How to Find a Load Center of Gravity Without a Drawing

Why this matters

You cannot find a centre of gravity in the field. You can only narrow a bound around it, and the rigging has to tolerate whatever bound you end up with. That reframing is the whole method: every stage below is judged by how much it narrows the band, and the last stage is pricing the band you did not close. A crew that treats the answer as a point will size the rig on a number they never earned; a crew that treats it as a band will size the rig on the worst case inside it, and will know exactly what a measurement would have bought them.

Before you start

Land the load and block it before anyone touches rigging, and keep it landed for every stage except the trial lift. Nobody works under a suspended load and nobody works inside the arc it could swing through; in construction that clearing duty for cranes and derricks sits at 29 CFR 1926.1425, and in general industry the load-handling requirements at 29 CFR 1910.179 carry it for overhead and gantry cranes.

Three more conditions, each attached to the act that needs it. Anything you slide dunnage or a load cell under gets cribbed and blocked so it cannot roll or settle onto a hand, and the cribbing goes in from the side with the load supported, never with fingers under an unsupported edge. Anything you jack gets a mechanical stand or crib under it before anybody reaches past the jack, because a hydraulic jack holding a load is not a support device. And where the load carries stored energy of its own, a charged accumulator, a compressed spring, a pressurised vessel or a live circuit, that energy is isolated and verified before the load is handled, under 29 CFR 1910.147 for mechanical and stored energy in general industry, or under 29 CFR 1910.333(b)(2) where the hazard is electrical, because 1910.147 expressly carves electrical utilisation work out to Subpart S.

Stage 0: the bound you start with

Write it down, because everything after this is measured against it. With no information at all, the centre of gravity lies somewhere within the load's own extent. On a 10.0 ft long skid that is a 10.0 ft band in the longitudinal direction. That is a useless bound, and naming it is what makes the next stages feel like progress rather than ceremony.

What it is emphatically not is the geometric centre. That is not a stage-0 estimate, it is an assumption with no evidence under it, and the worked example below shows it landing 17.5 percent of the span away from the truth.

Stage 1: reason from where the mass actually is

Walk the load and find the dense subassemblies. A motor and gear reducer at one end, a transformer near the top of a cabinet, a compressor bolted to one corner, a full tank down one side, a battery bank in a base. Each pulls the centre of gravity toward itself in proportion to its own mass, and you do not need any of those masses to know the direction.

Bound it, do not point it. On the skid above, a drive package at the right-hand end and an empty frame at the left is enough to say the centre of gravity sits between the middle and the drive package: call it 5.0 ft to 9.0 ft from the left end, a band 4.0 ft wide. That is a factor of 2.5 narrower than stage 0 and it took two minutes.

This stage is judgement, not measurement, and its band should be wide enough that you would be surprised to be wrong. A crew that narrows stage 1 to a foot has stopped estimating and started guessing.

Stage 2: documentation, and knowing when to stop

Check three sources before you get out any instruments: the equipment manufacturer's installation or rigging documentation, the shipping documentation, and the load itself for designated lifting points or a marked centre of gravity.

Designated lifting points are information even when no centre of gravity is published, because they were placed by somebody who knew where the mass was. They are also a constraint: a manufacturer that designates lifting points has generally not rated the equipment for a pick from anywhere else, and using other points is a question for that manufacturer, not a preference.

If a published centre of gravity exists, you are done. Record it with its source and skip to the last stage. Everything from here is for when nobody wrote it down, and on the example skid nobody had, so stage 2 narrowed nothing.

Stage 3: weighed reactions

This is the measurement, and it is straightforward statics. Support the load at two points a known distance apart with a scale or load cell under each. Read both. The centre of gravity's distance from one support is the opposite support's reading multiplied by the span, divided by the total.

The example, carried all the way through. Skid 10.0 ft long, supports at the extreme ends so the span is 10.0 ft. Cell at the left end reads 2,600 lb. Cell at the right end reads 5,400 lb. Total 8,000 lb.

• Distance from the left support: 5,400 x 10.0 / 8,000 = 6.75 ft.
• Cross-check the other way: 2,600 x 10.0 / 8,000 = 3.25 ft from the right support, and 6.75 plus 3.25 is 10.0 ft. The two agree, which is the only free check in the method and it catches a transposed reading.

Repeat with the load turned 90 degrees, or with the supports placed across the other axis, to get the second horizontal coordinate.

The instrument error, with its basis and its character, before it goes anywhere near the answer. Take load cells specified at plus or minus 1 percent of reading, and treat the two as independent random spreads, which is the right character when they are separate instruments with separate calibrations.

• Left cell: 1 percent of 2,600 is 26 lb. Right cell: 1 percent of 5,400 is 54 lb.
• Work out what each does to the answer. The sensitivity of the computed distance to a 1 percent of reading error turns out to be identical for both cells, because both contributions come out as the span times both readings times 0.01, divided by the square of the total: 10.0 x 2,600 x 5,400 x 0.01 / 8,000 squared = 0.0219 ft each.
• Independent random spreads combine in quadrature, not by addition, so the combined spread is 0.0219 x the square root of 2 = 0.0310 ft.
• If you would rather carry a worst-case bound, those two terms add linearly and the result is reported with one inequality sign rather than as an interval: the error in the computed distance is not more than 0.044 ft.

Two error characters behave completely differently here, and knowing which you have is worth more than knowing the magnitude.

A common multiplicative offset, both cells reading the same percentage high because they were calibrated against the same reference, cancels exactly. Multiply both readings by the same factor and the computed distance is unchanged, because the factor appears in the numerator and in the total. You get that cancellation for free and only when the offset is genuinely common.

A common additive offset, both cells reading a fixed amount high regardless of load, does not cancel but is small and directional: add 20 lb to each and the computed distance moves from 6.75 to 6.74 ft, toward the middle.

Where the instruments are hydraulic jack gauges rather than load cells, the basis changes and it changes badly. A pressure gauge is normally specified as a percentage of full scale, so at a low reading the error expressed as a percentage of reading is much larger, and seal friction adds a systematic offset that differs between two jacks and therefore does not cancel. A gauge-based reading gives a much wider bound and should be recorded as one rather than written down to two decimals.

What stage 3 bought. The band went from 4.0 ft wide at stage 1 to 0.088 ft wide at stage 3, taking the worst-case bound of 0.044 ft either side. That is a factor of about 45. Note also which of the two earlier estimates was any good: the stage-1 band's midpoint of 7.0 ft was 0.25 ft off, while the geometric-centre assumption of 5.0 ft was 1.75 ft off, which on a 10.0 ft span is 17.5 percent. Reasoning about mass was worth something. Assuming the middle of the box was not.

Stage 4: the trial lift, which confirms rather than measures

The trial lift does not produce a coordinate. It produces a yes or a no on the attitude the plan predicted, and where the answer is no it produces a direction.

Rig to the measured position, take up the slack slowly with no snatch, and stop at a small clearance, on the order of a few inches, enough to read the attitude and no more. Shock loading is prohibited outright at 29 CFR 1910.184(c) in general industry and 29 CFR 1926.251 is the construction home of the same duty, and hoisting through slack is the most common way it happens. Everybody stands outside the arc the load can swing through, nobody under it, nobody touching it, and the observation is made from outside that zone rather than from beside the load.

If it hangs level, the plan held. If it tilts, the centre of gravity is toward the low side and the bound has just been narrowed in a direction, which is real information. Land the load, block it, and move the pick point with the rig fully slack. Rigging is never adjusted on a suspended load, and a hoist brake holding the load while somebody reaches in is not a parking device.

The height coordinate stays a bound

Weighed reactions give you horizontal position and nothing else. Getting the height by measurement means weighing the load again while it is tilted through a known angle, which is neither practical nor safe on most field loads.

So the height stays a reasoned estimate with a wide band, used the way a wide band should be: take the high end of it for any stability check, because a higher centre of gravity shrinks the clearance to the suspension point and is therefore the conservative choice. Where the load is tall, top-heavy, or picked from anywhere near its own mid-height, that determination belongs to a qualified person, and where the configuration is unusual, to a registered professional engineer. A sibling article covers why the height decides stability.

What the band cost, priced in leg tension

The point of the whole procedure. Two-point pick at the two ends of the same skid, legs at 60 degrees from horizontal, and the angle convention is stated because half the trade quotes it from the vertical: all sling angles here are measured from HORIZONTAL.

Every tension above is rounded up, because rounding a tension upward is the conservative direction.

Read the table as three decisions rather than three numbers. The assumed centre would have sized the heavy leg at 4,620 lb against a real 6,240 lb, so the requirement is 35 percent above what was selected, and the selection is short in the direction that parts a sling. Rigging honestly to the stage 1 band means 8,320 lb, which is safe and is 33 percent more sling than the measurement needs. Measuring turned an 8,320 lb requirement into a 6,240 lb one, a 25 percent reduction, and that is what the load cells bought.

One qualifier from the earlier stages has to appear on the same sheet. If the heavy leg is choked rather than hung vertical, the sling tag's choker rating replaces its vertical rating before you compare, and it is a replacement rather than a deduction. A rig sized on the vertical rating and hung in a choker has skipped a correction the tag already printed for you.

The failure mode of skipping this is not dramatic, which is why it persists. The load comes up, hangs crooked, somebody says it always does that, and the pick completes. One leg carried a third more than anyone intended, on a sling whose design factor absorbed it silently, and the next time the shortcut runs it will be on a heavier load or a steeper angle.

References

• 29 CFR 1910.184, Slings (general industry), safe operating practices at (c) including the shock loading prohibition, and 29 CFR 1926.251, Rigging equipment for material handling (construction)
• 29 CFR 1926.1425, Keeping clear of the load (construction cranes and derricks), and 29 CFR 1910.179 for general industry overhead and gantry cranes
• 29 CFR 1910.147 for control of hazardous mechanical and stored energy, and 29 CFR 1910.333(b)(2) where the hazard is electrical, which 1910.147 expressly excludes
• Equipment manufacturer's rigging and installation documentation: designated lifting points, shipping weight, and any published centre of gravity
• See related: Why the Center of Gravity Decides the Whole Lift; What Happens When the Hook Is Not Over the Center of Gravity