How much solvent can the loop return?

Collecting a waste stream and recovering its solvent are two different steps. Change both to see how much fresh acetonitrile is still needed.

Teaching model
Closed solvent circuit with fresh input, collection, recovery and solvent returnedFollow 1,000 kg of solventOne demand cycle · illustrative mass balanceFresh bought571.6 kgDrug synthesis1,000 kg usedCollect the dry stream56.0% collectedEvaporate, crystallize76.5% recovery yieldReturned428.4 kgFresh + returned = demand440.0 kg not collected131.6 kg lost in recoveryWater-containing streams are outside this loop.

Enlarge for bigger labels, then swipe sideways →

Fresh bought571.6 kg
Returned428.4 kg
Reduction42.8 %

56.0% collection × 76.5% yield returns 428.4 kg. Fresh purchases are 571.6 kg per 1,000 kg demand, a 42.8% reduction on this illustrative common basis.

What this model assumes

Demand is fixed at 1,000 kg per cycle. Returned solvent = demand × collected fraction × recovery yield; fresh demand = demand − returned. Collection and yield use the same solvent-mass basis in this illustration. Uncollected solvent and recovery losses are excluded from reuse. The poster’s 56% collection and 76.5% overall recovery multiply to 42.84% on this common basis, not its separate ~75% procurement claim. Actual denominators, plant/laboratory consumption and allocation require clarification. The 75% preset is hypothetical and does not resolve that discrepancy. Non-aqueous streams only; no water separation, quality approval, operating economics or energy balance is modeled.

Collection and yield multiply.

The cited SK pharmteco poster describes recovery from non-aqueous oligonucleotide synthesis streams. Evaporation removes heavy material; melt crystallization then separates acetonitrile from the remaining mixture.

Its inputs include 56% collection and 76.5% overall recovery yield. On the common mass basis used here, those numbers return 428.4 kg per 1,000 kg demand. The poster separately claims about 75% less purchased fresh solvent. That headline cannot be reproduced from these two fractions alone; its denominator and process boundary need clarification.

The 75% preset shows one hypothetical combination that reaches that reduction. It is not a reconstruction of the pilot. Water-containing streams need additional treatment and remain outside this model.

If half the used solvent is collected and half of that is recovered, how much demand can reuse cover?

Taking a real process
beyond the lab?

Jon and I are building RARE Labs toward an autonomous pilot plant builder using AI and robotics. Join the waitlist for updates, and tell us what you’re trying to scale.

Join the RARE Labs waitlist ↗

Go a little deeper.