Keeping Deep Wells Alive

Abstract

A well begins aging during construction. Casing wears, cement experiences thermal strain, seals relax, corrosion consumes metal, scale narrows flow, fractures change conductivity, sensors drift, and repeated interventions add damage. I develop an integrity system that preserves a verified baseline, classifies barriers and functions by consequence, maps each degradation mechanism to capable detection, and prepares repairs before the failure becomes urgent. Continuous monitoring is combined with scheduled and event-triggered logs. Surface equipment is modular and replaceable. Downhole work uses qualified robotic and conveyance systems with explicit barrier states. Availability depends as much on restoration time as on component life, so spares, access, procedures, and trained contractors belong in the design. The objective is not an immortal well. It is a well whose changing condition remains known enough to repair, derate, suspend, sidetrack, or abandon deliberately.

Age begins at installation

Threads can gall. Casing can wear during drilling. Cement placement can leave channels. Thermal ramp can create the first debonding. A tool can score a sealing bore. The commissioning record must therefore preserve the initial condition rather than assume perfection.

The baseline includes casing tally, material, connections, torque, wear estimates, cement volumes and evaluation, pressure tests, geometry, temperature, annular response, fluid chemistry, fiber condition, valve states, software, and every accepted deviation. Later evidence gains meaning by comparison.

An asset hierarchy links the station to well, string, interval, component, and replaceable unit. Every item has service limits, exposure history, inspection, maintenance, approved substitutes, and abandonment relevance.

Classify consequence before choosing maintenance

Barrier and safety functions receive conservative assurance. Production functions receive reliability optimization. Low-consequence instruments may run to failure if redundancy and replacement are simple.

Class	Consequence	Default strategy
I	Loss of containment, well control, or public safety	Continuous evidence, proof testing, conservative limits
II	Loss of essential monitoring or safe shutdown	Redundancy, condition monitoring, rapid restoration
III	Loss of thermal service or reservoir control flexibility	Planned intervention and spare capacity
IV	Limited convenience or duplicated research function	Economical corrective maintenance

Criticality follows function, not price. A low-cost annular pressure gauge can be Class I evidence. An expensive turbine component may be Class III if the plant can stop safely.

Reliability-centered maintenance asks what function can fail, how, with what consequence, and which task changes that risk. Project Lithos combines condition-based work, scheduled replacement, proof testing of dormant protections, and conscious run to failure where consequence allows.

Map mechanisms to evidence

Casing can suffer uniform corrosion, local pits, erosion, stress-corrosion cracking, fatigue, wear, buckling, or connection damage. Cement can crack, debond, shrink, alter chemically, or lose zonal isolation. Scale can reduce flow and immobilize devices. Fracture conductivity can decline or short circuit. Pumps and valves experience wear, deposits, insulation damage, and seal failure.

No inspection detects all mechanisms. Visual evidence, caliper, multifinger geometry, electromagnetic and ultrasonic techniques, temperature, noise, pressure tests, annular monitoring, chemistry, debris, vibration, and flow response cover different defects under different well conditions.

For each mechanism, the integrity plan records susceptible location, initiation conditions, growth model, detection method, coverage, probability of detection, measurement uncertainty, interval, threshold, and response. An unlogged section is marked unknown rather than healthy.

Remaining life can be expressed conceptually as

$$t_r=\frac{a_c-a_m}{\dot a},$$

where $a_m$ is measured damage, $a_c$ is critical damage, and $\dot a$ is growth rate. Each term has uncertainty, often large. The inspection interval preserves margin to the conservative lower estimate and includes time to mobilize a repair.

Monitor between inspections

Continuous channels watch annular pressure, injection and production pressure, temperature, flow, mass balance, pump behavior, chemistry, seismic response, and selected distributed fiber signals. Trends can raise inspection priority before an alarm limit is crossed.

Event-triggered inspection follows unusual pressure, thermal excursion, earthquake, loss of circulation, well-control event, stuck tool, unexpected torque, chemical excursion, or evidence of ground movement. Calendar cannot anticipate all damage.

Sensor confidence is part of the integrity state. If the annular gauge drifts or fiber coverage is lost, the well may require derating or an intervention log. Operating without required evidence is itself a degraded condition.

BaselineKnow the constructed state

Preserve materials, geometry, tests, damage, and accepted deviations.

DetectFind change with time to act

Combine continuous monitoring, repeatable logs, tests, and event triggers.

RestoreClean, repair, replace, sidetrack, or derate

Use qualified barriers, tools, spares, and work packages.

LearnExamine failures and update the population

Revise models, intervals, materials, designs, and operating limits.

Scaling needs accessible sacrifice

Some deposition is inevitable when fluid cools or depressurizes. I prefer scale to form in accessible separators, vessels, filters, or exchanger surfaces designed for cleaning rather than in a production interval or deep valve.

Differential pressure, heat-transfer performance, chemistry, and recovered solids reveal accumulation. Mechanical cleaning, jetting, milling, chemical dissolution, or replacement is selected from deposit composition and barrier risk. Dissolution products must remain compatible downstream.

Scale-control chemicals are verified against reservoir rock, cements, metals, elastomers, and environmental disposition. Treatment cannot become a new damage mechanism.

Flow paths should admit pigging or mechanical cleaning where practical. Exchanger trains isolate individually. Spools and valves use standardized dimensions and lifting points. Maintenance-friendly surface design protects the much harder subsurface system by controlling chemistry before reinjection.

A workover is a barrier campaign

Before opening the well, the team defines its live pressure sources, existing barriers, required independent barriers, kill or circulation method, pressure-control stack, fluid, environmental containment, and contingency. A downhole intervention may alter reservoir pressure and seismic response, so geological limits remain active.

The sequence is stabilize, verify barriers, rig up and pressure test, enter, diagnose, perform bounded work, retrieve and examine, verify repair, restore completion, test barriers, and ramp service under observation.

SecureEstablish verified barriers DiagnoseLocate mechanism and extent InterveneRestore with recovery options ProveTest before thermal service

Temporary plugs and isolation devices have stated ratings and verification. Tools crossing a barrier are accounted for continuously. Lost equipment is described precisely for fishing or future abandonment.

The repaired well warms gradually. Pressure, annuli, chemistry, seismicity, and the repaired function are observed through hold points. Full production is an acceptance stage, not the first test.

Spares reduce geological downtime

Mean availability for a simple repairable function is approximated by

$$A=\frac{\mathrm{MTBF}}{\mathrm{MTBF}+\mathrm{MTTR}}.$$

Increasing time between failure helps. Reducing time to restore can be equally powerful. Deep-well repair time includes diagnosis, engineering, permits, specialist mobilization, pressure equipment, tools, weather, intervention, testing, and ramp.

Project Lithos stores long-lead wellhead parts, seals, gauges, fiber interrogators, pump modules, valves, chemicals, and critical electronics under controlled conditions. It maintains call-off agreements for rigs, coiled tubing, wireline, cranes, and laboratories. Interface standards allow later equipment to replace an obsolete vendor model.

Special tools are proof tested and exercised. A recovery fixture that has sat for ten years without fit verification is not a dependable spare.

Decide when not to repair

A well can be derated, converted to monitoring, recompleted in another interval, sidetracked, suspended, or abandoned. Repair is not automatically the safest or most economic choice.

The decision compares remaining barrier confidence, repair probability, thermal value, seismic implications, repeat-failure risk, intervention consequence, alternative wells, and final abandonment. Sunk drilling cost does not justify repeated exposure to an uncertain barrier.

The institution succeeds when retirement is deliberate. The well’s final evidence improves casing, cement, chemistry, tools, and inspection across the fleet. Keeping deep wells alive ultimately means managing their whole life, including the moment one well should stop so the institution can continue.