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.
Approach
The variable that decides the money is not the one you are measuring.
Eight industries, eight instruments, and the same sentence underneath all of them.
Nobody is steering badly. They are steering by a proxy.
Every process has a boiling point.
Some number that tells you something is happening, and not what. The instrument changes with the industry. The sentence underneath does not.
| Industry | What it is steered by | What that cannot tell you |
|---|---|---|
| Aroma distillation | Temperature at the column head | Which compounds are boiling, only that something is |
| Natural extracts | Still temperature, vacuum and a clock | How much solvent is still in the extract, and how much aroma has left with the rest |
| IPA and DIPE | The same reading, near an azeotropic pinch | Anything at all, over the range that decides the product |
| Solvent recovery | Pot temperature and a clock | How much solvent is still in the residue, and how much has degraded getting there |
| Sugar pans | Proof, movement, and the look of the massecuite | Supersaturation, which is the thing that decides the crystal |
| Soda ash carbonation | A periodic analysis of what left the tower | What is happening inside it, which is where the yield is decided |
| Surimi to crab stick | A gel strength grade on a certificate | What this paste will do today, after a year of aging |
| Boilers | Oxygen at the economizer | Whether that is the right oxygen for the coal on the grate today |
| Reaction selectivity | Temperature and elapsed time | Which product the last degree of temperature actually bought |
What a plant does when it cannot see, and why.
One kind of failure is loud
Off-spec product. An odor complaint, a held batch, a requalification, a CO alarm. It produces a laboratory result, a meeting and a memory.
The other is silent
Prime product in the wrong drum, solvent in the residue, steam spent on air, a slightly smaller crystal. No alarm, no analysis, no shift log. It is visible only in a mass balance most plants do not compute.
So the operator holds a margin, and holds it wide enough to survive the worst case he can imagine.
That is the correct response to asymmetric consequences, and it is expensive.
This is the part that matters and the part most often got wrong. The plant is not being run badly. The margin exists because the signal is missing, and the width of the margin is the size of the missing signal. Take a plant with no composition measurement and remove its margin and you do not get efficiency, you get off-spec product. The only honest way to narrow the margin is to supply the signal.
The first principle
Detect. Do not predict.
There are two ways to supply a missing signal. One of them quietly carries somebody else's error into your plant.
Estimate the material going in and you inherit the error in that estimate, multiplied by however sensitive your decision is to it. Read the process while it is running and that term is not in the arithmetic at all.
What is left is signal noise in your own plant, which is something we can measure rather than argue about.
The second principle
Physics fixes the shape. Your plant fixes the constants.
Why data alone is not enough
A model learned only from operating data knows the states the plant has been in. It has never seen the rare and expensive ones, which are exactly the ones worth controlling, and it has no way to tell you it is extrapolating.
Why physics alone is not enough
A textbook equilibrium does not know your tray efficiency, your holdup, your heat loss, your fouling, or how your packing has aged. The idealized answer and your column are two different things.
So the model is built from both
Thermodynamics, kinetics and transport give the structure of the answer, and it holds in states the plant has never visited. Your own data resolves the constants the physics cannot supply. Neither half is decorative.
The estimate does not need to be exact.
It needs to be better than a reading that is confounded, and to update continuously rather than every half hour.
The third principle
An estimate that publishes its own error.
The screen shows the estimate
Continuously, on every stream or every vessel, where before there was one reading and a laboratory result at intervals.
Beside the last lab result
And the running bias between the two, on the same screen, in the same minute. When your laboratory disagrees, the operator sees it without asking anybody.
Which is the larger claim
A number said to be exact can only be taken on trust. A number that shows its own disagreement with your laboratory is checked every shift, on your bench, against your instrument.
One engine, and where it sits.
TC-APC, Thermo-Cognitive Advanced Process Control, reads a process through its thermodynamics. Where the variable that decides the money is mechanical rather than thermal, the same engine reads it through its rheology, and that one is RC-APC, Rheo-Cognitive Advanced Process Control. The estimator changes. The principle does not.
Either way it is a supervisory layer. It sits above the control system you already have, reads process variables from it and writes setpoints back to it. The loop between your DCS and your plant is untouched, and the guardrails that hand control back live inside your DCS rather than inside our box.
What this approach cannot do.
This list is on every application page in its own terms, and it is the part of the argument we would most like you to hold us to.
It cannot loosen a specification
Quality limits enter as hard constraints, never as terms to be traded against yield. A scheme that improves recovery by relaxing an impurity limit has not solved the problem. It has moved it to your customer.
It cannot defeat a phase equilibrium
Where a separation is capped by an azeotrope, or a reaction by its thermodynamics, no control policy changes that. The answer is a different process, not better control of this one.
It cannot supply equipment you do not have
An under-staged, fouled or flooding column, a pan with poor circulation, a boiler leaking air. A good estimator reports accurately that the capability is not there, which is worth knowing and is not a yield improvement.
It does not replace your laboratory
It interpolates between laboratory results and holds the plant steady in between. The laboratory stays the reference and the arbiter, and the disagreement between the two is displayed rather than hidden.
And the commercial one, which matters more than any of the above. Where your material is approved against a reference sample, any change that shifts the impurity profile can trigger requalification even when it looks like an improvement. So the first objective in this work should be reduced batch-to-batch variation, not increased yield. Consistency is what your customer is paying for, it is the easier case to make internally, and the recovery tends to follow anyway.
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 the note
Cut-point determination in the vacuum fractionation of odour-critical products
The argument on this page, worked through properly on one kind of column, including a section on what it cannot do. Three pages. Free, no form, and we do not ask for your email address.
Or start with your own numbers
Send operating data from one unit
We will tell you how wide the margin you are holding actually is and what it is costing, with the uncertainty stated openly rather than buried. No charge, and no commitment either way. Including the answer that there is nothing here worth pursuing.
projects@yielddynamics.in Talk to IIT Bombay engineers Call +91 94230 07004Monday to Saturday, 9:00 to 19:00 IST