Emerson’s Sergei Kuznetsov will be presenting at the rapidly approaching Emerson Exchange Technical Conference. In his presentation, Three Non-conventional Level Control Applications, Sergei shares how to approach three quirky level control challenges:
- Particle bin level control with large deadtime presence
- Copper mine concentrator sump system
- Oil platform water injection surge vessel level control
The Oriented Strand Board (OSB) manufacturing process makes wood-type boards from 2-4″ wood flakes combined with epoxy resin. The basic process begins with a dry wood bin filled with the flakes. These are fed to a blender where they are mixed with the resin. Next, the coated wood chips are fed to a forming head where they are formed into mats. In turn, these mats are fed by conveyor belt to the press where the particleboard is created. Deadtime, on the order of 2-4 minutes, is the huge control challenge. The flow rate of the wood chips into the blend needs to be as stable as possible and the level of coated wood in the forming heads must be uniform or the quality of the OSB wood is greatly impacted.
Sergei applied PID control with a Smith Predictor to effectively remove the loop deadtime and allow the cascaded PID controllers with the flow loop as the slave loop and the wood level as the master loop to be aggressively tuned. The standard Smith Predictor algorithm had to be slightly modified to handle the integrating nature of the level loop. Sergei cites a book by Karl J. Aström and Tore Hägglund, Advanced PID Control, chapter 8.5 that describes the modified Smith Predictor for use in integrating processes with long deadtimes.
By applying this algorithm and tuning it the dynamics of the OSB manufacturing process, the plant was able to reduce the panel board weight variation by 29%. They were also able to reduce level-related shutdown rates from 15 per shift to less than 1 on average, and reduce raw material consumption by 7%, which translates into more than a million U.S. dollars per line per year.
Sergei’s second level control challenge was a copper mine water management system. The goal was to recover as much water coming from the thickening ponds as possible and pump it back into the “Head” tank where it could be reused by the concentrator plant. Water from the thickeners comes to two sump tanks. Each tank is equipped with a set of single-speed centrifugal pumps. The sump tanks overflow lines are routed downstream to the tailings ponds. Another challenge was that the head tank overflow piping was not operational. This caused additional constraints on this control problem. The control strategy needed to prevent head tank overflow, maximize its level to provide enough water to supply the concentrator during demand spikes, remain below the high-level safety interlock, and keep the sump tanks at minimum safe levels.
Sergei designed a simple, elegant solution by gradually limiting the number of pumps running by each of the sump tank level control algorithms when the head tank level approaches the high-level interlock setpoint. The first pumps to be shut down are the ones associated with the sump tank with the lowest level to help equalize levels over time. Also, when the head tank is recovering from a high-level situation, the pumps of the sump tank with the highest level are started first, again to help equalize the levels. The control strategy was intuitive for the operations and maintenance teams and provided robust control for the mine for the time needed to address their process equipment problems.
The third example was on an offshore platform enhanced oil recovery project using water flood techniques. A similar, simple solution was applied to a water surge tank used to inject oily water back into the wells. Tight platform real estate dictated a small volume surge tank, but the required flow rates were quite high. Sergei employed a simple calculation block to the set of conventional controls. The control strategy was able to reconcile the tank level control objectives with water injection goals set by the production team. Additionally, he employed energy-saving algorithms to the pump variable frequency drives (VFDs) that were used to manage the water injection process.
If you have some tough level-control applications, you’ll want to attend one of Sergei’s two workshops and hear about his approach in solving these. You may also want to bring your toughest level control challenges to ask him about.