# Backfill properties

Generally, the only time when engineers think about backfill is at the beginning of a new mine or when it becomes an “issue.”  An issue develops as a result of a crisis, like a fill failure or a blocked distribution system or when metal prices drop and cost savings need to be found.  Consequently, it becomes necessary to understand what needs to be “tweeked.”  The following section presents the many characteristics that describe backfill in ways that allows us to understand how this material performs.

Each type of backfill can be characterized by the following properties. Some of these characteristics can be modified by inexpensive means, while others are more costly to alter. In general, the two defining characteristics of any backfill are strength and flowability. All other characteristics play a role in either improving or impairing strength or flow.

There are several significant features of tailings, sand or waste rock that must be understood before serious work can begin on redesigning or optimizing an engineered backfill.

## Grain Size

A grain size distribution analysis of: tailings, sand or rock provides a “finger print” of the backfill material.  The analysis provides a count of particle sizes over the range of sizes for a given sample.  It allows for the comparison of different batches, as well as the establishment of benchmarks:

Example 1: Paste backfill is generally described as requiring 15% passing 20 microns.  A paste fill with lesser fines will have a higher probability of flow problems, especially in distribution systems where the ratio of the total vertical to horizontal flow, in full column flow systems is less than one.

Example 2: Most hydraulic tailings backfill lie in the range of 40 to 150 microns.  Natural sands tend to be courser, ranging from 150-600 microns (Thomas et al 1979).

The size distribution is commonly presented in the form of a cumulative distribution curve, as shown in Figure 1.  The graph shows the actual distribution of grain sizes found in a variety of base metal and gold tailings as well as natural sand.  Note that most of these tailings have had some portion of their fines removed through a process known as “desliming.”  A narrow range or steep curve indicates the presence of a dominant grain size.  A curve dominated by one grain size indicates a material that is poorly packed; it will have a high void ratio, a low packing density but have an excellent permeability and percolation rate.

Figure 1:  Coarse to fine mine fill

## Chemistry

The chemistry of the aggregate (i.e. sand, tailings or rock) material and any water used in the process should be determined.  Potable water is ideal, but any type of water—especially process water—should be tested, in backfill strength and flow tests, and then compared to potable water to determine if there are any adverse effects.

Although in most cases, natural sands have had tens of thousands of years or more of geologic processes to screen out soluble materials, this is not the case for tailings and rock fill.  It is highly recommended that all aggregate material to be used in backfill be run through x-ray diffraction testing(XRD) to determine its composition.  With respect to natural sands, the material should be screened to remove any organic material, large sized aggregate or agglomerations of wet or hardened material. Organic material will cause adverse reactions with any binders and over-sized aggregate can plug distribution systems.  Because they are generally inert, an abundance of silicate minerals are desirable in the aggregate.  The presence of sulfide minerals in amounts above roughly 5% can cause backfills strengthened with straight Portland cement to lose strength over time.  The presence of zinc in backfill, can delay the set time of binders from a few days to several weeks.  Removing or diluting the mineral or element causing the problem is one potential solution. An alternative solution could lie in the choice of binder.  Blended cements may play a role in reaching seven-day and long-term strengths.  More is written on this subject in the binder section.

## Moisture

The moisture content of aggregate is defined as:

$\mbox{Moisture Content} = \frac{ {\mbox{moist sample mass - dry sample mass}}}{{\mbox{dry sample mass}}}$

Knowing the moisture content of the aggregates allows the designer to calculate the correct amount of additional water required when cement has been added.

Mill personnel generally report pulp densities on a total weight basis:

$\mbox{Pulp Density} = \frac{ {\mbox{dry weight of aggregates plus binder}}}{{\mbox{dry weight of all solids plus weight of all liquids}}}$

Hydraulic and paste fill are often described in terms of their pulp density, for example: 72%, 81%, etc.

## Percolation

The percolation rate of hydraulic backfill, or more simply, the rate at which water passes through backfill, is significant because backfilled stopes have been known to fail when water cannot drain quickly enough while backfilling is in progress.  The backfill becomes saturated (the state where all the void space between particles is filled with water) to such a height that the pressure exceeds the design strength of the fill barricade. A large shock, resulting from a nearby blast, can cause the saturated fill to suddenly behave like a liquid, a term referred to as liquefaction.  Although the addition of as little as about 2% binder to the backfill almost eliminates the potential for liquefaction, the potential still exists if ponding is allowed to occur during the backfilling process.  Ponding occurs when backfill water builds up on the surface of the backfill as a stope is being filled.  As backfill discharges from the delivery pipes and hits the pond, the cement segregates from the sand and or tailings.  As a result, the backfill contains subhorizontal lenses of cement-rich and cement-poor zones.  This situation is to be avoided because it leads to a weakened backfill with unpredictable strength characteristics. In addition, to overcome the ponding problem, the backfill delivery rate must be slowed to accommodate for the slow permeability rate.  Slowing the delivery rate usually means stopping the pour and waiting for the pond to drain.

### Percolation rate

Thomas (1979) defines the percolation rate as follows:

$V_p = \frac {q}{A}$

where Vp = percolation velocity, in m/s

q = rate of flow in m3/sec, and

A = cross-sectional area of sample normal to the flow direction, m2

Although a rate of 100 mm/h is accepted as the benchmark, the means of measuring the percolation rate make it difficult to get repeatable results. A more pragmatic approach involves improving fill practices. Such practices may include:

• the addition of a minimal amount of cement to all stopes to be backfilled (excluding rock fill), approximately 2%
• the adjustment to the size distribution of the backfill, either removing fines and or adding course material
• the use of weeping tiles in the stope to assist fill drainage and allow adequate drainage at barricades
• prevention of ponding by monitoring pours and be prepared to stop the pour to allow drainage
• maximizing the pulp density of the slurry (i.e. reducing the water content) such that adequate slurry velocity in the distribution system is still achieved.
• Prevention of flush water from entering into the filled stope.

Note: In a true paste backfill there is no excess, or very little water, that settles out of the paste.  The hydration reaction in the binder requires a small proportion of water as well.  Consequently, percolation rate is not an important factor in cemented paste backfill.

## Solids Relative Density

Solids relative density is the ratio of the density of a unit volume of fill compared to the density of a unit volume of water.  For example, a backfill sample with a density of 2,500 kg/m3 and water at 20°C with an approximate density of 1,000 kg/m3 gives a relative density of 2.5.  Note that the value of the relative density is expressed without units since both numerator and denominator have units of kg/m3.

## Void Ratio

Void ratio is defined as the ratio of the volume of voids to the volume of solids. It is a measure of the packing density.  It can be determined as follows:

Based on the dry weight of the sample,

$Void Ratio = \frac{(sample~volume~x~density~of~solids) - 1}{mass~of~solids }$

McGary (1961) has determined that an ideally packed material—with, for example, three distinct grain sizes such as coarse, medium and fine components—occurs at volume percentages of 66:25:9 respectively, in the size ratio of 77:7:1 ( mesh sizes of 7,60, 400 respectively).  Such a system, with spherical particles, would attain a void ratio or packing density of 90% of its theoretical density.  Consequently, only 10% voids would remain.  A void ratio of 83% is considered to be a “good” density because should binder be required, very little of it would be necessary for void filling, making more binder available for coating particles, thereby enhancing cohesion.

## Strength

The strength of backfill can be estimated by determining its uniaxial compressive strength (UCS).  It is common practice to mix backfill samples in the lab using tailings, sand, and/or rock samples, based on the mix design used at the mine site.  The backfill is typically placed in 200mm x 100mm or 300mm x 150mm cylinders and cured at room temperature in 100% humidity until ready for breaking.  Previously mixed samples can also be collected from underground by installing a “T” connection and a valve in the delivery pipeline, close to where the fill enters the stope and allowing the fill to fall into a barrel with cylinders secured and prearranged such that they fill up. Once the binder has set, in approximately 24 hours, the cylinders can be moved to the lab for curing and breaking.  The average of at least three cylinder breaks is generally represented as the strength of the in situ fill.

In situ strengths are difficult to obtain depending on the homogeneity of the backfill. Drills with special bits can be carefully used to recover core from backfilled stopes. Alternatively, where fill is exposed, large chunks can be recovered and hand drilled in the lab to obtain cores. These cores can then be prepared and broken to obtain the UCS. Experience has shown that in situ strengths are often greater than the strengths provided by laboratory testing. Belem et al. (2002) reports that the combination of drainage and settlement pressure of the in situ paste backfill mass is likely the main reason for the difference in measured strengths. An additional contributory factor lies in the fact that in situ strengths are usually measured months or years after the initial pour while laboratory test results are obtained in usually 28 or 56 days. The time differential gives the binder additional time to strengthen, and therefore, the older the break the higher the differential in strength.

Several of the characteristics that have been discussed have an influence on backfill strengths. The grain size distribution, for example, plays a role not only in the strength, but also in the percolation rate, void ratio, moisture content and flowability of the fill. The interdependence of these characteristics means that if strengths are being optimized and some characteristics are changed for the benefit of strength, other characteristics may show deterioration, for example, in hydraulic backfill, if the void ratio of the tailings has been minimized—or conversely, if the density has been maximized— higher strengths with minimal amounts of binder are possible. However, the presence of too many fines (greater than 15% under 20 microns) can reduce the permeability which slows the drainage of water. This can lead to the ponding discussed in Percolation rate. Ponding may also lead to another phenomenon known as liquifaction, discussed in Percolation. Consequently, any changes must be thoroughly evaluated, usually in the lab followed by a bulk trial in an isolated stope.

The largest influence, by far, on backfill strength is the addition of binder or cement. However, the combination of water to binder, commonly referred to as the water to cement ratio plays a significant role. The over-addition of water, compared to binder has a negative effect on strengths.

Water is required:

1. For the cement hydration reaction
2. To provide a transportation medium when mixed with rock fill. This is accomplished by allowing the paste to become sticky and coat the rock, and by facilitating the flow in pipe lines for hydraulic or paste fill.

It must be remembered that binder is by far the most expensive ingredient in backfill. Consequently, the desire to improve backfill strength by adding more binder is resisted by the cost of additional binder. A cost vs. benefit analysis is recommended to determine the addition rate at which the binder offers maximum benefit to an operation.

Tell-tale signs that backfill strengths are inadequate include consistent high dilution during blasting or frequent fill failures during the mucking operations. Characterizing the backfill performance, in terms of strength, and further comparing this to the design strength, allows the backfill practitioner to determine the safety factor. The safety factor provides a starting point for improving backfill performance or making backfill more efficient.

## Rheology

An understanding of the flow or rheology of hydraulic and paste backfill is a key characteristic described by various parameters that are required for the design of a distribution system with minimal disruptions and longevity.  Rheology (from the Greek rheos –flow and logos – knowledge) encompasses the science of flow phenomena.  Fluids can behave in many different ways due to factors such as particle size, flow velocity, pipe diameter and density. In hydraulic backfill, the deposition of solids occurs when the flow velocity falls below a limiting value known as the critical velocity (Vc).  Problems related to Vc express themselves in the extreme form as plugged pipes.  The onset of plugging can occur:

1. in pipe sections furthest from the vertical, usually with an extensive horizontal run,
2. at any change in direction of slope, or
3. when a change in the consistency of a slurry occurs.

Thomas (1979) simplifies the classification of mixtures found in hydraulic backfill as either homogeneous or heterogeneous.  Homogeneous mixtures contain fine particles (<50μm) that are uniformly distributed across t he pipe cross-section.  During pipe flow, a homogeneous mixture behaves as a pure liquid of the same density as the slurry and with an apparent viscosity that depends on the solids concentration.

A heterogeneous flow is characterized by non-uniform distribution across the pipe cross-section.  The solids that are larger than those in homogeneous mixtures are supported in suspension by fluid turbulence.  The concentration increases towards the bottom of the pipe. Backfill in which less than 25% of the particles are finer than 50μm can be characterized as heterogeneous.  Most hydraulic fill is heterogeneous. It should be noted that within the heterogeneous mixture there are various flow regimes that are dictated by particle size and flow velocity. At high velocities, the distribution of solids within heterogeneous mixtures, at higher concentrations, approaches that of a homogeneous mixture. This regime is referred to as pseudo-homogeneous. At lower velocities, the solids concentration increases towards the bottom of the pipe and friction loss is higher than that of an equivalent liquid. At even lower velocities, larger, denser particles tend to fall to the bottom of the pipe, where they either move along in a series of intermittent jumps, in a flow regime known as saltation or slide along the bottom of the pipe en masse in a process known as moving bed flow. At yet lower velocities, the largest particles settle out completely, forming a stationary bed. Such a condition reduces the cross-section of a pipe and renders it completely plugged, a condition that must be avoided.

The friction factor, which is a measure of the friction resulting from the flow of backfill media through the pipe, is used to determine the head loss due to friction. The Darcy-Weisbach equation, below, is used in the calculation:

$H = f.(\frac{L}{d}). (\frac{V^2}{2g})$

where

f = friction factor
L = pipe length (m)
V = flow velocity (m/s)
d = pipe diameter (m)
g = gravitational acceleration (m/s2)

### Pipe wear

In a gravity flow system where the pipe is full throughout its length, the slurry will flow at the velocity where the friction head loss is equal to the static head difference between the surface plant and the pour point.  Consequently, the greater the ratio of vertical to horizontal lines, the greater the flow velocity.  Especially in deep mines, this velocity is often too great.  Several reports have shown that the wear rate of pipes is proportional to the velocity of the slurry being transported (McKibben and Shook, 1991), and that of the factors affecting pipe wear, velocity has the greatest effect (Steward and Spearing, 1992).  Consequently, methods to control the velocity of the flow are critical in order to control pipe wear.  Several strategies to reduce excessive slurry velocity include:

• increasing fill pulp density,
• using non-vertical pipe inclinations, reducing pipe diameters, and

Evenly distributed pipe wear across a section is considered normal and is to be expected. However, uneven pipe wear is a common symptom of burst pipes. Locations where uneven wear can be anticipated include near elbows or surge points, in vertical pipes, after free-fall.

Strategies that minimize pipe wear include:

1. minimizing the number of pipe direction changes,
2. making direction changes gradually using long transition pieces,
3. designating high wear pieces for sacrificial wear, and designing these sections such that they can quickly and regularly be changed,
4. letting fill wear against fill, i.e. the dead end portion of a “T” piece utilized at the transition between a vertical and horizontal will fill up with backfill allowing the vertical flow of slurry to deflect across it to the horizontal pipe, and
5. using wear-resistant materials or coatings in the pipe.