Scale-up and technology transfer
Why a safe lab reaction can be unsafe in the plant, and which process changes need new safety data.
Educational overview. The principles here are general and simplified, and the figures and rules of thumb are typical values, not limits for your plant. Decisions need process-specific data, the applicable codes and standards, and a qualified assessment.
Why a safe lab reaction can be unsafe in the plant
A 1 L flask and a 5 m³ reactor run the same chemistry under very different heat-transfer conditions. Volume, and so heat production, grows with the cube of the size; jacket area, and so heat removal, grows only with the square. The area per unit volume falls roughly as 1/D: a 5,000 L reactor has around a tenth of the cooling area per litre of a lab flask.
| Lab flask (1 L) | Plant reactor (5 m³) | |
|---|---|---|
| Area / volume | ~50–100 m²/m³ | ~2–4 m²/m³ |
| Heat loss when cooling fails | High: the flask cools itself in air | Almost none: practically adiabatic |
| Mixing time | Seconds | Tens of seconds to minutes |
| Time to heat / cool | Minutes | Hours |
So a reaction that “just got warm” in the lab can run away at scale, and a 10-minute lab addition may need 4 hours in the plant (lesson 4).
Things that change reactivity during scale-up
- Batch vs semi-batch. In an all-in batch there is no control of the rate: all the energy may be released at once. Proving that an addition is dose-controlled (sample just after the end of the addition: little unreacted reagent left) removes most of the risk.
- Concentration. Reactions in 7–10 volumes of solvent are rarely hazardous; below about 5 volumes, ΔTad rises and more testing is needed.
- Large excesses of reagent. Often a sign the chemistry is not understood; the excess may cause side reactions or decompositions and needs a planned quench.
- Longer times at temperature. Distillations, hold times and filtrations are much longer in the plant. A slow decomposition that was invisible in the lab can matter.
- Solids and mixing. Poor suspension or a stopped agitator lets a reagent layer accumulate unreacted, then react suddenly when mixing restarts.
- Materials of construction and contaminants. Rust, metal ions or water from utilities can catalyse decompositions (e.g. of peroxides, hydroxylamine, DMSO mixtures).
Changes that need new safety data
If the initial assessment was lean, many changes are fine without re-testing. These are not, and should go back to the process safety lab:
- Anything that increases heat or gas generation, or their rates: new reagents, molar ratios, higher concentration, faster addition, a significantly different reaction or work-up temperature.
- Changing a dose-controlled step into an all-in process.
- A change in the composition of a distillation residue.
- Charging a reactive reagent as a concentrated solution instead of a solid (compatibility: e.g. NBS in NMP or DMF, nitric acid with organic solvents).
Every change to chemistry, equipment, procedure or staffing that can affect a hazard must be reviewed before it is made, by people with the right expertise. Flixborough is the textbook case of what happens without it.
Key takeaways
- Heat production scales with volume, cooling with area: plants are nearly adiabatic.
- Dose-controlled operation, enough dilution and modest excesses reduce scale-up risk.
- Changes that raise heat or gas generation, all-in charging, new residues or concentrated reagent solutions need new data and MOC.