Operating Guidelines for Heavy-Media Cyclone Circuits

Operating Guidelines for Heavy-Media Cyclone Circuits

G. H. Luttrell


Prep plants use heavy-media cyclones (HMCs) to clean particles in the 50- to 0.5-millimeter (mm) size range. These high-capacity units, which are often operated in parallel banks, use centrifugal forces to enhance the separation of fine particles that cannot be efficiently upgraded using static density-based separators (i.e., heavy-media baths).

In the United States, HMC circuits account for an annual production of approximately 175 million tons of clean coal (16% of the total U.S. production) and represent an installed capacity of nearly 80,000 tons per hour (tph). Because of the high tonnages treated by these units, a small improvement in HMC efficiency can have a large impact on profitability. For example, a modest 2% increase in the efficiency of HMCs in the United States would produce an additional 3.5 million tons of clean coal. At a market price of $25/ton, the recovered tonnage represents additional annual revenues of nearly $400,000 for an average prep plant.

Performance evaluations conducted at industrial plants indicate that an improvement of this magnitude can often be realized at minimal cost through minor alterations in circuit layout or equipment maintenance procedures.


One of the more common problems encountered with HMCs is clean coal overload. The vortex finder of a heavy-media cyclone is in many ways analogous to the overflow lip of a heavy-media bath. In a bath, a minimum overflow depth of 3 to 4 inches must be maintained to ensure that the largest clean coal particles can be hydraulically carried into the clean coal product. Likewise, an adequate flow of media must pass through the HMC vortex finder to carry out the larger coal particles present in the feed. If the media flow to the overflow is too low, then the excess clean coal cannot be carried through the vortex finder and will instead report to the refuse stream. This problem is common in HMCs operated with too large an apex or too low of an inlet pressure.

Another common problem is particle retention. The centrifugal field within an operating HMC causes magnetite to classify and to preferentially report to underflow. The classification causes the underflow specific gravity (SG) to be higher than that of the feed and the overflow SG to be lower than that of the feed.

As a result, middlings that have a density between that of the feed SG and overflow SG tend to remain in the cyclone for a longer period of time than particles outside this density range. Retention is normally associated with only the coarsest particles and rarely occurs for particles finer than about 15 millimeters (mm).

The retention of coarser middlings may in some instances improve the separation by holding middlings until they are broken into smaller particles that are better. Particle retention can, however, become a serious problem when middlings enter the cyclone at a faster rate than they can be removed. The excessive buildup of middlings eventually leads to a sudden surge of solids to underflow that clears the accumulated material. Unfortunately, the surge also tends to carry out a portion of low-density clean coal to the refuse stream.

Surging is an extremely complex phenomenon that is difficult to predict. As a rule-of-thumb, surging may occur when the media differential (defined as the difference in the SG between the overflow and underflow) is 0.4 or greater. To avoid this condition, the operator can switch to a larger apex, lower the cyclone inlet pressure, use finer magnetite, or reduce the top size of the feed material. Care must be exercised, however, since some of these actions (such as the use of a larger apex or lower inlet pressure) can create a different problem such as overloading of the vortex finder.


Inlet pressure is one of the key variables that influences HMC performance. Good operating practice dictates that the inlet pressure should be maintained between nine and 12 cyclone diameters of media head. If the pressure is too low, coal may be misplaced to refuse as the air core becomes unstable and a higher portion of media splits to underflow. To check for adequate pressure, the HMC should be equipped with an operating and properly calibrated pressure gauge. The required gauge pressure (P [subscript]g ) can be calculated using:

P [subscript]g = I[superscript three] (SG)[N(D [subscript]c ) [+ or -] H]

in which I[superscript three] is the specific weight of water, SG is the specific gravity of the circulating media, N is the number of diameters of media head, D [subscript]c is the cyclone diameter, and H is the distance between the pressure gauge and the centerline of the cyclone.

For example, consider a 24-inch diameter HMC circulating 1.6 SG media with a minimum of 9 diameters of media head. A pressure gauge located 36 inches below the centerline of the cyclone should read at least 14.6 psig [i.e., 62.4 lb/ft [superscript]3 x 1.6 (9 x 2 ft + 3 ft)/144 in [superscript]2 /ft [superscript]2 = 14.6 lb/in [superscript]2 ]. At the maximum of 12 diameters of media head, the corresponding pressure would be 18.7 psig [i.e., 62.4 lb/ft [superscript]3 x 1.6 (12 x 2 ft + 3 ft)/144 in [superscript]2 /ft [superscript]2 = 18.7 lb/in [superscript]2 ]. Thus, an acceptable pressure gauge reading for this particular HMC configuration would be 14.6 to 18.7 psig.


Overloading of the vortex finder can cause large coal losses in HMCs. To avoid this problem, the volumetric media-to-coal ratio in the overflow should exceed 2.5. This condition can be checked by collecting samples of the overflow media as it discharges from the sieve bend feed box at the top of the sieve. Multiple samples across the entire sieve width and into the full depth of the sieve feed stream must be taken to ensure that the sample is representative. The required minimum media-to-coal ratio (R [subscript]mc ) can then be calculated using:

R [subscript]mc = Mm/Mc a I c/[c]Ima!

in which Mm is the mass of sampled media, Mc is the mass of sampled coal solids (plus 28 mesh only), Ic is the estimated density of the coal solids, and Im is the density of the circulating media. Mm can be calculated by subtracting the Mc from the total mass of original slurry sample (media and solids).

For example, a representative sample of slurry from the feed box discharge of a HMC clean coal sieve was collected and found to weigh 45 pounds (lb). The slurry sample was screened and found to contain 10 lb of plus 28 mesh solids (dry). The specific gravities of the coal solids and media were found to be 1.4 SG and 1.6 SG, respectively. Based on these values, the media-to-coal ratio is acceptable since it is greater than 2.5 (i.e., [(45-10)/10][1.4/1.6] = 3.1). If the value is less than 2.5, then the media flow rate should be increased or the clean coal tonnage rate should be reduced.


In most cases, a minimum of 2/3 of the volume flow of media that is fed to the cyclone should report to the cyclone overflow. The split can be determined by measuring the cyclone feed density, overflow density (through clean coal sieve), and underflow density (through refuse sieve) and using the formula:

I[superscript two] = SGu – SGf/SGu – SGo

in which I[superscript two] is the fractional split of media to the overflow and SG [subscript]f , SG [subscript]u and SG [subscript]o are the specific gravity values for the feed, underflow and overflow, respectively. For example, if the SG of the feed, overflow and underflow are 1.5, 1.4, and 1.7, respectively, then the media split to overflow is 2/3 [i.e., (1.7-1.5)/(1.7-1.4) = 2/3] and the split is acceptable. Corrective actions should be taken if the value is less than 2/3. In most cases, a smaller apex can be used to correct an overflow volume that is too small.


The quality of the circulating media can have a large influence on HMC performance. Quality can be measured in terms of the size distribution of the media solids and the degree of media contamination by nonmagnetic material. Most operators recognize the importance of purchasing makeup magnetite that has the proper size distribution. Grade “B” magnetite, which is 90% passing 325 mesh, is a common choice by operators in the United States. Unfortunately, the circulating media may be significantly finer or coarser than the as-received magnetite.

Media bleed systems that divert media from only the clean coal circuit will tend to create a circulating media that is coarser. This tendency is caused by the classifying effect within the cyclone that forces finer fractions of the magnetite to the magnetic separator where it may be lost. Likewise, a bleed system that diverts only media from the refuse circuit will tend to make the circulating media finer since the coarser fractions from the underflow have an increased chance of being lost in the magnetic separator.

In practice, bleeding from the refuse side will generally increase plant magnetite consumption since a larger overall mass of media is usually bled in this arrangement. Also, magnetite that is too coarse can become unstable and may lead to surging and losses in coal yield.

Magnetite that is too fine is generally not detrimental to good performance, although the finer magnetite would be more difficult to recover in the magnetic separation circuits. Also, there is some evidence to suggest that finer magnetite is suitable for circuits operated at lower SG cut-points (less than 1.45 SG). For higher SG separations (greater than 1.6 SG), coarser magnetite can be used to minimize viscosity without losing media stability because of the higher solids content. In any case, the particle size distribution of the media should be routinely monitored by means of electronic particle sizing techniques (e.g., Microtrac size analysis) to ensure that the magnetite is of a consistent and acceptable grade.

Media contamination can also adversely impact HMC performance by affecting the viscosity and stability of the media. Operators commonly define contamination as the percentage of nonmagnetic fines (less than 28 mesh) contained in a sample of dried media solids. The slurry sample is normally passed through a Davis tube to separate the magnetic and nonmagnetic solids prior to drying. A rule-of-thumb used by many plant operators is to maintain a contamination level below 10% by weight. However, the level of contamination that is acceptable varies greatly depending on the SG of the circulating media, i.e., higher levels of contamination are generally desirable for low SG separations to promote media stability, while the same level of contamination would be detrimental for separations conducted at a higher SG.

A more suitable method of quantifying contamination is to determine the total amount of nonmagnetic solids in the circulated slurry. For most plants, the circulating media should contain less than 7% by weight of non-magnetics in the total slurry (excluding any material coarser than 28 mesh). For example, if a 1,000-gram sample of circulating media (slurry) contains 65 grams (dry) of non-magnetic material, then the non-magnetic contamination is 65/1,000 (6.5%). Therefore, the contamination is acceptable. The bleed should be increased if the 7% limit is exceeded.


HMCs are frequently installed in banks of two or more parallel units in order to achieve the production requirements of a given plant. For all practical purposes, the maximum yield from such a circuit can only be achieved when the HMCs all operate at the same specific gravity cut-point. This optimization principle is valid regardless of the desired quality of the total clean coal product or the ratios of different coals passed through the circuit. To avoid differences in cut-points, all cyclone components (i.e., apexes and vortex finders) on the same bank of HMCs must be of the same size and type. Also, the feed distribution system must be configured so that each cyclone in a bank receives the same flow rate and quality of feed coal and media. Failure to do so can result in significant losses of clean coal yield.


Density tracers offer a quick and effective method for evaluating the performance of HMC circuits. Density tracers are simply plastic blocks (usually cubic) that incorporate high-density fillers to create particles with densities of 1.2 SG or higher with an accuracy of +0.005 SG.

The blocks can be introduced into the feed stream to a HMC circuit to mimic the behavior of coal, middlings, and rock particles. During a typical test, a number of tracers of different densities are selected based on the anticipated circuit operating conditions. The tracers are added to the process feed stream, i.e., de-slime screen oversize. Together with feed coal, the tracers pass through the separator and report to either the high- or low-density product streams as dictated by the particular characteristics of the separator.

The tracers are brightly colored so that they can be manually retrieved from the product drain and rinse screens. Those from overflow are recovered and counted separately from those that report to underflow. A simple procedure is then followed to develop a partition curve for the separator.

Since this procedure eliminates the need for sample collection and laboratory analysis, the partitioning data obtained using tracers are generally more accurate than data derived using conventional sampling and float-sink procedures. More importantly, the partition curve can be obtained after only about one hour. If conventional procedures (sampling and float-sink tests) are used, the data are typically available only after several weeks.

To illustrate the effectiveness of this analytical tool, density tracers studies were conducted around the HMC circuits at five eastern U.S. prep plants (seven HMC circuits). The evaluations were carried out using 32- and 16-mm density tracers. The normalized partition data for each plant are summarized in Figure 1. (Note that two HMC circuits were evaluated at plants A and B.)

Despite the wide range of coal types treated by these circuits, the partition data show that the separations provided by the HMC cyclones were relatively sharp (Ep

The full capabilities of HMC circuits are often not realized in industrial practice due to a lack of accepted guidelines for circuit design and operation. To help alleviate this problem, practical operating guidelines have been provided for plant operators that may improve the performance of their HMC circuits. Partition data collected at several industrial installations using density tracers suggest that good heavy-media cyclone performance can be achieved if these guidelines are successfully implemented.

Luttrell is a professor and Barbee is a graduate student at Virginia Tech’s Mining and Minerals Engineering Department. Wood is president for Partition Enterprises, an Australian firm specializing in coal preparation. Bethell is the coal preparation director for Massey Energy.


The authors would like to acknowledge the financial support for this work provided by the U.S. Department of Energy (DE-FC26-01NT41061).


Abbott, J., 1982. “The Optimization of Process Parameters to Maximize the Profitability from a Three-Component Blend,” 1st Australian Coal Preparation Conference, April 6-10, Newcastle, Australia, pp. 87-105.

Chedgy, D.G., Watters, L.A., and Higgins, S.T., 1986. “Heavy Medium Cyclone Separations at Ultra-Low Specific Gravity,” 10th International Coal Preparation Congress, Edmonton, pp. 60-79.

Davis, J.J., 1987. “A Study of Coal Washing Dense Medium Cyclones,” Ph.D. Thesis, University of Queensland, Australia.

Gottfried, B.S. and Jacobsen, P.S., 1977. “A Generalized Distribution Curve for Characterizing the Performance of Coal Cleaning Equipment,” USBM. Report 8238.

King, R.P. and Juckes, A.H., 1988. “Performance of Dense Medium Cyclone when Beneficiating Fine Coal,” Coal Preparation: An International Journal, Vol. 5, pp. 188-210.

Luttrell, G.H., Catarious, D.M., Miller, J.D., and Stanley, F.L., 2000. “An Evaluation of Plantwide Control Strategies for Coal Preparation Plants,” Control 2000, J.A. Herbst (Ed.), SME, Littleton, Colo., pp. 175-184.

Napier-Munn, T.J., 1984. “The Mechanism of Separation in Dense Medium Cyclones,” Ph.D. Thesis, University of London, England.

Rao, T.C., Vanagamudi, M., and Sufiyan, S.A., 1986. “Modelling of Dense Medium Cyclones Treating Coal,” International Journal of Mineral Processing, Vol. 17, pp. 287-301.

Scott, I., 1988. “A Dense Medium Cyclone Model Based on the Pivot Phenomenon,” Ph.D. Thesis, University of Queensland, Australia.

Wood, C.J. 1990. “A Performance Model for Coal-Washing Dense Medium Cyclones.” Ph.D. thesis, unpublished, University of Queensland, Australia.

Wood, C.J., 1997. “Coal Preparation Expertise in Australis: In-Plant Issues and the Potential Impact of Broader Applications,” Proceedings, Coal Prep ’97, Lexington, Ky., pp. 179-198.

Wood, C.J., Davis, J.J. and Lyman, G.J., 1987. “Towards a Medium Behavior Based Performance Model for Coal Washing Dense Medium Cyclones,” Aust. IMM, Brisbane, pp. 247-256.

COPYRIGHT 2003 PRIMEDIA Business Magazines & Media Inc. All rights reserved.

COPYRIGHT 2003 Gale Group