Note: Descriptions are shown in the official language in which they were submitted.
12~744
--1--
BACKGROUND OF THE INVENTION
The present invention is directed generally to a
method and apparatus for controlling a setpoint for a
process control variable, and more particularly, to a
5 method and apparatus utilizing a dynamic reconciliation
technique.
Numerous types of controllers and control systems
are known which take advantage of process models for
controlling a process parameter. ~n one type a simple
10 dynamic relationship is assumed between the manipulated
procass variable and the controlled process parameter. The
known value of the manipulated variable is then used with
the model to estimate the controlled process parameter. The
difference or bias between the predicted value of the
15 controlled parameter and the measured value of that
parameter is used to ad~ust the manipulated process
variable to move the controlled process parameter to the
desired value. This technique is called ~'Internal Model
Control" and was described by C.E. Garcia and M. Morari in
20 IEC. Proc. Des. and Dev., Volume 21 in 1982.
Another technique postulates a simple dynamic
relationship between a process parameter which can be
easily measured, such as a temperature, and the process
parameter to be controlled, such as a concentration, which
25 is more difficult to measure. The model and the easily
measured process parameter are then used to estimate the
controlled process parameter. Once again the value of the
controlled process parameter prédicted by the model is
compared with the measured value of the controlled process
30 parameter and a difference or bias is calculated. This bias
is then used to adjust the manipulated process variable to
744
move the controlled process parameter to its desired value.
This technique is called Dynamic Reconciliation and was
described in an article by Robert V. Bartman entitled "Dual
Composition Control in a C3/C4 Splitter" appearing in the
S September 1981 issue of CEP.
An alternative approach is to use a material or
energy balance model to estimate the controlled process
parameter. As in the above techniques the difference
between the model estimate of the controlled process
10 parameter and the measured value of the controlled process
parameter is used to adjust the manipulated variable to
move the controlled process parameter to its desired value.
This is the approach described herein.
SUMMARY OF THE PRESENT INVENTION
The pre5ent inventlon is directed to a method for
controlling a system parameter based on controlling a
setpoint for a process control variable. The method
includes the steps of measuring the rate of input to and
output from a vessel. The difference is then determined
20 between the input and output rates to provide an expected
rate of accumulation in the vessel. The actual rate of
accumulation in the vessel is measured. The difference
between the expected rate of accumulation and the actual
rate of accumulation is determined to provide a bias term.
25 A value for the setpoint for the process control variable
is calculated by solving one of a material or energy
balance equation which includes the bias term and a term
representative of a predetermined time period during which
the controlled parameter is to reach a desired value. The
30 setpoint for the process control variable is then
controlled based on the calculated setpoint value.
~77~
--3--
In one embodiment of the present invention it is
anticipated that the step of measuring the rate of input to
and output from a vessel includes the step of measuring the
rate of flow of material input to and output from the
5 vessel. In such an environment, the difference between the
input and output flow rates is determined to provide an
expected rate of accumulation of material in the vessel.
The actual rate of accumulation in the vessel is measured.
The difference between the expected rate of accumulatlon
10 and the actual rate of accumulation is determined to
provide a bias term. A material balance equation of the
general type wherein the actual rate of accumulation of
material in the vessel equals the expected rate of
accumulation of material in the vessel plus the bias term
15 is solved to provide a value for the setpoint for the
process control variable.
In another embodiment of the present invention
the step of measuring the rate of input to and output from
a vessel includes the step of measuring the rate of energy
20 added to and energy removed from the vessel. In such an
environment, the difference between the energy input and
energy output rate is determined to provide an expected
rate of accumulation of energy in the vessel. The actual
rate of accumulation of energy in the vessel is measured.
25 The difference between the expected rate of accumulation
and the actual rate of accumulation is determined to
provide a bias term. An energy balance equation of the
general type wherein the actual rate of accumulation equals
the expected rate of accumulation plus the bias term is
3Q solved to provide a setpoint for the process control
variable. The process control setpoint is then controlled
based on the calculated setpoint value.
12~7744
--4--
The present invention is also directed to ~r
apparatus for controlling a system parameter based on
controlling a setpoint for a process control variable. The
apparatus is comprised of means for measuring the rate o~
i~put to and output from the vessel, means for determining
the difference between the input and output rates to
provide an expected rate of accumulation in the vessel,
sensing means for sensing the actual rate of accumulation
in the vessel and means for determining the difference
10 between the expected rate of accumulation and the actual
rate of accumulation to provide a bias term. Means f~r
calculating are provided for calculating a value for the
setpoint for the process control variable by solving one of
a material or energy balance equation which includes the
15 bias term and a term representative of a predetermined time
period during which the controlled parameter is to reach a
desired value. Output means control the setpoint for the
process control variable based on the calculated setpolnt
value.
t~ e~ ~o~ ~
The p~eser~ _invcn~n compensates for sensor
errors by including a bias term in the equation which is
representative of errors in the system. In an ideal system,
the bias term would have a value of zero. However, under
real world con~itions, the addition of this bias term to
25 the equation provides a very accurate indication of the
control action necessary to achieve the desired result.
Additionally, by including a term in the equation
representative of a predetermined time period during which
the controlled parameter is to reach a desired value, the
30 system can be driven as quickly or as slowly as desired.
These represent substantial advantages over the prior art.
These and other advantages and benefits of the present
invention will become apparent from the description of 2
preferred embodiment hereinbelow.
12~774A
_ESCRIPTION OF THE DRAWI~GS
In order that the present invention may be
clearly understood and readily practiced, preferred
embodiments will now be described, by way of example only,
5 with reference to the accompanying figures wherein:
Fig. 1 illustrates a vessel together with varlous
sensors and control equipment controlled by a control
system constructed according to the teachings of the
present invention;
Fig. 2 is a flow chart illustrating the steps
carried out by the control unit shown in Fig. l; and
Fig. 3 illustrates a reactor vessel together with
various sensors and control equipment controlled by a
control system const~ucted according to the teachings of
15 the present invention.
DESCRIPTION OF A PREFERRED EMBODIMENT
The present invention will now be described in
con~unction with a particular emobodiment. However, it
should be recognized that the particular environment in
20 which the present invention is described, is intended for
purposes of illustration only, and not limitation. There
are numerous environments in which the Dynamic
R ~ p ~
Reconciliation Process Control Method and Aff~r$us of the
present invention can be used.
In Fig. 1, a vessel 10 contains a material 12
undergoing a process. The material 12 is inpu-t to the
vessel 10 through an input pipeline 14 having an input
valve 16. The position of the input valve 16 determines the
flow rate of material 12 into the vessel 10. The flow rate
30 of material 12 through pipeline 14 is measured by an
orifice plate 18 and a differential pressure gauge 20. The
differential pressure gauge 20 produces an output signal
representative of the flow rate of material 12 into the
1.~7 7744
~,
vessel lO which is input to a control unit 22 through
line 21.
~ aterial 12 can be removed fro~ vessel 10 through
an output pipeline 24 having an output valve 26. The
5 position of the output valve 26 determines the flow rate of
material 12 through the output pipeline 24. ~he flow rate
of material 12 through output pipeline 24 is m~asured by an
orifice plate 28 and a differential pressure ~auge 30.
Pressure gauge 30 produces an output signal representati~
10 of the flow rate of material 12 through output pipeline 24
which is input to a control unit 22 through a line 31.
The level of material 12 in the vessel lO is
measured by a differential pressuPe gauge 33. A signal
representative of the level of material 12 in the vessel 10
15 is input to the control unit 22 from the pressure gauge 33
through a line 35.
Those of ordinary skill in the art should
recognize that other devices for providing flow or level
indications may be used. The illustration of orifice plates
20 and differential pressure gauges is for purposes of
illustration only, and not intended as a limitation.
The control unit 22 performs a sequence of opera-
tions, described more fully hereinbelow in conjunction with
Fig. 2, to produce output signals for controlling the
25 process carried out in the vessel lO. An output signal may
be input to a controller 37 through a line 38 for
controlling the position of the input valve 16 or an outp~t
signal may be input to a controller 40 through a line 41
for controlling the position of the output valve 26, or
30both. In this manner, control unit 22 can precisely control
the flow of material 12 into and out of the vessel 10 as
well as the level of material 12 in the vessel 10.
~27 774A
--7--
The control unit 22 working with the raw data
received from pressure gauges 20, 30, and 33 solves, in the
embodiment shown in Fig. 1, a material balance equatiOn.
Th,e equation solved by the control unit 22 is derived from
5 the following basic material balance equation:
in Fout + Fgen . . . . . . . (1)
Where dM/dt is the actual measured rate of accumulation of
material 12 in the vessel, Fin and F t are the measured
flow rates of material 12 into and out of the vessel 10,
10 and Fgen represents any generation of material 12 within
the vessel 10. Equation ~1) can be expressed as a
difference equation:
M(t) - M(to) = Fin (t) ~ Fout (t) + Fgen (2)
t - to
I/l5/~ Equation (2) can b~ made into a basic material
r ~ f>V~
15 balance ~y~h~ Peeo~c~ n equation by adding a bias
term B(t) to compensaté for sensor as well as other errors
in the system.
M(t) - M(to) = F. (t) - F (t) + F (t) + B(t) . . (3)
1n out gen
t - to
Rearranging equation (3):
20 B(t) = M(t) - M(to) *
t - to (F (t) - F (t) + F (t)). . . . (4)
in out gen
where the superscript * is used to indicate that the input
and output flowrates have been adjusted (if necessary) to
the same dynamic frame as the controlled variable.
At this point, it may be advisable to pass the
25 bias term, B, through a lead/lag algorithm. Experience
--8--
indicates that a lead is seldom required but that filtering
is often advisable. Thus a filtered bias term can be
produced using
Bf(t) = Bf(to) ~ a (B(t) - Bf(to)] . . . . . . . . . (5)
Substituting the filtered bias term into equation
(3) and solving for the setpoint of the manipulated
variable, in this case FinsP (t):
sp sp pv
F (t) = M(t) - M(t) + F (t) ~ F (t) - Bf(t) . . . (6)
in Tau out gen
where the abbreviations sp and pv stand for setpoint and
present value, respectively, and the new parameter, Tau,
represents the time allotted for the controlled parameter
to reach the setpoint, i.e. desired value. Depending upon
15 the process involved, the generation rate Fgen(t) or the
output flow rate setpoint FoUtsP (t) may be the manipulated
variable.
Those of ordinary skill in the art will recognize
that the solution of equation (6) for the setpoint value
20 for the manipulated control variable can be simply and
quickly calculated by a suitably programmed microprocessor
or microcomputer. It is anticipated that control unit 22
could include a commercially available microprocessor or
microcomputer suitably programmed to solve equation (6).
A flow chart illustrating the steps carried out
by the control unit 22 is illustrated in Fig. 2. The flow
chart illustrated in Fig. 2 begins with decision step 44
wherein the control unit 22 determines if this is the
initial pass through the program. If that question is
30 answered affirmatively, the control unit 22 reads the
current, or new, level of material 12 in the vessel lO from
~277744
g
pressure gauge 33 at step 46. This level reading is then
stored in memory at step 47 for future use and the control
unit 22 exits the program.
If at decision step 44 the control unit 22
determines that this is not the initial pass through the
program, the control unit 22 then reads the current level
of material 12 in the vessel lQ at step 48. At step 50,
the inlet and outlet flow rates are read from pressure
gauges 20 and 30, respectively.
The control unit 22 then determines the expected
rate of accumulation by determining the difference between
the input flow rate and the output flow rate at step 52 and
adding any generation term. The value of the term Fgentt)
is determined. This value may be determined from either a
lookup table in which the control unit 22 merely lookc up
an appropriate value depending upon such parameters as
input flow, output flow, temperature, pressure, etc.
Alternatively, the value for the term Fgen(t) may be
calculated in a separate subroutine depending upon the
particular procegs being carried out in vessel 10.
At step 54, the old, or previous, level reading
of material 12 in the vessel 10 is brought from memory. At
step 56, the actual rate of accumulation of material 12 in
the vessel 10 is determined by determining the difference
between the current level reading and the old level reading.
At step 58, the bias term B is provided by
determining the difference between the expected rate of
accumulation of material in the vessel and the actual rate
of accumulation of material in the vessel. If required, the
bias term B may also be filtered. At step 60, the new or
current level reading is input to memory for use in the
next pass through the program.
~2~744
--10--
At step 64, the control unit 22 solves equation
(6) to determine a new value for the setpoint of the
controlled process control variable. In our example, the
control unit 22 solves equation (6) for the value of the
input flow control setpoint. When the control unit 22
reaches equation 64, it has all the information required
for solving equation (6). The term (MSP(t) - MP (t))/Tau
specifies that you want M to move from its present value
(pv) to its desired value (sp) in time Tau. The output
flow FoUt(t) was read at step 50,as was the generation term
figure. The bias term Bf(t) was determined, and filtered
if necessary, at step 58.
Once the calculated setpoint value is obtained,
it is output at step 65 to controller 37 which controls the
position of the valve 16. Clearly, the flow chart illus-
trated in Fig. 2 could be modified such that the position
of the output valve 26 rather than the input valve 16 is
controlled.
The present invention may also be used in conjunc-
tion with an energy rather than a material balance equation.The necessary instrumentation for effecting an energy
balanced system is illustrated in Fig. 3. In Fig. 3 a
vessel 68 is charged with a material or materials which will
undergo a chemical reaction within the vessel 68. To effect
the chemical reaction, the vessel 68 is to be maintained at
a predetermined elevated temperature. In order to add heat
to the vessel 68 a plurality of gas jets 70 are provided.
Gas is provided to the jets 70 through a pipeline 72 having
a valve 74 and an orifice plate 76. The orifice plate 76
30 acts in combination with differential pressure gauge 78 to
provide an indication of the rate of gas flow in the
pipeline 72.
In order to remove heat from the vessel 68 a pipe-
line 80 is provided to supply coolant to the vessel 68. The
~xm4~
--ll--
pipeline 80 includes an input valve 82, a flowmeter 84, a
temperature sensor 86 for measuring the temperature of the
coolant flowing into the vessel 68, a temperature sensor 88
for measuring the temperature of the coolant leaving the
5 vessel 68, a flowmeter 90, and an output valv,e 92.
The vessel 68 is provided with a temperature
sensor 94 which produces an output signal input to the
control unit 22. The signals produced by the pressure
gauge 78, flowmeters 84 and 90, and temperature sensors 8~
10 and 88 are also input to the control unit 22. The control
unit 22 produces output signals input to a controller
for controlling valve 82, a controller 96 for controlling
valve 74, and a controller 98 for controlling valve 92.
In operation, the dynamic reconciliation process
15 control method operates subs~antially as described above
except that energy in the form of heat, rather than
material flow, is used to control the system. In other
words, the actual rate of accumu}ation of energy, or heat,
within the vessel 68 must equal the expected rate of
20 accumulation of energy plus the bias term.
Referring to equation (3), the left-hand side of
the equation is the actual measured rate of accumulation of
energy (heat) as determined by successive measurements of
the temperature sensor 94. The energy input to the system
25, is determined by the amount of gas supplied to the jets 70
plus the temperature and flow rate of the coolant into the
vessel 68. The removal of heat from the system is
determined by the temperature and flow rate of coolant out
of the vessel 68. The term Fgen (t) is replaced by a term
30 representative of any heat generated by the chemical
reactions occurring within the vessel 68. The expected rate
of accumulation of energy is determined by determining the
difference between the rate of energy input to the
vessel 68 and the rate of energy output from the vessel 68.
~2q7~
-~2-
The bias term is obtained by determining the difference
between the expected rate of accumulation of ener~y and the
actual rate of accumulation of energy. The setpolnt of the
variable which is to be controlled can be determined in a
5 manner similar to that described above in conjunction with
Fig. 2.
S Example 1
B~ l)'~l9 The reconciliation process control system
of the present invention can be implemented in a wide
10 variety of environments. For example, in one application
~ ~o~y~4
the bed level of a 'reactor was controlled. In this
application, the setpoint of a product discharge rate
controller was adjusted to maintain the desired bed level.
Unfortunately, neither the polymer production rate
15 (generation) nor the product discharge rate (output) were
directly measurable. Instead, the production rate was
inferred from a heat balance around th~ cycle gas cooler.
Product discharge was a contlnuous, batch operation. A P~X
drop rate controller was used to initiate the product
20 dlscharge sequence at a specified frequency. The amount of
product removed in a single drop varied with reaction
conditions as well as the mechanical condition of the
discharge system. The discharge rate was inferred based
upon the drop rate (discharges per hour) and fluidized bed
25 density using a simple correlation.
The actual reconciliation was performed on bed
weight, not level. Thus the key variables for substitution
into equation (3) were:
C(t) = Bed Wt (t) = Bed Level * Fluid Bed Dens * 0.1653
30 F ge~t) = Production Rate (t)
F ou~t) = 0.1724 * Fluid Bed Dens * Discharge Rate
The reactor level measurement was very noisy. In
order to avoid excessive control action the bias was then
i277744
-13-
passed through a filter, equation (5), with the filter
factor set to 0.55. In addition the reconciliation
calculation, equation (4), was performed only once every
five minutes.
The best value for Tau (20 minutes) was
determined by trial and error. The control calculation,
equation (6), was performed once per minute. This system
resulted in very accurate control of the bed level.
Example 2
Another example involved an application wherein
the partial pressure of an ethylene feed gas was
controlled. Once again there were a few complications.
First, ethylene w~s fed to the reactor on pressure denland.
Therefore its partial pressure could not be directly
15 controlled. Instead, the buildup of inerts in the reactor
was controlled by periodic venting of the reaction mixture.
These inerts consisted of nitrogen (used to carry catalyst
into the reactor) lsopentane (the co-catalyst carrier) and
ethane (produced by a side reaction). Nitrogen and
20 isopentane feed rates were both measured but ethane
generation was inferred using a correlation based upon
hydrogen concentration and production rate. A final problem
involved the product discharge system. During a product
discharge sequence, reaction gas is lost. Most of this gas
25 is returned by a recycle compressor, but an unmeasured
quantity is lost. Once again this flow was inferred.
The key variables for substitution into equation
(3) were:
C(t) = (Nitrogen + Isopentane ~ Ethane) * Total Moles Gas
30 F ge~t) = Ethane Generation Rate (t)
F ln(t) = Nitrogen Feed + Isopentane Feed
F ou~t) = C(t)*(vent Rate ~ Discharge Losses)/Tota~as~oles
~2~7744
-14 -
The resulting bias was then passed through the filter,
equation (S), with the filter factor set to 0.2. This was a very
heavy filter, but it was necessary due to analyzer signal to noise
ratio. Again, the inventive control method worked well.
In addition to the aforementioned examples, the inventive
teaching may be used for controlling the following processes:
distillation columns lhip and bottom reboiler level control,
overhead condensor level control, steam addition to steam heated
reboilers, and fuel rate to direct fired reboilers), steam systems
(steam header pressure regulation and boiler feedwater level
control)~ furnaces (regulation of fuel to maintain temperature),
high pressure tu~ular or autoclave LDPE reactors (modifier
inventory control, ethylene vinyl acetate inventory control,
initiator addition rate control and hish and low pressure
~eparator polymer inventory control), and low pressure gas phase
polyolefin reactors (polymer inventory control, comonomer
inventory control, monomer inventory control, inerts inventory
control, cataly~t inventory control, co-cata~yst inventory
control, reaction temperature control and chain transfer agent
inventory control).
The inventive process and apparatus represents a
substantial advance over the prior art. Not only does it take
into account various errors occurring throughout the system, but
it also enables the system to be continuously controlled. The
accuracy will enable higher production rates, improved transitions
resulting in less offspec material and improved monomer
utilization, and reduced offspec material due to lower steady
state variability in resin density. While the present invention
has been described in connection with exemplary embodiments
thereof, it will be understood that many modifications and
variations will be readily apparent to those of ordinary skill in
the art. This application and the following claim~ are intended
to cover those modifications and variations.
