The Lucky Friday Mine, owned and operated by Hecla Mining Company since 1958, is located approximately 1.6 km east of Mullan, ID (Fig. 1). Lucky Friday is the oldest and deepest currently operating mine in northern Idaho’s Coeur d’Alene mining district. Utilizing underhand cut-and-fill stoping to mine narrow, sub-vertical veins of lead–zinc–silver ore at depths currently around 2300 m below ground surface, the mine produces an average 725 t/day. Recent completion of a new 1140-m winze, the No. 4 Shaft (Sturgis et al. 2017), has extended the mine to just over 2900 m below ground surface, making it the third deepest operating mine in the western hemisphere (Alexander et al. 2018).

The Lucky Friday Vein was historically the principal ore-bearing structure at the mine until production began in 1997 from several mineralized veins in the Gold Hunter property located about 1500 m northwest of the original Lucky Friday workings. The upper Gold Hunter deposit is hosted in the Wallace formation of the Precambrian Belt Series and transitions into the St. Regis formation below the 5900 level (about 1800 m below the shaft collar). The Wallace formation lithology consists of weak, highly foliated argillite, argillite alternating with silt caps, and siltite. The argillites of the transitional St. Regis formation below the 5900 level have increasing silt and quartzite content with depth.

The Gold Hunter deposit lies between two west–northwest trending faults, which are separated horizontally by about 457 m, and consists of a system of several definable veins striking west–northwest and dipping 80° to 90° south, parallel to foliation. Production is primarily from the 30 vein—a composite of closely spaced veins and veinlets averaging more than 1.2 m wide. A schematic of the current extent of mining in the 30 vein is shown in Fig. 2. The actual depth of cover is roughly 270 m greater than the depth below collar.

Depending on the rock type and bedding orientation, average unconfined compressive strength (UCS) values for the argillite host rock and vein rock range from about 97 to 122 MPa (Seymour et al. 2016). The major in situ stress is horizontal and oriented northwest with a magnitude about 1.5 times the vertical stress (Whyatt et al. 1995). As a result, the stress magnitude is comparable with deep, South African gold operations (Alexander et al. 2018).

After an undercut stope is excavated and supported, a layer of broken rock or “prep muck” with a thickness of 0.4–0.6 m is spread on the floor of the stope. No. 7 DYWIDAG^® bolts, 1.8 m in length, are driven vertically into the loose muck on roughly a 1.2 × 1.2 m² pattern to retain potential slabs that may form as the fill is compressed by wall closure. The bolts are fitted with steel plates and nuts and wired together as shown in Fig. 4.

Classified mill tailings are mixed with 8–10% binder (25% cement and 75% finely ground, granulated blast furnace slag) at a surface batch plant and gravity-delivered in a paste-like consistency to the stope via an underground pipeline distribution system. To contain the paste backfill during placement, a wooden fill fence is constructed across the width of the stope, limiting the lateral extent of the backfill pour to about 46 m and restricting the height of the pour. The process creates a backfill beam, having a vertical thickness of about 3 m, and leaves a void or gap, approximately 0.3 m in height, between the upper surface of the backfill pour and the bottom surface of the previously filled cut (Fig. 5).

After the east and west stopes on either side of the slot drift have been backfilled, the paste is allowed to cure and gain sufficient strength. A subsequent undercut stope is then mined in the vein beneath the newly formed backfill beam (Fig. 6). Loose muck that was placed on the floor of the previous cut protects the fill during blasting and falls away from the back as the heading is advanced. To support the CPB back, chain link mesh is installed overhead using the exposed DYWIDAG^® bolts and additional friction bolts as needed. Additional rock bolts and mesh are installed to support the stope walls. The CPB, bolts, and mesh thus form a stable back under which mine personnel can safely work (Fig. 7).

The CPB mix designs used for underhand stopes are provided in Table 1. Paste fill typically contains at least 15% by weight of particles less than 20 microns (Brackebusch 1994; Henderson et al. 2005). The results of a sieve analysis performed on tailings from the mine are provided in Table 2.

Based on past practices, Hecla requires the CPB in their underhand stopes to have a 28-day UCS of about 2.75 MPa. However, unpublished results indicate that the average UCS of CPB samples collected at the batch plant is typically around 3.4 MPa or higher, and that higher in-place strengths are also measured from CPB samples collected through underground coring. These results are also supported by additional tests with CPB samples obtained by coring large backfill slabs brought to the surface (Johnson et al. 2015). Brazilian and splitting tensile strength tests with CPB samples have shown that the indirect tensile strength of the paste fill is normally about 10% of its UCS (Johnson et al. 2015). In addition, the in-place density of the CPB is about 2050 kg/m³ with a porosity ranging from 35% to 40%.

As explained by Stone et al. (2019), there are currently no established standards for preparing and testing cemented backfill samples. As a result, standards for other materials such as concrete or rock core are loosely adapted for use with CPB, a much softer material. The CPB sample preparation methods and testing procedures, which were used for the unconfined compression and indirect tensile tests mentioned above, roughly followed guidelines developed for concrete or rock core by the American Society of Testing and Materials (ASTM) and the American Concrete Institute (ACI).

Rotational failure is not kinematically possible in vertical or near-vertical stopes and therefore does not need to be considered at the Lucky Friday Mine. Sliding failure is kinematically possible but has never occurred at the mine. Considering the Mitchell equation for sliding stability (Eq. 1) demonstrates why this is the case. (1)(σv+d×γ)>2(τfsin2(β))(dL), where σ_v is the vertical stress from loading above the sill, d is the thickness of the sill, γ is the unit weight of the paste fill, τ_f is the shear strength of fill/rock contact, β is the stope dip angle, and L is the span of the stope.

Assuming rough stope walls, sliding failure would require mobilization of the paste shear strength. Conservatively assuming that the shear strength of the contacts is due only to cohesion and neglecting normal stress yields factors of safety well in excess of 100 for typical stope geometries and backfill UCS as low as 1.37 MPa (tensile strength is assumed to be 10% of UCS). Any horizontal pressure induced on the fill from stope closure would further increase sliding resistance.

Likewise, caving stability—a function of fill tensile strength and span width—is also not of concern. The Mitchell equation that governs the caving stability of the fill is provided by (Eq. 2): (2)L×γ>8σtπ, where σ_t is the tensile strength of the paste fill.

For typical stope widths of 3–4.5 m, assuming a tensile strength equal to 10% of UCS results in safety factors well in excess of 20, even for a low strength fill of 1.37 MPa.

Flexural stability is of primary concern in slot intersections during the period prior to any significant undercutting of the previous backfilled stope. This is because the slot intersection is typically the widest open span (up to 6 m diagonally), with initially little closure occurring before mining the cut (Fig. 8).

The Mitchell flexural stability equation can be used to calculate factors of safety for paste fill beams (Eq. 3): (3)(Ld)2>2(σt+σc)σv+d×γ, where σ_c is the horizontal confinement stress.

Figure 9 presents the results of a parametric stability analysis using Eq. 3, assuming self-weight loading of the CPB beam, a tensile strength of 10% UCS, and neglecting the impact of wall closure, which initially tends to confine the fill and improve its flexural stability. Thickness is the primary factor that determines the flexural stability of a backfill beam for these limited mining spans. As a result, the stability is significantly impacted by any factor that tends to reduce the effective thickness of the beam, such as cold-jointing.

The quality control of the fill placement process is therefore vitally important for ensuring design strength and preventing cold joint formation. Stacking and surging of the paste fill can occur near the outlet of the paste pipeline while filling the stope, causing the backfill to be deposited in a discontinuous or intermittent manner. This results in horizontal layering and cold jointing within the overall mass of the paste fill pour. Stacking and surging problems were observed at the Lucky Friday Mine as mining progressed beyond the 5900 level (1800 m below shaft collar) and were attributed to premature hydration of the cement binder, resulting from very long pipeline transport distances. As mining depth increases, the temperature in the rock and underground workings also increases. These higher temperatures, in addition to the frictional heating in the long distribution pipelines, may result in acceleration of the cement hydration process, increasing the paste viscosity and thus, degrading workability and flow characteristics. The mine now uses a binder consisting of 25% cement and 75% finely ground, granulated blast furnace slag to prevent stacking and surging. The slag has been effective in retarding the hydration process, allowing the workability of the mix to be maintained throughout a pour.

Control measures, including monitoring and control of the moisture content of the tailings, schedules for calibrating scales and meters in the batch plant, procedures for sampling and testing the strength of the paste, field tests and procedures for validating flow characteristics and detecting cold joints, training and certification of batch plant operators, and instructions for documenting, recording, and reporting backfill-related information, are also critical for maintaining good quality control of the final placed product.

A unique challenge to underhand cut-and-fill mining at Lucky Friday Mine is dealing with horizontal stress induced in the CPB as a result of stope wall closure. During undercutting, the stope walls converge in response to the high horizontal ground stresses and compress the CPB. These horizontal loads cause crushing and extensional fracturing near the fill surface as shown in Fig. 10.

Results of coring have found that these fractures are primarily horizontal and occur typically within 0.3 m or less of the surface of the beam (Fig. 11). While the bolts and chain link mesh contain spalling on the underside of the beam where personnel are working, the upper surface of the paste fill is unconfined and deforms into the gap above (Fig. 12). The closures encountered in the mine typically exceed the elastic limit of the fill. Therefore, a very brittle fill should be avoided, and a more ductile fill with significant residual strength is desired.

As the underhand mining front continues to advance deeper, the cemented backfill is subjected to further horizontal closure with each additional undercut. This closure eventually compresses the fill. After substantial hanging wall-to-footwall closure, the backfill will, in theory, behave as a compacting material and begin to strain-harden, gaining stiffness as it is compacted and its void spaces and fractures are compressed. The number of undercuts required to initiate this strain-hardening process will depend on the porosity of the fill and confinement.

Custom-designed closure meters were built to measure hanging wall-to-footwall convergence in the backfilled stopes (Fig. 13). The closure meters are installed in the stope prior to backfilling and encapsulated during the paste pour. The body of the closure meter consists of telescoping sections of steel pipe and tubing attached to steel end plates, which are bolted to the stope walls. Linear position transducers are mounted internally in the closure meter to measure displacement as the stope walls converge. Pressure cells mounted on steel anchor plates measure the horizontal change in stress in the CPB.

Two types of closure meters were fabricated and installed: single-acting closure meters with one position transducer for measuring displacement across the full length of the instrument (hanging wall-to-footwall closure) and double-acting closure meters equipped with a second position transducer that is used for measuring displacement to a center plate located at the mid-span of the instrument (measuring closure from the hanging wall to the mid-span of the stope).

UniMeasure^® HX-P510 linear position transducers were chosen to measure stope closure. The position transducer consists of a rotary potentiometer that is encased within an environmentally sealed, waterproof housing. The rotary potentiometer measures a voltage output as a stainless steel wire cable is extended or retracted from the housing. These voltage measurements are in turn converted to displacement units to reflect a change in position of the cable. The selected model has a resolution of ± 0.01% and a measurement range of 63.5 cm.

Backfill horizontal stress is measured using pressure cells mounted directly to the closure meter’s steel anchor plates (Fig. 14). Geokon Model 4810 contact pressure cells with a measurement range of 0–7.5 MPa were selected. This type of earth pressure cell is specifically designed to measure soil pressures exerted on a structure.

This instrument consists of two circular stainless steel plates that are welded together around their edges, forming a narrow cavity which is filled with hydraulic oil. A thick back plate protects the instrument and mounts directly to the structure. A thin front plate is welded to the back plate, forming a flexible hinge for increased sensitivity to pressure changes. Pressure applied to the cell induces an equal pressure on the internal hydraulic fluid that is, in turn, sensed by a vibrating wire transducer connected to the cavity between the two plates by high-pressure stainless steel tubing. A measurement of the change in the frequency of the vibrating wire is converted to pressure using a calibrated gage factor supplied by the manufacturer. The pressure cell is also equipped with a thermistor to help account for the influence of changes in temperature on the instrument’s readings.

Although current data acquisition system (DAS) designs use the mine’s leaky feeder radio for data backhaul to the surface, the instruments themselves must still be wired to the dataloggers. Depending on the distance into the stope, and the location of the datalogger station, 60–180 m of trunk line must be connected to the instruments, hung through the stope and slot to the datalogger, and protected to minimize the potential for damage when mining the undercut.

To protect the transducer signal from electrical noise, each pair of lead wires in the instrument cable is wrapped in Mylar tape with an aluminum foil shield. A bare copper wire is routed with each pair of lead wires to drain any induced currents to ground. Direct burial cables with thick polyurethane jackets are used to protect the encased wires. For additional protection, the instrument cables are either: (1) placed in a flexible, plastic split-tube hose and hung from the back to prevent fill encapsulation or (2) run through polyurethane DriscoPipe^®, a product used by the mine for CPB distribution, and hung on the rib.

The closure meters and pressure cells are monitored every 2 h by Campbell Scientific dataloggers located at substations in nearby slot drifts. A typical datalogger station consists of the following components: CR1000 datalogger, AVW200 vibrating-wire analyzer, AM16/32B multiplexers, and various communication interfaces, depending on the specific link to the mine’s communication system—either a fiber optic cable or a leaky feeder radio. Both of these systems are linked to a computer server at the surface.

The DAS can be accessed remotely, and the instrument data can be viewed on Hecla’s corporate intranet website by mine management and NIOSH researchers on an almost real-time basis. This allows the mine staff to use the instrument data for daily operational decisions or safety concerns and for remote monitoring for maintenance and timely repairs.

Figure 17 shows typical measured closures from one of the monitoring sites. The approximate undercut start times are indicated on the figure by vertical dashed lines. With each undercut mining advance, the full-span closure meters generally measured a consistent increase in stope closure, averaging about 7.6 cm. The half-span closure meter measurements were a fraction of the total closure, but varied from site to site. Most of the stepped increase in closure occurred as underhand mining advanced directly below the locations of the instruments. The specific cause of the signal interruption in the half-span closure measurements shown in Fig. 17 has not been conclusively identified.

In 11 stope, a total of five undercut advances were monitored over a period of about 2 years. Measurements collected from all 11-stope backfill instruments from September 18, 2014 through September 7, 2016 are shown in Table 3, grouped by the monitoring locations noted in Fig. 15. Cuts for which data are not entered indicate that the instruments were not providing reliable measurements at the time of that undercut.

Two instruments (CM 1-L, CM 3-S) stopped working shortly after the first undercut. CM 2-L began providing reliable results again during Cut 4. Moisture intrusion may be the cause of this intermittent behavior. Three instruments (CM 2-S, CM 3-L, CM 4-L) provided continuous readings through all five cuts. Total measured stope closure ranged from 40 to 50 cm during this period.

Measurements recorded from the 15-stope backfill instruments from March 18, 2016 to March 13, 2017 are provided in Table 4, grouped by the monitoring locations shown in Fig. 16. After almost 6 months and three undercut advances, all of the instruments installed in the 15 stope were continuing to function, except CM 5-S and CM 7-S. Five closure meters (CM 5-S, CM 5-L, CM 6-L, CM 7-L, CM 8-L) were able to monitor through Cut 4. Three instruments (CM 5-S, CM 5-L, CM 8-L) were functioning after Cut 5, measuring as much as 33 cm of closure.

Figure 18 shows a typical example of the horizontal backfill pressures measured at one of the closure meter sites. As the first undercut heading is driven beneath the locations of the instruments, the horizontal pressure in the CPB increases rapidly. Loading of the CPB beyond its intact compressive strength occurs shortly thereafter, as depicted by the peaks in the pressure measurements during the first undercut. Average residual pressures of 1–2 MPa indicate, however, that the fill remains stable over several undercuts.

In 11 stope (Table 5), the maximum horizontal pressure measured during the first undercut ranged from 1.59 to 5.40 MPa and averaged about 3 MPa for the ten pressure cells, exceeding the 28-day target strength of 2.75 MPa. Fill pressure varied depending on the monitoring location in the stope and the specific placement of the pressure cells. The largest fill pressures were generally measured at the midspan (center) of the stope rather than at the hanging wall (south) or footwall (north), with a few exceptions.

Cemented paste backfill pressure measurements in 15 stope (Table 6) were similar to those of 11 stope, with maximum horizontal pressure measured during the first undercut ranging from 1.85 to 5.47 MPa and also averaging about 3 MPa for the 12 pressure cells. Design changes for the 15-stope closure meters provided better protection for the instruments and their lead wires, significantly improving the longevity of the instruments, especially the pressure cells.

Similar to a uniaxial or unconfined compression test, the maximum fill pressure appears to occur at the mid-span of the stope. Although the precise type of yielding and its initial location within the stope is unknown, the yield mechanism is likely a combination of tension and shear that is affected by stope geometry, local geology, and the presence of any lower strength zones within the fill.

While the 11-stope instruments were initially read manually for several months, the 15-stope instruments were monitored continuously with a datalogger before, during, and after the pour. This allowed determination of the effects of stope filling and paste curing on the readings.

Temperature changes can significantly affect pressure cell measurements in cemented backfill (Tesarik et al. 2006). As paste is poured and begins to cure, its temperature rises due to hydration of the curing cement. This increase in temperature causes the pressure cells to give a false indication of initial applied stress increase. Although the instrument is supplied with a temperature correction, this correction is for the vibrating-wire sensor itself, not the pressure cell bladder and contained oil.

The maximum fill temperature recorded during curing averaged about 49 °C ranging from 41 to 57 °C. Figure 19 shows a typical response of a pressure cell before, during, and after a pour. Once the stope is filled and the CPB begins to cure, a rapid increase in temperature is observed, which then levels off. This rise in temperature is accompanied by an apparent rise in pressure of about 0.2 MPa. After the temperature levels off, pressure changes can be attributed solely to stope closure. The temperature effect is negligible for long-term monitoring because the backfill temperature stabilizes, and the in-place CPB is usually subjected to pressures well in excess of the temperature response.

To further interpret the geomechanical response of the CPB to mining, in situ stress versus strain was plotted using the closure and fill pressure measurements. Many 11-stope instruments stopped functioning after the second undercut, making it difficult to construct stress–strain curves for each closure meter. Therefore, stress–strain curves were created for the east and west sides using measurements averaged from the instruments that were still functioning. The results are shown in Fig. 20.

The longevity of the 15-stope instruments allowed the in situ stress–strain behavior of the CPB to be constructed for each of the four 15-stope monitoring sites. Horizontal stress in the paste fill was calculated by averaging the pressure measurements obtained from the three pressure cells at each site, while horizontal strain was computed using the total stope closure data at each site. The 15-stope stress–strain results are shown in Fig. 21.

The 11-stope and 15-stope data show that peak strength occurs between 0.5 and 1% strain, consistent with laboratory testing of the paste fill. For a typical stope width of 3.4 m, this represents only about 2.5 cm of closure.