Why Rods Buckle in Curved Wellbores: A Practical Guide to Sideload Calculation
If a slender sucker rod is under any compressive load inside a tubing string, it has almost certainly already buckled — the Euler critical load for these length-to-diameter ratios is measured in tens of pounds, not thousands. So the question that actually matters isn't whether the rod buckled. It's how hard the buckled shape is pushing against the tubing wall, and what that contact force does to bending stress and fatigue life at the next coupling. That's the calculation most root-cause reviews skip, and it's usually where the answer is hiding.
The failure that looks like the wrong problem
A rod parts a few joints above a coupling, well inside a curved interval. The pumping unit's surface card shows nothing unusual — peak polished-rod load is within spec, the duty cycle looks like every other well on the pad. The instinct is to blame corrosion, a bad heat, or fatigue from the ordinary up-stroke/down-stroke tension cycle. Sometimes that's right. But in a curved or deviated section, there's a second load path that never shows up on a surface card: on the down-stroke, once the rod goes into compression, it buckles against the tubing wall, and every revolution bends it at that contact point. That's a fatigue driver stacked on top of the one everyone already accounts for — and if you only model axial tension, you're missing it entirely.
Why buckling isn't optional once you're in compression
The classic check is the Euler critical load for a pin-ended column: Pcr = π²EI / L². It looks like a threshold you might or might not cross. In practice, for a slender rod over a typical joint length, you cross it almost immediately. Take a 7/8" Grade D rod over a 25 ft segment: Pcr works out to roughly 95 lbs. A few thousand pounds of compressive load — trivial by sucker-rod standards — exceeds that by 30× or more. The practical rule: if the axial load in a segment goes negative at all, treat it as buckled. The interesting engineering question starts on the other side of that line.
From buckled shape to contact force: Mitchell's model
Once a rod is buckled inside a constraining channel, it doesn't spiral out to infinite deflection — the tubing wall stops it, and the buckled shape settles into a helix pressing against that wall. The question is how hard. Mitchell's helical buckling model gives the contact force per unit length as a function of the radial clearance and the compressive load:
fc = r · F² / (4EI)
where r is the radial clearance between the rod OD and the tubing ID, F is the compressive axial force, and EI is the rod's bending stiffness. Two things fall out of that relationship that matter for design. First, the force grows with the square of the compressive load — doubling the compression quadruples the contact force, not doubles it. Second, it scales with clearance: a looser fit between rod and tubing (a smaller rod in a larger ID, or added centralizer clearance) gives the buckled shape more room to develop force before the wall arrests it. Tighter is not always better here, but the direction of the effect is worth having in your head before you spec hardware. (Mitchell, R.F. (1986), "Simple Frictional Analysis of Helical Buckling of Tubing," SPE Drilling Engineering 1(6):457–465; Mitchell, R.F. (1988), "New Concepts for Helical Buckling," SPE Drilling Engineering 3(3):303–310.)
Chord-sagitta geometry: why the wall matters more than the rod
Buckling contact force is only half the sideload picture. The other half is pure geometry, and it doesn't care about load at all. Over a curved interval, the hardware — centralizers, couplings — rides against the tubing wall while the rod body spans each joint as a straight chord. The curved channel bulges inward relative to that chord by an amount called the sagitta: h = R − √(R² − (L/2)²), where R is the radius of curvature implied by the dogleg severity and L is the segment length. If that sagitta bulge exceeds the clearance the rod has to work with, the rod is forced into contact with the wall regardless of axial load — geometry alone puts it there. This is why a rod string can develop real sideload and mid-span wear in a curved section even during a stroke phase where the axial load is nowhere near buckled. The two effects — Mitchell contact force under compression, and forced geometric contact from curvature — add together, and a model that only checks one of them will underestimate the sideload whenever the other is active.
Worked example you can reproduce
A 7/8" Grade D rod, 25 ft segment, in a 2.441" ID tubing string with a 2.30" OD centralizer, running through an 8°/100 ft dogleg at 30° from vertical, carrying 150 lb·ft of drive torque with 15% ripple. On the down-stroke this segment sees 3,000 lbs of compression:
| Rod diameter | 0.875 in (7/8") |
| Segment length | 25 ft (300 in) |
| Axial load | −3,000 lbs (compression) |
| Dogleg severity | 8°/100 ft |
| Inclination | 30° from vertical |
| Drive torque | 150 lb·ft, 15% ripple |
| Tubing ID / centralizer OD | 2.441 in / 2.30 in |
| Euler critical load, Pcr | 95 lbs (compression exceeds this ~30×: buckled) |
| Mitchell buckling force, Fbuckle | 612 lbs |
| Gravity sideload, Fgrav | 26 lbs |
| Total lateral load, Plat | 638 lbs |
| Mid-span deflection (wall-limited) | 0.538 in |
| Peak bending stress (with joint SCF) | 4,703 psi |
| Von Mises stress | 25,607 psi |
| Yield safety factor | 3.9 |
| Combined fatigue utilisation | 0.36 (SF ≈ 2.8) |
Two things stand out. First, the buckling contact force (612 lbs) dominates the gravity term (26 lbs) by more than 20× — in a curved, compressed segment, Mitchell buckling is usually the sideload, not a minor correction to it. Second, the resulting fatigue utilisation (0.36) still leaves meaningful margin here, which is the point of running the number rather than assuming: a segment can be genuinely buckled and still be a non-issue, or it can be fine on paper for gravity alone and fail once you add the buckling term. You don't know which until you calculate it.
Run your own rod geometry through the Slender-Member Fatigue & Buckling tool to get the same buckling force, sideload, and fatigue utilisation instantly, with the full stress visualisation across the cross-section. It's a protected tool — contact us for access.
What this model does not account for
Worth stating plainly, in the same spirit as the worked numbers above: this is a single-segment, closed-form beam model, not a full-string finite element analysis, and it carries assumptions you should know before you lean on it. It treats the rod as a solid circular cross-section — no hollow or upset profiles. It analyses one segment in isolation and doesn't accumulate fatigue damage across the full string. It doesn't model buoyancy, fluid hydrostatic pressure, coupling weight, or drag forces. Material properties are minimum-specified values for the selected grade, not measured properties for your actual rod. The stress concentration factor at couplings is a fixed 2.5, which is representative for standard API hardware but will differ for special-clearance or upset connections. None of that makes the model wrong for its purpose — screening which segments deserve a closer look — but it does mean a result near the margin (a fatigue utilisation of 0.9, say, rather than 0.36) is a flag to investigate further, not a verdict to build a replacement schedule on unmodified.
The practitioner's decision rule
Three checks, in order, whenever you're troubleshooting a rod failure inside a curved or deviated interval:
- Is the segment ever in compression? If yes, assume it's buckled — the Euler threshold for slender rods is too low to treat as a real gate.
- Is the sagitta bulge larger than the available clearance? If yes, the rod is in forced geometric contact regardless of load, and that alone can drive mid-span wear.
- What does the combined sideload do to fatigue utilisation at the nearest coupling? This is where gravity, curvature-driven bowstring tension, and Mitchell buckling force all need to be summed — not just the largest of the three.
Skipping straight to a fatigue number without the first two checks is how a buckling-driven failure gets diagnosed as ordinary tensile fatigue, and the fix that follows doesn't address the actual load path.
FAQ
- What is helical buckling in a sucker rod string?
- When a slender rod is under compressive axial load inside a constraining tubing string, it buckles into a helical shape and presses against the tubing wall. The resulting contact force adds a lateral (sideload) component that the rod doesn't see when it's in tension.
- What is Mitchell's helical buckling model?
- A model developed by R.F. Mitchell (SPE Drilling Engineering, 1986 and 1988) that gives the contact force per unit length of a helically buckled tubular as a function of radial clearance and the square of the compressive load: fc = r·F²/(4EI). It's widely used in drillstring and rod-string design for exactly this reason.
- Does a rod need to be under heavy load to buckle?
- No. For typical slender sucker rod dimensions, the Euler critical load is small — often under a few hundred pounds — so almost any compressive load buckles the rod. The design question is the resulting contact force and sideload, not whether buckling occurs at all.
- Can a rod develop sideload without being in compression?
- Yes. Over a curved interval, the chord-sagitta geometry between hardware contact points can force the rod against the tubing wall purely from curvature, independent of axial load. This adds to — and can exist without — Mitchell buckling contact force.
Run your own segment geometry through the Slender-Member Fatigue & Buckling simulator — curvature-driven sideload, Mitchell helical buckling, chord-sagitta geometry, and fatigue utilisation, computed server-side.