1 September 2006
pH control nerves of steel
Software overcomes common process industry temperature and acid problem areas
By Jeff Braun and Matt Ashby
A steel caster is the most critical operating unit in a modern steel plant.
The caster transforms the liquid steel into solid slabs, ready for the rolling mill to produce the final product, steel sheets. The liquid steel cools to form a molded shell, with the shape, thickness, and width established by the mold. The mold consists of water-cooled copper plates attached to steel water boxes, forming a rectangle.
Several thousand gallons of caster cooling water pump in per minute at high pressure through the mold to cool the steel. Since the water temperature affects the condition of the steel slabs, it is one of the most critical quality related process variables, and it needs to be tightly controlled.
Let’s look at what it takes to control effectively the caster cooling water temperature and pickle-line rinse water pH value. This feature will also discuss model-free adaptive (MFA) control technology, which is helping Nucor Steel to improve product quality and plant efficiency.
The mold water process
Nucor Steel’s Decatur, Ala., plant has a caster cooling water system that supplies non-contact cooling water to two 90mm medium thickness continuous slab casters.
During normal operations, two of the three pumps run and supply water flow at 3000-4000 gallons per minute, depending on casting machine status. Since the mold water leaving the casters so hot, it passes through three heat exchangers for cooling.
Controlling of the mold water temperature takes place by manipulating the cooling water flow to the heat-changers. Previously, three PID controllers regulated each of the three cooling water valves.
The goal is to control the temperature of the supply mold water to the casters at 95 or 96°F with no deviation of more than +/- 2°F during any system transients.
During the steady state, the PID based control system could maintain the mold water temperature well. However, during a caster start-up or tail-out, there could be up to eight degrees Fahrenheit deviation, which could cause product quality problems.
In addition, the system is sensitive to the ambient temperature change from winter to summer. PID controllers need retuning seasonally to make up for the fluctuations in the cooling tower operating conditions.
Challenge and analysis
During system-transients when a caster is starting up or shutting down, the temperature disturbance to the mold water leaving the casters can be quite large. The PID controllers controlling the temperature for the mold water leaving the heat exchangers cannot react fast enough to compensate for the disturbance.
The temperature that needs to be controlled is not necessarily the temperature of the water leaving the heat exchanger, but the temperature of the supply mold water leaving the expansion tank. Therefore, it is better to add a controller to directly control the supply-mold water temperature.
The challenge is the supply-mold water temperature loop will have a large time delay of about five minutes. It is difficult to use a PID to control this loop.
The ambient temperature variations due to seasonal changes cause the cooling water temperature and other operating conditions to change. The elimination of the manual tuning of PID controllers would be best and is desirable.
We learned about MFA control at ISA EXPO. MFA control is attractive to us because:
MFA does not require the user to build a mathematical model for the process.
It can adapt to fit new operating conditions.
MFA can control complex systems.
It is easy to use and maintain.
Unlike PID, which is just one controller, there are actually many different types of MFA controllers available, and each one solves a specific control problem.
Some MFA controllers can solve the type of control problems commonly seen in a large process plant.
This means, we can simply select the appropriate controller, do some straightforward parameter configurations having to do with entering the sample interval, process-acting type, and estimated process time constant, and then we are ready to launch the MFA controller.
Single signal control water
In February, we launched a MFA control system with two controllers. We selected a single in single out (SISO) controller to control the caster mold water temperature by manipulating all three cooling water valves at the same time.
We also selected an anti-delay controller to handle the actual supply-mold water temperature, which has a large time delay. A feedforward controller as part of the SISO controller can produce quick control actions to compensate for the large disturbances occurring during process transient conditions.
The system works as a cascade control system with C1 and C2 as primary and secondary controllers, respectively. (See “Cascade control system for mold water temperature” figure.) C3 is the Feedforward controller. Processes 2 to 4 represent the sub-processes of the cascade system. The inner loop consists of C2 and P2, and the outer loop consists of C1 and P1, where P1 consists of C2, C3, P2, P3, and P4. Notice P1 as shown by the dotted line represents the process for C1 to control, where the process variable is the supply-mold water temperature. Since controller C2 is an MFA controller, the closed-loop dynamics of the inner loop will not change much, even though the process dynamics of P2 may change a lot. This means the interconnection of the outer loop and the inner loop becomes much weaker. A more stable inner loop contributes to a more stable outer loop, and vice versa.
The control system now can control the water temperature with +/- 1°F during normal operations as well as caster start-up and tail-out conditions. Product quality improved with a more consistent heat transfer rate.
pH control for pickle line
After our quick success in eliminating the long-lasting temperature control headache, we became a little greedy.
Using the same control software, we selected and configured two more controllers. This time, we chose pH controllers because we needed to control our pickle line rinse water treatment process that has two mixing tanks and needs two pH controllers.
Acidic rinse water cleans the surface of the steel in a pickle line. The used rinse water is dirty and acidic and needs treatment to reach a neutral pH.
Rinse water leaves the batch tank and enters Tank 1 to mix with 50% concentrated caustic water for a coarse treatment and then enters Tank 2 to mix with 10% concentrated caustic water for a fine treatment.
The goal is to control the Tank 2 water pH value at seven or a little lower. Depending on the operating conditions of the pickle lines, the rinse water entering Tank 1 can have very large variations in flow and pH value. Therefore, it is very difficult to control. Over dosing of caustic water is common and results in lower efficiency and caustic water waste.
A strong-acid, strong-base pH process is highly nonlinear. The pH value versus the reagent flow has a logarithmic relationship. Away from neutrality, the process gain is relatively small. Near neutrality where pH is 7, the process gain can be a few thousand times higher.
The large pH variation in effluent and varying time delays make a tough control problem much worse. Therefore, it is almost impossible for a fixed controller like PID to effectively control this type of pH process.
The configuration for a MFA pH controller is straightforward. We simply entered the rough estimated information about the pH-loop-titration curve.
MFA controller architecture
The core architecture of a SISO MFA controller uses a multilayer perceptron neural network that has one input layer, one hidden layer with N neurons, and one output layer with one neuron.
Within the neural network there is a group of weighting factors (wij and hi) that can be updated as needed to vary the behavior of the controller.
The algorithm for updating the weighting factors minimizes the error—e(t). Since this effort is the same as the control objective, the adaptation of the weighting factors can assist the controller in minimizing the error while process dynamics are changing.
In addition, the neural network based MFA controller remembers a portion of the process data providing valuable information for the process dynamics.
In comparison, a digital version of the PID remembers only the current and previous two samples, less intelligence than the MFA.
Fifteen minutes after we launched this control system, we saw a near flat line value for the pH on the trend chart for the caustic loop. Had we not known we had just changed controllers, we would have suspected the pH probes were broken.
Everybody in the control room was impressed and very happy because we used to have to constantly watch the pH system and manually adjust the pH control valves.
In the final analysis, the benefits of this installation to Nucor Steel are many. They are:
With tighter and more consistent control of water temperature, our steel product quality and plant efficiency has improved. In addition, the valve life will be longer due to the reduction of unnecessary cyclical movement when in PID control.
With effective automatic control of the pickle-line rinse water pH, there is a significant reduction in chemical reagent usage and our labor costs are less.
We estimate the return-on-investment for implementing the MFA control system to be one month.
Subsequent to these two quick successes, we upgraded the control software to include more controllers for other problematic loops in the plant.
As engineers working in a large process plant, we believe many of our colleagues in the iron and steel industry and other process industries face similar problems and can realize similar solutions.
ABOUT THE AUTHORS
Jeff Braun (JBraun@nsdecatur.com) has two engineering degrees and is a re-gistered professional engineer. Matt Ashby (MAshby@nsdecatur.com) is an auto-mation/control electrician. He has nuclear power and submarine maintenance experience. They work at Nucor Steel’s Decatur, Ala., facility.
pH is the degree of acidity or alkalinity measured on a scale from 0 to 14 with 7 the neutral point. Measurement of pH is important to quality control in any number of industries. From 0 to 7 is acidic, and from 7 to 14 is alkaline. Battery acid is about pH 1. Lye is a 13 ph, and ammonia is around 12.
Heat exchanger is a device for transferring heat from one fluid to another, where the fluids are separated by a solid wall so they never mix. They are widely used in refrigeration, air conditioning, space heating, power production, steel production, and chemical processing. One common example of a heat exchanger is the radiator in a car, in which the hot radiator fluid cools from the flow of air over the radiator surface.
PID control is a common form of single-loop controller, which can handle many situations encountered in control. PID refers to a three-term control mechanism combining proportional, integral, and derivative control actions.
SISO: We characterize a system by how it responds to input signals. In general, a system has one or more input signals and one or more output signals. Therefore, one natural characterization of systems is by how many inputs and outputs they have: SISO (single input, single output), SIMO (single input, multiple outputs), MISO (multiple inputs, single output), MIMO (multiple inputs, multiple outputs).
Closed loop control is control achieved by feedback, i.e. by measuring the degree to which actual system response conforms to desired system response and utilizing the difference to drive the system into conformance.
Pickle line: A pickle line removes surface impurities from hot rolled coiled steel via a wash and rinse process utilizing an acid solution. A pickle line is not necessarily one individual piece of equipment, but it can consist of equipment occupying 100,000 square feet of floor space, can be three stories high, and may operate 24 hours a day processing steel coils weighing up to 40 tons.
Chinese steel plant clears air with software: Model-free adaptive control puts Ling-Yuan Iron and Steel in a league of its own. www.isa.org/link/Steelclearair
Process pH measurement continues evolution: The acid-base relationship was important then … www.isa.org/link/ProcesspHevolution
Principles and guidelines of pH measurement www.isa.org/link/guidelines_pH