Cable Pulling Tension and Sidewall Pressure to IEEE 1185 and AEIC CG5

Cable installed badly and cable installed well look identical from the outside. The damage happens during the pull and shows up years later as a jacket split, a deformed core or a failure at a bend. The calculation is not difficult; what makes it unreliable in practice is that different documents give different limits, and the two most commonly cited authorities are cited for the wrong thing.

01The two limits, and why the tighter one is usually the bend

The reason sidewall pressure catches people out is that it scales with tension but inversely with radius. A pull that sits comfortably inside the conductor limit at 40 % can exceed the sidewall limit at a single tight bend, because the tension has already multiplied around the earlier route. The worked example in section 07 demonstrates exactly that, with the same cable and the same tension in both cases.

02Which documents actually set the limits

This section exists because the most common error in cable pulling calculations is sourcing the limits to the wrong standard. We have made this error on this page in a previous version, so it is worth being explicit about what each document does and does not contain.

Scope statements taken from the IEC webstore entries for IEC 60502-1 and IEC 60364-5-52, the IEC 60502-1:2004 table of contents, IEEE 1185-2019 and IEEE 576-2000, the AEIC CG5 reference in trade technical guidance, and the publisher preview of ANSI/ICEA P-45-482-2017.

03Step-by-step pulling tension calculation

Break the route into segments and carry three quantities: the tension entering the segment, the geometry of the segment, and the tension leaving it. Work from the cable end back towards the winch, or from the winch forward — the arithmetic is the same, but for short pulls the bend factor is applied to the tension already in the cable, which is what makes bends disproportionately expensive.

Straight and sloped runs

For a horizontal straight run, tension accumulates as friction against the cable's own weight:

T = μ · W · L

For a sloped run the weight component along the route is added or subtracted depending on direction, giving the two forms used in IEEE 576 and IEEE 1185:

slope up: T = L · W · (μ cos θ + sin θ) slope down: T = L · W · (μ cos θ − sin θ)

Where W is the cable weight per unit length in newtons per metre, L is the segment length in metres and μ is the friction coefficient.

Bends — the capstan effect

T_out = T_in · e^μθ

θ is the angle of wrap in radians. This is the one conversion worth double-checking: IEEE 1185's annex presents the slope angle in degrees, while European manufacturers and most installation software work in radians, and a factor of 57 between the two produces a spectacularly wrong answer. One caution on our own part: we present θ in radians here because that is the convention in the capstan relation as used in the manufacturer installation guides; if your calculation sheet uses degrees anywhere, convert before you compute.

For calibration, the multiplier is 1.48 at 45° and 2.19 at 90° for a friction coefficient of 0.5, and works out at about 1.60 for a 90° bend at μ = 0.30. Three 90° bends in series do not add — they multiply, and that is why route design matters more than winch capacity.

Friction coefficient: the input most likely to be wrong

Measured values from lubricant manufacturer friction-table test data for PVC conduit. Assumed values from a major European cable maker's installation friction table. Note the scale of the gap: measured lubricated values sit at 0.11 or below, while design tables commonly assume 0.25–0.50. Using a design-table value is conservative. Using 0.5 as a default when a lubricated pull is planned may overstate tension by a factor of four or more — which is wasteful but safe, and far better than the reverse. We found no published friction coefficient for direct burial or for LSZH sheaths, and we are not going to invent one.

04Sidewall pressure at the bend

Sidewall pressure is the force the cable applies to the outside of the bend, per unit length of cable in contact with it:

P = T / R

T is the tension in newtons and R the bend radius in metres, giving P in newtons per metre. Some references use the tension entering the bend and some the tension leaving it; we use the exiting tension, which is the higher of the two and therefore conservative. In imperial units the same relation gives pounds-force per foot.

Conversion used: 1 lbf/ft = 0.014594 kN/m. We have deliberately not filled this table with manufacturer-specific numbers beyond the general default, because the permissible sidewall pressure is a property of the specific cable: jacketed, lead-sheathed and interlocked-armour constructions differ, and the figures we could find for individual constructions could not all be re-verified against a live published source at the time of writing. The 300 lbf/ft figure is the most widely applied general default and is what we would use when a datasheet is silent — but where a datasheet or an installation specification gives a different figure, that figure governs.

05Bend radius and fill during the pull

The radius that matters during installation is not the radius the cable will have in service. Cable bent to its minimum installed radius and then pulled around that curve experiences a sidewall pressure that the same bend in its final position never sees. Manufacturer practice is therefore to double the requirement during laying: an SWA multicore cable with a minimum installed radius of eight times the outer diameter is commonly pulled at sixteen times.

Because the standard route gives no figure — IEC 60502-1 contains none, and IEC 60364-5-52 states only that cables shall not be damaged — the numbers you work from are the manufacturer's published minimums plus the project's own laying specification. For a route being designed, the design decision is the opposite of the calculation: choose radii large enough that the sidewall check passes at the tightest bend with margin, rather than discovering the failure at the winch.

Fill matters for the same reason. Several cables in one duct share the space, bear on each other, and produce a combined sidewall load on the duct wall that is higher than any single cable would generate. Design tables for lubricated multiple-cable pulls start at μ = 0.45 with three cables in a duct, against 0.25–0.30 for a single cable — a 50 % to 80 % increase in the friction term before any bend is encountered. If the duct is already installed and congested, that number should be used even if the pull is well lubricated.

06Maximum tension: conductor limit against grip limit

The maximum tension a pull may develop is not one number but two, and they rarely coincide: what the conductor can bear, and what the pulling attachment can transmit into the cable.

The imperial rule comes from an ICEA-sponsored study and is reproduced in distributor technical guidance; it is a manufacturer and industry rule of thumb rather than a mandatory standard clause. The metric equivalents shown are the ones European manufacturers publish, and we are conscious that they do not agree with the imperial rule — 50 N/mm² against roughly 70 N/mm² for copper is a 30 % difference in the permissible load. We are showing both rather than selecting one. For aluminium the rule depends on temper: one distributor's tabulation puts hard-drawn aluminium at the same 0.008 figure as copper, while soft solid aluminium can be considerably lower.

What the pulling attachment transmits

• Pulling eye fitted to the conductor transmits the full conductor allowance, so the conductor rule above applies directly.
• Basket or stocking grip over the jacket transmits tension into the sheath rather than the conductor. Standard practice limits a grip to about 1,000 lbf, approximately 4.45 kN, per grip, with the limit also capped by the conductor allowance — so the grip is the governing constraint on a short, light pull and the conductor is governing on a long one.
• Interlocked armour constructions may be limited by the armour rather than by the conductor, because the armour wire is what takes the load when the grip bears on the sheath. Published tabulations vary by construction, so this is a case where the datasheet figure has to be requested rather than assumed.

The consequence is that a calculation showing 5 kN of required tension can be perfectly acceptable against a 6.7 kN conductor allowance and still exceed the grip limit if the pull is being made on a stocking grip rather than a pulling eye. The attachment is part of the calculation, not an afterthought — and where armour, grip and conductor give three different numbers, the lowest of the three governs.

07Worked example: a 95 mm² four-core SWA pull

The route below is deliberately ordinary: a ducted pull with one 90° bend. All inputs are stated so the arithmetic can be recomputed.

Outer diameter and mass checked against three published datasheets for 4 × 95 mm² constructions: 42.3 mm and 5,245 kg/km for a flexible Class 5 build, and 41.3 mm / 41.7 mm at approximately 5,280 and 5,600 kg/km for hard Class 2 builds. We use 5.60 kg/m, the heaviest, because a heavier cable produces higher tension and we would rather overstate the requirement than understate it.

Step 1 — tension after the first straight run

T₁ = 0.30 × 54.9 × 120 = 1,976 N

Step 2 — tension after the 90° bend

90° is 1.5708 radians, so μθ = 0.4712 and the capstan multiplier is e^0.4712 = 1.60.

T₂ = 1,976 × 1.60 = 3,162 N

Step 3 — tension after the second straight run

T₃ = 3,162 + (0.30 × 54.9 × 40) = 3,162 + 659 = 3,821 N

Step 4 — check against the conductor

Allowance = 95 mm² × 70.2 N/mm² = 6,669 N

The required pull tension is 3,821 N, which is 57 % of the imperial rule of thumb. Against the European manufacturer figure of 50 N/mm² the allowance is 4,750 N, so the same pull uses 80 % of it. Both are inside their limits — but note that the answer to "is this pull acceptable?" changes materially depending on which rule the specification follows, which is exactly why the rule should be fixed in the method statement before the pull, not chosen afterwards.

Step 5 — check the sidewall pressure

P = T₂ / R = 3,162 / 1.2 = 2,635 N/m = 181 lbf/ft

At 181 lbf/ft this is roughly 60 % of the 300 lbf/ft general default — the figure we would use when a cable datasheet is silent. On that basis the bend passes with margin; if the datasheet for the cable in hand publishes a higher permissible value, the margin is greater still.

08Field checks, and the equipment side

Calculation gets you a route and a limit. These are the checks that decide whether the pull on the day matches the calculation.

• A tension-monitoring device in the line, not a winch gauge. Winch hydraulic pressure is not a reliable proxy for line tension, particularly through a capstan winch.
• Lubrication applied at the entry and re-applied at intermediate points on long pulls, not once at the start.
• Rollers at every change of direction on above-ground sections, sized for the cable rather than for whatever was on the van.
• Confirmed published limits for the actual construction — bend radius and, where published, pulling tension and sidewall pressure.
• A stop condition agreed in advance: the line tension at which the pull pauses and the route is reviewed, written into the method statement.
• Route condition checked before the pull: duct alignment, debris, sharp edges at the duct mouth, and the temperature of the cable — cold cable is stiffer and less tolerant of tight bends.

Two of those items are the subject of their own pages. Equipment selection and the pre-installation checklist are set out in the contractor's pre-installation checklist and tension guide, and the hardware for long pulls — winches, tension monitors, rollers and grips — is covered in cable pulling tools for large installations. This page deliberately stays on the calculation; general laying methods and requirements sit in our guide to cable laying methods.

09What we publish, and what we will not pretend to publish

We publish construction data. For our CU/XLPE/LSZH/SWA/LSZH 0.6/1 kV armoured cable, that means IEC 60502-1 and BS 5467 construction, a minimum bend radius of twelve times the outer diameter for multicore and fifteen times for single-core, and, on request, the outer diameter and mass per metre for a specific size — which is what you actually need for the arithmetic above. If you are running a pull calculation, ask for those two figures for the exact construction and size quoted, not for a generic table, because the difference between a Class 2 and a Class 5 build of the same nominal size is material to the answer.

What we will not do is quote a pulling limit we cannot substantiate. We do not publish a blanket maximum pulling tension for a construction family, because the limit belongs to the conductor build and the attachment, and it changes with the size. We could not locate any IEC, EN or BS test for pulling-induced damage — IEC 60502-1's test list contains a bending test but no pulling test, and the tensile test that exists in IEC 60794 applies to optical fibre cable, not power cable. So there is no certification that says a cable survives a pull; there is only the calculation, the route, and the discipline of stopping when the number is reached. We would rather say that plainly than imply a guarantee the standards do not offer.