The Lift That Was Within Capacity and Still Went Wrong

Why this matters

Nobody on this job did arithmetic that was wrong. Every check somebody made returned a pass, and the lift still came up crooked, swung into a parapet and put a sling out of service on the first pick of the day. The reason is that "within capacity" is a property of a product, not of any single factor, and the crew checked three factors one at a time against three different baselines. Derates compound. They multiply, they do not take turns, and they get applied in a fixed order. This is the reconstruction, worked from what the job actually left behind, which was less than anybody expected.

What the record contained

The lift was over before anyone asked a question, which is the normal case. The evidence was a job file, two phone photos and a tagged sling.

• The equipment data plate, photographed at the pick: operating weight 4,000 lb.
• A photo taken from the ground as the package cleared the curb, showing the load out of level and both bridle legs visible against the sky.
• The sling tag, photographed by the rigger before the pick as a habit: vertical 4,200 lb, choker 3,360 lb, basket 8,400 lb.
• The sling itself, tagged out with a cut through the cover and into the load-bearing yarns at one edge contact point.
• A lift plan, one page, listing the weight, the sling and a planned leg angle of 60 degrees from horizontal.

And the gaps, which are the more useful half. No angle was recorded at the pick. No centre of gravity was determined; the plan said "pick points per the unit's lifting lugs" and stopped. No hitch was written down anywhere, and the crew rigged a choker around the frame rails because the lugs did not line up with the sling lengths they had on the truck. Nobody wrote down that they had changed the hitch, because changing the hitch did not feel like changing the plan.

Reading the angle off the photograph

The one number the record did not carry was recoverable from the photo, and it is the number everything else turns on.

Measured against the horizontal frame rail of the package, which is a straight reference in the same plane, both legs came in at roughly 40 degrees from horizontal against a planned 60. State the convention every time you say it: from horizontal. Half the trade quotes sling angle from the vertical instead, and the two conventions give reciprocal answers, so "40 degrees" with no convention attached is not information.

Why it was 40 and not 60: the crew's slings were the length they had, and the choker they rigged took the pick points out to the frame rails, which sit wider apart than the lugs the plan assumed. Wider pick points and the same sling length is a shallower angle. Nobody chose it and nobody noticed it, because the choice that produced it was a choice about which hitch to use.

                    hook
                     /\
       leg A        /  \        leg B
                   /    \
      -----[======*======]------ frame
           |      |      |
        rail    C of G   rail
                 (off centre, sits
                  nearer rail A)

  angle measured from HORIZONTAL, at the frame
  load hangs so the C of G finds the hook, which
  tilts the package and loads leg A harder

The three checks that each passed

Here is what people actually said, in order, and what each one compared.

"The sling is rated 4,200 and the load is 4,000, so we are fine." This compares the total load against one leg's vertical rating. It is a comparison between two quantities that do not belong in the same sentence, and it passes by 5%, which is exactly the kind of narrow pass that makes people stop checking.

"There are two legs, so it is 2,000 each." True only if the centre of gravity sits midway between the pick points and only in a vertical hitch. Neither held.

"The angle is a bit shallow but we are miles under." No number was applied. "Miles under" was inherited from the first check, which had already compared the wrong pair of quantities.

Three passes, three different baselines, none of them the one the sling was actually working against.

Doing it in the right order

Derates compound multiplicatively, and the order matters because each one operates on the output of the last. Share, then angle, then hitch, then dynamics.

Line 1, total load. 4,000 lb, from the data plate, confirmed as an operating weight rather than a shipping weight.

Line 2, share on the heavy leg. The package's centre of gravity sits nearer one end, which is why it came up out of level: a suspended load rotates until its centre of gravity is under the hook. From the pick-point geometry, the planner's split is 60 / 40. That is a simplification and it needs saying: with unequal shares and equal-length legs the geometry is not symmetric and the two legs do not sit at the same angle, so the real distribution is a statics problem that belongs to the qualified person, not to a photo. Used here as a bound in the conservative direction, which means assigning the larger share to the leg being checked. Heavy leg vertical share: 0.60 x 4,000 = 2,400 lb.

Line 3, angle correction, per leg. Leg tension is the vertical share divided by the sine of the angle from horizontal. At the planned 60 degrees that factor is 1 / 0.866 = 1.155. At the measured 40 degrees it is 1 / 0.643 = 1.556. Applied to line 2: 2,400 / 0.643 = 3,734 lb, rounded up to 3,740 lb, because a tension is never rounded down.

Line 4, hitch correction, and this is a re-basing rather than an addition. The sling was in a choker. The tag's choker line is 3,360 lb, and that figure already contains the hitch derate. So the correction is not a multiplication applied to the tension; it is a change of which line on the tag the tension gets compared against. Multiplying the tension by a remembered choker percentage and then comparing it to the choker line charges the hitch twice, which produces a number that is wrong in the direction that stops the lift, and a crew that has been burned by that once will stop applying the hitch at all. Compare 3,740 lb against 3,360 lb.

Line 5, dynamic factor. 1.0. The load was raised smoothly and did not snatch. Printed because it is the term that would have taken this from a near miss to a drop, and a sibling card carries the multiplier and how fast it climbs.

The product. 3,740 lb against a governing 3,360 lb. That is 111% of the sling's rated capacity in the hitch it was actually rigged in, on a lift where every individual check returned a pass.

Against the plan. The plan's own figure was 2,000 / 0.866 = 2,309 lb, rounded up to 2,310 lb, at 55% of the vertical line the plan assumed. The lift executed at 1.6 times the tension the plan carried, and against a governing capacity 20% lower than the one the plan named. Both halves moved, in opposite directions.

Why the sling did not part

It matters that this ends in a near miss rather than a fatality, because the shape of the outcome is what people learn from and the wrong lesson is available.

At a design factor of 5, the tag's 4,200 lb vertical implies a breaking strength around 21,000 lb, and 3,740 lb is nowhere near it. Being at 111% of a rated capacity does not break a sling. What it does is spend the margin that the design factor exists to cover: manufacturing variation, wear nobody logged, the abrasion at an unprotected edge, and a weight figure that came off a data plate rather than a scale.

The actual sequence was mechanical and it started with the centre of gravity. The load rotated as it lifted, because it always will, and the rotation pulled the choker along the frame rail rather than letting it bite. The package came up out of level, swung as it cleared the curb, and touched the parapet. The sling's edge contact on the frame rail, unprotected, cut into the load-bearing yarns during the movement. Nothing dropped. One sling was destroyed, the pick was aborted, and the crew got a second attempt they had not planned for.

The wrong lesson available here is "we were fine, the sling held." The right one is that the crew never knew what number they were operating at, and the one that finally told them was written on a sling they were throwing away.

What the record would have needed to catch it

Not more paperwork. Three fields, all of which take under a minute.

The hitch, written down at the pick rather than at the plan. The hitch was the change that moved everything else, and it was the one decision made on the roof with no record. A plan field that says "hitch as planned / hitch as rigged" would have forced the second entry.

The angle, measured at the pick. Not estimated from the cab, not inherited from the drawing. An angle finder, a phone level against a leg, or the same photo the crew already took, read at the time rather than a week later.

The centre of gravity, or an explicit statement that it was assumed central. The plan said nothing, which reads as "central" and means "not considered." A field that forces a choice between "determined" and "assumed central" turns a silence into a decision somebody has to own.

Checking your own lift against this

Take the last lift plan you wrote and run one test on it: find the single number the plan compares the load against, and ask which tag line it came from and which hitch that line describes. If the plan names a capacity without naming a hitch, the plan has already made the mistake in this case, and it made it on paper before anybody got on a roof.

Then check the arithmetic direction. Every correction in the chain above makes the tension larger or the capacity smaller. If your plan's corrections ran the other way, you have applied at least one of them backwards, and the flattering direction is the one that drops a load.

References

• 29 CFR 1926.251 (rigging equipment for material handling), construction, including sling rated-capacity requirements
• 29 CFR 1910.184 (slings), general industry
• The sling manufacturer's tag and rating documents, which are the governing source for vertical, choker and basket capacities and for whether an angle-specific capacity is already derated
• ASME B30.9 (slings), in the edition your jurisdiction, contract or employer programme has adopted
• See related: What Shock Loading Does That a Static Load Does Not; Stopping a Lift and What Happens Next; How to Estimate a Load Weight When Nobody Knows It