Batch Scaling Calculator

The scale factor is a ratio of two batch sizes, and applying it to a component amount is a single multiplication. The assumption folded into that multiplication is that every property of the process that matters is proportional to batch size, so that the ratios between components, the order of addition, the temperature and the timings all stay exactly where they were. That is a strong assumption and it is the whole content of the calculation. The calculator applies it faithfully and has no way whatsoever to test it against the physical process.

It helps to be precise about what linearity predicts and what it does not. It predicts an amount: ten becomes fifty at a factor of five, for every component, without exception. It does not predict a yield, a rate, a purity, a particle size or a duration. Nothing in this arithmetic guarantees that a process which achieved a particular result at the base size will achieve it at five times the volume, and the yield percentage calculator measures that outcome rather than predicting it, which is the distinction this page exists to draw. A scaled recipe is an estimate of what to try, not a forecast of what will be obtained.

The factor is also easy to misread in the direction that matters. A factor of 0.2 is a scale down to a fifth, not a reduction of twenty percent, and a factor of 5 is five times the component rather than five percent more of it. Run the arithmetic on one component before reading anything else: if the scaled amount for that component looks wrong, nothing about the others is worth reading. And apply the same factor to every component. A recipe scaled selectively is not an arithmetic error but a different recipe, and it is the easiest way to produce a process nobody validated.

Several things stop being proportional as volume grows, and the first two bite almost immediately. Vessel geometry does not scale, because doubling every linear dimension multiplies volume by eight rather than two, so a vessel built to hold twice the material has a different depth, a different shape and a different pattern of flow. Mixing power per unit volume falls while the distance fluid must travel to become uniform grows, so the time needed to homogenise rises even though the written stirring time did not change by a second. Headspace changes too, and anything sensitive to a gas phase meets a different vessel.

Thermal behaviour is the other near universal failure, and it follows from the same geometry. A reaction rate per unit volume can be unchanged in a linear scale while the heat released per unit volume rises, because the surface available to remove it falls relative to the volume producing it. The result is a hotter core, a slower return to the target temperature, and, where the material is sensitive, a different product distribution. Mass transfer behaves the same way: a larger vessel gives a longer diffusion path, so anything limited by dissolution or by transfer between phases slows in proportion to length rather than to volume.

Solubility and nucleation sit further along the same axis. A concentration that was comfortably soluble becomes supersaturated when the cooling rate changes with scale, and a supersaturated solution that nucleated cleanly in a small vessel may stay clear or crash at once in a large one. Precipitate and crystal size are set by how fast supersaturation is relieved against how fast nuclei appear, and both of those shift with mixing and heat transfer. None of this is exotic. It is the ordinary reason a process that works at one size needs re-optimising at another, and none of it is visible in the factor returned here.

Not everything in a recipe is multiplied by the factor, and the test is what the instruction actually says rather than what it seems to mean. Anything specified as a ratio, a percentage, a concentration or a proportion is defined relative to the batch and must scale with it. Anything specified as an absolute amount, a fixed catalyst charge, a seed quantity or an adjustment made to a measured value rather than to a calculated volume is defined independently of size and must not. Where a stock solution supplies a component, the volume of that stock scales so that the mass it delivers scales with it, and the stock solution calculator is where that volume gets recomputed.

This is where a scaled recipe quietly stops being the recipe that was validated. A catalyst added as a fixed mass rather than as a loading fraction was probably fixed for a reason connected to the original size, and multiplying it changes the process instead of reproducing it. When scaling, write the new recipe out in full and check every line against the original wording rather than applying a blanket factor, and resolve anything ambiguous before starting. Where a component arrives as a dilution step rather than as a volume, the C1V1 dilution calculator is what keeps the resulting concentration fixed while the volumes move.

Scaling down is not safe merely because it is smaller. The surface to volume ratio moves the wrong way, so losses to glass, to filters and to transfer tools become a larger fraction of a smaller batch. Dead volume in the vessel and in the pipettes stops being negligible, and weighing error stops being negligible, because the absolute tolerance of a balance is a fixed mass and dividing that fixed mass by a smaller batch makes a larger relative error. A recipe that works in a beaker can fail in a vial for precisely these reasons.

Small volume arithmetic is the same point seen numerically, and the guide to rounding and small volume precision covers it in detail. The useful discipline is to decide which measurements will carry the result and keep them comfortably clear of the resolution of the equipment, rather than to compute with care and then weigh at the edge of what a balance can read. When many samples are needed from the smaller batch, the sample preparation calculator handles the volume side of that, and the honest test of whether the scale held is a per step recovery percentage measured at the new size rather than an assumption that it transferred.