FAB5 GEOSOLUTIONS | OCTOBER 2026

Designing proactive, trend-embedded TARPs for tailings facilities

Fabricio da Silva, Founder & CEO, Fab5 GeoSolutions Inc.

---

1. The problem: most TARPs report colour, not condition

A Trigger Action Response Plan (TARP) exists to convert a measurement into a pre-agreed decision, with a named owner, within a defined timeframe. In practice, most instrumentation TARPs on tailings facilities do something narrower: they paint each reading green, yellow, amber or red against a fixed threshold and wait for the colour to change. That is a reporting convention, not a governance control. It tells the operation that a limit has been crossed. It says nothing about where the reading is going, how fast, or whether that direction is one the design ever contemplated.

The consequence is predictable. Facilities with rich instrumentation and current TARPs still arrive at the red state by surprise, because the information that mattered, the trajectory, was never part of the trigger logic. A proactive TARP fixes this by embedding trend directly into the triggers, so that drift toward an initiating condition is surfaced while there is still time for an accountable human to act, rather than waiting for the within-the-green-zone reading to pass into the yellow zone.

2. One instrument is not the asset

The most common misreading of a TARP is to equate an instrument’s colour (TARP condition) with the structure’s condition. A single piezometer in amber does not place the embankment in amber. An instrument reports a local condition at one point, in one zone, for one parameter. The asset state is a judgement about a failure mode, and it requires corroboration: neighbouring instruments in the same zone, cross-parameter evidence (pore pressure against deformation, seepage flow and clarity), rainfall and deposition context, instrument health, and visual inspection.

The discipline cuts both ways. A single red reading is not dismissed; it triggers verification within a defined window (re-read, instrument check, corroborating data) and escalation if confirmed. Equally, a cluster of instruments all sitting green while trending in the same direction can be more significant than one isolated red. Two exceptions override the corroboration rule: where contractive, saturated tailings are confirmed, a credible single reading of rising pore pressure is treated as progression; and observed acceleration on any instrument governs regardless of colour. The declaration of asset state remains with the Engineer of Record (EOR) and the Responsible Person (RP), never with the colour of one cell on a dashboard.

3. The static TARP: thresholds anchored to Factor of Safety

The conventional and commonly used approach back-calculates instrument thresholds from stability analysis. The EOR determines the phreatic level at each critical piezometer that reduces the Factor of Safety (FoS) of the governing section to defined values, and those levels become the trigger elevations. The static TARP is an appropriate backstop because it ties every threshold to the design basis. Table 1 presents an illustrative example for a single piezometer.

Its limitation is structural. A static band is blind to everything below the first threshold. A reading can sit in green for months while rising steadily toward amber, and the reactive TARP approach registers nothing until the line is crossed.

4. The seasonal TARP: thresholds anchored to the facility’s own behaviour

Tailings facilities breathe with the hydrological cycle. Pore pressures rise during freshets, wet seasons, and near instrumentation deposition (for example), and reduce during dry periods, excavations, etc. A seasonal TARP uses the facility’s historical record to define the expected performance envelope for each period, typically the monthly mean (μ) plus or minus a selected multiple (k) of the standard deviation (σ), and triggers on departure from that envelope. The static FoS limit remains in place as a cross-section (FoS related) evaluation ceiling: whichever trigger is more conservative governs.

A July reading of 311.0 m is green on the static TARP, a full metre below the FoS trigger. On the seasonal TARP it sits six standard deviations above the July norm. One plan says “all clear”. The other says the drainage system is no longer behaving as it has for eight years. That is the value of the seasonal TARP: it detects change in behaviour long before change in safety margin, which is exactly the lead time a governance system exists to buy.

5. Instrument and data infrastructure aging: the control that erodes

Every trigger in Sections 3 and 4 assumes that the reading is true, current and representative. That assumption rests on a physical and digital chain: sensors, cables, readout units, loggers, telemetry, databases, and processing scripts, and that chain is itself a form of control. Like a drain or a filter, it degrades with time. Unlike a drain, it degrades silently: an aging piezometer does not alarm; it drifts, lags or flatlines, and a flatlined reading in the green is indistinguishable on a dashboard from a stable facility. The erosion runs along four dimensions, summarized in Table 3.

Dimension	How the control erodes	What the TARP must require
Time span and service life	Sensors installed at start-up are expected to report through operation, closure and post-closure, often beyond their service life. Raises, settlement and corrosion damage cables; buried sensors cannot be recalibrated or recovered.	Register of installation date, expected service life and condition for every TARP-critical instrument; replacements installed and overlapped before end of life, not after failure.
Frequency of readings	Manual frequencies are relaxed as the facility “settles”; telemetry and logger outages go unnoticed; reading intervals are not matched to how fast the failure mode can develop.	Frequency set by the failure mode’s rate of development and the time-to-threshold window; missed or late readings logged and escalated as a condition in their own right.
Quality of readings	Zero drift, offset shifts after maintenance, temperature and barometric correction errors, noise, frozen or repeated values, and datum or unit errors introduced during data migrations.	Automated validation and quality flags at ingestion; periodic verification against twin instruments, manual readings or response to known forcing; unvalidated data may raise a trigger but never clear one.
Continuity and coverage	Instruments lost to raises or construction, replacements started on new baselines, thinning coverage of the critical section, and records broken by platform or database changes.	Minimum number of functioning instruments per failure-mode zone; overlap period for every replacement; any change to the data chain managed through MOC.

Table 3. Erosion of the monitoring control and the minimum TARP response. Data-health status of TARP-critical instruments is reviewed by the RTFE and reported to the EOR.

The statistics inherit the defects. The seasonal envelope in Section 4 is computed from the record, so it absorbs whatever the record contains. A drifting sensor shifts μ; a noisy one inflates σ and widens the band; an outage removes precisely the season that mattered. A long record is evidence of stable behaviour only if it is also evidence of reliable measurement. The σ multiplier should therefore be selected on verified record length, not calendar age: no statistical band until three hydrological cycles of validated data exist, k = 2.0 thereafter, widened toward 2.5 – 3.0 only where the record is stationary, the instruments are verified, and no material change has occurred. An instrument that fails validation reverts to the static FoS TARP plus rate-of-rise triggers until its record is re-established.
Silence is not green. Loss of signal, a missed reading, an unexpected vertical bump, a failed validation check or a flatlined trace on a TARP-critical instrument is a trigger, with a named owner and timeframe like any other: verification by the Surveillance Lead within a defined window, and escalation to the RTFE if the reading is not restored. Where a failure-mode zone loses enough instruments to no longer demonstrate its condition, the asset state for that zone is unknown, and unknown is managed conservatively and not reported as green, as the risk profile for that facility now has changed. Aging monitoring infrastructure is therefore a capital and governance item, planned and funded across the life of the facility, not a maintenance afterthought.

6. What a proactive TARP embeds

Rate triggers: rate of piezometric rise per month and per metre of raise, independent of level.
Time-to-threshold: projection of the current trajectory to the static limit, triggering on forecast lead time rather than measured breach.
Seasonal deviation: departure from the period-specific envelope, as in Section 4.
Response-to-forcing signature: a change in how an instrument responds to rainfall or deposition (faster, larger, slower to recover) is a drainage signal in its own right.
Cluster and cross-parameter coherence: multiple instruments in one cross-section, linked to a failure mode, moving together, or pore pressure corroborated by deformation or changes in seepage.
Event triggers: rainfall, snowmelt, construction rates outside the expected range, or seismic thresholds that initiate special inspection irrespective of instrument readings.

Each trigger must name the decision it unlocks, the role that owns it, and the timeframe. Escalation must be socially free: if invoking a level requires courage or permission, it will not be invoked.

7. The governance position
A proactive TARP does not predict failure, and no analytical layer should claim to. It surfaces drift toward initiating conditions earlier, with the evidence and its uncertainty stated, so that the accountable human, the EOR and the RTFE, can decide while options still exist. Trend triggers are proposed by the technical team, ratified by the technical authority, tested by exercising notification chains, and reviewed whenever the facility changes.
The question every owner should ask of their own TARP is simple: if a piezometer rose steadily for six months and never crossed a line, would anyone be told? If the answer is no, the facility has a colour scheme, not a trigger action response plan.

About the author. Fabricio da Silva is Founder & CEO of Fab5 GeoSolutions Inc., a geotechnical governance consultancy serving the mining sector, with 15+ years of experience across tailings, water, and ground control in Brazil, Canada, the US, Australia, Indonesia, New Caledonia, and Burkina Faso. Fab5 offers TARP governance reviews as part of its GEMS™ (Governance Excellence & Maturity System) diagnostic. fabricio.dasilva@fab5geosolutions.com