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.
Soda ash, Solvay carbonation towers
The conversion is known when the analysis comes back. By then the tower has moved on.
Everything that decides the yield happens inside a hundred feet of steel that nobody can see into.
So the tower is run conservatively, and the cost of that is never counted.
One vessel doing three jobs at once.
A carbonation tower is a gas absorber, a reactor and a crystallizer, and it is all three simultaneously, along its own height.
Brine and ammonia enter at the top and fall. Carbon dioxide enters at the base and at the side and rises. The two streams exhaust each other gradually over the full height of the tower, and solid sodium bicarbonate is collected at the bottom.
The chemistry is simple to write down. What it is not is simple to control.
2NH3 + CO2 + H2O → (NH4)2CO3 (NH4)2CO3 + H2O + CO2 → 2NH4HCO3 NaCl + NH4HCO3 → NaHCO3 + NH4Cl
Deployed on fourteen towers and validated over six months of plant operation
What cannot be measured.
The state is a distribution
Roughly twenty-three ionic species, spread across twelve vertical zones, each zone at a different composition, temperature and degree of supersaturation.
The operator sees none of it
A handful of temperatures, some flows and pressures, and a laboratory analysis that arrives periodically.
And the analysis is already history
It describes what left the tower, not what is happening inside it. By the time it comes back the tower has moved on.
A narrow band, with a real penalty on both sides.
Below the band
Conversion goes sluggish. Product is lost and throughput falls. Quiet, continuous, and it appears in no alarm.
Above the band
The exotherm takes over. The tower fouls, crystal quality degrades, and recovery is slow and expensive. Loud, and everybody remembers it.
The band is narrow, the penalty on both sides is real, and the position inside it is invisible.
So the tower runs somewhere safely inside, permanently.
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.
What was built
A model of the inside of the tower, solved continuously.
A hybrid model of the tower: thermodynamics, reaction and crystallization kinetics, heat and mass transfer and multiphase flow, written in C++ with machine learning modules for the parameters the physics cannot supply on its own.
It read all fourteen towers through the existing control system, took laboratory analyses as manual entries, and simulated the internal state of each tower in real time. Over three hundred simultaneous equations, solved continuously, to reconstruct a composition and temperature profile that no instrument could report.
It then issued setpoints back to the existing control system. Nothing in the plant was replaced. The towers, the chemistry and the control system stayed exactly as they were, which is the same arrangement we would bring to a column today.
What was verified
Six months on trial, across the fleet.
The finding
The towers placed on the system held their parameters in a tighter band and showed an increase in conversion efficiency. That is the finding as it was recorded at the time, in a signed letter from the General Manager who coordinated the project. He continues to take reference calls.
And the number we do not publish
We do not publish a percentage, because none was established to a standard we would be willing to defend. The certificate says better control of parameters and an increase in conversion efficiency, and it says nothing more. Neither do we.
The signed certificate, and how to reach the man who signed it
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.
Ask for the reference call
Speak to the General Manager who ran the trial
He has agreed to take calls from anyone seriously considering this work. Ask and we will put you in touch. We do not need to be on the call.
projects@yielddynamics.in Talk to IIT Bombay engineers Call +91 94230 07004Monday to Saturday, 9:00 to 19:00 IST
Or start with your own tower
Send operating data from one tower and the laboratory results beside it
We will tell you how tight the band you are running is, and what the margin is costing you, with the uncertainty stated openly. No charge, and no commitment either way.