A part-used spool is one of the more annoying unknowns in 3D printing. Weighing it needs the empty reel weight, which is usually not printed anywhere. Measuring it with calipers needs no such information — the geometry gives the answer directly.
Three measurements
Measure across the filament at its widest point — that is the outer diameter. Measure the diameter of the bare core the filament is wound on. Then measure the width of the wound block, which is usually narrower than the gap between the flanges.
The filament occupies a ring. Its cross-sectional area is pi over four times the difference of the squares of the two diameters, and multiplying by the width gives the gross volume the winding takes up.
Calipers are ideal but a steel rule works. The outer diameter is the measurement that matters most, because it enters as a square.
Packing factor is the one estimate
Round filament cannot fill a rectangular space. Wound side by side in neat layers, circles in a square grid occupy 78.5 percent of the space. Wound so that each layer nests into the valleys of the one below, the theoretical maximum is 90.7 percent.
Real spools land between, and where depends on how carefully the manufacturer winds. 80 percent is a good starting point and the field is adjustable for a reason.
This is also the only figure here you cannot measure directly, which makes the calibration step below worth doing once.
Calibrate on a full spool
Take a brand new spool of a brand you use often. Measure it, run this calculation, and see what weight comes out.
If it says 1,000 grams, your packing factor is right. If it says 1,120, lower the packing factor until it reads 1,000 — for that brand, that is the correct number, and every part-used spool of the same brand will then read accurately.
Different brands wind differently, so the factor is per brand rather than universal. Writing it on the box takes five seconds.
Why the last quarter reads badly
Near the core, small errors in measuring the outer diameter have a large effect, because the ring is thin and the squares of two close numbers are close together.
A spool that is nearly empty is also often unevenly wound, with filament crossing over itself rather than lying in layers, so the packing assumption breaks down.
For the last hundred grams, weighing is more reliable than measuring — if you know the reel weight by then.
The runout question
The real reason to know this is the overnight print. A twelve hour print that runs out at hour ten wastes the filament, the time and often the part.
The rule of thumb worth adopting is to not start a print that needs more than 80 percent of what you believe is left. The measurement is good to a few percent, not to the gram.
Printers with a runout sensor change this calculation, because a pause is recoverable. Without one, a conservative margin is cheaper than a failed print.