-cuts 21,23 and 25 start at the top face of the cylinder and are cut to 90% of the depth of the cylinder. Saw-cuts 22,24 and 26 start at the bottom face of the cylinder and are cut to 90% of the height of the cylinder. Thus, none of the segments is completely separated from its neighbour. However, for electrical currents induced by the solenoid to circulate completely around the outer circumference of the cylinder they must follow the circuitous path indicated by the arrows 30.

This path presents a substantially higher resistance than the paths which pass through the knife-edge of the valve 1 on its inner circumference, so that the currents will preferentially flow around the knife-edge rather than around the outer circumference The main features and advantages of the valve of the invention can perhaps be summarised as follows. Firstly, it is a fully proportional control valve in which a radial compressive force is applied to a jet of liquid metal, so changing the diameter of the jet and hence the flow rate of the liquid metal. Secondly, by compression the liquid metal jet it causes the liquid metal jet to separate from the walls of its container or the associated pouring nozzle and so reduces the possibility of contamination of the liquid metal. Thirdly, it enables the separation point of the jet

from the walls of its container or associated pouring nozzle to be accurately defined, so increasing the accuracy of flow control and further reducing the risk of contamination. Fourthly, it enables the electrical currents induced in the valve by the surrounding solenoid to be concentrated in a narrow well-defined region of the inner circumference of the valve, so increasing the current density in that region and increasing the local magnetic field strength and hence the radial force on the liquid metal jet. Furthermore, by creating a magnetic field which extends axially below the intersection of the inner conical walls it stabilises the jet in this region.