Processing of Silica Sand Introduction Silica Sand—the most common ingredient in glass. Whether the glass is container glass, window glass, boro-silicate glass, fiberglass or water glass (sodium silicate), silica sand makes up approximately 60-70 percent of the glass batch and therefore has a significant affect on the glass quality.

For all glasses, the consistency of the chemical components and size characteristics are important to producing high quality glass. The three main chemical contaminates in silica sand are usually the iron content expressed as Fe2O3, the alumina content expressed as Al2O3, and Titania expressed as TiO2. If the alumina content is high due to a presence of feldspar, then the other alkalis such as CaO, NaO2 and K2O may also be significant contaminants.

Each of the contaminates mentioned above will impact the end product. The impact of the contaminates may be either positive or negative depending on the percentage of the component in the silica sand and the level of the same component in the glass. For example, generally, lower iron content is better however, to produce amber containers, iron is often added to the glass batch. Therefore, higher iron content could be beneficial as long as the iron content was consistent. Higher alumina due to a Na-feldspar mineral can help lower the batch cost since the lower priced silica sand is substituted for a portion of the more expensive feldspar or soda ash. However, too much alumina, generally greater than 1.8 –2.0 percent, cannot be used for container glass without blending with lower alumina sand. In addition, Al2O3 levels greater than 0.3 percent are generally unacceptable for window glass.

Regardless of the contaminate concentration, the consistency is extremely critical. Consider a glass container company producing amber bottles with inconsistent levels of iron and alumina in their sand. The results would be a bottle with too high or too low iron content. Too little will affect the color, too much will result in very brittle glass. As alumina levels change, the viscosity and density of the glass change making it impossible in today’s high speed bottling machines to make a consistent product.

In addition to the chemical components, the other key component of glass sand is the lack of refractory heavy minerals. These minerals are generally aluminum silicate minerals but there can be others such as chromite. Whereas the chemical components are measured in tenths or hundreds of a percent, the refractory minerals are measured in the number of grains in a particular size sample. As few as 1-2 grains of chromite in 500 grams of silica sand can render the sand unfit for glass production. If a silica product has 1 grain of refractory mineral per 500 grams, a typical glass plant using 350 tpd of sand would have in excess of 700,000 defects per day; far too many to make the sand a desirable component.

Since silica sand is a low priced commodity, high shipping cost would result in an uneconomical deposit with freight cost being more valuable than that of the silica sand. Thus, it may be necessary to process a lower grade deposit that is closer to the market place than process a higher-grade deposit that is much farther away. Processing can often be achieved at a fraction of the freight cost.

Background This paper discusses the many processing techniques used in processing silica sand in Western countries. The techniques consist of both wet and dry processes, and at times, process combinations to produce an acceptable product.

Although the introductory portion of this paper discussed the various chemical contents of the sand, mineral processing techniques do not separate chemical components but rather the minerals that make up these components. For minerals to be separated, they must first be liberated. Once liberated, it is possible to develop processing techniques to meet glass sand specifications. If significant iron bearing minerals are located within the sand grain, or there is severe iron staining on the exterior of the grain, achieving the glass specifications may not be possible.

Below, Figure 1 shows a good example of high and low quality silica sand. For the most part, the high quality grains are very clean with little or no inclusions. The one inclusion is a minor rutile grain that does not significantly contribute to contamination level of the sand. The poor quality sand shows significant iron

bearing inclusions within numerous grains. In addition, a significant percentage of the grains have iron staining. Note the surfaces of the grains are also very irregular which would make removal of the surface staining (using an attrition scrubber) even more difficult.

Figure 1 Silica Sand Grain Comparison

PROCESS DESCRIPTION

WASHING Washing is the simplest and lowest cost method of cleaning silica sand. In some of the very pure deposits that are void of heavy minerals, high amounts of clay and silt, and no surface staining, washing is sufficient to produce acceptable grade product.

In the process, water is added to the sand and is generally pumped to a cyclone for desliming. The movement of the slurry passing through the pump and pipeline is sufficient to loosen the small amount of fines or clay that are in the ore body. Once the minor amounts of fines or clay have been released from the silica sand they can be removed via a variety of methods.

ATTRITION SCRUBBING Attrition scrubbing is used when the clay or silts are more tightly bound to the silica grains, or clay particles are similar in size to the silica sand grains. In this process, silica sand is introduced to an attrition-scrubbing unit similar to the one shown in Figure 2.

Figure 2 Outokumpu Technology Inc. FLOATEX® LPF 1300 Attrition Scrubber

For proper attrition scrubbing, it is important that the solid percentage of slurry be within a range of 72-75 percent solids. At this content, there is good particle-to-particle contact and the viscosity of the slurry is low enough to allow the slurry to move freely in the attrition-scrubbing tank. When the solids percentages are lower than 72 percent, there is sufficient water in the slurry to allow the particles to stay apart and prevent the necessary particle-to-particle contact required to scrub the clay from the surfaces. When the solids are higher than 75 percent, the slurry to too viscous and the impellers are not able to move the slurry. When the slurry does not adequately move, particle-particle contact is not possible.

In order to process the high percent solids slurry on a consistent basis, the power transmission system employed is extremely critical. The best possible solution is a gear driven machine as shown in Figure 3. Although v-belt systems tend to work fine when the units are new, as they age and are not properly maintained, slippage occurs and the impeller will not rotate with the resistance caused by the high percent solids. At this point, operators tend to add additional water to lower the percent solids and the scrubbing becomes ineffective

Typical scrubbing times are approximately 5 minutes. However, the actual retention time can vary considerably depending on the amount and type of material that needs to be liberated. Attrition scrubbing circuits may also include more than one attrition scrubbing stage. When there is significant clay content, two to three shorter scrubbing times with desliming steps in between is more effective than one long scrubbing time. Once the clay is liberated, the clay acts as a lubricant between the silica grains and therefore reduces the effectiveness of the scrubbing. By removing the liberated clay slimes, and then adding another stage of scrubbing, the scrubbing circuit is more effective.

The number of attrition scrubbing cells in the circuit is also important. Although one cell may have sufficient volume to meet the retention time needed, one cell will result in significant short-circuiting of the feed material. That is, if the average particle retention time is 5 minutes, there may be particles that will only have retention times of 2 minutes and others at 8-10 minutes. It is important that at least two attrition scrubber cells be used to provide the necessary retention time. Three or four cells are preferred whenever possible.

Capital costs per ton for attrition scrubbing varies widely due to the retention time requirements and the feed rates. Though higher feed rates have lower capital cost per ton than lower feed rates, costs in the $1000-2000/ton/hr range are typical.

DESLIMING By definition with silica sand, desliming is the process that removes the -100 micron material. These slimes are generally clay type minerals or very fine silica that is detrimental to the glass making process. Although there are many different methods of desliming, only two are widely used in the industry: cyclones and hydrosizers It should be noted that some older processing plants may still contain screw classifiers, but due to their high capital and maintenance costs they are not found in new plants.

Cyclones Cyclones are low cost and effective in removing the -100 micron slimes from silica sand when the -100 micron material does not exceed 3wt percent. Cyclones are effective in removing a majority of the slimes but, as the underflow product typically contains 30-40 percent water, some clay slimes remain in the underflow product. Typically, cyclones remove 80-90 percent of the –100 micron in the feed. Multiple cycloning stages will reduce the residual amount of slimes in the underflow product.

Hydrosizers For feeds that have more than 4-5wt percent -100 micron material, or when there is a need to remove all or part of the -150 micron material, the best equipment is the hydrosizer such as the FLOATEX® Density Separator shown Figure 3.

Figure 3 Photograph of Outokumpu Technology Inc. FLOATEX® Density Separator in a Silica Sand Plant

For all hydrosizers, the basic principles of operations are similar, but some have better control systems or water distribution. Figure 4 below shows a cut away of the hydrosizer.

Figure 4 Cut away diagram of a FLOATEX® Density Separator

The hydrosizer is a hindered settling devise. Therefore, it utilizes a current of water, introduced across the bottom of the hydrosizer, to expand sand slurry into a state of teeter. In this teetered state, the sand grains will classify themselves so that the coarse grains report to the bottom where they will stay relatively close to each other with high water velocities flowing between them. The finer particle will be dispersed to the higher levels where they will stay in more open suspension and therefore, the water velocity between them will be low. The principle of operation

Figure 5 Principles of operation of a hindered settler

The hydrosizers are equipped with a pressure-sensing device inserted into the teeter zone to give an indication of the specific gravity. For any pre-set upward current of water, the specific gravity is indicative of the average particle size of the sand above the sensing position, and therefore may be used to provide a variable signal to operate a valve to control the discharge of the coarse material at the bottom of the hydrosizer.

The hydrosizer has a greater separation efficiency when compared to a cyclone. This is due, in part, to the mode of operation and the addition of “clean” water added in the teeter zone. The net result is fewer fines in the sand and a more consistent level of those fines. If the sand is to be dried, the fines are removed prior to drying. Since the fines have more surface area, fuel savings in the dryer operation can be significant. In addition, with less fines there is less dust, which lowers worker exposure to silica dust and reduces maintenance on the dust collection system.

Capital cost per ton varies with feed rates but the range is typically $1000-2000/ton/hr with the lower cost being for the higher feed rates.

SIZING Sizing for glass sand is extremely critical in order to make high quality glass product. Sizing must be discussed in two different areas. One is the ability to size the ore at near 0.5 mm the other is to prevent oversize grains (+1mm) from entering the product.

Sizing to provide a sand product that is less than 0.5 mm is typically conducted with a screen or hydrosizer. The allowable material above 0.5 mm has been greatly reduced in the last 10-15 years. It was not uncommon in the 1980’s for the glass producers to allow 5wt percent +0.5 mm in the specifications for glass sands, Today, they are pushing the limit to 0 percent on +0.5 mm with specification now being proposed with limits on the 0.4 mm screen.

The lower amount of coarser grains allows the glass producers to operated lower furnace temperatures and maintain high production levels. The lower furnace temperature not only saves in fuel costs, but, due to lower furnace temperatures, additional savings are realized with longer furnace life.

The other sizing aspect is the amount of oversize grains. These are generally measured as +1 mm grains and are limited to a few grains per kg. These grains generally are too large to melt in the furnace and cause defects in the glass. Like the refractory minerals discussed above, a few grains per kg results in 1000’s of defects per day, and can significantly impact the glass production cost.

Screens High frequency screens supplied by companies like Derrick or Rotex are generally used to make the 0.5 mm separation. The latest developments are high capacity screens that can size 250 tph with one screen. The high capacity screens employ several screen decks within one machine unit. The system employees a feed distribution system that evenly feeds all screen decks simultaneously. Figure 6 shows a photograph of the screen produced by Derrick for this application.

Figure 6 DERRIK® Stack Sizer

Screens are generally sized to produce a good either undersize or oversize product. That is, if the goal is to have a product that is 100 percent -0.5 mm, the underflow of the screen will meet this goal. However, the oversize of the screen will contain some -0.5 mm. A typical size distribution of the products and feed are shown below in Table 1.

In plants that incorporate drying, screening is often accomplished after drying as the screening operates more efficiently when dry. However, if high percentages of +0.5 mm material were in the ore body, then

Table 1

Typical Screen Size Distribution

Size (micron) Feed (wt%) Oversize (wt%) Undersize (wt%)

1000 35.7 89.6

710 28.3 78.4 0

500 35.2 92.0 3.6

425 39.4 95.3 7.31

355 44.1 96.8 13.1

300 48.7 97.6 18.9

250 54.1 98.0 26.7

125 71.8 100 54.0

90 80.0 65.6

wet screening would be preferred to eliminate as much oversize as possible prior to drying. This approach saves the drying cost and allows for higher production rates of final product from the dryer.

Typical capital cost for a screens ranges from $1000-2000/ton/hr.

Hydrosizer In addition to screening, the 0.5 mm separation can also be accomplished with a hydrosizer such as the FLOATEX® Density Separator that has been previously discussed. These units offer some advantages over screens since the capacities for the same footprint are higher than for screens. They offer more flexibility in sizing since a cut size can be changed with change to the sensor set point compared to the necessary screen cloth changing in screens. A typical size separation is shown in Table 2 below.

Typical capital cost per ton for a FLOATEX® Density Separator will vary with feed rate but in the range of $500-1000/ton/hr with the higher feed rates at the lower cost per ton.

GRAVITY SEPARATION Gravity separation for glass sands are limited to spirals since other gravity separation processes, such as enhanced gravity separators that are too expensive to operate or processes such as shaking tables that are too low capacity for glass sands, are not cost effective.

Spirals are capable of separating heavy mineral particles from glass sands to lower the iron content. Generally, specific gravity differences of greater than 0.5 to 1.0 units are required to make an effective separation. Table 3 below shows the specific gravity of typical minerals associated with glass sands.

Table 2 Typical FLOATEX® Density Separator Size Distribution

Size (micron) Feed (wt %) Underflow (wt %) Overflow (wt %)

1175 1.4 4.3

850 3.2 9.8 0

595 7.6 22.9 0.2

420 35.8 71.8 15.2

300 62.5 94.8 46.8

200 84.3 100.00 77.1

The Table 3 above shows that the iron bearing heavy minerals have specific gravities that are significantly higher than the 2.65 of silica. Therefore, this group can be separated using spirals. The last two minerals known as aluminum silicates have specific gravities that are only 0.55 units higher than silica. Although this differential is high enough for spirals to be effective, it is at the minimum differential needed. Therefore, the efficiency is low especially if there is a broad size range.

Also, since these aluminum silicates are refractory heavy minerals, there is a zero tolerance of these minerals for glass production. That is, only 1-2 grains/kg will be above the permissible limits. To further complicate matters, it is the coarser grains that cause glass stones and these are the most difficult to remove with spirals. However, there are methods that can be employed to help eliminate the coarse aluminum silica grains, and will be discussed later. Spirals are considered flowing film separators and as such, the forces employed cause the lighter specific gravity particles to be pushed to the outside regions of a spiral. The heavier particles tend to stay close to the inner portion of the spiral as shown in Figure 7 below.

Table 3

Specific Gravity of Various Minerals

Mineral Specific Gravity

Silica Sand 2.65

Feldspar group 2.6-2.8

Ilmenite (FeTiO2) 4.7

Magnetite (Fe3O4 5.2

Leucoxene (FeTiO2 altered) 3.6-4.3

Andalusite (Al2SiO5) 3.2

Sillimanite (Al2O(SiO4) 3.2

Figure 7 Photograph of an Outokumpu Technology Inc. CARPCO® Model LC3700 Spiral Processing Glass Sand

In addition, forces increase as the distance from the spiral surface increases as shown below in Figure 8.

Figure 8 Forces employed on Various Size Particles in a Spiral

Figure 8 shows that the larger particles are impacted by the greater forces and the smaller particles are impacted by the lower forces. For the most part, this improves the separation efficiency of glass sands since the heavy minerals tend to be finer in size than the silica sands. The finer heavy minerals have lower forces acting upon them and therefore tend to stay closer to the inner race, the coarser sand particles have the greater forces and travel outward toward the outer region of the spiral.

Refractory heavy minerals are difficult to remove, as they tend to be coarser and have specific gravities much closer to silica sand. Typically, spirals alone do not produce acceptable results unless the sand is preclassified prior to the spiral. A narrow particle size range will increase the removal efficiency. If the

preclassification is accomplished with a FLOATEX® Density Separator, another benefit is derived. Not only will the FLOATEX® unit make a separation based on size, it will also make a separation based on specific gravity. Therefore, some of the coarse refractory minerals will report to the waste stream (+0.5 mm underflow) and not need to be removed with the spiral.

Capital cost for spirals are low with typical cost in the $1000/ton/hr range.

A typical spirals performance for silica sand is shown in Table 4.

The data shows that two stages of spirals lower the iron content to acceptable levels. On the material from the USA, the feed was dried and then reported to magnetic separation to achieve the final iron of 0.025 percent. Without the spiral process, the best iron achievable with magnetic separation was 0.032 percent.

FLOTATION The flotation process is primarily used to remove the iron bearing heavy minerals in glass sand. The minerals are generally the same as the iron bearing minerals described in the gravity section. At times, flotation can also be used to remove mica.

For iron bearing heavy mineral removal, the feed is conditioned with a fatty acid at high percent solids. In this process, the pH is generally in the 2-3 range, the percent solids at 70-72 percent and the fatty acid addition rate is 1.0 #/ton. Fatty acids, supplied by companies like Cytec Industries, are tailored for each particular deposit and often contain frothers and other proprietary chemicals that enhance the performance.

After conditioning, the feed is introduced to a series of flotation cells. To prevent short-circuiting, there are usually 4-6 cells in series. Flotation time is usually less than 5 minutes.

Typical cells used in the silica industry are in the 3-8-m3 range. Capital cost for a typical flotation circuit ranges of $6000-8000/ton/hr. Operating costs are approximately $1.00-1.25/ton for flotation reagents and power consumption is equal to 10 kw/ton including the conditioning stage.

DRYING In North America, all glass sand is dried at the silica processing plant prior to shipping to the glass manufacturing plant. The advantage of dried sand to the glass producer is the better ability to accurately measure the batch ingredients, better material flow characteristics, and better mixing of batch ingredients. The cost to dry the sand, however, is borne by the producers of the silica sand. Drying is the single most expensive unit operation in the silica plant. As the cost of fuel continuously increases, it becomes more of a cost burden.

Table 4

Typical Spiral Performance Single Pass Double Pass

Sample Wt % Recovery % Fe2O3 Wt % Recovery % Fe2O3

USA 97.7 0.066 90.3 0.049

Mexico 89.7 0.039 84.4 0.028

Europe 95.7 0.038 91.6 0.033

For the most part, the silica sand is dried using fluidized bed dryers. The fluid bed dryers are much more fuel-efficient than the rotary dryers that were formerly used in the silica industry. Typical fuel oil consumption for fluid bed dryers is 5.2 liters/ton compared to 8.7 liters/ton for rotary dryers. There are still some rotary dryers used in the silica industry, but they are old units and most likely be replaced in the near future.

Typical capital costs per ton per hour for fluid bed dryer are in the $4000-6000 ton/hr.

MAGNETIC SEPARATION Magnetic separation has changed considerably in the last 10-15 years. Prior to this, electromagnets were used in glass plants to remove the iron bearing minerals. Today, the rare earth roll magnetic separators dominate the industry.

When the rare earth magnets were first introduced, there were problems associated with the magnets including low production rates, low sand temperature requirements, and high costs. However, today, the rare earth magnets have improved greatly. Temperature ratings are above 120 C with better instrumentation to shut down the process if temperatures exceed the maximum rating. Costs have been reduced significantly with capital cost of rare earth magnets being less than 50 percent of the capitals cost/ton of the older electromagnets. Figure 9 shows a close up of the actual separation and 3 pass INPROSYS® rare earth magnet designed for the silica sand industry.

Fig 9 Outokumpu Technology Inc. INPROSYS® Rare Earth Roll Magnet Triple-Stage Unit

Production rates have continued to climb per meter width of separating zone due to the increase of the roll diameter. The earlier magnets were 75 mm in diameter, and, until recently, the 100 mm diameter was the standard. Production rates in glass sand were in the order of 3-7 tph/meter depending on purity of feed and the desired end product. Rare earth roll magnetic separators now have diameters of 150 and 300 mm diameters with 2-4 times the production rates of the 100 mm roll.

Typically, the rare earth roll magnetic separators are configured in a two or three pass system with the non-magnetic product from the first pass reporting to the second and third roll for an additional pass. In some cases, four passes are required to achieve the results. However, the vast majority of all minerals are removed in the first two passes and the subsequent passes are only required when the ore quality requires more processing to meet a specification. Several examples of results from magnetic separation tests are shown below in Table 5.

The table shows that with decreasing feed grades for the USA sample, the iron contents continued to decrease. This feed material is the same as the example shown in the spiral data. For the European data, increasing feed rates resulted in higher iron content but even at 5.0 tph, the iron content was within the desired specification of <0.018 percent Fe2O3.Typical capital costs for rare earth roll magnetic separators are $6,000-10,000/ton/hr

TRIBOELECTRIC SEPARATION New process or old? Triboelectric separation has been commercially exploited since the mid 1940’s. However, until recently, it has been only used by 1 or 2 companies in the salt industry. Triboelectric separation results when one mineral gains an electron from another mineral. When this occurs, the mineral that has gained an electron becomes negatively charged and the one that gave up the electron becomes positively charged. When the minerals are dropped between two oppositely charged electrodes, the negative mineral is attracted towards the positive electrode and the positive mineral towards the negative electrode and a separation occurs. Figure 10 shows an Outokumpu Technology’s T-Stat Separator.

Table 5

Typical Magnetic Separator Performance Product TPH/Meter Feed %

Fe2O3 Wt % Product %

Fe2O3

USA 5.0 0.089 97.9 0.039

USA 5.0 0.066 98.1 0.031

USA 5.0 0.053 97.5 0.031

Europe 3.0 0.085 96.5 0.012

Europe 4.2 0.085 97.6 0.013

Europe 5.0 0.085 97.1 0.014

Figure 10 Outokumpu Technology Inc. CARPCO® T-Stat Separator.

For the silica industry, the triboelectric separation process is used to separate feldspar from quartz. Traditionally, this separation has been conducted using flotation. In the conditioning stage prior to flotation, hydrofluoric acid is used to activate the feldspar and depress the quartz. The amine is added to float the feldspar from the quartz.

Although the process is widely used, it is not ideal. There are many factors that influence feldspar removal efficiency, such as: water quality, percentage of feldspar in the feed, feed rate, and size distribution. As these variables change, the amount of feldspar remaining in the quartz changes and therefore the Al2O3 content changes. As discussed in the introduction, changes in Al2O3 content in the glass batch results in changes in the glass density and viscosity, which affect the downstream process.

In addition, it should be noted that, the hydrofluoric acid is environmentally unfriendly. Since the HF is added to the water, the entire water circuit of the plant is contaminated with HF and water discharge from the plant cannot be tolerated.

With the triboelectric separation process, HF is still required, however, since this is a dry process, the HF can be better controlled. In the triboelectric process, fumed HF is added to dry, hot (100-120C) feed material in a rotary mixer. As the feed material is being mixed in the presence of HF, electrons are transferred from the feldspar to the quartz. The feldspar then becomes net positive and the quartz net negative. When this material is introduced between to highly charged electrodes (+50, -50 KV), the quartz reports to the positive electrode and the feldspar to the negative electrode.

Compared to flotation, the HF is much easier to control. Any excess HF from the rotary mixer reports to a wet scrubber. These commercially available wet scrubbers utilize only minor amounts of water to scrub the air free of HF fumes. They have efficiencies of greater than 99 percent and use only 4 liters/minute of new water. The HF laden water from the scrubber reports to a system that cleans the water to 10-12 ppm F ion and produces a CaF filter cake that can be land filled.

In addition to the improved environmental aspects, the process is more stable since there is no influence from the change of process water quality. In addition, changes in feldspar content or size distribution do not have as significant an influence as flotation since there is no amine (collector) being used that must be adjusted to reflect these changes.

Table 6 shows the result of a typical quartz-feldspar separation that has been conducted using Outokumpu’s T-Stat. Recycling of the middling product will improve the recovery. Although these products met the specification of the customer, an additional cleaning stage would increase the grade of the products.

GRINDING/AIR CLASSIFICATION Grinding is widely used in the silica industry to produce ground silica for the textile fiberglass industry. For the textile fiberglass industry, consistency of chemical components along with size distribution is as important as with other glass products. A typical grinding mill such as Outokumpu Technology’s 375kW is shown in Figure 11 below.

Figure 11 Outokumpu Technology Inc. Nordberg Grinding Mill

For the size distribution, it is important that the ground silica is void of any oversize grains and that the extreme fines be as low as possible. Typically, the textile fiberglass industry uses a product that is either 95 percent passing –75 micron or 95 percent passing 45 micron. For these sizes, the oversize is defined as grains coarser than 250 microns and extreme fines a less than 5 microns. Both of these can cause unmelted silica stones and result in breakages of the fiber during the drawing process. Although it is obvious why the over size grains are difficult to melt, the possibility of the fines not melting does not seem to be valid. In this case, when there are excess amounts of fines, the fines tend to agglomerate and not mix with the fluxing agents. Therefore, the agglomerates do not melt. To maintain a consistent particle size distribution, void of oversize grains and low percentages of extreme fines, the grinding mills are used

Table 6 Typical Results of T-Stat separation PRODUCT Wt % % Quartz % Feldspar Qtz % Dist Feld % Dist

Feed 100.0 48.7 51.4 100.0 100.0

Feldspar Prod 41.7 11.9 88.1 18.2 71.5

Middling 17.3 57.3 42.7 20.4 14.4

Quartz Product 41.0 82.4 17.6 69.4 14.1

with high efficiency air classifiers such as the one produced by Progressive Industries Inc., as shown in the Figure 12 below. The unique patented seal prevents oversize from entering the product and the high-speed rotor, combined with the secondary air washing, prevents product size particles from returning to the mill for more grinding and production of the extreme fines.

Figure 12 Progressive Industries, Inc. Air Classifier

Since consistent chemistry is important, the use of natural flint grinding media and liners has been on the decline during the past 10 years. These natural products were used for many decades but due to a combination of declining product quality and the demands for better silica products, they have been replaced with high alumina ceramic materials.

These engineered products are significantly more costly per kg than the natural materials but due to their better wear characteristics offer lower cost per ton of silica. The ceramic liners are approximately 20 percent as thick as the natural flint liners but last 5 times longer. As important to the wear characteristics, is that since the liners are not as thick, they allow greater mill volume to be filled with grinding media and therefore allow for higher production rates.

SLILICA SAND PROCESSES The previous portion of this paper has discussed the unit operations that are used in a silica plant. These unit operations can be used in various combinations to achieve the desired end product based on the starting ore quality.

Below, the flow sheets for two recent projects by Outokumpu Technology Inc. will be discussed.

Plant A In the first process, shown in Figures 13 and 14 below, the objective of the plant was to produce two different glass sands with premium glass sand having low iron content and a higher iron standard glass sand that would be sold damp. In addition, the company also wanted to produce a ground silica product for use in the textile fiberglass market.

Figure 13 Wet Processing Diagram of Glass Sand Plant A

In this plant, attrition scrubbing is used to break down some loosely consolidated sand grains and to liberate any clay from the silica grain surfaces. In the next step, the FLOATEX® Density Separator is used to make a 200 micron separation. For this deposit, this separation stage was needed since the coarser silica grains tend to be more pure and the minor inclusions have less effect on the iron content compared to the finer grains. The FLOATEX® Density Separator underflow product reports to two stages of Outokumpu Technology’s CARPCO® Spirals to remove the free iron bearing minerals. The iron content of this product will be less than 0.015 percent Fe2O3 prior to further treatment in the dry process (see Figure 14).

The overflow of the first Density Separator also reports to two stages of Outokumpu Technology’s CARPCO® Spirals to remove any free iron bearing particles. The flowsheet could be simplified by conducting the spiral separation prior to the Density Separator. However, the process produces higher-grade products by first classifying the sand into two more narrowly sized products. The product from the spirals then report to another Density Separator to remove the –100-micron fines generated by the attrition scrubbing stage. The underflow of the product is then stockpiled, allowed to drain and then shipped as damp glass sand with a Fe2O3 content of <0.020 percent.

Figure 14 shows the remainder of the process, which incorporates the dry processing section of the plant.

Figure 14 Dry Processing Diagram of Glass Sand Plant A

In this section of the plant, the premium glass sand from the wet processing section is dried and then reports to a 2 stage INPROSYS® Magnet for final cleaning. The spiral removed the free iron bearing minerals and some rutile that was present in the sand. Rutile has a high specific gravity but is non-magnetic. The rare earth roll magnetic separator removes the iron bearing minerals that were not removed by the spiral. For the most part, these minerals are present as inclusions in the silica grains. After magnetic separation, the iron content of the glass sand is reduced from 0.015 to <0.010 percent Fe2O3.

In addition to the magnetic separators, the plant will also produce ground silica for the textile fiberglass market. The mill, grinding media, and air classifier have been chosen to assure the rigid product quality needed for this industry.

Plant B In the second plant design, the ore quality was considerably lower than that of the first plant design. This deposit was substantially closer to the market place than the competition. Therefore, although the processing is very extensive, and operating cost higher than average, the delivered price to the customer was still substantially lower than the competitors.

Figure 15 shows the processing diagram for Plant B. In this plant, a screen was used to not only remove the oversize, but also break down the loosely agglomerated sand particles. Due to the high percentage of clay, a FLOATEX® Density Separator is employed as a desliming process after the initial washing stage. The pumping action helps liberate the sand grains from a majority of the clay. In the plant process, approximately 15wt percent of the feed is lost as –100-micron material in the desliming stage. After the initial desliming, three stages of attrition scrubbing with desliming by cyclones after the first and second stage are used to liberate and remove the remaining clay and silt. A FLOATEX® Density Separator is used for the final desliming stage, instead of a cyclone, to assure that the –100 micron material is removed.

Figure 15 Processing Diagram for Plant B

This deposit contained higher amounts of feldspar than desired by the glass customer. The feldspar resulted in alumina contents too high to be used for float glass. After the sand is dried, it is activated using fumed HF in a rotary mixer. Once activated, it reports to the T-Stat separator where the majority of the feldspar is removed as a waste product. The middling fraction is recycled back to the activation process to allow for a second pass through the unit.

The sand fraction from the T-Stat then reports to the INPROSYS® rare earth roll magnetic separators to reduce the iron content. The non-magnetic fraction is then stored and shipped to the customer. The process results in a product with an iron content of <0.030 percent Fe2O3 and <0.3 percent Al2O3. Although the iron content is much higher than in Flowsheet A, the close proximity to the market place makes this an acceptable product.

The previous flowsheets show two examples of glass sand processing. Each process is unique to the ore body and the glass sand customer’s needs. This is true for all sand deposits. However, there is always a common need for any sand process. The sand process must be the most economical process possible since the selling price of sand is very low. At the same time, the process must be capable of producing a product that has a consistent chemical composition with iron contents generally lower than 0.035 percent Fe2O3 and at times lower alumina levels. In addition to the chemical composition, the process must also be capable of eliminating the oversize and fines so that the sand particles are within the –500 +100 microns size range.