RU2366911C2 - Method for automatic control of nickel tetracarbonyl (nt) decay process in device with electric heating of walls - Google Patents

Abstract

FIELD: physics, control.

SUBSTANCE: invention is related to methods of automatic control of technological processes, in particular to system of nickel tetracarbonyl decay process control in hollow cylindrical device. Statistic adaptive model of 2^nd order is used for control, which makes it possible to predict these parametres with sufficient metrological characteristics. Generated adaptive statistic model of the 1^st order is used to calculate settings for temperature mode stabilisation circuits in device. Calculation is carried out as solution of optimisation task of mathematical programming with limitations in the form of inequalities. Obtained solutions are used as settings for temperature stabilisation controllers by zones of furnace that vary supply of electric energy and maintenance of specified values, besides special algorithmic solutions make operation of control circuits mutually independent on each other and invariant to external disturbances: flow and temperature of return gas CO and air temperature in workshop.

EFFECT: improved parametres of produced nickel powder quality.

1 dwg

Description

The invention relates to methods for automatic control of technological processes, in particular to a control system for the decomposition of nickel tetracarbonyl in a hollow cylindrical apparatus.

In Russia at the moment there was no working system for automatic control of the decomposition of nickel tetracarbonyl. Manual control is used according to the technological regulations (instructions). The technological instructions indicate strict temperature limits inside the case, as well as the boundaries of the control actions (temperature of the heaters), which cannot be exceeded.

This process is difficult to control, and even more so for creating an automatic control system, since the process is stochastic, i.e. continuously changing, and under the same conditions of the process at the output, it is possible to obtain nickel powder with different quality indicators. In addition, a method for obtaining continuous reliable indicators of the quality of the powder has not yet been invented. Thus, compliance with the technological instructions does not guarantee the receipt of the powder of the required quality.

In addition, there is a method for automatically controlling the decomposition of nickel tetracarbonyl based on a model calculated from experimental data. The coefficients of this model are calculated in advance and are not automatically adjusted. In addition, it should be noted that by creating a model on a sufficiently large range of possible values of process parameters, its accuracy is deteriorating. This model controls the process temperature (temperature inside the decomposer housing) and tries to stabilize it. The disturbances that occur in the system are not directly monitored, but are recorded when the process temperature changes, which stabilizes due to changes in fuel consumption. In addition, the indicator of the quality of this system is only the bulk density of the nickel powder, it does not take into account the average particle diameter, which does not allow to obtain a fully conditioned nickel powder. This solution is the closest to the proposed technical solution, so we take it as a prototype. (Automatic control of the process of nickel carbonyl decomposition; V.A. Ivanov, D.I. Lisovsky, V.V. Stopkevich; TsIINtsvetMet Non-ferrous Metals Bulletin No. 21, 1966)

The aim of the invention is to eliminate the disadvantages of the prototype, i.e. improving the quality of the obtained Nickel powder in the process of decomposition of Nickel tetracarbonyl, using the objective function:

where Ves ^* , Fisher ^* - set values of quality indicators;

Ves, Fisher - predicted quality indicators;

ω ₁ , ω ₂ - weighting factors (ω ₁ + ω ₂ = 1), set by the master technologist.

Nickel powder quality is understood to mean product quality indicators: bulk density (Ves) and average particle diameter (Fisher).

To solve this problem, a method is used to automatically control the decomposition of nickel tetracarbonyl (TKN) in an apparatus with electric wall heating, including temperature control of sections of the outer wall of the apparatus, characterized in that

- simultaneously measure and stabilize the temperature of the wall of the apparatus in the heating zones,

- form a statistical model of the following type:

where Ves is the bulk density of the nickel powder;

Fisher — average particle diameter of nickel powder;

Т _{к1 ... 6} - case temperature in zones 1-6 (preliminary tasks to local stabilization loops);

P _TKN - pressure Ni (CO) ₄ in the decomposer;

Q _CO is the consumption of CO in the decomposer;

P _O2 is the oxygen pressure in the decomposer;

and ₁₀ ... and ₂₉ are the coefficients of the model,

- measure the air temperature in the workshop, the temperature and flow rate of the circulating CO,

- then continuously correct the tasks calculated by the statistical model and corresponding to the objective function to the local stabilization contours according to the air temperature in the workshop, to the temperature and flow rate of circulating CO, using the expressions:

where i = 1,2,3 ... n;

- температура нагрева в i-той зоне (скорректированное задание локальным контурам стабилизации);

- heating temperature in the i-th zone (adjusted task to local stabilization loops);

- величина коррекции первоначального задания в i-той зоне;

- correction value of the initial task in the i-th zone;

T _Ki - initial task for the heater in the i-th zone, satisfying the objective function, calculated according to the model (2);

T _mi (j) is the temperature of the material in the j-th zone;

T _ki (j) is the body temperature in the j-th zone;

T _mi (j-1) is the temperature of the material in the (j-1) -th zone;

T _mi-i (j-1) is the temperature of the material in the (j-1) -th zone at the previous moment in time;

T _CO (j-1) is the temperature of the circulating CO in the (j-1) zone;

Q _CO (j-1) - consumption of circulating CO in the (j-1) zone;

T _OC (j-1) is the ambient temperature in the (j-1) zone;

and ₁₀ ... and ₆₅ are the coefficients of the model,

- predict the current quality indicators of the process according to the following formulas:

Where:

- прогнозное значение насыпного веса;

- forecast value of bulk density;

- прогнозное значение среднего диаметра частиц;

- the predicted value of the average particle diameter;

x ₁ ... x ₆ = T _{k1 ... 6} - case temperature in zones 1-6;

x ₇ = P _TKN - pressure of Ni (CO) ₄ in the decomposer;

x ₈ = Q _CO - _CO consumption in the decomposer;

x ₉ = P _O2 - oxygen pressure in the decomposer,

, подают в качестве задания на регулирование температуры участков внешней стенки аппарата.- and the obtained values

, served as a task for regulating the temperature of the sections of the outer wall of the apparatus.

The proposed method uses both analog information received from sensors located on the device and discrete information entered manually according to the results of analysis of the final product. The control algorithm allows us to solve the problem with a minimum of impacts, since the assessment of the need for decision-making is made after a sufficiently large period of time. In addition, in order to avoid sharp changes in the tasks of the actuators, the temperature of the case calculated using the optimization task in different zones of the furnace is adjusted depending on the value of their difference with the previous values and on whether one or both process quality indicators differ from their set values by more than allowable value.

The work of the invention is as follows: a database of normal operation is collected for 48 hours with a resolution of 5 minutes, then the coefficients of the mathematical model are calculated by the formulas (2), (3) and (4).

We use the following notation in formula (2):

y ₁ = Ves;

y ₂ = Fisher;

x ₁ ... x ₆ = T _k1 ... T _k6 ;

x ₇ = P _TKN ;

x ₈ = Q _CO ;

x ₉ = P _O2 .

Then the system of equations will take the form:

The search for the coefficients of these equations is carried out using the least squares method. To do this, minimize the sum of the squared deviations

Equating the partial derivatives with respect to the coefficients a _{(i) j} to zero, we obtain the following system of equations:

We rewrite these equations by introducing the following notation:

Then the system of equations is written as:

Where

a _i = (a _0i , a _1i , a _2i , a _3i , a _4i , a _5i , a _6i , a _7i , a _8i , a _9i ,) ^T

B = (b _0i , b _1i , b _2i , b _3i , b _4i , b _5i , b _6i , b _7i , b _8i , b _9i ,)

Equation (7) is solved by the Gauss method, which consists in reducing the matrix D _i to a triangular form.

We use the following notation for formula (3):

Then equations (4) are rewritten as follows:

The search for the coefficients of these equations is carried out using the least squares method. To do this, minimize the sum of the squared deviations:

where N is the number of measurements, i = 1 ... 6.

Equating the partial derivatives with respect to the coefficients a _{(i) j} to zero, we obtain the following system of equations:

We rewrite these equations by introducing the following notation:

Then the system of equations is written as:

Where

a _i = (a _0i , a _1i , a _2i , a _3i , a _4i , a _5i ) ^T

B = (b _0i , b _1i , b _2i , b _3i , b _4i , b _5i )

Equation (11) is solved by the Gauss method.

The search for the coefficient values for equations (4) is carried out by the least squares method. To do this, minimize the sum of the squared deviations

Equating the partial derivatives with respect to the coefficients a _{(i) j} to zero, we obtain the following system of equations:

We introduce the following notation:

Then the system of equations is written as:

Where

a _i = (a _0i , a _1i , a _2i , a _3i , a _4i , a _5i , a _6i , a _7i , a _8i , a _9i , ... a _55i ) ^T

B = (b _0i , b _1i , b _2i , b _3i , b _4i , b _5i , b _6i , b _7i , b _8i , b _9i , b _54i )

Equation (14) is solved by the Gauss method.

After that, the current quality indicators are calculated by the formula (4), using the obtained model coefficients and current process parameters. If these values differ from the set ones by an amount greater than the tolerance, the control actions are calculated according to the formula (1), substituting the calculated model coefficients and the necessary quality indicators into it and using the objective function (2). Next, the calculated values are corrected by the formula (3). The calculated values of the housing temperatures in various zones are checked for compliance with the permissible limits:

where T _ki is the calculated value of the body temperature in the i-zone;

- минимальное значение температуры корпуса в i-зоне;

- the minimum value of the case temperature in the i-zone;

- максимальное значение температуры корпуса в i-зоне.

- the maximum value of the case temperature in the i-zone.

If the values of the calculated temperatures are outside the permissible limits, then they are taken equal to the limiting values and the optimization problem is again solved by formulas (1), (2) and (3) with five variable temperatures, then, if necessary, with four, etc. .

After that, in order to avoid sharp changes in the tasks of the actuators, the correction value of the current temperatures in the zones is calculated depending on the difference between the newly calculated and the previous temperature and on whether one or both quality indicators differ from the set ones by more than the permissible value. If both quality indicators are outside the acceptable zone relative to the set value, the correction value is defined as

T '= ΔT / k,

where ΔТ is the difference between the current and newly calculated temperature of this zone,

k is the tuning parameter.

With an unacceptable deviation from the set value of only bulk density, the temperature correction is calculated as

T '= ΔT / k + 1,

where ΔT is the difference between the current and newly calculated temperature of this zone,

k, 1 - tuning parameters.

And in the latter case, when only the value of the Fisher indicator is outside the permissible zone, the correction value is

T '= ΔT / k + m,

where ΔT is the difference between the current and newly calculated temperature of this zone,

k, m - tuning parameters.

(k, l, m - standard values, which are tuning parameters, 1 <m).

Numerical example

Let's take the database from 04.25.2006 6:39 a.m. to 04.25.2006 7:14 a.m.

Time 6:39 6:44 6:49 am 6:54 6:59 a.m. 7:04 7:09 a.m. 7:14 a.m. The weight 0,034 0.19 0.19 0.19 0.19 0.19 0.19 0.19 Fisher 2,51 2,51 2,51 2,51 2,51 2,51 2,51 2,51 Tkor1 536.7 536.7 536.3 533.7 533.5 531.9 530.8 530.5 Tkor2 521.1 520.7 518.5 517.9 517.6 516 514.9 514.5 Tkor3 420.6 420 418.4 417.7 417.3 416.9 415 414.6 Tkor4 406 405.4 405 404.8 404.6 402.3 402.1 401.3 Tkor5 305 304.6 304.7 302.8 301.9 301.9 301.6 301.3 Tkor6 300.9 300,2 300,2 299.8 299.5 297.7 297 296.6 Tmat1 267.5 266.3 265.5 264.6 263.7 262.8 261.7 260.7 Tmat2 274.7 273.8 272.6 272.1 270.9 269.8 269.3 268 Tmat3 267.7 266.3 266 264.9 264.2 263.3 262.5 261.6 Tmat4 257.7 256.8 256,4 255.4 254.7 253.8 253.1 252.2 Tmat5 237 236.3 235.4 234.9 234.2 233.3 232,4 231.8 Tmat6 209.5 208.1 207.8 207.2 206.4 206 204.8 204.1 ExSCO 6.1 5.9 6.7 6.9 6.9 7.5 7.1 7.3 Pressure O2 0.2 0.199 0.198 0.198 0.197 0.196 0.195 0.195 DavlTKN 0,034 0,033 0,034 0,035 0,034 0,035 0,034 0,035 Rass about CO 121.5 120,4 118.5 119.9 118.8 120.3 121.6 121.5 T about WITH 26.3 26 25,4 25.3 25 25 24.3 24.2 Tos 10.06 10.06 10,2 10.25 10.25 10.45 10.16 10.25

For model (2), according to formula (7), we calculate the coefficients

The weight Fisher freedom -0.159614551 2,966396298 Tkor1 0,000339887 6.81055E-07 Tkor2 0,000228099 -0.003440634 Tkor3 0.002550118 0.00319666 Tkor4 -0.002334715 0,003433622 Tkor5 -0.001831462 -0.01445 Tkor6 0,001514299 0,010088988 ExSCO -0,000756139 -0.002119386 Pressure O2 0.220661353 0.377208688 DavlTKN 0.965923275 3.392790128

By the formula (2) we obtain the forecast Weight - 0.227, Fisher - 2.715.

A more accurate forecast by formulas (4) gives the following results: Weight - 0.217,

Fisher - 2,696, which for given values (and tolerances) for Weight 0.18 (0.02) and Fisher 2.5 (0.15) requires the calculation of new values for the temperature of the buildings.

By the formula (2), we calculate the new values of the housing temperatures using the objective function (I):

Tkor1 = 526.84;

Tkor2 = 498.32;

Tcr3 = 418.31;

Tkor4 = 406.02;

Tkor5 = 301.31;

Tkor6 = 280.86.

We check the occurrence of the calculated temperatures of the enclosures within the established limits:

The temperature of Tkor2 is below the lower limit (500), thus Tkor2 = 500.

We recalculate the temperatures:

Tkor1 = 526.84;

Tkor2 = 500;

Tkor3 = 416.75;

Tkor4 = 406.53;

Tkor5 = 300.60;

Tkor6 = 280.18.

The correction of the calculated new values of the temperature of the buildings according to formulas (3) will be:

ΔTcor1 = 0.86;

ΔTcor2 = -0.13;

ΔTcor3 = 0.06;

ΔTcor4 = 0.45;

Δ Тcor5 = 0.31;

ΔTkor6 = -0.03.

Thus, the corrected values of the temperature of the shells will be (correction according to Tkor2 again takes Tkor2 out of the acceptable limits, therefore Tkor2 remains equal to the lower boundary):

Tkor1 = 527.7;

Tkor2 = 500;

Tkor3 = 416.81;

Tkor4 = 406.98;

Tkor5 = 300.91;

Tkor6 = 280.15.

Next is a step-by-step transfer of the calculated control actions on the actuators:

Step 1 Step 2 Step 3 Tkor1 529,5667 528.6333 527.7 Tkor2 509,6667 504.8333 500 Tkor3 415.3367 416.0733 416.81 Tkor4 400,5267 399.7533 398.98 Tkor5 296.5033 291.7067 286.91 Tkor6 291.1167 285.6333 280.15

Predicted values of quality indicators for the calculated (final) values will be:

Weight = 0.194;

Fisher = 2.641.

Thus, the performance of the objective function is ensured.

Brief Description of the Drawings

The drawing shows a diagram of a system for automatic control of the decomposition process TKN. Information on the current process parameters: bulk density and average particle diameter of nickel powder (10), current temperature of the decomposer housing by zones (14.1-14.6), pressure of Ni (CO) ₄ in the decomposer (11), pressure О ₂ (12) in the decomposer , CO consumption (13) in the decomposer, enter the mathematical model (3) of the process, which, depending on the values of these parameters, generates preliminary tasks for local controllers (15). Further, these tasks are compared with the current temperature of the decomposer housing by zones (14.1-14.6) and adjusted according to the values obtained in the control correction block (4).

Calculation of the correction of the task to the controller is carried out in the control correction block (4), which receives the current values of the following process parameters: temperature inside the decomposer housing in zones from thermocouples (8), flow rate and temperature of the return СО (9), air temperature in the workshop, near the bottom a non-lined zone from a temperature sensor (7).

After that, the corrected tasks are sent to the controller (5), which, through the tasks to the heaters (16.1-16.6), sets the necessary values of the temperature of the decomposer housing by zones, ensuring minimal deviations of the quality indicators from the set values.

Claims (1)

A method for automatically controlling the process of decomposition of nickel tetracarbonyl (TKN) in an apparatus with electric wall heating, including controlling the temperature of sections of the outer wall of the apparatus, characterized in that
at the same time measure and stabilize the temperature of the apparatus wall in the heating zones,
form a statistical model of the following form

Where

- bulk density of nickel powder;

- the average particle diameter of the Nickel powder;
Т _{к1 ... 6} - case temperature in zones 1-6 (preliminary tasks to local stabilization loops);
P _tkn - pressure Ni (CO) ₄ in the decomposer;
Q _WITH - consumption of CO in the decomposer;
P ₀₂ - oxygen pressure in the decomposer;
and ₁₀ ... and ₂₉ are the coefficients of the model,
measure the air temperature in the workshop, the temperature and flow rate of the circulating CO,
then continuously correct those calculated by the statistical model and corresponding to the objective function of setting the local stabilization contours for the air temperature in the workshop, for the temperature and flow rate of the circulating CO, using the expressions

where i = 1,2,3 ... n;

- heating temperature in the i-th zone (adjusted task to local stabilization loops);

- the correction value of the initial task in the i-th zone;
T _Ki - initial task for the heater in the i-th zone, satisfying the objective function, calculated according to the model (2);
T _mi (j) is the temperature of the material in the jth zone;
T _ki (j) is the body temperature in the jth zone;
T _mi (j-1) is the temperature of the material in the (j-1) th zone;
T _mi-i (j-1) is the temperature of the material in the (j-1) -th zone at the previous moment in time;
T _CO (j-1) is the temperature of the circulating CO in the (j-1) zone;
Q _CO (j-1) - consumption of circulating CO in the (j-1) zone;
T _OC (j-1) is the ambient temperature in the (j-1) zone;
and ₁₀ ... and ₆₅ are the coefficients of the model,
Predict current process quality indicators using the following formulas:

Where

- forecast value of bulk density;

- the predicted value of the average particle diameter;
x ₁ ... x ₆ = T _{k1 ... 6} - case temperature in zones 1-6;
x ₇ = P _tkn - pressure Ni (CO) ₄ in the decomposer;
x ₈ = Q _CO - _CO consumption in the decomposer;
x ₉ = P ₀₂ - oxygen pressure in the decomposer,
and the obtained values

served as a task for regulating the temperature of sections of the outer wall of the apparatus.