Thermodynamics of inductor saturation.

As molten iron fills the voids in the refractory lining of an inductor, it forms a saturation network. When this iron penetration gets too close to the bushing joint, a runout can occur.

The design of a channel induction furnace should include every consideration possible to minimize or eliminate inductor runouts. When inductor runouts persist, however, foundrymen must determine: "Is the source of the problem the refractory lining or the equipment?

As pointed out last month in Part 1 of this two-part series, 95% of inductor runouts occur at the bushing joint insulator. What is missing is some means to prevent the molten metal saturation network from coming in contact with and destroying the bushing joint insulator. Thus, the subjects of why the saturation network forms, how the saturation network moves through the lining, and what can be done to keep the saturation network from contracting the bushing joint insulator are important ones to address.

Historically, chromate-bonded castable magnesia refractories and mullite-bonded wet ram alumina refractories were the linings of choice for melting or holding iron in channel furnaces. The life of an inductor lined with these refractories was a function of how long it took the inductor lining to wear or to build-up.

Inductor Lining Wear

Inductor lining wear occurs when the refractory is worn away by chemical reduction, by solutioning in other low melting point oxides and/or by other means. When wear occurs, the channel cross section increases in diameter while the channel loop diameter decreases. As wear takes place, both the power and the current increase for a given applied voltage. When the current exceeds the rated ampacity of the electrical components, the inductor must be shut down and relined.

Channel Plugging

Channel plugging occurs as the result of an oxide build-up on the surface of the refractory surrounding the channel. When plugging occurs, the channel cross section decreases in diameter while the channel loop diameter increases. As plugging takes place, the power and current decrease for a given applied voltage. When the power level falls to the point that the melt rate or the ability to superheat can't be sustained, the inductor must be shut down and relined.

To determine whether the refractory in the inductor was wearing or plugging, some means of evaluating the status of the inductor became necessary. One approach was to monitor changes in the furnace conductance. An increase in the conductance ratio indicated that wear was taking place while a decrease in the conductance ratio indicated that plugging was taking place.

Another approach was to monitor changes in the furnace reactance. A decrease in the reactance ratio indicated that wear was taking place while an increase in the reactance ratio indicated that plugging was taking place. Either of these two methods was intended to provide the user with a tool to monitor the condition of the inductor so that a proper decision could be made as to when to shut down the inductor.

Lining Saturation

Now that dry vibrated linings seem to be the linings of choice, monitoring of the conductance and/or reactance ratios doesn't provide metalcasters with all the information needed to determine the condition of the lining. The problem is that even without any changes in either the conductance ratio or the reactance ratio, an inductor will unsuspectingly run-out. These run-outs occur as a result of refractory saturation by molten iron.

When a refractory lining is installed, not all of the lining volume is occupied by the refractory grain. Some of the lining volume is occupied by air-filled voids. The ratio of the void volume to the total volume is the measure of porosity of the lining material. Since the voids are interconnected, they fill with molten iron and form a saturation network.

Why the saturation network is more pronounced in some refractories than in others is explained as follows. Roughly, the installed porosity is 16-22% for castable refractories; 22-25% for wet ram refractories; and 24-27% for dry vibrated refractories. The higher the porosity, the larger the saturation network.

Since neither the conductance ratio nor the reactance ratio enables the determination of how close the saturation network is to the bushing, a new measuring stick is required. By installing thermocouples in the refractory, the temperature gradient between the bushing and the saturation network can be monitored.

Using a simple clock diagram, the thermocouples are taped to the outside of the bushing at the 3, 6, 9 and 12 o'clock positions. They are then run longitudinally from the edge of the bushing to the center of the bushing, bent 90 [degrees] to protrude 1 in. radially away from the bushing and in line with the center of the channel. When the temperature of any of these thermocouples exceeds 1500F (816c), the inductor should be shut down.

Mechanism of Saturation

Figure 1 shows a vertical section of the FeCSi ternary phase diagram at 2% Si. The initial freezing of a cast iron containing 3% carbon begins at the liquidus temperature of about 2300F (1260C) (point A) and continues to freeze until the beginning of eutectic freezing at 2060F (1127C) (point B). The amount of liquid iron remaining at this temperature is about 70%. Continued freezing then occurs in the eutectic solidification range of 2060-2010F (1127-1099C), with final freezing occurring at the solidus temperature of 2010F (1099C) (point C).

When the refractory lining in an inductor comes into contact with molten iron for the first time, the iron flows into the voids around the refractory grains and continues to do so until the temperature of the leading edge of the molten iron decreases to the solidus temperature of 2010F (1099C), at which it freezes. The temperature profile for a 6 in. thick magnesia lining between the channel and the bushing suggests that the saturation network would be limited to a depth of about 1.66 in. as shown in Fig. 2. However, examination of spent inductors shows that saturation generally occurs to much greater depths.

Thermal Conductivity

While the premise that saturation will continue until the temperature of the leading edge of the saturating iron decreased to the solidus temperature is correct, the data presented in Fig. 2 doesn't address the effect metal saturation has on the thermal gradient. The thermal conductivity for a saturated lining must be higher than that for the refractory, yet lower than that for the molten iron.

The thermal conductivity of the saturated portion of the lining can be estimated as follows. If the refractory has 25% porosity and if the molten iron totally fills these voids, the thermal conductivity of the saturated portion of the lining should be directly related to the partial densities of the two components. The estimated thermal conductivity for a magnesia lining saturated with iron ranges between 124-137 Btu's per square foot, per hour, per degree Fahrenheit, for one inch of thickness.

When the temperature profile is calculated using the thermal conductivity for the saturated portion of the lining in conjunction with the thermal conductivity for the unsaturated portion of the lining, the results are quite different. Figure 3 shows that a 6 in. thick dry vibrated magnesia refractory with 25% porosity will have a calculated depth of the saturation network of 3.2 in., which more closely approximated field observations of spent inductors.

Saturation Heating

To this point, the saturation network has been considered to get its energy solely by thermal conduction from the molten iron in the channel. However, the saturation network is also subject to inductive heating. How this internal heat generation influences the depth of the saturation must be addressed.

To review, the manner in which the channel furnace works is as follows. When the magnetic flux produced by the coil passes through the surrounding channel, it induces a voltage in the channel. Since the channel is a short circuited secondary winding with a low resistance, a large current exists in and heats the metal in the channel.

Likewise, if the magnetic flux produced by the coil passes through a surrounding saturation network, a voltage will be induced in the saturation network. Since the saturation network is also a short circuited secondary winding, but with a high resistance, a small current exists in and heats the metal in the saturation network. Although the internal heat generation in the saturation network is considerably less than the heat generation in the channel, it is sufficient to alter the temperature profile so that metal saturation will proceed deeper into the refractory.

How close to the bushing will the saturation network get before freezing is a function of the heat extraction capabilities of the bushing. For a water-cooled, copper bushing with a 6 in. thick magnesia lining, the depth of the saturation network has been found to be about 4.25 in. In other words, molten iron was within 1.75 in. of the bushing.

Start-up data for a 3000 kW single loop inductor showed that just after the introduction of molten iron, the resistance was 0.093 Ohms, the reactance was 0228 Ohms and the iron temperature was 2450F (1343C). After 72 hours, the resistance was 0.094 Ohms, the reactance was 0.208 Ohms and the iron temperature was 2650F (1454c). While the resistance remained virtually unchanged, the reactance dropped 8.8% during the first three days.

The resistance of the channel is directly related to the resistivity of the molten iron, directly related to length of the current path and inversely related to the area of the current path. In order for the resistance not to change, the channel length had to decrease and/or the channel area had to increase to offset the increase in the resistivity with temperature. Because it is highly unlikely that any channel wear occurred in three days, these conditions are more likely explained by the formation of a saturation network.

The reactance of the channel is directly proportional to the area bounded by the inside of the secondary current path (the internal diameter of the channel) and by the outside of the primary current path (essentially the outside diameter of the power coil). In order for the reactance to decrease, there had to be a slight reduction in this area. Again, because it is highly unlikely that any channel wear occurred in three days, these conditions are more likely explained by the formation of a saturation network.

Since the log data showed little or no change over the next week, it appears that the saturation network formed within three days of start-up. Based on an examination of an inductor that had a premature failure due to a power outage, a full saturation network was found to be in place within three days of start-up. Since the start-up conductance or reactance readings are generally chosen a day or so after the initial log data settles down, the start-up data can represent a fully saturated lining that accounts for reasons runouts occur with no change in the conductance or reactance ratios.

The saturation network freezes symmetrically at some distance around the bushing except at the insulated joint. Since electrically insulating materials are less thermally conductive than the metallic bushing, the thermal gradient in the refractory around the bushing gap is altered such that the saturation network can come in contact with the bushing joint insulator. When it does, it thermally destroys the bushing joint insulator and results in a molten metal runout through the bushing joint as shown in Fig. 4. Statistically speaking, the conventional bushing gap is an area awaiting a molten metal runout. The majority of channel furnace runouts occur at the bushing joint insulator, and occur whether or not the refractory in the inductor is wearing or plugging.

Fortunately, not all linings run-out and not all linings penetrate to the depth as determined by the above calculations. To understand what is occurring, the leading edge of the saturation network must be viewed, When molten metal enters a magnesia refractory, the iron displaces air from the voids around the refractory grains. Since the air contains oxygen, the leading edge of the iron is oxidized to iron oxide.

The iron oxide will go readily into solution in the magnesia lining forming magnesioferrite or magnesiowustite, depending upon the partial pressure of the available oxygen. If either of these phases form in sufficient quantities, they will plug the voids and stop further penetration. However, if the carbon in the iron reduces the newly formed iron oxide, the magnesioferrite or magnesiowustite barrier won't form and metal saturation will continue unimpeded.

Minimizing saturation in refractory linings, especially in dry vibrated linings, requires the development of a reliable slag precursor to seal off the voids. Until this refractory development occurs, an alternate solution is needed to eliminate saturation runouts.

One alternative in heading off the adverse thermal discontinuity due to the bushing joint insulator is water cooling the insulated joint in the bushing as shown in Fig. 5. When heat extraction becomes uniform around the circumference of the bushing, there is no discontinuity in the temperature gradient around the bushing. As pointed in Part 1, inductor runouts have virtually been eliminated after fitting the bushing with a water cooled joint cap.

In Review

Saturation of refractory linings is a thermodynamic process wherein molten iron flows into the voids in the refractory lining to a depth determined by the thermal conductivity of the saturated refractory and the solidus temperature for the iron in question unless a slag precursor seals off the voids thereby preventing further metal penetration.

The thermal discontinuity at the bushing joint insulator encourages deeper saturation in this vicinity of the inductor, accounting for the many inductor runouts. Water cooling the bushing joint can help eliminate bushing joint runouts. Monitoring the temperature gradient with thermocouples imbedded in the refractory provides the operator with much more reliable data than is provided by either conductance or reactance calculations.