From deep geotechnical investigations to real‑time sensors and a push toward dry‑stacked “cake,” best practice for coal tailings storage is moving fast away from big wet impoundments — and toward tighter water control and higher stability.
In tailings storage, the most dangerous failures often trace back to a simple culprit: too much water in the wrong place. As one expert review puts it, “the main cause [of failures] is too much water in the wrong place” (link.springer.com). That’s driving a two‑track shift in coal tailings management: tougher front‑end design and construction rooted in geotechnical data, and a downstream operational pivot to dewatered, stacked tailings that drastically cut ponded water.
This isn’t just theory. Design guides from the US EPA’s Office of Solid Waste lay out the playbook—spanning site investigation, embankment zoning, water balances, spillways, and surveillance—and the industry literature now emphasizes a “robust surveillance program” integrating conventional instruments with UAVs and satellite imaging (nepis.epa.gov; www.mdpi.com).
Geotechnical investigation and foundation data
Safe dams start underground. Engineers map subsurface geology and groundwater via boreholes, test pits, or CPT (cone penetration test) soundings to identify soil/rock layers, faults, buried channels, and hydraulic conditions (nepis.epa.gov). Lab testing of samples covers moisture content, grain‑size distribution, Atterberg limits (plasticity indices), consolidation, shear strength, and permeability; rock specimens are tested for shear along weak seams (nepis.epa.gov).
In situ tests—SPT (standard penetration test), CPT, vane shear, and pressuremeter—capture field stiffness and strength to complement lab data (nepis.epa.gov). The investigation flags weak or compressible zones, selects borrow materials (e.g., sand or gravel for stability zones), and determines if foundation treatment such as compaction, grouting, or cut‑off trenches is required. One guideline is explicit: “An investigation will assess site geology (strata depth, thickness, continuity, composition), hydrogeology, [and] geotechnical properties of soil and rock affecting design” (nepis.epa.gov).
Embankment configuration and construction control
Best‑practice embankments use zoned cross‑sections: a low‑permeability core or upstream liner (clay or geosynthetic) to block flow, flanked by permeable shells and internal drains to collect seepage (nepis.epa.gov). For raising methods, downstream or centerline raises are favored for high‑saturation tailings and seismic areas because they do not rely on tailings for support—unlike upstream raises, which historical data show were prone to liquefaction during earthquakes (nepis.epa.gov).
Designers commonly set gentle slopes (3H:1V or flatter) and target factor‑of‑safety (FS) > 1.3 under static loads; seismic checks use pseudo‑static or dynamic analysis, with some guidelines calling for FS ≥ 1.1 under seismic loading. Construction quality control locks this in: filter gradations must prevent piping, and soil fills are compacted in lifts to meet density and permeability specifications. Where upstream or centerline raises are used, tailings deposition is tightly managed so beaches form a competent, coarse foundation. The EPA guide cautions upstream construction is only acceptable if “the tailings beach forms a competent foundation” for the next dike (nepis.epa.gov).
Internal drainage—chimney drains, blanket drains, and toe drains—lowers the phreatic surface (the internal water table) and relieves pore pressure. Seepage collection is routed to monitored discharge points, and, where needed for suspended solids removal ahead of reuse circuits, facilities may draw on common industrial unit operations such as clarifiers (clarifier) and physical separation systems (screens and oil removal) consistent with site water balance objectives.
Flood hydrology, freeboard, and run‑on control
Water design starts with a mass balance: inflows from ponded process water, rainfall and run‑on, and outflows via evaporation, spillage, and seepage (nepis.epa.gov). Hydrologic modeling of storm hydrographs and frequency curves informs spillway sizing and pump capacities (nepis.epa.gov).
Standards are tightening. An Indonesian update (SNI 3432:2020) explicitly requires sizing earth/rockfill dams—including those retaining mining waste—for the 100‑year flood peak (www.bsn.go.id; www.bsn.go.id). A case study cited by the EPA provided ~0.9 m (3 ft) of operational freeboard above calculated flood storage (nepis.epa.gov). Designers warn that higher freeboard means more water stored—and more seepage to manage—so run‑on diversions around the basin are standard, from ditches to bypass channels (nepis.epa.gov).
Impoundment water and seepage management
Inside the impoundment, the goal is a thin “exercise” pond—enough water to aid consolidation, without acting as a surge reservoir (nepis.epa.gov). Decant systems (fixed towers or floating platforms) return water via pumps back to the plant, sized to the site mass balance (nepis.epa.gov); where recycled flows require polishing to meet internal reuse targets, plants may incorporate ultrafiltration as a pretreatment to membranes (ultrafiltration) within broader industrial water trains (RO, NF, and UF systems), configured to site conditions.
Drains—horizontal base blankets, chimney drains, toe trenches—keep the phreatic surface well below the crest. Runoff diversions at the rim minimize stormwater inflow (nepis.epa.gov). One cited operation (Stillwater Mine) set a fixed spill volume for the design flood and maintained a 3 ft crest freeboard, with routine surveys to verify the water level; operating criteria require >0.5–1 m freeboard at all times to account for waves and precipitation (nepis.epa.gov). Upstream of treatment, facilities also rely on coarse primary removal where needed—screening and oil separation are typical industrial measures consistent with stormwater controls (waste‑water physical separation).
Real‑time instrumentation and TARP governance
Continuous monitoring is now the norm. Piezometers (instruments that measure pore‑water pressure) are grouted in nested arrays to map the 3D phreatic surface, with vibrating‑wire sensors logging hourly data via telemetry (link.springer.com). Inclinometers in borehole casings track lateral movement; settlement points and castellations on the crest record vertical deformation with millimeter‑scale leveling or differential GPS; seepage weirs and automatic flow logs capture any increases that could indicate internal erosion; and water‑level indicators on decant structures verify pond elevation relative to design (nepis.epa.gov).
Visual inspections are backed by drones and satellite tools; guidelines emphasize a “robust surveillance program” combining conventional instruments with UAVs and even satellite InSAR to catch precursors to failure (www.mdpi.com; www.mdpi.com). Data feed a Trigger–Action Response Plan (TARP): threshold exceedances trigger predefined steps, such as lowering pond levels via pumps. Documented cases show settlement surveys and foundation piezometers prompting maintenance before damage occurs (nepis.epa.gov; link.springer.com).
Dewatered and dry‑stack tailings trend
The biggest structural change is getting rid of the big pond. Thickened, paste, or filtered tailings lift solids to >50–60% by weight—creating a moist “cake” that’s stacked and compacted like earth fill, without a large water‑retention dam. Literature reports that “low water contents” allow “greater support” and “reduce the risks of instability/earthquake liquefaction” (www.mdpi.com).
Safety and environmental gains are direct: no overtopping hazard, far less potential for toxic slurry release, and a smaller footprint. Research projects that “mining operations with high production of tailings will apply dry stacking without dams to guarantee sustainability and community safety” (www.mdpi.com). With stacked deposits, dam heights are scaled down “considerably” because only smaller containment berms are needed (www.mdpi.com), and progressive reclamation can start earlier.
The water savings are headline‑worthy: dewatering can recover on the order of 80–95% of process water. One large automatic filter press announces up to 95% water recovery for reuse (fls.com); a South African mine reported ~80% recovery via filter presses, sending clean water back to the plant (www.engineeringnews.co.za; www.engineeringnews.co.za). Over 90% water reuse is now achievable, helping arid‑region mines approach “zero discharge” operations. Where plants do polish recovered water to match process needs, they draw from familiar industrial toolkits such as membrane trains (membrane systems) configured to the site.
Adoption is accelerating despite higher capex and energy for filtration. Analyses cite a surge in filter presses and vacuum belt filters across new projects, and majors have signaled plans to move new tailings storage facilities toward filtered tailings. In Indonesia, while coal tailings have traditionally been deposited as slurry, pressure is expected to mount for dewatered methods as regulations tighten around B3 (hazardous) waste (www.prosiding.perhapi.or.id; www.prosiding.perhapi.or.id).
Design guidance and sources
For comprehensive design and evaluation guidance—covering geotechnical investigation, design variables, water control, and case examples—see the US EPA Office of Solid Waste’s “Design and Evaluation of Tailings Dams” (1994) (nepis.epa.gov; nepis.epa.gov; nepis.epa.gov; nepis.epa.gov; nepis.epa.gov). For monitoring systems and surveillance advances, see Zare et al. (2024) in Minerals 14(6), 551 (www.mdpi.com; www.mdpi.com) and Clarkson & Williams (2021) in Mining Technology (cataloging sensors and telemetry). For dewatered and dry‑stack practice, see Godwin et al. (2022) and Cacciuttolo & Atencio (2023) in Minerals 13(11):1445 (www.mdpi.com; www.mdpi.com; www.mdpi.com), plus industry performance notes from FLSmidth (95% water recovery; fls.com) and Engineering News (~80% recovery case study; www.engineeringnews.co.za; www.engineeringnews.co.za). Indonesian regulatory context is outlined by Nur Anbiyak et al. (2020) in PERHAPI proceedings (www.prosiding.perhapi.or.id; www.prosiding.perhapi.or.id) and BSN’s SNI 3432:2020 announcement (www.bsn.go.id; www.bsn.go.id).
