WO1993004358A1

WO1993004358A1 - Moisture measurement in a granulated mix

Info

Publication number: WO1993004358A1
Application number: PCT/AU1992/000429
Authority: WO
Inventors: Kenneth Garry Kemlo; Neil Barry Morris
Original assignee: The Broken Hill Proprietary Company Limited
Priority date: 1991-08-15
Filing date: 1992-08-14
Publication date: 1993-03-04

Abstract

Determination of the moisture level in a granulated mix is based on detecting an electrical characteristic of the granulated mix, such as voltage drop or current detected across spaced apart electrodes (A8) and calculating an attribute of the electrical characteristic which is representative of the physical properties of the granulated mix. Moisture level control is also disclosed.

Description

MOISTURE MEASUREMENT IN A GRANULATED MLX

The present invention relates to a method and a system for determining the moisture level in a granulated mix and to a method and a system for controlling the moisture level in a granulated mix.

In particular, the present invention relates to a method and a system for determining the moisture level in a green feed of a sinter plant and to a method and a system for controlling the moisture level in a green feed of a sinter plant.

In the preparation of a green feed for a sinter plant water additions are made to a dry mix to obtain suitable physical properties, such as granulation and permeability, for satisfactory sintering of the green feed.

In the preparation of a green feed, when water is added to a dry mix and the material is agitated the finer particles become attached to the surfaces of the larger particles. The resulting mix has a lower bulk density owing to the increase in air volume within it and this facilitates air flow through the bed during the sintering operation.

The transportation of fine material from the interstices between the larger particles to the surfaces of the larger particles continues as water is added, and the permeability of the bed with respect to the passage of air flow increases as a consequence. At a critical water level (typically, around 6% by weight for iron ore sinter used in steelworks) the granulation and permeability, hereinafter referred to as the "physical properties", of the green feed reach an optimum level and the addition of further water produces no significant improvement in these properties. It is undesirable for the water level to be above or below the critical water level. In this regard, by way of example, the addition of water above the critical water level only serves to flood the interstices with water which has to be evaporated during sintering and thus increases unnecessarily the operating costs of the sinter plant.

As a consequence, it is important to determine the critical water level and to control the preparation of the green feed so that the water level of the green feed is as close as possible to the critical water level. An important part of this process is the accurate determination of the water level of the green feed. One known method for determining the water level in a green feed is to dry a sample of the green feed and to calculate the loss in weight of the sample. However, the method is not satisfactory because of the time delay between taking the sample and calculating the weight loss. Also, the method is not satisfactory because the amount of water needed depends on a range of variables, including how water is absorbed by the fine material, the mix temperature, the physical properties of the dry mix, and the level of return fine material, and in practice the variables do not remain constant over a period of time.

Another known method for determining the water level in a green feed is based on the use of infra-red instrumentation. However, this method has been found to be unsatisfactory in terms of signal drift, electrical maintenance and also the relatively high cost of the instrumentation.

An object of the present invention is to provide a method and a system for determining the moisture level in a green feed of a sinter plant which alleviates the disadvantages of the known methods described in the preceding paragraphs.

According to the present invention there is provided a method for determining the level of moisture in a granulated mix, the method comprising:

(a) detecting an electrical characteristic of the granulated mix;

(b) calculating an attribute of the electrical characteristic which is representative of the physical properties of the granulated mix; and

(c) determining the level of moisture in the granulated mix from the attribute.

The term "attribute" is herein understood to mean a calculated statistic.

It is preferred that step (c) comprises determining the level of moisture in the granulated mix in accordance with a predetermined relationship between moisture level and the attribute.

According to the present invention there is also provided a method for controlling the level of moisture in a granulated mix in a process for preparing the granulated mix, the method comprising, determining in accordance with method steps (a) to (c) described above the level of moisture in the granulated mix and controlling the supply of water to the preparation process as required to adjust the level of moisture in the granulated mix to a critical water level for optimum granulation and permeability.

According to the present invention there is also provided a system for determining the level of moisture in a granulated mix, the system comprising:

(a) a sensor for detecting an electrical characteristic of the granulated mix;

(b) a means for calculating an attribute of the electrical characteristic which is representative of the physical properties of the granulated mix; and (c) a means for determining the level of moisture in the granulated mix from the attribute.

According to the present invention there is also provided a system for controlling the level of moisture in a granulated mix in a process for preparing the granulated mix, the system comprising, the system described in the preceding paragraph, and a means for controlling the supply of water to the preparation process as required to adjust the level of moisture in the granulated mix to a critical water level for optimum granulation and permeability.

It is preferred that the electrical characteristic be voltage drop or current detected across spaced apart electrodes in the granulated mix. It is preferred particularly that the electrical characteristic be voltage drop.

It is preferred that the attribute be independent of the magnitude of the electrical characteristic.

It is preferred that the electrical characteristic be selected as a means of providing an indication of the electrical conductivity and the electrical resistance

(hereinafter referred to as "the electrical conductivity" in view of the inverse relationship between electrical conductivity and electrical resistance) of the granulated mix between the electrodes.

In one preferred embodiment of the method and system the attribute is based on spatial autocorrelation and/or 2-d pattern recognition of the 2-dimensional shape of a trace of the electrical characteristic versus time. In a preferred method and system there are 3 autocorrelation attributes and 3 2-d pattern recognition attributes.

In another preferred embodiment of the method and system the attribute is calculated from the following formula:

C = Σ f(t) - f(t + ε)

where f(t) is the value of the electrical characteristic at a time t and f(t + ε) is the value of the electrical characteristic at a time t + ε.

The present invention, as defined broadly above, is based on the following observations made during experimental work carried out at the sinter plant at BHP Steel Rod and Bar Products Division, Newcastle, to investigate the relationship between electrical conductivity and the level of moisture in a granulated green feed for the sinter plant.

(a) There is a general or average (hereinafter referred to as "the mean") level of electrical conductivity (DC component) at any instant in time which is insensitive to the physical properties of the green feed and, for example, does not provide an indication of the critical water level where optimum granulation and bed permeability are achieved.

(b) There is a variation of the level of electrical conductivity on a short time scale, due to the flow of granulated green feed past a sensor for measuring conductivity, which produces a pattern of changes (AC component) which varies markedly with small changes in water level and is a sensitive indicator of the physical properties of the green feed.

It was realised from the foregoing that the variation of the pattern of changes of the electrical conductivity with time is an intrinsic characteristic which provides information about the effect of the water level on the physical properties of the green feed. In particular, it was realised that there were different patterns of changes when the water level was below, equal to, and above the critical water level for optimum granulation and permeability. It was also realised that recognition of the pattern of changes could be achieved by calculating one or more attributes of a detected electrical characteristic and that a relationship between the attributes and the water level could be determined.

The present invention is described further with reference to the accompanying drawings in which:

Figure 1 is a schematic diagram of the sinter plant at BHP Steel Rod and Bar Products Division, Newcastle;

Figure 2 is a selection of traces from a chart recorder which indicate the changes with time of the rectified voltage across an electrode and the earthed metal structure of the sinter plant during the course of experimental work in which the electrode was located in samples of a flowing granulated feed mix at different water levels and a wave form was applied to the electrode;

Figure 3 is a water controller chart recording over a ten hour period generated during a trial of a first preferred embodiment of the method and the system of controlling water level in accordance with the present invention carried out at the sinter plant of BHP Steel Rod and Bar Products Division, Newcastle; and

Figures 4 and 5 are plots reflecting the results of experiments to assess the performance of the first preferred embodiment of the method and the system of the invention carried out during the trial.

With reference to Figure 1, in the sinter plant, iron ore, lime and coke are mixed with water in a primary mixer (not shown) and are transferred by means of a conveyor 3 to a 150 tonne capacity bin 5. The green feed is transferred from the bin 5 by means of a short conveyor 7 and a chute 9 to the inlet end of a secondary mixer 11. The secondary mixer 11 is a large cylindrical vessel having internal baffles and/or spiral shapes to promote tumbling action and is inclined at approximately 3 degrees to the horizontal. The green feed and water are fed into the inlet end and move downwardly to the outlet end, and there is progressive granulation of the green feed during such movement through the secondary mixer 11. The granulated green feed is transferred from the secondary mixer 11 by a conveyor system 13 to a feed distribution bin 15 and drops from a narrow slot in the lower end of the feed distribution bin 15 onto a rotary distribution drum 17 which feeds the granulated green feed onto a 6 metre wide conveyor 19. Typically, the granulated green feed forms a 350-450 mm layer on the conveyor 19.

The experimental work on the sinter plant on which the present invention is based was carried out in two stages. In the first stage a reference data set was generated by detecting an electrical characteristic indicative of the electrical conductivity of the green feed by means of a sensor, indicated by the term "PROBE (A8)" in Figure 1, at a location immediately downstream of the feed distribution bin 17. The sensor PROBE (A8) comprised an electrode in the form of a mild steel rod (12 mm OD) vertically sunk about 75 mm in the bed. The sensor PROBE (A8) further comprised an electrical circuit which included the electrode, the green feed, a voltage source, and a standard resistor (220 ohms) . A 12 volt 50 Hz waveform was passed through the electrical circuit and back to the voltage source such that approximately 4 volts appeared across the green feed and the remaining 8 volts appeared across the standard resistor. An electrical characteristic, the voltage drop across the green feed, was monitored.

The reference data set comprised 32 three minute samples of output voltage signal of the sensor PROBE (A8) covering a range of water levels encountered in day to day operations.

A selection of typical traces of voltage drop versus time for the samples is shown in Figure 2. It can be seen from Figure 2 that the recorded traces show a series of pen movements above and below a mean signal level.

It was found that, whilst the mean signal level is related to the general water level in the green feed, there were serious problems with attempts to use this mean signal level as a base for accurate measurement of the water level requirements of the green feed.

For example, it was found that tramp material caught on the sensor PROBE (A8) caused changes to signal strength which were not related to actual changes in water level. In addition, it was found that the signal level increased without water changes - for example when the proper flow of green feed past the sensor PROBE (A8) reduced by the presence of a large rock caught upstream. Furthermore, it was found that during periods of significant changes in the amount of return fines on the sinter plant, changes in water level were observed which did not correspond to changes in signal strength. This behaviour was also observed to a lesser extent when there was variation in the height of the green feed at the sensor PROBE (A8) . In addition, it was found that water quality and green feed constituents changes and probe wear also contributed to lowered confidence in the use of average signal level as a means for determining water level.

However, notwithstanding the disappointing result with the mean signal level, it was observed that short term variations in the signal levels were related to the water level.

In particular, it was observed that the shape of the inked section of the traces in Figure 2 was similar when the water level of the mix was constant and the shape changed when the water level changed. It was realised that the changes in the shape of the traces, ie the pattern of changes, were related to the effect of the water on the physical properties of the green feed, ie granulation and permeability, and thus could provide a basis to determine a more meaningful measure of water level than could be obtained from the mean signal level.

In order to recognise differences in the patterns it was necessary to calculate attributes of the signal which were freely constructed to measure the perceived differences apparent to the eye of an observer by observing the chart recording of the traces and the appearance of the green feed and to known changes deliberately made to water flow rates to the sinter plant during the experimental work.

In accordance with a first preferred embodiment of the method and system of the invention, a set of attributes based upon spatial autocorrelation or 2-d pattern recognition of the traces was used. Three attributes of each type were shown to have a high correlation with hand assessment by the operator of the granulation and permeability properties of the green feed. All six attributes (a_x, a₂, ...a₆) were independent of the mean signal level. For instance, multiplying the signal values in any trace by 100, 1 or 0.01 and then calculating the attribute from the scaled data lead to three identical values of the attribute.

The three spatial autocorrelation attributes (a_x, a₂, a₃) were calculated by looking at the way in which the voltage changed with respect to time, this providing significantly more information than using, say, the moments of variance which only look at the way the data is distributed about the mean signal level (ie one dimensional) . The attribute a_x is the Gates Coefficient Spatial Autocorrelation, the attribute a₂ is the average deviation from the mean signal level over the average deviation from the next data point, and the attribute a₃ is the autocorrelation coefficient for the signal with a time span of 5 seconds. The 2-d pattern recognition attributes (a₄, a₅, a₆) were calculated by first converting an array of 900 voltages into 33 columns (time zones) . Each column represented the chart recorder's pen marks. The width of the column represented the width of the drawn pen line, and the length of the column represented the extent of variation of the signal in a time frame equal to the pen width. Further calculations operated directly on these columns. The attribute a₄ is the column to column overlap, the attribute a₅ is the proportion of columns over and under the mean signal level, and the attribute a₆ is the average of the signal range in common to three adjacent columns.

The 6 attributes for each of the 32 samples were calculated and individually plotted against a hand estimated value of water level and a value of water level estimated by eye from the corresponding portion of trace. With the exception of attributes a₃, a₄ clear linear relationships were observed. Using linear or second order polynomial regressions, equations were obtained relating estimated water versus attribute and, in addition, the appropriate error estimates were noted. On the basis of the errors and an understanding of the physical properties reflected by the attribute a weighing factor w_i was assigned to each attribute. On the basis of the regression equations, the attributes a_± were converted to the least squares best fit water estimate, b_±. The final calculation of water was of the form:

Tonnes of Water from Set-Point = Σ

In accordance with a second preferred embodiment of the method and system of the invention, a single attribute C_L expressed in terms of the following algorithm was used to examine the structure of the traces.

C = Σ f(t)-f(t+ε)

where f(t) is the signal at time t and f(t+ε) is the signal at time t + ε, and the summation runs from the first sample to the last sample minus ε samples.

The attribute calculates the average absolute change in amplitude of the signal over a time interval ε. It is possible to calculate the attribute a number of times using different values of ε. This then provides information on the shape of the signal.

Four attributes C_± were calculated for each of 25 of the 32 samples and a comparison, by multiple linear regression, against hand estimated values of water level provided weighing factors (d_±) . The final calculation of water was of the form:

Tonnes Water from Set Point = Σ G_jά.^ .

When determining moisture level by the second preferred embodiment the four attributes were calculated for each of the 25 samples from the 900 latest conductivity measurements.

The calculations are set out in Table 1 below. In the table, the time shifts ε are in numbers of data points, and the values of water level (WaterP) from the attributes C_t and the hand estimated values of water level (WaterE) are in tonnes/hr. from the Set-Point. 14

Table 1

With reference to Sample "28", by way of example, it was found that ^=1.33777 when ε=100; c₂=l.23790 when e=125: c₃=l.18359 when ε=150: c₄= 1.23270 when ε=200. The result of the multiple linear regression of c₁..c₄ with WaterE of all samples from the reference data set gives coefficients of d0=l.23779. d-=0.69554, d₂=0.68472, d₃=l.34421 and d₄=l.98450.

Therefore,

WaterP=d₀+d₁c₁+d₂.c₂+d₃.c₃+d₄.c₄

+1.23779-0.69554*1.33777+0.68472*1.23790 +1.34421*1.18359-1.98450*1.23270 =0.50342 tonnes/hour.

The second preferred embodiment of the method and system of the invention requires only one parameter to be calculated to examine the structure of the trace rather than a set of parameters which must be calculated in the first preferred embodiment of the method and system of the invention, and thus potentially the second preferred embodiment has advantages in terms of reduced program size and computation time compared with the first preferred embodiment.

In the second stage of the experimental work the computer calculated water levels from the reference data set in accordance with the first preferred embodiment of the method and system of the invention were used as a basis for automatically controlling the water supply to the sinter plant to control the water level in the green feed during a trial carried out at the BHP Steel Rod and Bar Products Division, Newcastle. In broad terms, the method for automatically controlling the water level in the green feed comprised the following steps:

(a) detecting the voltage drop by means of sensor PROBE (A8) shown in Figure 1; (b) determining six attributes of the voltage drop;

(c) determining the level of moisture in the green feed using the relationship between the moisture level and attributes determined from the reference data set; and

(d) controlling the flow rate of water to the green feed in the secondary mixer 11 in accordance with the moisture level determined in step (c) above.

The calculation of the attributes and the determination of the water level in steps (b) and (c) above was carried out by an IBM PC compatible computer. The output voltage signal from the sensor PROBE (A8) was rectified, amplified and filtered and then passed to the computer via an analogue-digital converter (Analogue Devices RTI-815) .

The control of the flow rate of water in step (d) was carried out by a Moore Instruments MYCRO 352 Single loop Digital Controller (SLDC) which operated by comparing the calculated water level against a predetermined set point and adjusting a water supply valve opening on the basis of this comparison.

Figure 3 is a ten hour section of a trace from the water controller chart recorder for the secondary mixer 11 taken during the course of the trial. The right hand trace, which has a span of 5% to 6% water shows the calculated water value during the ten hour period. A set-point of 5.8% was set by the operator on the water controller console, such that calculated water levels above 5.8% lead to a partial closure of the valve and values less than 5.8% lead to the controller increasing the valve opening. The left hand trace shows the actual water flow into the secondary mixer 11 on a 0 to 5 tonne per hour scale. This water flow was determined by the on line control program output.

On Figure 3 there is marked a series of labels A-F and the following is a brief explanation of each of the labels.

A. The secondary water mixer 11 was shut off by an operator for routine plant maintenance.

B. During this time the system calculated levels of water in excess of the set point requirement (ie wet). The corresponding signal to the valve caused the flow rate to the secondary mixer 11 to be reduced.

C. An operator selected manual control for one hour.

D. When returned to automatic, the computer determined wet conditions persisting for half an hour.

Throughout this period the system reduced the water supply to the secondary mixer 11 at the rate of 0.4 tonne/hr.

E. In the following 4 hours the calculated water level was consistently low (dry) and the system increased the water supply to the secondary mixer 11.

F. A third wet period with a corresponding reduction in the water supply.

In order to test the performance of the method and the system deliberate changes were made to the amount of water on the sinter plant. This enabled the response of the automatic method and system to be assessed as it brought the water level back to the desired level. Two such examples are provided in Figures 4 and 5. Changes to primary water levels required 15 to 20 minutes to reach the sensor PROBE (A8) . Changes to the water level supplied to the secondary mixer 11 required about 5 minutes to reach the sensor PROBE (A8) .

In Figure 4 the water level in tonnes/hr was reduced on the mixer primary from 10.5 to 10.0 at time zero. Since the level of water in the mix had been reduced it was expected that the method and system would increase the water supply to the secondary mixer 11 by this amount some time after the change had reached the sensor PROBE (A8) . The computer calculated water level showed a decline, from about 5.80% to 5.70% as the effect of the reduced primary flow became evident and there was a subsequent increase in calculated water as the controller fed more water in to the secondary mixer 11 to compensate for the reduction at the primary mixer.

With reference to Figure 5, in this experiment the water flow rate to the secondary mixer 11 was suddenly reduced by 1.5 tonne/hr manually. Shortly afterwards computed water values went from 5.80% to 5.68%. The automatic control system increased the water flow rate and by window number 17 the calculated water value had reached 5.77% and at this stage had put 0.8 tonne/hr back onto the flow to the secondary mixer 11. It is noted that there were corresponding changes to the average signal level and the pattern on the trace sections.

In summary, the trial established that the first preferred embodiment of the method and system of the present invention for determining and controlling the water level in a granulated mix operated successfully.

Many improvements and modifications may be made to the present invention as describe in the foregoing without departing from the spirit and scope of the present invention. In this regard, whilst the preferred embodiments of the method and system of the present invention are described in relation to a sinter plant in a steelworks, it can readily be appreciated that the invention is not so limited and extends generally to applications where water is added to a granular mix in order to obtain a desired rheological property, such as consistency, kneadability, or viscosity. Examples of other applications include concreting, non-ferrous sinter plants, and food production.

Claims

CLAIMS :

1. A method for determining the level of moisture in a granulated mix, the method comprising:

(a) detecting an electrical characteristic of the granulated mix;

(c) determining the level of moisture in the granulated mix from the attribute.

2. The method defined in claim 1, wherein step (c) comprises determining the level of moisture in the granulated mix in accordance with a predetermined relationship between moisture level and the attribute.

3. The method defined in claim 1 or claim 2, wherein the electrical characteristic is voltage drop or current detected across spaced apart electrodes in the granulated mix.

4. The method defined in claim 3, wherein the electrical characteristic is voltage drop.

5. The method defined in claim 3 or claim 4, wherein the attribute is independent of the magnitude of the electrical characteristic.

6. The method defined in any one of the preceding claims, wherein the attribute is based on spatial autocorrelation and/or 2-d pattern recognition of the 2- dimensional shape of a trace of the electrical characteristic versus time.

7. The method defined in claim 6, comprising 3 autocorrelation attributes and 3 2-d pattern recognition attributes.

8. The method defined in any one of claims 1 to 5, wherein the attribute is calculated in accordance with the following formula:

C = Σ f(t) - f(t + ε)

9. A method for controlling the level of moisture in a granulated mix in a process for preparing the granulated mix, the method comprising, determining the level of moisture in the granulated mix in accordance with the method defined in any one of the preceding claims, and controlling the supply of water to the preparation process as required to adjust the level of moisture in the granulated mix to a critical water level for optimum granulation and permeability.

10. A system for determining the level of moisture in a granulated mix, the system comprising:

(a) a sensor for detecting an electrical characteristic of the granulated mix;

(b) a means for calculating an attribute of the electrical characteristic which is representative of the physical properties of the granulated mix; and

(c) a means for determining the level of moisture in the granulated mix from the attribute.

11. A system for controlling the level of moisture in a granulated mix in a process for preparing the granulated mix, the system comprising, the system for determining the level of moisture in the granulated mix defined in claim 10 and a means for controlling the supply of water to the preparation process as required to adjust the level of moisture in the granulated mix to a critical water level for optimum granulation and permeability.