How we build and check these calculators

Last reviewed: July 28, 2026 · D. Shin

A calculator is only worth using if you can tell where its numbers come from. This page describes the process every tool on this site goes through, so you can judge the output instead of taking it on faith.

1. The problem comes first, not the formula

Each tool starts from a real task — measuring what is left on a part-used roll, reading a volume off a dip stick, stepping a stock acid down to a working strength. We write down what the user actually knows (usually a few measurements taken with a tape or a ruler) and what they need out. That determines the inputs; the formula is chosen to fit them, not the other way round.

This is why, for example, the roll length calculator asks for outside diameter, core diameter and material thickness rather than asking for the wound length you are trying to find, and why the tank tools accept a dip reading instead of a fill fraction. If a tool demands a number nobody has on the floor, it is the wrong tool.

2. Formulas are derived, then stated on the page

The mathematics behind these tools is standard: the area of an annulus for wound length, circular-segment geometry for horizontal cylinders, mass balance for dilution and mixing. We derive the working form ourselves from those first principles rather than copying an expression from another site, because a copied formula carries any error and any hidden unit assumption with it.

Every calculator page then shows the formula it uses. If a result looks wrong to you, the formula is right there to check against — and most pages include a worked example with real numbers that you can reproduce by hand or in a spreadsheet.

3. Reference data is cross-checked before it ships

Some tools need published physical data: density-to-concentration tables for sodium hydroxide, hydrochloric acid and sulfuric acid, the density curve of ethanol-water mixtures, material thickness conventions. That data is the most likely place for a silent error, so it gets the most attention.

Our rule for reference tables:

• At least two independent published sources — typically standard engineering and chemistry references such as Perry's Chemical Engineers' Handbook, the CRC Handbook of Chemistry and Physics, and NIST data — must agree before a value is used.
• Where sources disagree beyond rounding, we do not average them. We either find a third source to break the tie, or narrow the published range of the table to the region where the sources agree.
• We never transcribe a table from another calculator site. Aggregator sites propagate each other's errors, and a table copied from one is unverifiable.
• Interpolation is stated. Where a tool interpolates between tabulated points, the page says so, because interpolation is exactly where a value can drift away from its source.

4. Testing against known answers

Before a calculator goes live it is run against reference cases with independently known answers — a hand calculation, a published worked example, or a geometric case with an exact closed-form result (a half-full horizontal cylinder is exactly half its total volume; a roll wound to twice the core diameter has a known annular area). Edge behaviour gets explicit attention: zero and negative inputs, a dip reading at exactly the tank crown, unit switches at the boundary between metric and imperial.

Calculations run entirely in your browser in double-precision floating point. We round displayed results to a sensible number of significant figures for the input precision — showing eight decimal places on a measurement taken with a tape measure would be false confidence, not accuracy.

5. Units and where errors really come from

In practice, most wrong answers in this domain are not formula errors — they are unit errors and measurement errors. So every tool works in both metric and imperial, converts internally through a single canonical unit rather than pairwise, and labels every field with its unit.

It is worth being blunt about the second source of error, because no calculator can fix it: the output cannot be more accurate than the input. A roll diameter measured to ±2 mm and a material thickness quoted to one significant figure will not produce a length good to the metre, however many digits the screen shows. Where a result is sensitive to a particular input, the page says which one.

6. Review dates and corrections

Pages carry the date they were last reviewed. A review means the formula, the reference data and the worked example were re-checked against sources — not that a typo was fixed.

When a reader reports an error we reproduce it first, then correct the page and note the change. Corrections are credited to the person who reported them if they want to be named. We would rather publish a correction than defend a number, and a reported error is the cheapest quality control a site like this can get — if you find one, please tell us with the page and the inputs you used.

What these tools are not

They are calculation aids. They are not engineering advice, not a substitute for measurement, and not appropriate as the sole basis for safety-critical, structural, regulatory or commercial-settlement decisions. For that work, verify the figure physically and, where it matters, have a qualified engineer confirm it. Our full disclaimer is in the terms of service.