To understand the flow characteristics of sand, we built a flexiglass model silo 8
by 8 by 13 inches with a conveyor and bin to simulate sand handling practices commonly
found in industry. The model has interchangeable loading and discharge gates to demonstrate how
silo design and sand handling methods affect the distribution and consistency of sand. For the
first two demonstrations, the sand is dyed red white ash and was loaded simply to identify its
position in the silo. The first demonstration shows that in a single discharge system,
the last sand in is the first sand out. Because of this funneling effect, an object, a half dollar
for example, placed on top of the sand in a 300 ton silo will be discharged in the first four
tons. Although the sand appears to blend, this mixing is taking place at the interface patches.
There is in fact no significant blending as the sand flows through the silo. Except for minor
mixing at the layer interfaces, each color remains solid and unblended as it discharges. The fact
that in single discharge systems, the last sand in is the first sand out is important to remember
when the sand is to be used with temperature sensitive chemical binders. A chemical reaction
between the binder and catalyst can be increased or decreased by variances in the sand temperature.
Cool sand requires a certain ratio of binder and catalyst to react in a given time. If a truckload
of warmer sand enters the system and bottles through, the ratio must be changed. Changing
this ratio decreases production. Failure to change may cause scrapped cores. As the load of
sand discharges, we observe another characteristic of sand. Its tendency to come to rest in a cone
whose sides slope at a 30 to 35 degree angle with respect to horizontal. Sand lying beneath
this angle, whether in a free-standing cone or against the walls of a silo, remains immobile.
As we shall see in the demonstrations that follow, this tendency of sand that lies under
the angle of repose, not to move, has a significant effect on the efficiency of silo load and discharge
operations and on the usable capacity of the silo itself. Our first demonstration has emphasized
two important characteristics of sand. First, that there is no significant reblending of sand
as it discharges. Second, that in a single discharge system, the last sand in is the first sand out.
Our second test demonstrates that in a multiple discharge system, the order of discharge is the
same as the loading sequence. That is, the first sand in is the first sand out. The evenly distributed
holes simulate a system where the sand is discharged through a number of pipes at the bottom of the
silo. In contrast to the flow in our first demonstration, the first sand to discharge is
white sand from the bottom layer, the first sand in. The significance in this order of discharge
is that the sand has time to cool in the silo. During the summer, it's often difficult to deliver
sand at temperatures below 100 degrees Fahrenheit. Blower trucks increase the sand temperature five
to 20 degrees by transporting it in air at temperatures of up to 270 degrees Fahrenheit.
Multiple discharge silos provide more time for the sand to cool prior to use. The sand does not
funnel or rattle down from the top layer, which remains level. The only evidence of funneling
can be seen in the ridges at the bottom interface of the blue and white layers, which indicate that
the bottom white layer has emptied to its angle of repose or resting level. At this time, about 60%
of the sand has discharged. Funneling has begun to appear at the top of the sand, and as the sands at
the interfaces all begin to follow the path of least resistance, some blending takes place. In
some silos, the sand remaining at the bottom may tend to become contaminated and is best left in
place rather than removed. In our second demonstration, we have seen that multiple discharge systems
allow sand to cool and to discharge evenly in a first in first out sequence. Our next demonstration
is like the first. Discharge is through a single hole in the center of the bottom plate directly
beneath the loading point. For this and all the remaining demonstrations, we dyed the grains of
sand according to size rather than by the batch to demonstrate how sand segregates itself. The
coarse 30, 40, and 50 mesh grains are red. The medium 70 mesh grains are their natural color,
white, and the fine 100 mesh through pan material is blue. Sand segregates itself according to size
because the coarse red grains, which tend to be heavier and more rounded, roll over the finer blue
grains, which tend to be angular and to lock together and resist movement. The coarse grains
roll over the finer grains and fall to the lower edge of the cone and to the outside of the silo
wall. Thus, the silo load appears to be red or coarse. The test sand is a four screen, 62 grain
fineness silica sand, selected because it is one of the most widely used distributions. But because
of sand's natural tendency to segregate itself according to size, samples taken at regular
intervals during discharge ranged in this load and discharge configuration from a grain fineness
number of 69 to a sample of fineness number 60 taken at about this point in the discharge. Our
test samples, therefore, had a range of nine fineness numbers. As the discharge stops,
the segregation of the coarse red grains on the outside and of the finer blue sand in the center
is very easily seen. The third demonstration showed how sand self-segregates in single center
load and discharge systems. The graph summarizes the flow characteristics illustrating the average
grain fineness number of seven samples taken periodically during discharge. The fourth
demonstration simulates a loading practice common in industry, a bucket elevator or side
loading system. The discharge point is in the near right hand corner directly beneath the loading
point. Our transparent model also shows segregation at the center of the cone of repose. In effect,
the sand segregation pattern that you now see is a cross section of the center of the load in the
previous demonstration. Notice that the interface or border between the reds and blues is like
interlocked fingers. The reason for this pattern is that the fines, being angular, tend to lock
before they can slide very far. The coarse red grains roll over them to the lowest extreme. The
buildup of fines directly over the discharge gate will be the first sand out. Samples taken at this
time are about 72 mesh, 10 points finer than the average grain fineness of the load. As we shall
see in the summary chart, the average fineness number of each successive sample will decrease,
reflecting the fact that the coarser sand is the last to be discharged. This view of what happens
at the center of the cone of repose also makes it easier to understand how sand funnels or raffles
out the single discharge point. Although some of the coarse red grains are beginning to discharge,
obviously the greater part of it cannot leave the silo until the angle of repose favors its
rolling down toward the discharge hole. This close look at the discharging sand shows that
the coarse sand is now coming out in surges. A sample taken at this time registered an average
grain fineness number of 64. Many foundries use gravity discharge systems that are formed by
putting a discharge pipe into the side of the silo. Depending on the location of the inside end
of the pipe, a number of different grain fineness sands can be taken from the same silo. Where such
discharge systems are used with continuous mixers, for example, the variation in fineness of the load
discharges can result in scrap castings. For the end of this discharge, the sample indicated an
average grain fineness of 60, a 12-point spread. The sand remaining in the silo after discharge
indicates that off-center loading and discharge reduces silo capacity and efficiency. As we have
seen, the fine grains tend to discharge first, and about one-third of the load discharges before
the coarse fraction of the load begins to flow. The fifth demonstration shows the flow through a
silo where the sand is loaded at a corner diagonally opposite the discharge opening. Because the coarse
red grains roll to the lowest point, this system forces the coarse sand to a position over the
discharge gate so that it discharges first. A finer sand lies beneath the loading point,
away from the discharge opening. This, in the previous arrangement, reduced the silo's effective
storage capacity because when the crest of the cone reaches the loading point, much of the volume
of the silo above the cone of repose remains empty. The initial discharge screen analysis
indicates a coarse sand of fineness number 48 due to the positioning of the coarse grains over the
discharge outlet. This represents a departure of 14 points from the original grain fineness number
of 62. Because such a great quantity of the coarse grains discharge first, the fines, due to their
angularity, build up to a steeper angle of repose, then tend to slide down as a mass, creating a
surge effect. For this reason, the next analysis indicates a grain fineness number of 64. As the
discharge continues, the variance of grain size and distribution continues its cyclic pattern.
This pattern of discharge is very detrimental to core production and casting quality. Concentrations
of coarse material form very porous cores, allowing the metal to penetrate between the grains,
causing rough finishes and burn-in type defects. A concentration of fines decreases the permeability
of the core and they lead to pinholes and other gas-related defects. Because of its
volumetric inefficiency and the cyclic behavior of grain size at the discharge point, this silo
design is not recommended. The sixth demonstration shows the flow and distribution from a single
discharge system with a deflector over the discharge gate. Loading is from a single source,
directly above the deflector. Theoretically, the fine sand that builds up in the center of the cone
of repose will be forced to migrate halfway to the final wall during discharge, to blend with the
coarse material which migrates in from the outer wall. The fine layer on the bottom is the result
of scattering that is exaggerated in a small model like ours. This layer is beneath the angle
of repose and can be ignored because it never discharges. This silo design represents one of
the sand and bin manufacturers' attempts to control by rebounding the segregation of sand.
Two methods of loading the silo to prevent segregation have been tried in the past. The
first approach, the use of multiple inlet loading systems, has been tried with some success,
but this method slows the rate of silo loading. Consequently, it hasn't won much acceptance. The
second method, using a rotary reblender, has proven satisfactory in the grain industry,
but the abrasiveness and density of sand precludes its use in our work. As the sand flows out,
it funnels down through a single rat hole, discharging the fines concentrated above the
deflector. Samples taken during the first two-thirds of the discharge reveal the grain
fineness number of 67, five grain fineness numbers higher than the original analysis.
Samples taken during the last third of the discharge indicated a grain fineness number
of 53, a total change of 14 grain fineness numbers. The remaining sand is predominantly
fine and should be allowed to remain in the silo. Attempts to use it could result in scrap
cores and castings. The deflector or baffle performs as badly as any single discharge
silo. The seventh demonstration will show the reblending characteristics of sand flowing
from a single center loading through a square silo and discharging from a multiple opening
system. The shape of the silo, incidentally, has no significant effect on sand segregation.
What you see in here in this film holds true generally for round and rectangular silos
as well. In this sequence, the lighting will show that the distribution of coarse red grains
is greatest at the farthest distance from the loading point, that is in the corners
of the silo. The thin light layer at the bottom of the reference number six was caused by
interrupting the loading. It's the result of the force of the resumed loading pushing
the fines that remain on the top of the cone of repose down over the cone. As it flows,
sand almost tends to self-segregate. The objective of this discharge system design then must
be to get the sand to re-blend into a consistent distribution. In this respect, multiple discharge
systems are generally far better than single discharge systems. Samples taken throughout
discharge showed a variance of about only three grain fineness numbers. Not perfect,
but far superior to any single discharge system that we have seen. As the silo empties, notice
that the cone of repose very gradually levels off until the sand reaches free-flow conditions
near the bottom. A good way to check multiple discharge systems is to look at the top of
the sand. A pileup over any area could indicate a plugged outlet. The sand should flow evenly
through the multiple openings. The corner load discharge configuration seen in this
film used 36 percent of the available silo capacity. The center load discharge system
used 65 percent, while the center load multiple discharge used 80 percent of the available
silo capacity. Silo's incorporating this multiple discharge system have three advantages over
single discharge systems. As we have seen, the first in, first out characteristic allows
sand the maximum opportunity to cool. The tendency of the sand to re-blend to a closer
consistency and the more efficient utilization of storage capacity gives multiple discharge
systems a clear advantage over alternative systems. Now let's consider another aspect
of sand segregation, the loading and transportation of sand in a railroad box car. The most common
method of loading is by a slinger or high-speed belt placed through the center door. In our
demonstration the pipe introduces the material into the corner of the car so that the sand
segregation can be seen in cross section. In normal loading the sand is thrown against
the center of the end wall of the car and drops into the same coning and segregation
pattern seen in the silo demonstration. After one end of the box car is loaded the slinger
is turned to load the other end. It is conceivable that if enough random samples of the sand
were taken from either cone before the car is moved that the sand would test out to an
average grain fineness of 62 and the samples could be representative. After all the sand
is loaded there is about eight feet of space between the bottom of the two cones. But after
the railroad has subjected the car load to coupling, switching, bumping and numerous
starts and stops the fines at the top of the cones will slide down and bury the coarse
grains at the lower extreme. When the car is open the sand often falls out of the doors
illustrating the degree of movement that has occurred during shipment. After shipment it
is impossible to obtain a good reliable test sample of the sand. We also built a scale
model 110 covered hopper car to see how its configuration affects segregation of the sand.
Before loading the car is inspected for overall cleanliness to ensure that the material previously
shipped in the car is not adhering to the sides of the roof. Acceptable cars are moved
below the loading station, the loading chute is lowered and the sand pours in. The inclined
wall of the car amplifies the segregation pattern seen in previous demonstrations. Because
their configurations are similar to hopper cars blower trucks affect sand segregation
in a similar manner. So much of this demonstration holds true for blower trucks as well. As the
coarse grains roll down to the lowest point the fines build up beneath the loading point.
This is significant because of the common practice of drawing test samples from the
sand directly below the loading hatch. It is one of the two places where sand is accessible
but obviously the sample is not going to be representative due to the concentration of
fine. Moving the loading chute to the second hatch buries the coarse sand with the remainder
of the load. Because of the layering effects and the burying of the coarse grains accurate
sampling from the top of the load after it has been delivered is impractical. No matter
how the car is loaded segregation of the sand will occur. A view of the top of the loaded
compartment illustrates the impracticability of obtaining representative samples from the
top of the hopper car or blower truck. When the car gets to its destination the load will
have settled with the fines sliding down the cones but there will be no significant re-blending.
To show how the other side of a typical loaded hopper car looks we poured sand through the
hatches on the far side of the car for an all around look at the segregation pattern.
Sand is the least expensive product in the foundry yet it is blamed for the most defects
in castings and other end products. Perhaps these demonstrations have brought to light
some of the reasons. Martin Marietta ships sand by many different carriers all of which
are subject to the self-segregating behavior of sand. The shipment as a whole is consistent
but the segregation of sand in motion as we have seen makes it impractical if not impossible
to take a representative sample that reflects the properties of the whole shipment as specified
by the customer and shipped by the supplier. Both compartments show a concentration of
fines at the crest of the cones with the coarse fraction rolling to the lowest extreme against
the rail car wall. Sampling from the bottom at the discharge gate is also impractical
due to the buildup of fines above the gate. The segregation pattern is the same in both
compartments. Fines covering the discharge gate then a layer of coarse sand then more
fines on the top of the cones. The analysis of the sand based on samples taken at unloading
is easy to predict. When the discharge gate is opened the fines lying on the bottom drop
out in the first two tons. Then the coarse red grains begin to funnel out as if through
a single discharge system. The first sample taken after the initial surge of fines registers
a grain fineness number of 58, four points below the norm of 62. As the sand continues
to discharge the samples indicate a fineness of 68 reflecting the discharge return to the
fine side of the spectrum with the last sample being 49 mesh average. The demonstrations
that you've just seen show that the self-segregated characteristic of sand is natural and inescapable
and that it makes sampling for acceptance testing generally impractical and unreliable.
What can you do about this situation? How can you be sure the sand that's going into
your process is of the consistency you need? The answer is twofold. First, select a reliable
supplier. Second, be sure your own sand handling process reblends the sand and minimize segregation
and provide the consistency you require.