By Dennis I. Phillips

Each of the common methods for density control in a heavy-media circuit has its advantages and disadvantages and each is affected by bleed and level fluctuations. What effect do upsets to the circuit have on density control and sump level? To find out, the effect of coal feed size changes and feed on/off conditions were modeled and evaluated with respect to media density, sump level, media bleed, and magnetite loss potential.

Raw coal feed is far from a homogenous material since the size consist and coal and rock makeup are constantly varying causing difficulty in maintaining precise control of media density and level. Media density is made up of fine magnetite and water and any variation in the amount of water coming with the coal or from the sprays will affect the volume and media density of a sump and must be removed or added as needed.

Media is removed from the circuit by means of carryover on the coal products to the dilute side if the drain-and-rinse (D&R) screen. Hopefully the magnetite in this carryover is returned from the magnetic separators and helps to maintain the media density if the water is controlled. All of the interactions explored as to their affect on the level and density in a media sump.

Both a heavy media cyclone and a heavy media vessel circuit are affected by these interactions but a heavy media cyclone circuit normally includes the finer coal sizes and is more affected by changes than a vessel circuit. A heavy media cyclone circuit will be discussed here and the reader should know that any conclusions can also be applied to a coarser coal vessel circuit but probably to a lesser degree.

The premise for why a tight density control is so important is based on the idea that if a plant operates at 1.54 specific gravity (SG) for an hour and then at 1.50 SG for the next hour (which is an average of 1.52 SG for two hours) that is not as efficient as operating for two hours at a steady 1.52 SG. Partition curves are used to show the sharpness of separation in a cleaning circuit; the steeper the curve the better the separation. A good separation at one density mixed with another good separation at another density is not equivalent to a good separation at the average of the two densities. This is due to the average of two good partition curves being poor partition curves (See Figure 1). This concept has been discussed many times over the years and enforces the importance of steady density control.

Heavy Media Control
The two things that must be considered in the control of a heavy media circuit are density of the circulating media and the heavy media sump level. There are various methods to control media density and there are various methods to control the heavy media sump level. Density and level controls are often combined in various ways and one may heavily influence the other. The first thing to determine about a heavy media sump, regardless of whether it is for a coarse coal vessel or a heavy media cyclone circuit, is whether the sump level is set to rise or drop. Since the amount of magnetite is generally a fixed amount in a circulating media circuit, a dropping sump level results in a rising density and a rising sump level results in decreasing density.

Intentionally removing media from a heavy-media circuit to the dilute circuit is said to increase the “bleed.” If the intentional  bleed is set to zero and there is no intentional water addition then whether the sump level rises or drops depends on how much water is coming with the raw coal, how much media is adhering to the coal and refuse, and on the location of the dividing point in pans under the D&R screens. If the dividing point is close to the feed, and then a considerable amount of the media will drop through into the dilute portion causing the media sump to drop in level. If however the dividing point is further from the feed end and there is much water coming with raw coal then the sump level may be a rising sump.

A dropping sump usually has an automatic water valve  to maintain level and a rising sump must have a controlled bleed in order to maintain a reasonable level. The density is usually controlled by:
•    adding magnetite;
•    adding a concentrated media from some form of overdense system;
•    indirectly by trying to maintain level in the media sump; or
•    the currently preferred method is by causing  the sump to be in a rising density condition and then to add water to maintain the density.

The latter method has the best response time since it involves adding water to the suction side of the media pump where it is immediately mixed with the media and the density is changed rapidly.

Since a fairly constant sump level results in a relatively stable density much emphasis has historically been placed on sump level control. It is often forgotten that it is the density of the media and not the sump level that determines the separation point and thus the ash and yield of the product. As long as the level is within an operable range the coal doesn’t see any affect from level, only from density. For this reason, the tightest control should be on the density and the level can be more variable as long as it does not drastically affect the cyclone or vessel pressure.

A spreadsheet based model was developed that accounted for all of the input and output streams around a heavy media circuit. Both the magnetite and water flows were calculated for each stream to determine when the circuit was balanced or to calculate the effect of the upsets.

For the heavy media cyclone (HMC) circuit, several parameters were held constant for all of the evaluations. The circuit involved a 250-ton-per-hour (tph) HMC circuit with 6,000 gallons of 1.60 SG media in the operating sump as the base condition. A single 10-ft long x 36-inch diameter magnetic separator is assumed for the dilute magnetite recovery circuit. The media sump pumps to the heavy-media cyclone where the overflow goes onto a single clean coal D&R screen with an underpan having a primary section for return of drained media to the media sump and a secondary section where the rinsed media and the rinse water goes to a dilute sump and is pumped to the magnetic separator. The cyclone underflow goes to the refuse D&R functioning identical to the clean coal screen. The magnetic separator concentrate returns to the media sump and the separator tailings go elsewhere. Gallons and gallons per minute (gpm) are the units of volume and flow used in the tables, but one can see the relationships intended without converting to other units.

Since the raw coal stockpiles are seldom well-blended, there is often a change in the particle sizes of the feed due to a change in feed stock or to reclaiming from different parts of the feed pile where coal segregated itself by size during stockpile buildup. When the feed changes to a finer particle size, most parts of a coal preparation plant are heavily affected. For an HMC circuit, the finer material create more carryover of media to the dilute side of both the clean and refuse D&R screens. This increased fines condition was explored (See Table 1) where a thick seam deep-mined coal of typical hardness with an average grain size of 6.75 mm on the plus 1 mm material is compared to that from the softer deep mined coal with an average grain size of 3.70 mm.

Condition 1 in Table 1 has the circuit balanced at steady state on the coarse coal. Condition 2 represents the change to finer sized coal with no adjustment to the bleed. Since the media carryover increases, the level in the sump drops and density increases to 1.603 after 1 minute and to 1.62 after 5 minutes. Condition 3 shows the bleed must be adjusted from 246 gpm down to 197 gpm to balance the circuit for the finer sized coal. Of special note is that with the increased media carryover with the increase in fines in Condition 2 the amount of magnetite returning with the magnetic separator concentrate increased  from 50 tph to 65 tph.

This circuit had a 10-ft magnetic separator which is only capable of recovering 5.5 tph per foot or 55 tph total. The magnetic separator is overloaded by 10 tph and, while not all of the additional 10 tph of magnetite will be lost, it is probable that much of it will be. During normal operation, this magnetite loss problem would not be noticed until considerable amounts of fine coal  are encountered and at this point magnetite loss may be blamed on difficulty in rinsing rather than also on the overloaded magnetic separator.

An increase in fines normally means an increase un water coming with the coal from the deslime screen into the HMC pump. In Table 1, both sizes of coal had 15% moisture while in Table 2 an 18% moisture is used for the fine coal as it discharges from the deslime screen into the HMC sump. Condition 1 in Table 2 is the initial starting point on coarser coal with the circuit in balance while Condition 2 indicates the status after one minute if finer coal with no adjustments to the bleed. The water in the additional media carryover that leaves the sump is less than the water coming with the coal and returning with the magnetite concentrate. The difference is a gain of 14 gpm of water to the sump. Although not a large amount, it is enough to lower the density by nearly 0.01 in 5 minutes without some other adjustment. Condition 3 shows the bleed is increased to bring the sum back into balance.

This example indicates how a shift in fines or water with the fines can reverse the direction that a media sump is moving. Whereas Table 1 shows an increase in fines can cause a sump to lose level and increase in density, Table 2 shows the same fines can cause the opposite effect on the sump if the water with fines increases even a relatively small amount. In both Conditions 2 and 3, the 10-ft magnetic separator would be overloaded with magnetite.

Density Control & Water
The conditions in Table 3 hold the feed water constant at 15% moisture, but show what can be done with an automatic density control using water on the pump suction. Condition 1 is the balanced circuit on the coarse coal as before and Condition 2 represents the change to finer feed coal with the bleed fixed to that of Condition 1 except that 28 gpm of water is added to the pump suction to maintain a constant density and thus a stable level.

The problem with this is that with additional media carry over of the fines and the high bleed there is too much magnetite going to the magnetic separators. Even if they were oversized to handle the load, it is unnecessary to send that much magnetite to the magnetic separators since over time the small losses add up. Condition 3 indicates the reduced bleeds which then allows the density control water to be reduced to 10 gpm and still maintain density and level.

Feed On/Off & Non-magnetic Materials
Another common source of upset for media systems occurs when the coal feed is turned off temporarily. In Table 4, where Condition 1 is a balanced sump on fine coal and Condition 2 is the sump one minute after the feed has stopped coming into the sump, but the media circuits are still running. Without any adjustments to the bleed, the level will drop fast and the density will increase. The sump volumes in the table list the media volumes only and do not show that the actual sump drops by 320 gallons just from the volume of 250 tph of coal no longer in the circuit (assuming 3,000 gpm of pumped media). Condition 3 has the bleed and density control water to zero. With nothing coming in or out, the sump should balance and remain stable while circulating.

Actual plant performance, however, confirms that Condition 3 is theoretical only and in real plant conditions the density usually drops and the sump level drops or raises but does not remain stable. In actual plant conditions, two things are happening to cause imbalance. For one, the sump level will raise if the spray water on the D&R screen is close enough to the feed end of the screen to go into the drain portion or it will drop if not enough spray water goes into the drain section to replace the media that continues to ride the screen panels onto the rinse section of the screen. Poor screen feed methods or insufficient screening area intensifies this problem and in addition some plants have bleeds that mechanically will not go to zero. Many plants simply do not try to run the systems without coal for any length of time and will adjust things manually to get by during  those few times they go idle.

The other cause of an imbalance is due to the effect of non-magnetics in the circulating media. Coal fines, clays and fine sands will build up in the circulating media and contribute to the density of the other media. Without the fresh coal coming to replenish the non-magnetics, the media bleed, intentional or not, sends media to the magnetic separator where the non-magnetics are removed and only the most pure magnetite returns. The effect of this is the density is significantly lowered over time while a media system is “idling.” Condition 4 is an attempt to simulate an idling condition where media carryover is continued but the amount of magnetite returning is less than the solids in the carryover media, thus allowing the non-magnetics to be removed and the sump level to drop while also lowering the density. This actual situation will vary considerably from plant to plant and should be evaluated specifically at each location.

Changes in coal particle size can cause multiple changes in the heavy-media circuit such as more media carryover and more water into the circuit. The net effect of these changes will sometimes increase the density and sometimes decrease the density. If there is sufficient rinse water then finer grain size may cause the sump density to go high because water is carried over to the dilute but the magnetite returns. If there is insufficient rinse water, the density drops because some magnetite is lost. An increase in water that comes with the fines in the feed from the deslime screen may counter the loss of water to dilute from carryover.

One of the most significant findings of these evaluations is how easy it is to drastically increase the load on the magnetic separators. When either more fines and/or more bleed results in higher magnetic loading on the magnetic separators, it is quite possible that the magnetic load is more than the separator is designed to handle and will result in considerable lost magnetite. This probably happens more often than many plant operators realize and the magnetite losses are blamed on difficulty in rinsing rather than to consider the results of carryover and bleeding. Higher operating media densities require much higher bleeds to remove even more water from the circuit and thus the load going to the magnetic separators is greatly increased.

The simplest density control method is to cause a rising density sump and then to add water to the pump suction automatically. For this to work well without contributing to excess magnetic losses, one must monitor the bleed vs. density control relationship. Positioning the bleed system manually and letting the sump level float provides a tighter density control. If the density control water valve stays at a high percentage of open then the amount of bleed may need to be reduced.

Some plants prefer to maintain the sump level by automatically controlling the bleed for level. The problem with this is that it can result in frequent slugs of magnetite to the magnetic separators and it also works against the density add water system since the increased bleed tries to throw away what the density controller is adding. The two controllers are constantly fighting each other. A media system can be controlled this way and will maintain a level and density, but the density will have more variation than if the sump level is allowed to float within some range.

Some operators like this auto level system since it is less work for them and requires less training, but remember that the density makes the separation and is the most important parameter. If the density is tightly controlled why does it matter if the level varies as long as it does not cause an overflow or drop to a point that cyclone pressure is affected?

It is difficult to determine how much non-magnetics is in the circulating media. How drastically the density drops when the feed is taken off the plant may be an indication of non-magnetics. If there is little or no bleed while operating or there is considerable undersize material coming across the deslime screens, then more bleed is probably needed. If sources of additional water into a media circuit are minimized the only amount of media that has to be bled is enough to keep the sump level from gaining. How much to bleed above that is a continuing conflict between the desire to bleed enough to prevent excessive non-magnetics from building in the circuit which causes poor separation due to high media viscosity and possibility of bleeding so much that too much magnetite is lost due to natural inefficiencies or grow overloading of the magnetic separator

Author Information
Phillips is principal, Phillips Process Engineering, a coal preparation consultancy based in Blacksburg, Va. He can be reached at 540-951-3201 (E-mail: He presented this paper at 2010 International Coal Preparation Congress (ICPC), which was held in conjunction with Coal Prep 2010 during April in Lexington, Ky. The paper is reprinted with permission from the Society of Mining Engineers (SME), which holds the copyright on the 2010 ICPC proceedings. To get a copy of the proceedings contact SME (