TYPICAL O2 SET POINT CURVE

This is a typical O2 content to boiler load set point curve.

PA FLOW: A CAUTIONARY TALE

This power plant called me to come out because they were having trouble maintaining control as the load was lowered. At higher loads the control of throttle pressure and megawatts seemed fine. But on a load decrease the boiler master and the turbine master would lose the ability to control in a way that looked suspiciously like the phenomena of hysterisis in a valve. Time was spent tuning and de-tuning the boiler and turbine controls. But no matter what we did the results were disappointing. See the figure below.

As can be seen, as the load demand and the load decreased there came a point when the throttle pressure began to rise precipitously. Various tuning was tried but all to no avail.

But then, in conversation, it came to light that this problem had shown itself in late January after testing had been performed at the plant to develop new primary airflow curves for the mills. The old and new PA flow curves are shown below:

The interesting portion of this graph lies between 45% and 60% feeder demand. The old curve, prior to testing, had begun to rise at 45% demand. This curve leveled out at about 92%. The new curve did not begin to rise until the feeder demand reached 60%. The feeder minimum speeds are about 45% at this plant. This means that the motive power provided by changing the PA flow was lacking from 45 to 60% feeder demand. Between these demands the feeder speed was changing, but the PA flow remained constant. This means that the ability to modulate the amount of coal going into the furnace was severely restricted.

We decided to modify the PA flow as shown below:


There is a fine line here between the desire to allow the PA flow to modulate between 45 and 60%, and the need to keep the air close to the amount required to maintain proper velocities in the coal pipes.

The results of this small change were immediate and very dramatic. See the figure below:



As a result of this change the unit megawatt control, and the throttle pressure control became well behaved.

HARDNESS OF COAL VS. PRIMARY AIR FLOW

There are several control loops associated with mill, or pulverizer control. These typically include the control of feeder speed, primary air (PA) flow, and mill outlet temperature. Depending on the available instrumentation and the particular type of mill other variables might be included. A non-exhaustive list of these includes mill differential, automatic wheel loading (AWL) pressure, classifier speed, and primary air fan speed. Several other factors are associated with mill control. These include the heat, or BTU, content of the coal, the chemical makeup of the coal, and the hardness or grindability of the coal.
Most recently, we focused on the grindability factor of the coal and its effect on the requirements for PA flow and AWL pressure. The type of mills at this plant are B&W MPS-75’s. The primary air flow is controlled by the PA rating damper; the temperature is controlled by the hot air and the tempering air dampers. There is automatic control of the wheel loading and the plant operates in automatic mode. There is a station for classifier speed but it is a manual loader, that is, there is no automatic control available.
The Hardgrove Grindability Index (HGI) is an index that pertains to the hardness of the coal, or the ease with which it can be pulverized. An HGI of 45 is harder coal, an HGI of 60 is a much softer coal therefore it grinds much faster. The original PA flow set point curve had been developed for an HGI of 45. But after some years had passed, the type of coal that was being burned had been changed to a softer coal. The HGI for the coal now normally runs up around 51 to 55 HGI. The rating dampers had been replaced some years before. The original parallel rating dampers were old and worn, so curves had been manipulated to improve performance at the end of service life with blades damaged or partially missing. The minimum PA flow set point had been increased by the plant in reaction to the condition of the PA dampers and the resulting leakage. The new dampers are opposing in nature and control well. But the PA flow set point curve remained elevated all across the full range of operations. As a result, the operators were forced to use extremely negative bias to get the classifier to reject any pyrites at all. These biases were on the order of -16 to -20 KSCF/HR.
Only with this kind of operator action, was the AWL pressure control able to maintain any sort of coal bed in the pulverizer. Coal was being swept out of the mill at primary air velocities that had been developed for much harder coal. The mills vibrated and rumbled a lot. The mill manufacturer, B&W, was contacted.
Calculations were performed by B&W for coal of varying hardness. The data that was generated showed a substantial difference in the PA flow and AWL pressure requirements for the mills at differing loads.
Let us consider the PA flow. Typically one set point curve is used to generate the primary air flow demand as a function of feeder speed. A second, lower curve is a function of actual coal flow, and is used to cross-limit the PA flow. There is also a minimum demand that is a constant and reflects the minimum amount of air that is necessary to transport coal.
The curve that was being used for one of the mills is shown below:


However, the data that was provided by B&W resulted in a family of curves. Now PA flow was a function not just of feeder speed but also the HGI or grindability factor. What this means is that instead of a single PA flow being associated with a given feeder demand, depending on the grindability factor, the PA flow can vary across a family of curves.
The HGI is provided on the data sheets that are associated with each coal shipment. As a result the operators should be able to estimate when any significant change in hardness is about to occur. The operator is able to enter the grindability factor for each of three coal bunkers at any time. The remote set constant for these is available on each pulverizer sequence page and the AWL control page for each mill. This is labeled “HGI”. The value is rate-limited such that a change from 51 to 52 will take one minute. This will then shift the PA flow set point curve as shown below:

Please note that this drawing is not rigorously correct, but is meant to show the general shape of the new family of set point curves. A second curve, of the same shape but a 3% lower minimum was generated as a function of actual coal flow to serve as the cross-limit for the mill.

The AWL pressure set point was changed in a similar fashion. Note that a higher grindability factor decreases the pressure set point.
This change should allow the operator to operate with little, or no, bias on the primary air flow control with the rating dampers and no bias on the AWL pressure control. There will be an increase in the amount of rejected pyrites from the mill (from none to some). It is important to realize that all of the biasing functionality that was there before this change still exists. This has been done in addition to the controls as they were.
It will take some time to see the long term effects of this change in the PA flow set point.

Hopes for this blog

I will be posting observations on the odd things that I run in to at various power plants from time to time. Hopefully anyone who visits this site will find it entertaining and/or thought-provoking. If any of you have a control-related question, I hope that you will feel free to post it. Any suggestions on subject matter will also be appreciated. If you like what you read please let others know about it. Possibly you might even be moved to by my book. Only time will tell if this will become the forum for ideas and stories concerning power plant controls that I hope it will be.
Thanx
Tim l.

AIR FLOW AND OXYGEN CONTROL

AIR FLOW AND OXYGEN CONTROL


This story concerns air flow and oxygen trim control. I also comment on the kind of things that can happen when years pass before a tuner is brought on site.

…Several years had passed since my last visit to this site. During the course of my work I discovered that the fuel master was no longer interlocked to manual when the FD fan maser was in manual. There is a hierarchy in the control loops that make up the combustion controls. It is as follows:

1. In order to place the air flow controls in automatic mode and keep them there the ID fan master (furnace pressure controls) must be in automatic mode.
2. In order to place the fuel master in automatic mode and keep them there the air flow controls must be in automatic mode.
3. In order to place the boiler maser in automatic mode and keep it there the fuel master must be in automatic mode.

This is a matter of safety. We want the furnace pressure and the air flow controls to be able to respond to fuel flow changes. If not, one can foresee a tired operator forgetting that his controls are in manual, increasing load beyond what the air flow can support, and if he panics, very bad things can happen.

The reason that this had been done became obvious rather quickly. There had been a long standing problem with the air flow control, especially at lower loads. The air flow characterization curve was a mess, the relationship between the air and fuel demands were hard for the operators to see, and at low loads the burner management system would inhibit firing or trip the unit if the air dropped too low. By the way, the BMS was acting correctly, the combustion control needed help.

Over the course of years, and lost in the shroud of antiquity by the time I returned to site, people unnamed had tried to address the situation. I am sure, that after trying their best, in exasperation, it was decided to disable the manual interlock to the fuel master that required that the air flow remain in auto.

Of course the proper thing to do at that point was to let the plant know the situation. Of course they knew that they were able to operate with the air in manual. They agreed that this was not how they really wanted to operate. But before I re-instituted the manual interlocks on the fuel master we had other fish to fry. The air flow and O2 trim problems had to be addressed, and the operators confidence in the controls improved, or else at the first instance of trouble, and with me gone from the plant, the controls would be put right back the way they had been.
The first thing to do was to address the air flow characterization curve. In the figure above the original curve and the curve as it came to be are displayed.

As you can see there is a discontinuity, or knee, in the original curve and it is very non-linear at lower loads. This is just where the problems that caused the manual interlocks to be ripped out occurred. This curve was changed by going to a few load demands, ranging from about 30% load to full load, and modifying the characterization curve, with the O2 trim in manual and at an output of 50%. This corresponds to a trim factor of 1.0, or no trim at all. As you can see, the resulting curve is much more linear, and therefore more believable. Then with a good air flow to operate on, the air flow demand was addressed. It turned out that the minimum air flow setting was at the exact value that triggered a furnace firing inhibit in the burner management system. So if the air flow dropped just a smidgeon below this value and then returned to the minimum, or slightly higher, the operator was not allowed to start any burners for five minutes. This was probably the straw that broke the camel’s back and had forced the manual interlocks to be removed from the fuel master, but we will never know for sure. Regardless, the low limit for the air demand curve was raised so that there was a cushion there for normal control deviation.
After this the O2 trim control loop was addressed. Because O2 trim control is an integral only controller, it does not have the dynamic capabilities of most controllers. As a result there are times when the controller should not be allowed the full range of control. At low loads, less than 30 to 35 %, the output from the O2 trim controller should not be allowed to go below 50%. At these low loads the air flow demand is at some minimum setting. The O2 trim controller should not be allowed to decrease air flow below this amount. As this control loop is intended to trim the air flow at steady state, it is a good idea to block any decrease in the controller output on any load change. Why is that? The question really should be; what is it that we want the boiler oxygen content to do? The answer to this is that on an increase in load we want the O2 to spike up and then return to set point. On a decrease in load we want the O2… to spike up and then return to set point. Blocking the decrease in the controller output on any change in load accomplishes this. Think of it this way, on a load decrease what will naturally occur? Due to the lag function in the cross-limited air demand the air will lag behind, that is the air will remain elevated for a period of time as the load, and the fuels, decrease. As a result the oxygen in the flue gas will spike up. If the O2 trim controller is not stopped from decreasing, the controls would see the O2 higher than set point and start cranking down. Then when the load gets where the operators have sent it, and the fuels are no longer decreasing, the air flow demand will catch up with the boiler demand and the O2 will, quickly, begin to fall. Now the O2 controller is not, nor should it be, equipped with proportional control. It will see the O2 falling and begin to crank the other way, but it cannot easily or quickly respond to the falling oxygen level, and then the O2 goes low. This tends to make operators nervous and unhappy. And if they get too unhappy, the controls are often placed in manual. So these proper limits and other normal bells and whistles were added or turned on. Finally, after the air flow and O2 trim control loops were placed in auto, and the operators saw, and more importantly believed that the controls were not going to bite them, the manual interlocks on the fuel master were reinstituted.

Coal and primary air flows in ancient pulverizers

Over the last month I have been working at a plant site in Alabama. The boiler controls were built in 1995. Some interesting quirks have come to light.
The plant engineers were somewhat apologetic about their coal flow measurement when I began working on their boilers. There are five units at this site. There are four B&W boilers that I have begun tuning. On all four of these boilers the coal flow is inferred from pulverizer differential. My reply to the engineers was that although this is not necessarily the most accurate method of developing the boiler coal flow, it was used for quite a long time in the power industry. Controls do not depend on accuracy, but rather, repeatability.
One of the first issues that were brought up by the operators was the propensity of the mills to “plug up” on load changes. They were absolutely white-knuckled whenever the mills began to unload, watching the mill current, ready to slap the controls into manual mode to prevent a high amp condition. These are B&W table top pulverizers. This seemed odd to me.
On investigating the pulverizer controls I realized that the controls were not typical. The fuel master provided the pulverizer demand. This went directly to the primary air (PA) flow control where it became the PA flow demand. The measured PA flow then became the pulverizer differential pressure set point. This means that whenever the PA flow moved, the feeder moved. This added noise to an already noisy system. Not only that, there was no cross-limiting of the air and fuel supply to the mill. Normally the fuel master demand goes to the feeder controls, and the feeder control output is then used to determine the PA flow demand through a minimum PA flow curve, that is provided by the mill manufacturer.
The reason for the mill plugging issue when the mills were unloading lay in the control configuration. As the load increased the PA flow came up ahead of the coal flow. But when decreasing, the PA flow decreased ahead of the coal flow. This drop in the transport medium (PA flow). If the demand moved fast enough, coal would build up in the mill, increasing the differential, and eventually causing high amps and a mill trip.
We modified the feeder demand to look directly at the fuel master demand. In addition we added cross-limiting on the mill fuel and air controls. This allows the air to lead the coal on an increase in load, and the coal to lead the air on a decrease in load. The plugging issue was substantially reduced and the mill controls were rendered more stable and easily controlled.