Workover & well-service candidate selection — a systematic, multidisciplinary workflow
Every producing field is a portfolio of wells quietly underperforming their potential. The hard part is not knowing that workovers add barrels — it is knowing which well, with which intervention, will pay. Pick wrong and you spend a rig month chasing oil that was never there. This is a structured, physics-first workflow for choosing candidates, built from the three disciplines that have to agree before a rig moves: geology and geophysics, reservoir engineering, and production engineering.
Why candidate selection is the whole game
A workover — re-perforating, stimulating, shutting off water, changing the lift system, repairing the wellbore, or recompleting to a new zone — is one of the few levers that adds production from existing wellbore without the cost of a new well. But rig time is scarce and expensive, and a failed job carries a double cost: the money spent, plus the opportunity lost on the well that should have been done instead. The objective of candidate selection is therefore not to find wells that could produce more — almost all of them could — but to rank the finite list by risked value: the incremental volume actually recoverable, multiplied by the probability the job succeeds, net of cost and downtime.
Two failure modes dominate in practice. The first is treating the symptom instead of the cause: acidizing a well whose problem is reservoir depletion, or installing a bigger pump on a well that is loading up with liquids. The second is selection bias toward the wells that are easy to reach or already on someone's wish list, rather than the wells the data actually points to. A disciplined workflow exists to suppress both.
A problem that belongs to three disciplines
No single discipline can qualify a candidate alone. Geology and geophysics answer whether hydrocarbon remains in reach of the wellbore. Reservoir engineering answers whether the reservoir can still deliver it. Production engineering answers what, mechanically, is limiting the well today and which intervention removes that limit. A candidate is only real when all three line up — remaining hydrocarbon, reservoir deliverability, and a removable well-level constraint. The integration is the method.
Step 1 — Quantify the production gap
The entry point is always the same question: how far is this well below what it should be producing? The production gap is the difference between a defensible estimate of the well's potential and its current actual rate.
where qpotential comes from analog type-wells, the well's own historical trend, or its theoretical inflow capacity at current reservoir pressure.
There are three complementary ways to establish qpotential, and they should be cross-checked rather than trusted individually. Decline-curve analysis (Arps, 1945) fits the well's established trend and flags the point where the actual rate departs below the fitted decline — a sudden drop usually signals a mechanical or near-wellbore problem, while a gentle sag below trend points to reservoir or lift causes. Type-well or analog comparison places the well against offsets completed in the same interval; a well producing well under its analogs is carrying a recoverable gap, provided the analogs are genuinely comparable in reservoir quality. Material balance and remaining-reserves work asks the prior question — is there volume left at all — by comparing cumulative production against the estimated ultimate recovery for the drainage area.
A gap is necessary but not sufficient. A well can sit far below its analogs because the rock around it is genuinely worse, not because anything is wrong with it. That is why the gap only opens the investigation — the next three steps test whether the gap is real and removable.
Step 2 — The G&G view: is there hydrocarbon left to reach?
Before any intervention is justified, geology has to confirm that producible hydrocarbon exists within reach of the wellbore. Several classic opportunities live here. Bypassed pay is hydrocarbon-bearing rock that was never perforated — often thin or low-resistivity intervals missed on the original log interpretation, or zones excluded because they looked marginal at the time. A modern petrophysical re-evaluation of legacy logs, sometimes with cased-hole pulsed-neutron or carbon/oxygen logging to read current saturation through casing, routinely finds intervals worth perforating.
Attic and structural oil sits updip of existing perforations where the well trajectory or original completion left it undrained. Compartmentalization — sealing faults or stratigraphic barriers — can leave whole blocks at near-initial pressure while the producing compartment depletes; a recompletion or sidetrack into an undepleted compartment behaves almost like a new well. Seismic attributes, structure maps, and pressure data from offset wells are the evidence base. The deliverable from this step is a yes/no on remaining accessible volume and, if yes, its location relative to the current completion.
If geology cannot point to where the additional hydrocarbon is — a specific interval, a structural position, an undepleted compartment — then the candidate is a production-engineering job on the existing completion (de-skin, lift, deliquify), not a recompletion. Mixing the two is how recompletions get drilled into depleted rock.
Step 3 — The reservoir-engineering view: can the reservoir still deliver?
Reservoir engineering converts "there is oil left" into "the reservoir can push it into the wellbore at a useful rate." The first screen is pressure. A reservoir near its abandonment pressure will not repay a stimulation, no matter how clean the perforations get — there is no energy left to move fluid. Current reservoir pressure relative to bubble point and to economic drawdown limits is the gate.
The second screen is deliverability, expressed through the inflow performance relationship and the productivity index:
A well producing far below the J implied by its rock quality is constrained by something removable — skin, partial penetration, or lift — not by the reservoir itself.
The IPR — built with Vogel's correlation for solution-gas-drive oil wells (Vogel, 1968), Fetkovich's deliverability form (Fetkovich, 1973), or a straight productivity index above the bubble point — defines the ceiling: the maximum the rock will give at a given flowing pressure. Comparing the IPR ceiling against actual production is the cleanest single test of whether the problem lives in the reservoir or in the well. The third screen is the drive mechanism and saturation behaviour: an active aquifer or expanding gas cap changes both the remaining-reserves picture and the water/gas-handling consequences of producing harder. Pushing drawdown on a well with a nearby water contact invites coning; the reservoir-engineering view has to flag that before the production engineer designs the lift.
Step 4 — The production-engineering view: what is choking the well?
This is where the specific intervention is identified, and the master tool is nodal (systems) analysis. The producing system is a chain of pressure drops from reservoir to separator — through the completion, up the tubing, across the wellhead and choke, into the flowline. Nodal analysis splits the system at a node (commonly the bottomhole) into an inflow side (IPR, what the reservoir delivers) and an outflow side (the vertical lift performance, what the tubing and lift system can carry). The operating point is their intersection. Whichever side is steep at that intersection is the binding constraint — and the right intervention is the one that moves that constraint (Economides et al., 2013; Beggs, 2003).
The production-engineering checklist runs through the usual suspects. Skin — near-wellbore damage from drilling, perforating, or fines and scale — shows up as a flowing pressure far below the IPR ceiling and is the textbook stimulation or acidizing candidate (Hawkins, 1956; McLeod, 1983):
Positive s is damage (a stimulation candidate); the extra pressure drop it causes is Δpskin ∝ s.
Liquid loading is the dominant gas-well constraint: when gas velocity falls below the critical value needed to lift liquids, the well begins to accumulate water in the wellbore, chokes itself, and dies — often while the reservoir still holds plenty of gas. The Turner droplet model (Turner, Hubbard & Dukler, 1969) and Coleman's lower-pressure revision (Coleman et al., 1991) set the critical rate; below it, the candidate is a velocity string, plunger lift, surfactant, compression, or a combination, not a stimulation.
Below the corresponding critical gas rate, liquids accumulate and the well loads up.
Artificial-lift evaluation asks whether the current lift method is correctly sized and whether a different method would unlock the well (Clegg, Bucaram & Hein, 1993): an undersized or worn pump, gas interference, or a well that has outgrown beam pumping and needs ESP or gas lift. Finally, wellbore integrity — scale, wax, asphaltene, sand fill, collapsed or corroded tubing, leaking packers — produces sudden, often dramatic rate losses that no reservoir or inflow model explains, and that a slickline clean-out, tubing change, or casing repair restores. One caution closes this step: the producing system does not end at the wellhead. A candidate’s achievable uplift can be capped by surface constraints — water- and gas-handling capacity, flowline backpressure, available compression — so the nodal model must be carried through to the separator before any rate is promised.
Matching the diagnosis to the right intervention
The diagnosis dictates the job. The most common mistake — and the most expensive — is correlation without causation: a well has low rate and high water, so it gets a water shutoff, when the real problem was a worn pump letting the wellbore load. The table below maps the dominant diagnostic signatures to the interventions that actually address them.
| Dominant signature | Most likely cause | Indicated intervention |
|---|---|---|
| pwf far below IPR ceiling; high skin | Near-wellbore damage | Matrix acidizing / re-perforation |
| Low rate, low skin, low-permeability rock | Reservoir tightness | Hydraulic fracturing |
| Gas well dies below critical rate | Liquid loading | Velocity string / plunger / compression / surfactant |
| Rising water cut, channeling signature | High-permeability water path | Water shutoff / conformance / recompletion |
| Rising water cut, coning signature | Water cusping below perfs | Reduce drawdown / shut off lower perfs |
| Sudden rate loss, no inflow explanation | Fill, scale, wax, or integrity | Clean-out / tubing change / casing repair |
| Inflow good, outflow constrained | Lift undersized or wrong type | Lift resize / convert to ESP or gas lift |
| Depleted producing zone, pay behind casing | Bypassed / undrained interval | Recompletion / add perforations / sidetrack |
Deep dive: artificial-lift selection
When the reservoir can no longer push fluid to surface at an economic rate — or never could — the intervention is artificial lift. It is the single most common workover class on mature oil fields, and the most commonly mis-specified. Selection is a screening problem in its own right: each method occupies an operating envelope defined by rate, depth, deviation, gas, sand, viscosity, temperature, and the realities of power and surface footprint (Clegg, Bucaram & Hein, 1993; Brown, 1980; Takacs, 2015).
The envelope chart sets the first cut; the secondary screens decide inside the overlaps. Sucker-rod (beam) pumping dominates shallow, low-rate stripper oil but gas-locks on high free gas and is limited in deviated wells. Progressive cavity pumps (PCP) excel on viscous and sandy heavy oil but are temperature- and depth-limited. Electric submersible pumps (ESP) move large volumes from depth — ideal for high water-cut, high-rate wells — but tolerate sand and free gas poorly and depend on reliable power. Gas lift handles high gas-liquid ratios, sand, and deviation gracefully and is the workhorse of offshore and high-GOR wells, provided injection gas is available. Plunger lift is the natural answer to liquid-loading gas wells. The screening table below is the practical reference.
| Method | Typical rate | Depth limit | Sand | High GOR | Best fit |
|---|---|---|---|---|---|
| Sucker-rod pump | < 1,500 bpd | ~11,000 ft | low–mod | poor (gas lock) | shallow low-rate oil |
| Progressive cavity (PCP) | < 4,500 bpd | ~6,500 ft | high | poor | viscous / heavy / sandy oil |
| Electric submersible (ESP) | 200–60,000 bpd | ~12,000 ft | low (abrasive) | poor–mod | high-rate, high water cut |
| Gas lift | 50–30,000 bpd | deep | high | excellent | high-GOR, deviated, offshore |
| Plunger lift | low (gas wells) | — | moderate | excellent | liquid-loading gas wells |
| Jet / hydraulic pump | wide | deep | mod–high | moderate | deep, deviated, remote wells |
Deep dive: sand control
In weakly consolidated formations, producing fluid drags formation grains into the wellbore. Sand erodes tubulars and chokes, fills the rathole, collapses casing, and can destroy surface equipment in hours. Sand control is therefore both a candidate-selection question (which wells will sand, and when) and a completion-design question (which method fits the rock). The screen begins with sand prediction: formation strength from sonic logs and core (uniaxial compressive strength), the in-situ stress state, and the drawdown and depletion the well will see. A useful first-order rule ties onset of sanding to rock strength and the flowing drawdown; once depletion raises effective stress, wells that produced clean for years can begin to fail.
If sand is expected, method selection turns on grain-size distribution and sorting. The uniformity (sorting) coefficient and fines content decide whether a bare screen will work or will plug:
Well-sorted sand (Uc < 3) can be controlled with standalone screens; poorly sorted sand with fines (Uc > 5) plugs screens and needs a gravel or frac pack. Gravel is sized to the formation: D50,gravel ≈ 5–6 × D50,sand (Saucier).
| Method | How it controls sand | Best fit | Main limitation |
|---|---|---|---|
| Standalone screens (SAS) | mechanical filter at sandface | clean, well-sorted sand; open-hole horizontals | plugs with fines / poor sorting |
| Gravel pack (open / cased hole) | sized gravel + screen | broad range; the proven default | rate / skin penalty |
| Frac pack | short frac + gravel pack | moderate–high perm, cased; negative skin | cost, execution complexity |
| Expandable screens | compliant screen to borehole | open hole, maximises ID | mechanical / expansion risk |
| Chemical consolidation | resin bonds grains in place | short, thin intervals; rigless | strength vs permeability trade-off |
| Rate / drawdown control | stay below critical drawdown | mild, intermittent sanding | caps deliverability |
Deep dive: water shut-off & conformance
Excess water is the most common reason mature wells become uneconomic — not because the oil is gone, but because lifting, separating, and disposing of water overwhelms the margin. Water shut-off is also the intervention most often done wrong, because the treatment must match the mechanism, and the mechanism is invisible without diagnosis. The discipline here is to diagnose first — production logs, pressure surveys, and the Chan water-oil-ratio diagnostic plot to separate coning from channeling from a simple watered-out layer — and only then choose a treatment (Seright, Lane & Sydansk, 2003; Sydansk & Seright, 2007; Chan, 1995).
The single most important screening principle, from decades of conformance field experience, is that treatability depends on where the water moves and whether layers crossflow. Problems confined near the wellbore, or in layers that do not crossflow, are treatable mechanically or with gels. Problems spread far into the matrix, or where layers crossflow, are difficult or impossible to treat chemically — and three-dimensional coning into a single matrix interval generally cannot be fixed with a chemical at all. Spending a gel job on a 3D coning problem is the classic, expensive mistake.
| Water problem | Diagnosis | Treatment |
|---|---|---|
| Casing / tubing leak | logs, pressure test | mechanical patch or squeeze |
| Flow behind pipe | cement-bond / temperature log | cement squeeze |
| Watered-out layer, no crossflow | production log (PLT) | mechanical shutoff: plug, packer, patch |
| Channeling along high-perm streak | Chan plot, PLT, tracer | conformance gel / polymer |
| Fracture / fault injector–producer path | tracer, Chan plot | gel conformance treatment |
| Coning / cusping near perforations | Chan plot, completion review | reduce drawdown; shut off lower perfs |
| 3D matrix coning | diagnosis of exclusion | generally not chemically treatable |
Step 5 — Ranking and screening the candidate list
Diagnosis produces a list of technically valid jobs. Ranking turns the list into a program. The currency is risked value, not raw uplift. A modest, near-certain gain often beats a large but speculative one, because the program has to deliver across many wells, not bet on one.
Ps = probability of mechanical and reservoir success; NPVsuccess from the nodal-predicted uplift and price deck; Cfailure = cost plus deferred production if the job fails.
Plotting candidates on two axes — probability of success against incremental potential, with bubble size scaled to net present value — turns the portfolio into a picture. The top-right quadrant (high confidence, high gain) is the program's backbone and gets executed first. The bottom-left (low confidence, low gain) is deferred or dropped. The off-diagonal quadrants are where judgement lives: a high-gain, low-confidence job may justify cheap diagnostic data (a production log, a pressure survey) to move it rightward before committing a rig.
Where interventions act on the well
It helps to keep a mental map of where each intervention physically acts, because a single well can carry more than one removable constraint, and the cheapest winning program sometimes combines two jobs in one rig entry.
Why candidates fail — and how to not be that statistic
Post-job reviews of disappointing workovers cluster around a few repeatable errors. Misdiagnosis is first: the symptom was read without confirming the cause, and the job treated the wrong constraint. Ignoring the reservoir limit is second: a beautifully executed stimulation on a well that had no pressure left to flow. Optimistic uplift is third: the nodal model assumed a skin reduction or a water cut the rock never delivered. Mechanical surprises are fourth — corroded casing, a stuck packer, fill deeper than logged — which is why integrity data belongs in the screen, not in the post-mortem. The antidote to all four is the same: insist that geology, reservoir, and production each sign off on their part of the story before the candidate enters the ranked list, and treat cheap diagnostic data (a pressure survey, a production log, a fluid sample) as the highest-return spend in the whole program.
A candidate is ready when you can complete this sentence with evidence for every blank: “There are ___ units of hydrocarbon in ___ reachable from this wellbore; the reservoir can deliver them because ___; the well is currently limited by ___; and ___ intervention removes that limit, for ___ cost at ___ probability of success.”
The workflow, in sequence
Pulled together, the method is a funnel that narrows a whole field to a defensible program. First, screen every well for a production gap against decline, analogs, and remaining reserves. Second, send the gap wells to geology to confirm reachable hydrocarbon and its location. Third, test reservoir deliverability — pressure, IPR ceiling, drive, and water/gas risk. Fourth, run nodal analysis and the production-engineering checklist to name the binding constraint and the intervention that removes it. Fifth, estimate uplift, cost, and probability of success, compute risked value, and rank. Execute the top-right of the chart, buy data to de-risk the promising-but-uncertain, and defer the rest. The discipline is unglamorous, but it is the difference between a workover program that compounds the field's value and one that quietly burns rig days. The same logic scales from a single well to a several-hundred-well portfolio — which is precisely where a structured surveillance system earns its keep, turning the screen from a spreadsheet exercise into a continuous, ranked queue.
One step remains, and it is the one most often skipped: close the loop. After each job, measure the actual uplift against the nodal prediction and feed the result back — into the type-well models, the probability-of-success estimates, and the diagnostic assumptions that drove the selection. A program that never compares predicted with actual repeats its own misdiagnoses; one that does becomes measurably better at choosing the next candidate. Candidate selection is not a one-time screen but a learning cycle.
References
Arps, J.J. (1945). Analysis of Decline Curves. Transactions of the AIME, 160, 228–247.
Vogel, J.V. (1968). Inflow Performance Relationships for Solution-Gas Drive Wells. Journal of Petroleum Technology, 20(1).
Fetkovich, M.J. (1973). The Isochronal Testing of Oil Wells. SPE Annual Meeting, SPE 4529.
Hawkins, M.F. (1956). A Note on the Skin Effect. Transactions of the AIME, 207.
Turner, R.G., Hubbard, M.G. & Dukler, A.E. (1969). Analysis and Prediction of Minimum Flow Rate for the Continuous Removal of Liquids from Gas Wells. Journal of Petroleum Technology, 21(11).
Coleman, S.B., Clay, H.B., McCurdy, D.G. & Norris, H.L. (1991). A New Look at Predicting Gas-Well Load-Up. Journal of Petroleum Technology, 43(3).
McLeod, H.O. (1983). The Effect of Perforating Conditions on Well Performance. Journal of Petroleum Technology, 35(1).
Clegg, J.D., Bucaram, S.M. & Hein, N.W. (1993). Recommendations and Comparisons for Selecting Artificial-Lift Methods. Journal of Petroleum Technology, 45(12).
Economides, M.J., Hill, A.D., Ehlig-Economides, C. & Zhu, D. (2013). Petroleum Production Systems, 2nd ed. Prentice Hall.
Beggs, H.D. (2003). Production Optimization Using Nodal Analysis, 2nd ed. OGCI / Petroskills.
Economides, M.J. & Nolte, K.G. (2000). Reservoir Stimulation, 3rd ed. Wiley.
Dake, L.P. (1978). Fundamentals of Reservoir Engineering. Elsevier.
Craft, B.C. & Hawkins, M. (1991). Applied Petroleum Reservoir Engineering, 2nd ed. Prentice Hall.
Lake, L.W. (ed.) (2007). Petroleum Engineering Handbook. Society of Petroleum Engineers.
Brown, K.E. (1980). The Technology of Artificial Lift Methods. PennWell.
Takacs, G. (2015). Sucker-Rod Pumping Handbook. Gulf Professional Publishing.
Penberthy, W.L. & Shaughnessy, C.M. (1992). Sand Control. SPE Series on Special Topics, Vol. 1.
Tiffin, D.L., King, G.E., Larese, R.E. & Britt, L.K. (1998). New Criteria for Gravel and Screen Selection for Sand Control. SPE 39437.
Seright, R.S., Lane, R.H. & Sydansk, R.D. (2003). A Strategy for Attacking Excess Water Production. SPE Production & Facilities, 18(3).
Sydansk, R.D. & Seright, R.S. (2007). When and Where Relative Permeability Modification Water-Shutoff Treatments Can Be Successfully Applied. SPE Production & Operations, 22(2).
Chan, K.S. (1995). Water Control Diagnostic Plots. SPE 30775.
Civan, F. (2007). Reservoir Formation Damage, 2nd ed. Gulf Professional Publishing.