Thermo-Cognitive Advanced Process Control (TC-APC) models, simulates and optimizes your process in real time to improve yield, conversion and throughput, and reduce energy use, with a quick payback.
Concretes, absolutes, resinoids and oleoresins
The extract passed on residual solvent. What it lost getting there appears in no reading.
Solvent removal is run to a recipe and judged by one laboratory number, the solvent left behind.
It is a distillation with your aroma in the pot, and it is decided in the last hour.
The last of the solvent is bought with heat, and paid for in aroma.
Every natural extract is made the same way. A solvent takes what you want out of the flower, the resin or the spice, and then the solvent has to be taken out of what you want. Concrete or absolute, resinoid or oleoresin, it is the second step that decides what the batch is worth.
Most of the solvent comes off easily. The last of it does not. As the extract thickens, its boiling point climbs and what remains has to work its way out of a viscous mass, so the final hour is the hottest and the slowest of the strip. And the limit is written in parts per million: in a spice oleoresin sold into the United States, the hexane left in it may not exceed 25 parts per million, and the acetone 30.
That hour is also when the extract is most exposed. The lightest aroma molecules leave with the solvent, and heat darkens color and dulls odor. So there is a point where another hour of stripping costs more in aroma and color than the specification needs. Every plant is somewhere either side of it, and nothing on the panel says which side.
Engineering proposal. Begins with a free check.
The science under TC-APC
An extract's result is decided in the particle and in the still.
How much the solvent takes out is settled inside every particle of flower, resin or spice. How much aroma survives, and how much solvent stays behind, is settled between the vapor and the liquid as the solvent comes off, not by the temperature read on the panel. That's the science TC-APC is built on, and AI makes it sharper.
The curve is the same shape every time and the end point is never in the same place. It moves with the lot, the size of the charge, the wax and resin it carries, and the vacuum on the day.
A fixed recipe cannot stop on a moving end point. It can only stop safely past the worst one.
And being wrong costs differently on each side.
Stop short of it
The extract fails on residual solvent. The lot is held and stripped again, which costs it more heat than getting it right the first time would have, and if a customer finds it, everyone remembers.
Push past it
The extract passes. The top notes that left with the last of the solvent, and the color the extra heat took, appear in no alarm, no analysis and no shift log, because the extract is tested against a limit, not against what it could have been.
One failure is loud and the other is silent, so after a few years every recipe is written for the hardest batch.
Sample one strip through its last hours. The answer is in your own laboratory.
Four reasons the end point will not stay put.
01
No two lots are alike
Flowers picked after rain, resin from a new source, a spice from another district or another year. Each lot yields differently, carries different waxes and gives up its solvent at a different pace. Most of this is known, on a lot sheet nobody feeds into the recipe.
02
The extract thickens as it goes
Near the end the batch is a viscous mass, and the last solvent has to work its way out of it. How fast depends on how much wax and resin the batch carries and how full the still is, and both change from batch to batch.
03
It is heat history, not temperature
What aroma and color respond to is the accumulated exposure, not the reading at any moment. A batch held near the limit for two hours is in a different condition from one that touched it once, and the panel shows the same number for both.
04
Under vacuum, the reading itself moves
A small pressure drift shifts the boiling point by several degrees, so the temperature you are steering by changes without the composition changing at all. The same problem as any vacuum column. Set out in full here.
This is the same problem as every other application on this site. The variable that determines the money is not measured, so the plant runs a fixed recipe with a margin built into it.
One step earlier, in the extractor, the same story.
How much the solvent takes out of the flower, the resin or the spice is decided by time, temperature, the solvent ratio and the number of washes. The first wash takes most of it. The last of the extract comes slowest, from deep inside each particle, so every extra wash buys less than the one before and costs solvent, heat and time.
Where to stop depends on the lot: how finely it was ground, how wet it is, how long it was stored. The recipe stops in the same place for all of them, and what stays behind in the spent material appears in no reading.
Where it applies
Four names, one process.
Concretes
Fresh flowers such as jasmine, tuberose and rose, extracted with a solvent, usually hexane. The waxy concrete is what remains once the solvent is gone.
Absolutes
A concrete or resinoid taken up in ethanol, chilled and filtered to drop the waxes, then stripped of the ethanol. Several hundred kilograms of flowers go into one kilogram of a flower absolute.
Resinoids
Dry resins and balsams such as benzoin, olibanum, labdanum and myrrh, extracted with a solvent that then has to come out of a thick, heat-sensitive mass.
Oleoresins
Spices such as pepper, capsicum, paprika, turmeric and ginger, extracted with acetone, ethanol or hexane, and sold on strength, color and residual solvent.
What TC-APC does about it
It works out when this batch is ready, and stops there.
It estimates the batch as it runs
How much solvent is still in the extract, and how much aroma is leaving with the vapor, continuously, from the instruments the still already has rather than from a new analyzer.
It carries a heat budget
Accumulated heat exposure is tracked through the batch, so the limit that applies in the last hour reflects what the extract has already been through.
It uses what you know about each lot
Moisture, assay and origin, from the checks you already run on each lot, set up the extraction and the strip before the batch starts, instead of being filed while the batch runs the same recipe as the one before it.
It reads from your existing DCS, PLC or SLC and writes setpoints back to it, so the strip ends when this batch is ready, not when the hardest one would be. Nothing is replaced, and it writes setpoints only after you have watched its estimate track your own laboratory results, for as long as you want. The deployment sequence in full.
What this cannot do.
It cannot put back what has gone
Top notes that left with the solvent, and color the heat has taken, are gone. The gain is in not losing them, which means the value is in the batches ahead rather than the drums behind.
It cannot improve the raw material
Poor flowers, resin or spice make a poor extract. The engine will show you how each lot behaved, which is useful in the conversation with your supplier.
It cannot beat the vacuum system
If the pump and condenser cannot hold pressure, the operating window is set by them. The estimate will report that accurately, and it is a mechanical answer, not a control one.
It does not replace your laboratory, or your nose
Your residual solvent, color and odor checks stay the reference. The estimate is checked against them every time they arrive, and the bias between the two is on the screen.
The mark
Find the optimum. Hold it while it moves.
The curve is a yield or selectivity response. The amber point is the optimum, and the dashed drop is where it sits on the operating axis.
The second half of the line is the part that matters. Finding an optimum is a study, and studies end up in drawers. Holding one while the feed, the fouling, the weather and the wear keep moving it is a control problem, and that is the work.
Start with one batch
Test one batch for residual solvent every half hour through its last hours
It costs a few laboratory runs and it answers the question before we do. If the solvent reaches its limit only at the very end of your recipe, the strip is already tight and there is little here worth pursuing. If it gets there well before, the time after that is the size of the prize, now a number rather than an argument.
projects@yielddynamics.in Talk to IIT Bombay engineers Call +91 94230 07004Monday to Saturday, 9:00 to 19:00 IST
Or start with the note
Cut-point determination in the vacuum fractionation of odour-critical products
Written for an aroma column, and the measurement argument underneath it is the same one. Three pages. Free, no form, and we do not ask for your email address.