When You Remove the Straight Part Between Two 90° Bends You Create a New Geometry

Removing the straight section between two 90° bends turns two separate turns into a single compound offset whose behavior depends on the bend radius, the distance between bends, and the direction of each turn. If both bends turn the same way, the removed spacer creates a tighter, higher-stress 180° reversal; if they turn opposite ways, it becomes a compact offset or jog with its own setback, gain, and fitting-to-fitting length.

This is true in sheet-metal ductwork, electrical conduit, pipe routing, and optical or structural layouts that use bent segments. The geometry is not merely "two elbows closer together." The short straight run between bends is a designed lever arm. It controls developed length, bend deduction, springback accumulation, and the way loads or flow reattach. Removing it changes all four at once.

Engineers often call the result a "close-coupled" bend pair, a "zero-tangent" bend, or a "compound bend." In conduit work, the same idea appears as a back-to-back bend. The term varies by trade, but the underlying geometry does not.

Mechanism: why the straight run matters

A bend is not a point. Every 90° turn consumes arc length and shifts the centerline of the part. A bend has an inside radius, a centerline radius, and a developed arc length equal to one quarter of the circumference at the neutral axis, roughly 1.571 times the neutral radius. Two bends separated by a straight segment each consume their own arc and also preserve the orientation and offset created by that straight segment.

The straight run between bends does three jobs. It sets the distance between bend tangent points, it determines the net direction change of the pair, and it absorbs springback and tolerance stack-up from the first bend before the second is formed. Remove it, and the tangent points of the two bends coincide or nearly coincide.

When the two bends turn the same direction, coincident tangent points produce a continuous 180° turn. The material now experiences a much longer total arc with only a brief or nonexistent neutral zone between bends. Strain accumulates rather than relaxes. The neutral axis shifts outward, the outer fiber stretches more, and the inner fiber compresses more. In thin-wall tube or duct, this is where ovalization and wrinkling concentrate.

When the two bends turn opposite directions, coincident tangent points produce an S-shaped offset with no straight leg. The pair behaves like a single fitting with a fixed setback. The centerline makes a lateral jog whose offset distance is governed by the bend radius and the angle, not by any straight length the fabricator can adjust.

There is a counterintuitive detail here. Removing the straight section does not always shorten the part. In a same-direction pair, the developed length can increase because the two quarter-circle arcs merge into a longer continuous arc with a larger effective bend allowance. The physical envelope shrinks while the cut length grows. That mismatch trips up fabricators who estimate material from the finished shape.

A second subtlety: the neutral axis radius is not the inside radius. For a given thickness, the neutral axis sits somewhere between the inside and outside surfaces, often near 0.3 to 0.5 of thickness from the inside face depending on the material and bend ratio. So the bend allowance you calculate from the inside radius alone will be wrong unless you convert.

Quantified context: radii, ratios, and allowances

The behavior of a close-coupled bend pair is dominated by the ratio of centerline radius to wall thickness or sheet thickness. Thin-wall parts, where that ratio is high, tolerate compound bends better because strain is distributed over more material. Thick-wall or low-ratio parts concentrate strain and are far more sensitive to a missing straight run.

For sheet metal, a common rule is that the minimum bend radius should be at least one material thickness for soft alloys and two to three thicknesses for harder tempers. For tube, practitioners often avoid centerline radii below about 1.5 times the outside diameter for cold bending without a mandrel. These are rules of thumb, not universal limits, and they shift with temper, wall thickness, and tooling.

Bend allowance for a 90° bend is approximately 1.571 times the neutral axis radius. Bend deduction, the amount subtracted from the flat pattern for each bend, equals twice the outside setback minus the bend allowance. When two bends are close-coupled, their setbacks overlap, so the flat pattern is not simply the sum of two independent bend deductions. Fabricators who apply single-bend deductions twice will over- or under-cut the blank.

Springback adds another quantified problem. Most metals recover a small angular amount after the tool releases, typically a fraction of a degree up to a few degrees depending on alloy, temper, and radius-to-thickness ratio. With a straight run between bends, the first bend's springback is partly absorbed before the second bend is formed. With no straight run, the two springbacks compound along a continuous arc, and the final angle can drift more than either bend alone would predict.

In ductwork, the straight run between two 90° elbows is often specified to preserve airflow. A close-coupled pair creates a sharper effective turn and a shorter pressure recovery zone. The pressure loss of the pair is not twice the loss of one elbow; it is higher because the flow never reattaches between turns.

Edge cases that change the answer

The same-direction close-coupled pair is the most mechanically demanding case. It produces a tight U-turn or 180° reversal with high outer-fiber strain. Thin-wall tube may ovalize; sheet metal may wrinkle on the inside of the combined arc; pipe may need a mandrel or hot bending to avoid collapse.

The opposite-direction close-coupled pair is more forgiving geometrically but harder to lay out. It creates a fixed offset with no adjustment leg. If the offset dimension is wrong, there is no straight run to trim or shim. The part must be remade.

A third edge case is the mixed-plane pair. If the two 90° bends are not in the same plane, removing the straight run creates a compound-angle transition rather than a simple reversal or offset. The layout becomes a three-dimensional problem, and the flat pattern or cut length cannot be computed from a single 2D projection.

Material matters as much as geometry. Annealed copper and soft aluminum tolerate close-coupled bends far better than cold-rolled steel or hard-temper stainless. Wall thickness relative to radius decides whether the bend is "thin-wall" or "thick-wall" in the sense that governs ovalization and wrinkling.

Warning signs that a close-coupled pair is outside safe practice include visible wrinkling on the inside of the combined arc, measurable ovalization of tube beyond a few percent of diameter, cracking or orange-peeling on the outer fiber, and springback that pushes the final angle outside tolerance. Any of these means the straight run was doing structural work, not just spacing work.

Misconceptions worth correcting

The most common misconception is that removing the straight run always saves space and material. It saves envelope space, but it can increase developed length and almost always increases local strain. The flat pattern and the finished shape move in opposite directions.

A second misconception is that two 90° bends are interchangeable with one 180° bend. They are not. A single 180° bend has one continuous arc with one set of tooling and one springback event. Two close-coupled 90° bends have two forming events, two springback events, and a transition zone that may not be fully formed. The results differ in ovalization, residual stress, and dimensional repeatability.

A third misconception is that the bend deduction for each bend can be added independently. Close-coupled bends share setback geometry, so the flat pattern must be computed from the combined tangent geometry, not from two isolated bends.

A fourth misconception is that duct or conduit close-coupled elbows behave like cataloged fitting pairs. Catalog pressure-loss and pull-resistance data assume a developed straight run between fittings. Remove it and the published coefficients no longer apply.

Practical actions: how to design and check a close-coupled pair

Start by deciding the pair type. Same-direction pairs need strain and ovalization checks. Opposite-direction pairs need offset and setback checks. Mixed-plane pairs need 3D layout verification.

Compute the combined bend allowance from the neutral axis, not the inside radius. Convert inside radius to neutral radius using the material's neutral-axis factor, then apply the arc-length formula to the total angle of the pair.

Check the minimum straight tangent requirement for your process. Many tube-bending shops require at least one to two tube diameters of straight between bends to grip and support the part. Below that, tooling changes are required.

Verify wall-thickness-to-radius ratio before committing to a close-coupled design. Model springback for each bend and check whether the compounded result stays in tolerance. For ductwork, treat a close-coupled elbow pair as a new fitting and source loss data accordingly. For structural or load-bearing bent parts, run a strain check at the outer fiber of the combined arc. Inspect the first article for wrinkling, ovalization, and angle drift before running a batch.

A simple decision table helps compare the three main configurations.

The time course of a close-coupled bend job follows a predictable sequence.

Reconciling thresholds is the practical crux. A tube shop may require one diameter of straight between bends, a sheet-metal rule may allow zero straight with a larger radius, and a duct standard may require a minimum straight for pressure recovery. These thresholds answer different questions. The shop threshold protects tooling and grip; the sheet-metal rule protects strain; the duct rule protects flow. Apply the strictest threshold that matches the failure mode you care about, and document which one governs.

Limitations and what the geometry cannot tell you

Geometry predicts where strain and layout problems concentrate, but it does not predict the exact failure point for a given alloy, temper, thickness, and tooling set. Those require material testing or the fabricator's qualified process data. Do not treat any single radius-to-thickness rule as a universal safe limit.

Published pressure-loss, pull, or flow coefficients for fittings assume a developed straight run between them unless the source states otherwise. A close-coupled pair is a distinct fitting with distinct coefficients, and those coefficients should be sourced or tested rather than assumed.

Springback estimates are process-specific. Tooling condition, lubrication, bend speed, and material lot all shift the recovered angle. A close-coupled pair amplifies that shift because two springback events compound with no neutral zone to absorb them.

Finally, the keyword answer has a direction-dependent caveat that thin explainers miss. Removing the straight part between two 90° bends does not create one predictable new shape. It creates a reversal if the bends turn the same way and a fixed offset if they turn opposite ways. The design, tooling, and inspection steps diverge from there.