A Distributed Control System Software
and Control Strategy for Blast Furnace Stove Combustion
G.F. Lauer, A.J. Barchfield Jr., M.J. O'Leary and R.A. Szantay
USS/Kobe Steel Company
R.P. Murray
Applied Control Systems, Inc.
USS/Kobe Steel Company recently modernized its No. 3 blast furnace
to make it a world class competitor. The furnace, which was bought
on stream again on May 4, 1992, now is operated using a total
furnace control system that permits one operator to monitor the
entire furnace complex from a central control room.
This high degree of automation is made possible by a fully integrated
and automated control system architecture, a distributed control
system (DCS) and a programmable logic controller (PLC). The DCS
controls the traditional analog functions. It also acts as the
main window to the blast furnace process for the stove, furnace,
gas cleaning and water cooling areas. It does this by using the
DCS's color graphics features. The PLC, meanwhile, handles the
traditional digital or sequential functions.
This control strategy is not an entirely now development. There
were past contributions to its success. During the late 1970's
and early 1980's, engineers from what was then U.S. Steel Corporation
developed a blast furnace stove combustion control strategy at
its research laboratories and Gary Works. This strategy was grounded
in sophisticated calculations that established operating setpoints
for the stove burner flame temperature and the stove heat demand.
The calculations were based on the hot blast requirement needed
to support the iron production schedules.
The major objectives of the flame temperature/stove heat demand
control strategy were to:
-Maximize fuel efficiency.
-Reduce combustion related refractory problems.
-Establish combustion control operating setpoints based on the
hot blast requirements that were set to achieve iron production
goals.
-Minimize operator interventions.
More recently, Applied Control Systems, Inc. transformed the flame
temperature and zone heat demand calculations into customized
control algorithms. The algorithms were implemented using Leeds
& Northrup (L&N) Excel Programming language. Excel is
an acronym for "extended controller engineering language."
The customized control strategy is now configured in each of the
L&N MAX1 DCS controller files at USS/Kobe, residing in the
programmed functions section of the file. Each stove has a primary
controller file supported by a hot redundant backup file.
The stove on-gas/on-blast cycle times have been established in
the DCS and are included in the heat demand calculations. The
DCS also monitors stove operating mode cycle times. Automatic
stove changing is initiated within the L&N system and is executed
through the sequencing logic of a PLC designed by Allen-Bradley
Co., Inc. hence, the stoves are maintained in a fully automatic
mode without the need for operator intervention. More operators
are only needed only if full automatic control is not available
because of a mechanical or electrical problem within the stove
facility. Included in the overall control system is the capability
to provide tagged variables and status conditions over a communications
network to a Level 2 system for statistical report generation.
The No. 3 furnace's entire control system has been on-line since
May 1992 and has achieved the stated objectives.
PROCESS CONTROL ARCHITECTURE
The No. 3 blast furnace control system architecture is comprised
of a fully integrated and automated PLC based controlled system
that performs processes as well as batch and sequential functions.
And, as already stated, it has a DCS performing process and sequential
control.
At the top of the Level 1 control architecture is a pair of Allen-Bradley
pyramid integrators and a MicroVAX 3000, functioning primarily
as a data concentrator and a local report generator. A pair of
L&N MAXport computers mainly serve as a gateway to provide
redundant data communications from both the PLC and the DCS systems
to a Level 2 terminal server. Figure 1 shows the process control
architecture.
Note that the stoves and some of the common control loops on the
No. 4 furnace are incorporated into the No. 3 furnace system architecture
and are available to the operator stations via fiber optic highway
No. 1. Previously, a limited DCS/PLC system was provided on the
No. 4 furnace, which came on-line in November 1989.
The No. 3 furnace stove combustion control system is contained
in controller files 36, 38 and 40 (Figure 1). Each primary controller
is supported by an on-line hot redundant backup file (i.e. 37,
39 and 41). Should the primary file fail, the transfer to the
backup file is instantaneous and smooth. No hard-wired manual
stations are provided.
STOVES
The three blast furnace stoves on the No. 3 blast furnace were
completely replaced with new Martin & Pagenstecher (M&P)
engineered stoves. The new stoves have a height of 124 feet 8
inches and a diameter of 25 feet and 6 inches. The checker heating
surface area encompasses and area of 390,773 ft2 per stove. The
stoves have internal, vertical ceramic burners (Figure 2), and
are fired with combustion fans capable of producing 40,000SCFM
at a discharge pressure of 45 inches W.C. The fans are driven
by 450 horsepower motors.
Detailed operating specifications were included with stove specifications
provided by M&P. Table 1 shows an excerpt of some of the M&P
specifications and the operating status of the No. 3 stoves as
set by USS/Kobe.
During the early construction stages of the Furnace No. 3 rebuild,
the decision was made to use the existing top gas analyzer and
gas sampling system that was in service on the No. 4 blast furnace.
This decision was made because the stoves on both furnaces receive
their top gas from a common blast furnace gas main. The stoves
on the No. 4 furnace had been controlled from an L&N DCS since
November 1989. These stoves use the same stove heat demand/flame
temperature combustion control program now in use on the No. 3
furnace stoves.
The top gas sampling system has proven to be very reliable. The
sampling system was specified by Perkin Elmer and was furnished
by JNM Corporation. It provides a clean sample to the analyzer
and includes an automatic Blow Back regulated by a built-in cycle
timer.
The top gas analyzer is a Perkin Elmer MGA1200 magnetic sector
mass spectrometer. It is equipped with a special feature that
provides individual 4-20 mA signals to the DCS, rather than the
typical RS232 or 422 communications links. This feature permits
continuous inputs for each top gas constituent (CO, CO2, and H2)
needed for flame temperature calculations.
The blast furnace gas is sampled in the gas main after the gas
from the No. 3 furnace is cleaned by a Davy cone scrubber system
and the gas from the No. 4 furnace passes through the Venturi
scrubber. The sampling point is common to both stove control systems
(i.e., the top gas sample is a combination of the gas contributed
from each furnace).
STOVE COMBUSTION CONTROL
Each stove on the No. 3 blast furnace, as well as on the No. 4
furnace, has its own separate combustion control program that
is initiated when the initial stove is selected for its on-gas
period.
Figure 3 shows the combustion control flow diagram and the sequence
of events contained in the calculations for setting the stove
heat demand in Btu/minute and the flame temperature in degrees
Fahrenheit.
The calculations and algorithms presently in use at USS/Kobe have
been modified to some extent from the original version prepared
for Gary Works' No. 7 furnace. The control flow has six steps:
1. Calculate the stove heat demand.
2. Correct the dry top gas analysis from the gas analyzer for
the wet gas analysis to provide the true Btu/ft3 value of the
blast furnace top gas.
3. Calculate the stove burner operating flame temperature setpoint.
4. Calculate the required enrichment Btu/ft3 provided from natural
gas.
5. Calculate the air flow setpoint for stoichiometric combustion.
6. Calculate the actual burner flame temperature.
Calculate the stove heat demand -
The stove heat demand is calculated in Btu/minute as a function
of wind rate, hot blast moisture and temperature content, stove
cycle times and stove efficiencies. Stove heat demand limits then
are established based on stove utilities and capacities. If the
limits are reached, alarm is posted and the system assumes a safe
stove heat demand setpoint. The result of this calculation is
established as the numerator in a later calculation.
Correct the dry top gas analysis from the gas analyzer for
the wet gas analysis to provide the true Btu/ft3 value of the
blast furnace top gas -
The individual top gas constituents, (i.e., CO, CO2 and H2) are
subjected to a five minute rolling average by the MAX 1 DCS processor.
The average analysis is error checked and used. In the event that
the average analysis is not acceptable, or if the top gas analyzer
is out of service, the system reverts to preselected stored values
and an appropriate alarm is sounded. The Btu/ft3 of the wet top
gas is calculated.
Calculate the stove burner operating flame temperature setpoint
-
The stove burner operating flame temperature setpoint is calculated
based on the required heating value of the mixed gas and is stated
in Btu/ft3. This result is provided to the mixed gas flow setpoint
as the denominator. Btu/minute (heat demand) divided by Btu/ft3
(flame temperature) provides ft3/minute, which serves as the mixed
gas flow setpoint.
Calculate the required enrichment Btu/ft3 provided from natural
gas -
The required enrichment Btu/ft3 provided from natural gas is calculated
based on the additions necessary to the Btu/ft3 of the wet top
gas. Then, the ration setpoint for enrichment is developed based
on the Btu/ft3 of the natural gas. Once the enrichment ration
is calculated, the natural gas flow controller setpoint is determined
based on the mixed gas flow control setpoint.
Calculate the air flow setpoint for stoichiometric combustion
-
The air flow setpoint for stoichiometric combustion is calculated
based on the mixed gas flow setpoint and mixed gas Btu/ft3. A
factor for excess air then is added and the air flow control setpoint
is calculated in SCFM.
Oxygen analyzers are provided on each of the two stacks on every
stove. The analysis either is automatically or manually selected
by the MAX 1 DCS. A flue stack oxygen setpoint is entered to an
oxygen controller. The controller is configure d to provide its
weighted output to establish a positive or negative oxygen bias
to the air flow controller setpoint in and increasing or decreasing
direction. The bias is reset to zero when the stove is not in
the on-gas period. The final operating air flow setpoint is the
calculated value, including the excess air factor, plus or minus
the bias factor from the selected stack oxygen analysis.
As a further check on combustion control accuracy, each stove
stack on the No. 3 furnace is equipped with a combustibles analyzer.
Historical trend screens comparing the oxygen and combustibles
analysis are available on the MAX 1 Multi-Operator Stations.
Calculate the actual burner flame temperature -
When the stove is in the on-gas period, the actual burner flame
temperature is calculated based on the wet blast furnace gas and
the natural gas flow. Should the actual burner flame temperature
exceed the aim flame temperature setpoint, the limit setpoint
is assumed and an alarm is posted. Before passing the calculated
flow setpoints on the controllers, the calculated heat demand
is compared with a limit setpoint. If the calculated setpoint
reaches or exceeds the limit, the limited setpoint is assumed
and an alarm is sounded.
On-Gas Period
The stove on-gas period is established in three phases that are
determined by the actual stove dome temperature control setpoint.
The three phases are defined as:
1. Stove is on gas and the stove dome temperature is greater that
15 F below the setpoint.
2. Stove is on gas and the stove dome temperature is within 15
F of the setpoint, and enrichment natural gas is available
3. Stove is on gas and the stove dome temperature is within 15
F of the setpoint, and enrichment natural gas is equal to zero.
Figure 3 outlines the three phases
A ramp program is configured in the microprocessor and applied
to the stove dome temperature setpoint. It provides an accurate
approach to the setpoint without the setpoint overshoot. When
the stove dome temperature reaches a preset temperature below
the operating setpoint, the approach ramp program is initiated.
The operating setpoint assumes the preset value and is increased
to the operating setpoint at a specific ramp rate in degrees Fahrenheit
per minute.
During the phase II of the on-gas period, the dome temperature
control provides bias to the operating Btu/ft3 setpoint calculation.
This ultimately results in a recalculation of the mixed gas flow
setpoint. Air flow still is calculated as before, and stack oxygen
analysis and bias control of the air flow setpoint remain active
during phase II.
Phase III of the on-gas period seldom is reached when natural
gas enrichment is required. Phase III begins when the stove dome
temperature controller has removed all of the natural gas enrichment
or if enrichment is not required. During phase III,, the oxygen
bias controller is inactive.
The system is configured to permit reentry of any phase as a function
of stove temperature deviation temperature from the setpoint.
Special attention is given to the M&P stove design with consideration
for the maximum allowable temperature of the ring wall that supports
the checker refractory. A maximum limit temperature is configured
and an alarm is posted if this limit is reached. A control ramp
program is initiated and the mixed gas flow setpoint is reduced
at a present rate.
Table II represents one of many reports generated for the blast
furnace complex. Data is provided to the VAX system for these
reports from the MicroVAX and L&N MAXports.
Summary
A blast furnace stove combustion control strategy that utilizes
the sophisticated capabilities of a state of the art microprocessor
based DCS has been installed on the No. 3 blast furnace of USS/Kobe
Steel Company. Since the May 4, 1992 blow in of the furnace, the
reliability of the entire DCS has been proven. The stove operating
performance has been met and is within the M&P specified guidelines.
The objective of a one person control room operation has been
achieved.
Acknowledgments
We gratefully acknowledge the past and present contributions made
to this control strategy. We also acknowledge the equipment and
services supplied by the following calculations:
Allen-Bradley Corporation, Cleveland, OH
Applied Control Systems, Inc., Pittsburgh, PA
JMN Corporation, Kent, OH
L&N, North Wales, PA
Perkin Elmer Corporation, Applied Science Division, Pomona, CA
USS/Kobe, Lorain, OH
U. S. Steel Corporation
USS Gary, IN, Works
USS T.I.P Group
U.S. Steel Research Center, Monroeville, PA.
