A stage cementing collar occupies a structurally critical position in any two-stage or multi-stage cementing operation: it must survive the full hydrostatic and mechanical loading of the casing string during run-in, open reliably under the hydraulic differential applied by the cement pumping crew, allow cement slurry to pass through its ports without leakage at the collar body, and then close and seal permanently under the wiper plug. If the collar's threaded connections to the casing string fail at any point in that sequence, the entire stage cementing operation fails — and the cement job for the affected interval cannot be completed without pulling the string, an operation that may itself be impossible after partial cement placement. The direct cost of a stage cementing failure can run into millions of dollars when the combined cost of the lost equipment, the remedial cementing attempt, the additional rig time, and in the worst case the loss of the wellbore is considered.
Given this consequence profile, understanding what caused a stage cementing collar to pull over — and specifically, whether the cause was a material or dimensional non-conformance in the collar itself or a systemic problem in the casing handling and make-up process — is not merely an academic exercise. It is the foundation for preventing the same failure mode from recurring in subsequent wells using the same casing batch and the same operating procedures. The investigation described in this article is instructive precisely because the collar's material and thread geometry were found to be fully conforming: the failure did not originate in the manufactured product. It originated in the condition of the casing coupling that the collar threaded into, and in the field decision to proceed with run-in after the make-up position had shown an unmistakable departure from the specified range.
Incident Summary
Incident Data Sheet
How Stage Cementing Collar Thread Connections Work: The Make-Up Position Standard
API Spec 5B governs the threading, gauging, and inspection of casing, tubing, and line pipe threads. For Φ339.72 mm API Buttress (偏梯形, BC) thread, the specification defines the nominal make-up position as the condition where the coupling face reaches the △-mark baseline on the pin — with an upper tolerance of the △-mark apex and a lower tolerance of 5.08 mm below the △-mark baseline. Put differently, the acceptable make-up window spans approximately two turns of thread engagement beyond what is casually called the "hand-tight" position, and the entire acceptable window is only about one turn wide.
This tight positional window exists because the thread form's load-bearing capacity is highly sensitive to the number of turns actually engaged in full-profile thread contact. API BC thread is a trapezoidal (buttress) form with an 0° load flank angle, designed to resist pull-out loads through thread flank bearing over the full engaged length. If the connection is made up to the correct torque value but falls short of the correct position — as occurred in this incident — the number of turns in full contact is reduced, the load path distributes over a shorter engaged length, and the connection's pull-out resistance is a fraction of its rated capacity even though the make-up torque log shows a compliant value.
Make-Up Position: Actual vs. Specified (Φ339.72 mm BC Thread)
−7.04 turns Lower limit
(−5.08 mm) Nominal
(△ baseline) Upper limit
(△ apex)
The actual make-up position was 5.04 turns short of even the lower tolerance limit — placing the connection well outside the range tested and rated by API TR 5C3. Physical test data on Φ177.8 mm BC casing shows that connections made up to the lower tolerance limit achieve 1.27× the API rated pull-out load; the under-made-up connection in this incident carried only a fraction of that capacity.
Sequence of Events Leading to Pullover
Pre-run inspection
Casing protection rings were found partially dislodged, deformed, or displaced upon arrival at the well site and in the yard. Despite these visible protection failures, no formal incoming inspection of coupling ovality or thread gauge was performed on the large-OD casing before use.
Make-up attempt 1
Float make-up of the #152 collar external thread into the #151 casing coupling internal thread. Make-up torque reached 18,011 N·m but significant remaining turn count (余扣) was noted — the connection had not reached the specified position.
Make-up attempt 2
Torque increased to 20,000 N·m; no further thread advance was observed. This is the first clear indication of abnormal interference — the torque is rising but the connection is not advancing to position. At this point, investigation of the coupling condition should have begun.
Make-up attempt 3 — anomalies observed
Torque increased to 21,000 N·m (maximum allowed). The #151 coupling was reported hot to touch and its factory-end thread rotated — indicating the coupling was in an abnormal interference condition. The crew discussed and concluded that make-up requirements were satisfied.
Run-in decision
Despite the anomalies (coupling heat, factory-end rotation, no confirmed position check against the △-mark), the crew ran the casing string to total depth. The make-up position of the stage cementing collar external thread was in fact 7.04 turns short of specification — but this was not verified before run-in.
Failure during cementing
Under the combined axial load from the string weight and the piston effect of internal pressure during cement displacement, the under-engaged collar external thread pulled out of the #151 coupling. Pump pressure dropped 3.9 MPa instantaneously; mud splash at the wellhead confirmed complete string separation at that depth.
Systematic Cause Elimination: What the Investigation Ruled Out
The investigation systematically evaluated five candidate causes for the pullover, working through each with quantitative analysis before converging on the root cause. Understanding why the other four candidates were eliminated is important for correctly framing the root cause and avoiding misallocation of corrective actions.
Calculated tensile load at the pullover location was 1,378 kN — only 14.9% of the Φ339.72 mm × 12.19 mm P110 BC casing's rated tensile capacity. Composite stress including piston effect reached only 41.3% of yield strength (safety factor 2.42). Overload as a cause was conclusively eliminated.
The thread locking compound used has a 60-minute open time after mixing and requires 48 hours for full cure. The entire float make-up operation for both collar connections took under 4 minutes. Solidified grease blocking make-up was thermodynamically impossible in this timeline.
Of the three connections made up simultaneously in the float make-up configuration, two reached or exceeded the △-mark tolerance limit (the #153 coupling advanced 2.31 turns past the △-mark baseline, exceeding the upper tolerance by 0.43 turns). Maximum torque 21,000 N·m was reached. Insufficient torque was eliminated.
Throughout the entire run of Φ339.72 mm casing, workers only needed to lightly guide or push the pin to engage the coupling — evidence that the string was essentially coaxial. The pulled-over external thread showed no galling. Non-coaxiality as a significant contributor was eliminated.
Post-pullover examination found 16 turns of clear make-up engagement marks on the 21-turn external thread — but no galling was observed. If galling had been the cause of the interference, it would have left characteristic surface damage. Galling was ruled out.
The #151 coupling was deformed (excessive ovality) from inadequate protection during transport and handling. The deformed coupling created abnormal thread interference that consumed make-up torque without advancing the connection toward the specified position. Torque-controlled make-up cannot detect or compensate for this condition. The connection reached maximum torque while still 7.04 turns short of specification.
The Core Mechanism: Why Coupling Deformation Defeats Torque-Controlled Make-Up
The causal chain in this incident begins with coupling deformation and ends with a connection that is torque-compliant but position-non-compliant. Understanding why these two measurements diverge is essential for recognizing the same failure mode in future operations.
Under normal conditions, the torque required to advance a buttress-thread casing connection increases progressively as more thread turns come into bearing contact, reaching the specified make-up torque at approximately the same time as the specified make-up position. This consistency is the basis for using torque as a proxy for position — which is the standard field practice at most operations, partly because the △-mark on the pin is sometimes difficult to locate in a field setting and partly because checking the mark adds time to each connection. For non-deformed couplings, torque-controlled make-up works because the relationship between torque and position is predictable within the tolerances of the thread form.
A deformed coupling breaks this relationship. When the coupling's internal geometry deviates from a circular cross-section (ovality), the interference between the internal and external thread flanks becomes non-uniform around the circumference: some sectors of the thread engagement are over-loaded while others are under-loaded. The over-loaded sectors generate disproportionately high friction torque at a stage in the make-up sequence when the connection has not yet reached its designed position. The torque indicator reaches the specified make-up value, but the actual engaged length — the quantity that determines pull-out resistance — is still far below its designed value. The coupling's hot temperature after three make-up passes is direct evidence of this abnormal friction: energy is being dissipated in heat from the irregular thread interference rather than being stored as mechanical pre-load in the connection.
The Role of Casing Protection Rings: Where the Failure Chain Actually Begins
A Φ339.72 mm × 12.19 mm P110 BC casing joint weighing approximately 1,113 kg for an 11-meter length is not a precision instrument that handles deformation gracefully. At that outer diameter and mass, the coupling is the most vulnerable part of the joint during transportation and yard storage: it is the outermost-diameter component, it concentrates bending stress at the pin-coupling transition zone whenever the joint is loaded at an angle, and it is the component most likely to make contact with adjacent joints when a string of casing rolls or slides in a pipe rack. API 5CT specifies protection ring requirements precisely because of this vulnerability. The customer's supplemental technical agreement for this well went further, specifically requiring protection rings at both ends of couplings on casing with OD ≥ 244.5 mm.
The investigation found that protection ring failure was extensive and systematic, not isolated to a single joint. Inspection of casing on the delivery trucks found partially dislodged, deformed, and displaced rings across the batch. Yard inspection found the problem to be even more severe in stored pipe. Critically, even casing joints that still had their protection rings in place showed coupling-to-coupling contact and impact deformation on the outer coupling surface — meaning the rings that were physically present were no longer providing effective protection, having themselves been deformed to the point where adjacent couplings could make metal-to-metal contact through or around them. No incoming inspection for coupling ovality against the user's specified tolerance (≤0.50% ovality) was performed on any of the casing in this batch. For OD ≥ 244.48 mm pipe in general, prior inspection data cited in the investigation noted a high proportion of thread pitch-diameter non-conformances attributable to plastic deformation — suggesting that coupling deformation in this size range is a recurring issue, not a one-off event.
Recommendations
For all API BC-thread casing connections, verify the △-mark make-up position visually at every joint — not only at the initial calibration set — particularly at connections involving accessories (stage collars, float collars, centralizers) where the float make-up configuration creates multiple simultaneous connections with potentially divergent responses. If make-up torque is achieved but the position indicator is not within tolerance, do not run the joint.
Implement incoming ovality inspection for all casing with OD ≥ 244.5 mm against the user-specified tolerance (≤0.50% ovality) as a condition of acceptance, not as a discretionary quality step. Couplings exceeding the ovality limit should be rejected before the pipe enters the yard inventory.
Develop and enforce a protection ring specification that includes a minimum structural performance requirement — not merely a dimensional specification. A ring should be capable of absorbing a defined impact load (representative of the dynamic loading on the coupling during transport and yard handling) without allowing coupling-to-coupling metal contact. Consider ring materials, geometry, and the minimum engagement length on the coupling body.
Establish and train to an explicit decision protocol for make-up anomalies: if make-up torque is increasing but thread advance has stopped before reaching the △-mark position, do not continue making up. Back out the connection, inspect the coupling bore for deformation, and — if the coupling ovality check shows non-conformance — replace the coupling before proceeding. A hot coupling and a factory-end thread rotating under load are both stop-work indicators under this protocol.
The float make-up configuration for a stage cementing collar connects three thread joints simultaneously with a single torque reading, making it inherently less diagnostic than single-joint make-up. For collar installations, supplement torque monitoring with independent position check of both the collar's external thread and the collar's internal thread before running below the rotary table.
Enforce pipe-to-pipe isolation during transport (dunnage spacing and load restraint), limit stacking height for large-OD casing to prevent lower-tier crushing, and conduct post-transport protection ring inspection on receipt before any pipe is moved to the yard rack. Document and photograph ring condition at each inspection stage.
Frequently Asked Questions
Q: What is a stage cementing collar, what does it do in a well completion, and why is its threaded connection to the casing string so structurally important?
A stage cementing collar (分级箍) is a downhole tool installed at a designed depth in the casing string, used in multi-stage cementing operations where a single conventional cement job from the shoe to surface would be impractical — for example, because of lost circulation zones, an annular gap too large for a single-stage slurry volume, or the need to protect weak formations from cement column hydrostatic pressure. The collar contains a body with sliding sleeve components and side ports: during the cement job, hydraulic pressure is used to open the ports, cement is pumped out into the annulus above the collar, and a wiper plug then closes and seals the ports. Because the stage cementing collar must function as a full-bore flow restriction, a cementing manifold, and a permanent pressure-sealing element simultaneously, its design involves considerably more complex internal geometry than a standard coupling or float collar. This complexity makes it heavier and longer than the standard casing joints it connects to — the 815 mm collar height in this incident is nearly twice the coupling height of a standard joint — which amplifies the bending moment that the lower thread connection must resist during float make-up and means that any under-engagement of the lower thread connection is particularly consequential for the integrity of the string below the collar.
Q: The investigation found that the stage cementing collar's own material and thread parameters met API Spec 5CT requirements. What specifically was measured, and what does this tell us about how to approach failure analysis for casing connection pullover events?
The investigation confirmed that the external thread of the stage cementing collar (#152) had thread pitch, taper, and tooth height within API specification; that the coupling bore ovality was within the user-specified 0.50% limit; and that tensile strength, yield strength, elongation, and Charpy impact energy all met the API Spec 5CT P110 grade minimums. These results are significant not only as a confirmation that the collar itself was not defective, but as a methodological point for any pullover failure analysis: thread geometry and material conformance in the pulled-over component does not establish the root cause — it only eliminates the collar as a direct source of weakness. The investigation would have been incorrect if it had stopped at confirming the collar's conformance and concluded "no defect found." The conforming thread parameters on the collar are actually the clue that the cause must lie elsewhere — specifically in the condition of the mating component (the coupling) and in the make-up process. Post-failure forensic evidence such as the △-mark position and the pattern of thread locking compound distribution on the external thread proved to be more diagnostic than any laboratory measurement on the collar itself.
Q: How is API BC (Buttress) thread different from API Long Round (LC) thread and from premium connections, and does the thread form affect the sensitivity of pull-out resistance to make-up position?
API Buttress thread (BC, 偏梯形螺纹) has a trapezoidal profile with a near-zero load flank angle (nominally 3° on the stabbing flank, near vertical on the load flank), which means that virtually all of the pull-out resistance is provided by the compressive bearing stress on the load flanks of the engaged threads — there is essentially no wedging or interference contribution from the thread form geometry itself under tensile loading. This characteristic makes BC thread highly efficient in pure tension but also means that pull-out resistance is very directly proportional to the number of turns in full-profile contact. If the make-up position is short by several turns, the pull-out resistance decreases nearly linearly with the turn deficit. API Long Round (LC) thread, by contrast, has a rounded profile that allows some limited tooth deformation under load, and its design make-up relies partly on the coupling's elastic spring-back to maintain flank contact pressure — so its sensitivity to positional deviation is somewhat different, though still significant. Premium connections (such as VAM, Tenaris TenarisHydril, or NK-3SB types) typically include a metal-to-metal seal and sometimes a torque shoulder in addition to the thread engagement, which provides additional resistance to over-torque and creates a more definitive make-up indicator — but even premium connections can be made up short of the designed position if the mating component is deformed, using the same mechanism described here. For BC thread specifically, the absence of a torque shoulder means there is no hard stop in the make-up sequence, making positional verification even more important.
Q: The incident reports that the coupling became hot during the third make-up attempt and the factory-end thread rotated. Why weren't these treated as stop-work conditions at the time?
This is the operational and human-factors dimension of the incident that the investigation's recommendation on "using operation" (使用操作) section addresses. The crew's reasoning appears to have followed a logic that is understandable but incorrect: torque had been reached (the universal field criterion for a made-up connection), two of the three simultaneous connections had confirmed good position, and the coupling heating and factory-end rotation were interpreted as signs that the connection was "working hard" rather than as signs that something was wrong with the coupling's geometry. The underlying problem is that the torque criterion, which is the dominant operational decision criterion at most wellsites for large-diameter casing, does not contain information about make-up position when the mating component is deformed — it only measures the aggregate rotational resistance of all thread joints in the power tong's grip simultaneously. A coupling that is generating excess friction from ovality-induced interference looks identical on the torque chart to a coupling that is generating correct progressive make-up load from advancing thread engagement. Without an independent check on the △-mark position, there is no field-accessible way to distinguish these two conditions from the torque log alone. The recommendation to establish an anomaly response protocol addresses this by defining specific observable conditions — continued torque increase with no advance in thread position, coupling heat, factory-end rotation — as stop-work triggers rather than as tolerated anomalies.
Q: What inspection and quality assurance steps should a casing user perform on received casing pipe — particularly large-OD casing — before it enters the well programme?
The investigation highlights a systematic inspection gap for large-OD casing that this case study should motivate users to close. For casing with OD ≥ 244.5 mm (approximately 9-5/8 inch and larger), the following inspections are most critical. First, protection ring condition should be checked on the delivery vehicle before offloading — rings that are displaced, deformed, or missing should be documented and the exposed couplings set aside for coupling ovality measurement before use. Second, coupling ovality should be measured on all joints with visible ring damage and on a representative sample (minimum 10% or per the applicable standard) of all joints received; any coupling exceeding the user's ovality specification (the customer standard in this case specified ≤0.50%) should be rejected. Third, thread profile gauging should be performed with appropriate API gauge sets — a deformed coupling will frequently also show thread pitch-diameter errors that can be detected by gauge before the joint reaches the floor. Fourth, for critical completion components such as stage cementing collars, float collars, and liner hangers, the receiving inspection should include a test make-up of the accessory's external thread into a dummy coupling of the same specification to confirm that normal progressive make-up behavior is achieved — identifying coupling mismatches before the component reaches the rig floor.
Summary
The stage cementing collar pullover analyzed in this investigation resulted from a non-conforming make-up position — 7.04 turns short of the API Spec 5B lower tolerance limit — caused by abnormal thread interference from a deformed #151 casing coupling. The collar's own material and thread geometry were fully conforming to API Spec 5CT P110 requirements; the failure originated not in the collar but in the casing coupling's condition and in the field decision to accept a torque-compliant, position-non-compliant make-up. The deformation traced back to inadequate casing protection rings that failed to prevent coupling-to-coupling contact during transportation and yard storage. The incident demonstrates the fundamental limitation of torque-only make-up control for large-OD API BC-thread casing: a deformed coupling can absorb the full specified torque through abnormal thread interference while the connection remains far short of the specified engagement position. Preventing recurrence requires both an upstream intervention — rigorous incoming inspection of coupling ovality and protection ring condition — and a make-up procedure intervention: mandatory △-mark position verification for all stage cementing collar installations and explicit stop-work criteria for the observable anomalies that appeared during this incident's make-up sequence. For enquiries about ShunFu Metal's casing and OCTG product range, visit gaslinepipe.com.
