- Email: [email protected]
Component wear in vertical spindle mills grinding coal
ELSEVIER
Int. J.
Miner. Process.44-45 ( 1996)569-58 1
Co:mponent wear in vertical spindle mills grinding coal J. Sligar ’ Pacific Power, Sydney, Australia
Abstract This paper reports on the contribution of a number of material coal properties and machine operating parameters to the wear of vertical spindle mill components. This assists in the evaluation of mills and coals for particular duties and enables prediction of the likely changes in maintenance regime if these variables are changed. Coal abrasiveness was found to be the most critical coal property with moisture next in importance. Feed rate was of importance with some interesting interactive affects. Roll grinding pressure was by far the most important machine variable with recirculation and air/coal ratio of lesser importance. Alloys used were characterized by Brinnell Hardness Number (BHN) and chromium alloy content.
1. Introduction
A paper on pulverizing mill component wear arising from size reduction of coal was presented at the World Congress on Particle Technology in Nuremburg in 1986. This surveyed the majority of coal fired power stations in Australia in a major industrial project on mill wear and the contribution of coal properties to this. This project found that the coal property giving the best initial indication of likely component wear in high speed, medium speed, or low speed mills was the coal abrasiveness as measured by BS1016:19. This project accepted that mills were operating near their optimum and no attempt was made to study potential mill variables, especially as all the mills were part of productive plants with economic and financial commitments. As an extension of this project a further study was undertaken, exploring mill variables and further coal
’Prerent address: Sligar and AssociatesPty Ltd, 10 Bond St., Mosman, 2088 N.S.W., Australia. 0301-7!~16/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved SSDI 0301-7516(95)00066-6
570
J. Sligar/Int.
J. Miner. Process. 44-45 (1996) 569-581
properties identified in the first project. Full industrial plant could not be subjected to the range of variables economically so a pilot scale medium speed vertical spindle mill was used, being the smallest version of one of the more frequently used mills in the industrial test. The aim was to better define the factors concerned with wear and the relative contribution of those associated with the milling machine and those concerned with coal properties. It was further decided to use a series of coals as a subset of the previous project but with as broad a range of likely contributing properties as possible. The operation of coal pulverizers in power stations and other applications is always a compromise between operating for maximum output and minimum maintenance cost. This project was thought to provide numerical input to this problem for effective decision making. 1 .I. Coal sample selection
In order to address the project objectives in as economic manner as possible, six (6) Australian domestic coals were chosen for comprehensive pilot scale mill testing. The coals were chosen as far as possible to satisfy the following criteria: - the coals had been part of the previous project * the coals covered a range of coal moisture levels from 3.6 to 29.1%; * the coals covered a range of Hardgrove Grindability Indices from 40 to 82; - the coals covered a range of Abrasion Indices from 7 to 66; I .2. Pilot scale test program The milling test work was conducted using a pilot scale mill, ICAL/EVT Model No. RP153X bowl mill with a 400 mm bowl diameter, three (3) rolls and an integral twin cone classifier. The procedures used in this project were developed as part of a separate research project. 1.3. Test program A number of mill operating variables were examined: material feed rate roll grinding pressure air/fuel ratio mill internal recircuation rate A range of alloys commonly used together with a range of aloys believed to have potential were tested in the table segments. For every test coal the mill was run at a range of gradually increasing roll grinding pressures from the minimum to the maximum available with nominal full load coal feed rate. This enabled identification of the first grinding pressure which produced 70% of product material less than 75 micron, which is the nominal standard for the industry. 9 * -
J. Miner. Process. 44-4.5 (1996) 569-581
.I. Sligar/lnt.
571
Tests were then carried out with a range of coal feed rates from 60 to 120% of capacity. In a limited number of cases air/fuel ratio and mill recirculation rate were also varied. Mois,ture had been previously identified as the coal property with anomalous effects on abrasiveness of coal, possibly contributing some corrosive component to the abrasive effect. Moisture was therefore tested at three levels, air dried, as received and as received plus 3% added moisture. The two main wearing components in a medium speed mill are the grinding rolls and the rotating table segments. Wear of the these components is the primary consideration for mill maintenance. The roll and table each have a finite thickness of wear resistant material which when exhausted requires the mill to be taken out of service and the elements replaced. It would be expected that the rate of wear of these components would be a function of the properties of the coal being ground, the settings and operation of the mill and of the met.al alloy used in the wearing components. These will be dealt with in turn in Sections 2, 3 and 4.
2. Coal property effects 2.1. Coal abrasiveness From previous work (Sligar, 1986) the abrasiveness of coal (BS 103819) was found to be the best overall measure of the coal characteristic causing wear in the three basic types of mill. Prediction equations using this factor were developed for a preliminary estimate of likely wear. It would be expected that the major contribution to wear prediction would be from this coal property together with a possible correction for moisture. As a first estimate Table 1 sets out the relevant variables. All these are based on comparing pilot mill tests using coal with as received moisture, operating at full load
Table 1 Roll wear rates at mill full load
Wear (g/ton) Mar Free SiO,(%) HGI AI hg/kg) GP AIXGP AIXHGI Marx SilD,
Bay
Cal
Con
Muj
Tar
Wal
4.1 6.3 6.2 55 49 3.5 172 2695 39.1
2.2 12.5 1.2 80 7 2.0 14 560 15.0
4.75 3.6 6.1 82 66 2.0 132 5412 22.0
7.2 29.1 2.2 60 68 5.4 367 4080 64.0
7.0 7.1 12.7 47 44 6.5 286 2068 90.2
10.9 8.7 8.3 49 81 6.0 486 3969 72.2
J. Sligar/Int.
572
and delivering micron. Comparing form:
pulverized
J. Miner. Process. 44-45 (1996) 569-581
coal with standard product
coal Abrasive
particle size of 70% less than 75
Index and mill roll wear the relationship
War = 0.092 AI + 1.2
is linear of the
( r = 79%)
War = wear rate (g/ton) AI = Abrasiveness
(BS1038:19)
2.2. Free silica Free silica had been proposed on a number of occasions as the major coal property factor contributing to mill component wear. Values of free silica for every test coal are shown in Table 1. A linear regression of these values indicated a linear equation with the following constants: War = 0.36 SiO, + 3.8
( r = 49%)
where SiO, = free silica content
(%)
The low correlation indicates that free silica is not a good measure. This conclusion is supported by previous work (Sligar, 1986) which showed that free silica alone was a poor measure but that a combination of silica content and particle size distribution gave a better correlation. 2.3. Hardgrove Grindability Hardgrove Grindability had been put forward on a number of occasions as a measure of mill component wear. Values for HGI and wear were extracted from Table 1 and a linear regression carried out. This resulted in the following relation: War = -0.138HGI
(r = 69%)
+ 14.58
Again this is a relatively poor correlation indicating the possibility of a correlation but no more than this. This agrees with previous project of Sligar (1986). 2.4. Moisture For as received
coal the correlation
War = 0.04Mar
+ 5.53
of wear with moisture
from data in Table
1 is:
(r = 14%)
where Mar = as received moisture
%
The two cases of air dried coal and as received
coal plus 3% moisture
are shown in
J. Sligar/lnr.
J. Miner. Process. 44-45 (1996) 569-581
573
Table 2 Roll wear rates with varying moisture
Bay
Cal
Con
Muj
Tar
Wal
Mad AI (mg/kg.) GP AIXGP Wear (g/ton) Mad X SiO,
4.0 41 4.05 166 3.98 24.8
9.8 I 2.2 15.4 3.56 11.8
1.2 25 2.0 50 4.0 1.3
18 29 7.0 203 8.84 39.6
5.6 45 3.8 171 6.5 71.1
2.4 45 _ _ _ -
Mar + 3% AI (mg/kg) GP AIXGP Wear (g/ton) M, X SiO,
9.3 _ -
15.5 2.0 14 2.74 18.6
6.6 2.05 51 7.37 40.3
32.1 3.74 102 4.8 70.6
10.1 6.2 360 8.0 128.3
11.7 _
Table 2. These relationships dried coal
of mill roll wear rate are similarly
Wad = 0.052 Mad + 3.85
linear and of the for air
( r = 34%)
where Mad = air dried moisture For as received
%
coal with 3% extra moisture
W! = 0.13 M, + 2.27
(using as received
Abrasiveness):
( r = 84%)
where WZ = extra moisture case From previous temperature measurements in the mill (Sligar, 1976) and analysis of coal in various parts of the mill most of the surface moisture is removed in the fast fluidized bed surrounding the rotating table. As received coal passes down the pulveriser feed chute on to the rotating table. It is mixed with partly dried, partly ground coal on the table and passes under the roll for a first pass at size reduction. It then passes over the rim of the table into the fast fluidized bed surrounding this. Hot primary air at 250-300°C passes upward through the overflow, reduces its moisture content by evaporation and lifts the drier coal into suspension, some to pass to the mill product outlet and the larger particles to return to the table for further size reduction. The upward flow of air is reduced from this 250-300°C range to about 70°C at the rim of the table, the same temperature as at the mill product outlet. Because of this change in moisture, actual roll and table segment wear takes place at a moisture level which is certainly less than the as received value but not in the dried state. The resulting wear rate will lie between the two values and the effect of additional moisture will accentuate this. From the equations linking roll wear with abrasiveness at different moisture levels the effect of moisture is significant when excess is present but not otherwise which is what would be expected.
J. Sligw / Int. J. Miner. Process. 44-45 (1996) 569-581
574
Table 3 Roll wear rates at varying mill capacity (as received coal) Feed rate (kg/hr)
Bay
Cal
con
Muj
Tar
Wal
300 400 500 600
7.22 5.27 4.98 4.37
4.90 3.68 3.03 1.60
4.97 4.72 4.76 8.52
8.26 6.97 3.86 3.14
9.42 6.35 6.18 5.94
13.82 11.33 10.20 10.56
2.5. Feed rate Table 3 sets out the wear rate arising from different coal feed rates with the mill output as close to 70% minus 75 microns as possible. This shows a linear relationship between wear and feed rate. An increase in feed rate leads to a decrease in wear as grams of material per kilogram of coal feed. This can be represented by an equation of the form:
War= -0.0023F+2.15
(r=72%)
where F = feed rate
(Note CAL 600 run neglected as mill drying capacity was exceeded). 2.6. Interactions With the exception of coal abrasiveness none of the single factors correlated well with wear. However, some combinations of these fared better. If the as received moisture content and free silica are combined to form a single factor then the correlation improves significantly over the individual correlations to: War = O.O80(Mar X SiO,) + 1.99
( r = 78%)
and W, = O.O3(M, X SiO,) + 3.48
(r = 68%)
A further possibility lay in AI and HGI forming a combined factor. This gave an equation which was not significant: War = O.O008l(AI x HGI) + 3.49
(r = 46%)
3. Pulverizer effects The prime factor in the effect of mill mechanics on component wear is the roll grinding pressure followed by the rate of recirculation within the mill which is controlled by the vane setting in the cyclone at the mill outlet. This induces various levels of recirculation of oversize coal in the mill. About 5% of the feed coal is reduced
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
575
Table4 Roll wear rates at different recirculation rate (as received coal) Bay 5.68
Cal 4.76
Con 5.10
Muj
0
10.42
Tar 11.87
Wal _
7.5 IS 22.5
5.30 5.92 7.28
6.3 1 5.93 7.96
3.17 4.20 6.74
13.32 11.86 11.06
8.93 11.22 10.63
0 -
Recirculation vane angle (“)
to product sized material per pass under the roll (Sligar, 1976). A high level of recirculation is therefore necessary in this type of mill to achieve reasonable output of desired product. 3.1. Roll grinding pressure From Table 1 a correlation between wear of mill components and roll grinding pressure was developed: War = 1.24 GP + 0.77
( r = 82%)
where GP = roll grinding pressure (MPa) 3.2. Recirculation TablIe 4 shows the roll wear rate as a function of recirculation within the mill at 60% load. Recirculation tests were carried out at this reduced rate because the range of vane angle for efficient mill performance near full load is very limited. Taking all results a correlation was obtained: War = 0.0059R + 0.094
(r = 33%)
where R = vane angle (“) was obtained. This was repeated using only vane angles 7.5 to 22.5 as the zero vane angle would never be used in practice. This gave: War = 0.019R + 0.76
(r = 56%)
This was repeated with vane angles 15, and 22.5 giving a similar correlation: War=0.03R+0.58
(r=51%)
Attempts were made to link classifier vane angle with real recirculation rate within the mil’l. This proved to be extremely difficult and also appeared to be specific to each type and capacity of mill. Classifier vane angle is adjustable and is usually set as part of the commissioning procedure with a mill. At full load other mill performance criteria limit the range of vane angle adjustment severely.
J. Sligar/Int.
576 Table 5 Mill air/coal
J. Miner. Process. 44-45 (1996) 569-581
ratio tests
Air: coal ratio
Classifier vane position (“)
3.0: 3.0: 3.0: 3.0:
1 I I I
0 7.5 15.0 22.5
2.5 : 1 2.5: 1 2.5: 1 2.5: 1
0 7.5 15.0 22.5
3.3. Air/coal
ratio
Tests were carried out on one coal using two air/coal ratios of 2.5 : 1 and 3 : 1 at 60% mill capacity. The results are shown in Table 5. From these results it can be seen that the air/coal ratio in combination with classifier vane angle has a significant effect on mill component wear. This occurs because at the lower air/coal ratio the upward velocity in the air gap surrounding the rotating table is reduced. This, in turn, reduces the particle size of coal or mineral matter which can be lifted into suspension. This effectively results in a bed of finer material as the coarser material cannot be lifted into suspension until it is ground to a smaller particle size. The effect is magnified with higher density minerals such as quartz which would result in the increased proportion of quartz on the mill table under reduced air flow. There were insufficient data collected to separate the vane angle and air/coal ratio effects and predict a relationship. 3.4. Interactions All the above factors, whether coal properties or mill adjustments, separately show varying degrees of correlation. Some combinations of these show superior correlations. Considering the basic approach of trying to concentrate the coal effects in one factor or area and the mill effects in another. The most important factor for coal is coal abrasiveness and for the mill is roll grinding pressure which give a very good combined correlation. Using values in Table 6 the equation representing these results is of the form: War = O.O17(AI x GP) + 1.8
(P = 97%)
Previous work on full scale mills (Sligar, 1986) showed a linear relationship between mill component wear and coal abrasiveness. The grinding pressure effect would presumably have been hidden in the relevant equation constant as the full scale mills would have been operating somewhere near their optimum grinding pressure and level of recirculation at the time.
J. Sligar/Inr. Table 6 Determination Coal
Bay
Cal
Con
of optimum
grinding
Moisture (W)
J. Miner. Process. 44-45 (1996) 569-581
pressure AI
GP (MPa)
P
54 55 _
41 49 -
4 3.5 _
8.66 10.1 _
4.0 4.1 -
80 _ -
7 -
2.2 2 2
7.3 6.2 7.5
3.6 2.2 2.7
82 82 -
25 66 _
2 2.02 2.05
7.69 7.64 7.65
4 4.77 7.37
HGI
4.0 (ad) 6.3 car) 9.3 (+ 3%) 9.8 (ad) 12.5 car) 15.5(+3%)
I .2 (ad) 3.6 (ar) 6.6 (+ 3%)
577
(kWh/ton)
W (pm/ton)
Muj
18.0 (ad) 29.1 car) 32.1 (+3%)
50 60 _
29 68 -
7 5.4 3.7
12.59 11.2 8.6
8.8 8.0 4.8
Tar
5.6 (ad) 7.1 car) 10.1(+3%)
50 47 -
45 44 -
3.8 6.5 4.9
10.3 11.2 11.3
6.5 7.0 6.72
The other factors obviously provide a minor correction to this basic equation. As an alternative using the other coal properties available the best of these was a combination of moisture and free silica. Thus the following correlation was possible where abrasiveness was not measured and the other factors were: War = O.O12(Mx In addition,
X
an equation
SiO,)GP
+ 2.15
was developed
(r = 72%) linking
grinding
pressure with HGI:
1 GP = (0.0091 HGI - 0.265) Com.bining
(‘=
94%)
this with the above, correlations 0.017AI
’[
War =
(0.0091 HGI - 0.265)
or:
+ 1’8
of the form were developed:
1
O.Ol(Mar X SiO,)
4. Mill wearing
component
alloys
As part of this program it was possible to compare a limited range of metal alloys for wear resistance. The pilot scale mill has a number of table segments and these were manufactured out of different alloys and subjected to identical coal wear conditions.
J. Sligar/lnt.
578
J. Miner. Process. 44-45 (1996) 569-581
Table 7 Table segment alloy properties Element (%o)
NiCrl
Cr2
CrMol5
Fe C Mn Si S P Cr Ni Mb CU Al V
balance 3.30 0.54 0.5 1 0.015 0.029 2.13 3.73 0.01 0.02 0.01 -
balance 2.5 1.3 0.8 0.02 0.03 28.0 0.05 1.30 0.03 -
balance 2.77 0.24 1.2 0.02 0.03 13.9 0.05 2.43 0.05 0.01
3
balance 3.36 0.29 1.09 0.02 0.03 19.9 0.07 1.10 0.03 0.085 -
balance 2.0 0.17 1.5 0.22 0.03 4.5 0.07 4.0 0.03 0.09 4.0
Hardness(BHN)
627
512
682
712
461
_
CrMo20
1
CrMoV 5 5 5
While the performance of individual alloys with specific coals could not easily be separated the relative wear of the alloys after being treated with all coals could be determined. 4.1. Table segment wear 4.1 .I. Alloy selection
A range of alloys was chosen which included the nominal industry standard, Nihard 1 (AS2027- 19851, a number of more controversial high chrome alloys and an experimental alloy. The high chrome alloys were chosen because of their anomalous results in industrial coal pulverizing mills. The alloys, together with their properties are set out in Table 7. 4.1.2. Sample wear
At the start and conclusion of the test the sample table segments were weighed and a weight loss per ton of coal throughput determined. The results are set out in Table 8. As a first comparison the wear results were compared with the surface hardness (BHN) of
Table 8 Table segment wear Alloy
Weight loss (g/t
NiCrl-550 Cr27 CrMol5 3 GM020 CrMoV 5 5 5
0.37 0.17 0.21 0.19 0.42
coal)
J. Sligar / In!. J. Miner. Process. 44-45 (1996) 569-581
579
the original alloys. This suggested a linear relationship between alloy hardness and resulting wear rate. A linear regression of the results suggested an equation of the form: Wau = - 0.000493 BHN + 0.57
( I = 90%)
BHN = hardness (BHN) A further comparison was made with the alloy constituents and a relation was found with chromium. A further linear analysis was made resulting in an equation of the form: War = - 0.0099 Cr + 0.406
( r = 92%)
Cr = chromium % in alloy Using the relationship BHN X Cr: War= -O.O000174(BHNX
Cr) +0.42
(r= 96%)
From these results it would appear that higher surface hardness alloys are more wear resistant in a relatively predictable manner. This obviously depends on the rate of change :in hardness with depth in the sample. Alloys with higher proportions of chromium can be metallurgically treated to be harder and hence provide superior wear performance. 4.2. Roll wear All the tests in Sections 2 and 3 were for roll wear rather than table segment wear as in Section 4. This was limited to a replaceable roll wearing ammlus of a proprietary alloy “Welten 360C” with a Brinnell Hardness of 405 BHN. This is not as hard as the conventional Nihard roll cast material and would be expected to wear at a greater rate as suggested from the equation developed for predicting wear from table segment material BHN. 4.3. Industrial mill perjtormance The results of this work on specific alloys is of great interest. However, it must be remembered that the test samples were small and capable of very precise metallurgical treatment to develop desired properties. The relative depth to which consistent high hardness exists in the samples would be greater than in industrial mills. Industrial mills require castings of up to 9 tons for mill rolls and up to 100 kg for table segments. These components cannot be held to such fine metallurgical treatment tolerances over the whole casting. Because of the above, industrial scale wear levels will not be as good as pilot scale results. In fact the higher chromium alloys have had a clouded introduction to the power industry, Rolls cast by some manufacturers have had an enviable reputation for low wear rates while rolls manufactured in the same alloy provided by other manufacturers have performed poorly. It should be noted that as an alloy becomes harder there is a greater tendency for brittle failure. The correlation of wear vs. hardness for small samples obviously has an upper limit where brittle failure becomes the dominant mode of failure rather than the
580
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
component wearing out. Brittle failure implies unplanned forced outage of the mill rather than a planned outage resulting from steady wear. A balance must be maintained between metallurgical factors to minimize forced outages but maximize life. From the results of this project a relatively reliable indication of likely wear in roll and table segments due to material used can be obtained from the mean hardness (BHN) of the wearing layer of material on the roll or table segment and the Chromium content of the alloy.
5. Mill wear prediction
A broad range of experimental results have been obtained linking pilot mill component wear with a number of variables. Correlation coefficients for every relationship give an indication of their reliability. Most are linear relationships where the slope of the line provides a measure of relative importance. The most effective relationships were found to be a function of coal properties and of machine adjustments. Interactions between these were not additive but multiplicative. From previous work with mills with fixed settings the wear rate was found to be proportional to the abrasion index of the coal with the mill at full load and near optimum grinding pressure. The present project is a logical extension of this leading to a predictive tool using coal properties and mill adjustments and including the properties of the component wearing alloys. An equation of the following form is appropriate: war = D(coa1) X
(mW1LWw)l
[f( std. alloy)]
I[
In addition the relationship linking grinding pressure with coal HGI may be used: 0.017 AI war=
0.0091 HGI - 0.265
or war=
O.O12(Mar X SiO,) 0.0091 HGI - 0.265
0.000049 BHN + 0.57
+ 1.8
0.55
I
- 0.000049 BHN + 0.57
+ 2.15 I[
0.55
1
These relationships may be used together to get separate estimates of likely wear depending upon what basic data is available on the coal and on the grinding machine. Two other factors were investigated at mill part load. This was necessary because one was the feed rate to the mill and was obviously at a range of loads. The other was the degree of recirculation in the mill. At full load the possible adjustment of this in an industrial mill is very limited, more flexibility being available at lower load. These two factors cannot be used directly in the prediction of wear as they were not measured at full load. However, if a mill is to run at part load for a considerable period they can be used to predict variation in wear with load and to determine the best vane settings for recirculation providing the full scale mill has the necessary flexibility.
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
581
In this case: W,ar = -0.0023F
+ 2.15
where F = feed in kg/h
or war = 1.15c + 2.15 where C = mill capacity ratio In addition, where major vane angle changes can be made: war = 0.0059R + 0.94 where R = vane angle (“) The feed rate function would be expected to be common for similar mills. The vane angle, however, is a much more variable factor, depending on the internal mechanics of the particular cyclone used within the mill body.
Referertces Sligar J., 1976. Simulation of a medium speed mill. Ph.D. thesis, University of Newcastle. Sligar J., 1986. Mill component wear rising from size reduction of coal. In: World Congress Particle Technology (6th European Symposium Comminution), Nuremberg, Part II, pp. 109- 119.
Int. J.
Miner. Process.44-45 ( 1996)569-58 1
Co:mponent wear in vertical spindle mills grinding coal J. Sligar ’ Pacific Power, Sydney, Australia
Abstract This paper reports on the contribution of a number of material coal properties and machine operating parameters to the wear of vertical spindle mill components. This assists in the evaluation of mills and coals for particular duties and enables prediction of the likely changes in maintenance regime if these variables are changed. Coal abrasiveness was found to be the most critical coal property with moisture next in importance. Feed rate was of importance with some interesting interactive affects. Roll grinding pressure was by far the most important machine variable with recirculation and air/coal ratio of lesser importance. Alloys used were characterized by Brinnell Hardness Number (BHN) and chromium alloy content.
1. Introduction
A paper on pulverizing mill component wear arising from size reduction of coal was presented at the World Congress on Particle Technology in Nuremburg in 1986. This surveyed the majority of coal fired power stations in Australia in a major industrial project on mill wear and the contribution of coal properties to this. This project found that the coal property giving the best initial indication of likely component wear in high speed, medium speed, or low speed mills was the coal abrasiveness as measured by BS1016:19. This project accepted that mills were operating near their optimum and no attempt was made to study potential mill variables, especially as all the mills were part of productive plants with economic and financial commitments. As an extension of this project a further study was undertaken, exploring mill variables and further coal
’Prerent address: Sligar and AssociatesPty Ltd, 10 Bond St., Mosman, 2088 N.S.W., Australia. 0301-7!~16/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved SSDI 0301-7516(95)00066-6
570
J. Sligar/Int.
J. Miner. Process. 44-45 (1996) 569-581
properties identified in the first project. Full industrial plant could not be subjected to the range of variables economically so a pilot scale medium speed vertical spindle mill was used, being the smallest version of one of the more frequently used mills in the industrial test. The aim was to better define the factors concerned with wear and the relative contribution of those associated with the milling machine and those concerned with coal properties. It was further decided to use a series of coals as a subset of the previous project but with as broad a range of likely contributing properties as possible. The operation of coal pulverizers in power stations and other applications is always a compromise between operating for maximum output and minimum maintenance cost. This project was thought to provide numerical input to this problem for effective decision making. 1 .I. Coal sample selection
In order to address the project objectives in as economic manner as possible, six (6) Australian domestic coals were chosen for comprehensive pilot scale mill testing. The coals were chosen as far as possible to satisfy the following criteria: - the coals had been part of the previous project * the coals covered a range of coal moisture levels from 3.6 to 29.1%; * the coals covered a range of Hardgrove Grindability Indices from 40 to 82; - the coals covered a range of Abrasion Indices from 7 to 66; I .2. Pilot scale test program The milling test work was conducted using a pilot scale mill, ICAL/EVT Model No. RP153X bowl mill with a 400 mm bowl diameter, three (3) rolls and an integral twin cone classifier. The procedures used in this project were developed as part of a separate research project. 1.3. Test program A number of mill operating variables were examined: material feed rate roll grinding pressure air/fuel ratio mill internal recircuation rate A range of alloys commonly used together with a range of aloys believed to have potential were tested in the table segments. For every test coal the mill was run at a range of gradually increasing roll grinding pressures from the minimum to the maximum available with nominal full load coal feed rate. This enabled identification of the first grinding pressure which produced 70% of product material less than 75 micron, which is the nominal standard for the industry. 9 * -
J. Miner. Process. 44-4.5 (1996) 569-581
.I. Sligar/lnt.
571
Tests were then carried out with a range of coal feed rates from 60 to 120% of capacity. In a limited number of cases air/fuel ratio and mill recirculation rate were also varied. Mois,ture had been previously identified as the coal property with anomalous effects on abrasiveness of coal, possibly contributing some corrosive component to the abrasive effect. Moisture was therefore tested at three levels, air dried, as received and as received plus 3% added moisture. The two main wearing components in a medium speed mill are the grinding rolls and the rotating table segments. Wear of the these components is the primary consideration for mill maintenance. The roll and table each have a finite thickness of wear resistant material which when exhausted requires the mill to be taken out of service and the elements replaced. It would be expected that the rate of wear of these components would be a function of the properties of the coal being ground, the settings and operation of the mill and of the met.al alloy used in the wearing components. These will be dealt with in turn in Sections 2, 3 and 4.
2. Coal property effects 2.1. Coal abrasiveness From previous work (Sligar, 1986) the abrasiveness of coal (BS 103819) was found to be the best overall measure of the coal characteristic causing wear in the three basic types of mill. Prediction equations using this factor were developed for a preliminary estimate of likely wear. It would be expected that the major contribution to wear prediction would be from this coal property together with a possible correction for moisture. As a first estimate Table 1 sets out the relevant variables. All these are based on comparing pilot mill tests using coal with as received moisture, operating at full load
Table 1 Roll wear rates at mill full load
Wear (g/ton) Mar Free SiO,(%) HGI AI hg/kg) GP AIXGP AIXHGI Marx SilD,
Bay
Cal
Con
Muj
Tar
Wal
4.1 6.3 6.2 55 49 3.5 172 2695 39.1
2.2 12.5 1.2 80 7 2.0 14 560 15.0
4.75 3.6 6.1 82 66 2.0 132 5412 22.0
7.2 29.1 2.2 60 68 5.4 367 4080 64.0
7.0 7.1 12.7 47 44 6.5 286 2068 90.2
10.9 8.7 8.3 49 81 6.0 486 3969 72.2
J. Sligar/Int.
572
and delivering micron. Comparing form:
pulverized
J. Miner. Process. 44-45 (1996) 569-581
coal with standard product
coal Abrasive
particle size of 70% less than 75
Index and mill roll wear the relationship
War = 0.092 AI + 1.2
is linear of the
( r = 79%)
War = wear rate (g/ton) AI = Abrasiveness
(BS1038:19)
2.2. Free silica Free silica had been proposed on a number of occasions as the major coal property factor contributing to mill component wear. Values of free silica for every test coal are shown in Table 1. A linear regression of these values indicated a linear equation with the following constants: War = 0.36 SiO, + 3.8
( r = 49%)
where SiO, = free silica content
(%)
The low correlation indicates that free silica is not a good measure. This conclusion is supported by previous work (Sligar, 1986) which showed that free silica alone was a poor measure but that a combination of silica content and particle size distribution gave a better correlation. 2.3. Hardgrove Grindability Hardgrove Grindability had been put forward on a number of occasions as a measure of mill component wear. Values for HGI and wear were extracted from Table 1 and a linear regression carried out. This resulted in the following relation: War = -0.138HGI
(r = 69%)
+ 14.58
Again this is a relatively poor correlation indicating the possibility of a correlation but no more than this. This agrees with previous project of Sligar (1986). 2.4. Moisture For as received
coal the correlation
War = 0.04Mar
+ 5.53
of wear with moisture
from data in Table
1 is:
(r = 14%)
where Mar = as received moisture
%
The two cases of air dried coal and as received
coal plus 3% moisture
are shown in
J. Sligar/lnr.
J. Miner. Process. 44-45 (1996) 569-581
573
Table 2 Roll wear rates with varying moisture
Bay
Cal
Con
Muj
Tar
Wal
Mad AI (mg/kg.) GP AIXGP Wear (g/ton) Mad X SiO,
4.0 41 4.05 166 3.98 24.8
9.8 I 2.2 15.4 3.56 11.8
1.2 25 2.0 50 4.0 1.3
18 29 7.0 203 8.84 39.6
5.6 45 3.8 171 6.5 71.1
2.4 45 _ _ _ -
Mar + 3% AI (mg/kg) GP AIXGP Wear (g/ton) M, X SiO,
9.3 _ -
15.5 2.0 14 2.74 18.6
6.6 2.05 51 7.37 40.3
32.1 3.74 102 4.8 70.6
10.1 6.2 360 8.0 128.3
11.7 _
Table 2. These relationships dried coal
of mill roll wear rate are similarly
Wad = 0.052 Mad + 3.85
linear and of the for air
( r = 34%)
where Mad = air dried moisture For as received
%
coal with 3% extra moisture
W! = 0.13 M, + 2.27
(using as received
Abrasiveness):
( r = 84%)
where WZ = extra moisture case From previous temperature measurements in the mill (Sligar, 1976) and analysis of coal in various parts of the mill most of the surface moisture is removed in the fast fluidized bed surrounding the rotating table. As received coal passes down the pulveriser feed chute on to the rotating table. It is mixed with partly dried, partly ground coal on the table and passes under the roll for a first pass at size reduction. It then passes over the rim of the table into the fast fluidized bed surrounding this. Hot primary air at 250-300°C passes upward through the overflow, reduces its moisture content by evaporation and lifts the drier coal into suspension, some to pass to the mill product outlet and the larger particles to return to the table for further size reduction. The upward flow of air is reduced from this 250-300°C range to about 70°C at the rim of the table, the same temperature as at the mill product outlet. Because of this change in moisture, actual roll and table segment wear takes place at a moisture level which is certainly less than the as received value but not in the dried state. The resulting wear rate will lie between the two values and the effect of additional moisture will accentuate this. From the equations linking roll wear with abrasiveness at different moisture levels the effect of moisture is significant when excess is present but not otherwise which is what would be expected.
J. Sligw / Int. J. Miner. Process. 44-45 (1996) 569-581
574
Table 3 Roll wear rates at varying mill capacity (as received coal) Feed rate (kg/hr)
Bay
Cal
con
Muj
Tar
Wal
300 400 500 600
7.22 5.27 4.98 4.37
4.90 3.68 3.03 1.60
4.97 4.72 4.76 8.52
8.26 6.97 3.86 3.14
9.42 6.35 6.18 5.94
13.82 11.33 10.20 10.56
2.5. Feed rate Table 3 sets out the wear rate arising from different coal feed rates with the mill output as close to 70% minus 75 microns as possible. This shows a linear relationship between wear and feed rate. An increase in feed rate leads to a decrease in wear as grams of material per kilogram of coal feed. This can be represented by an equation of the form:
War= -0.0023F+2.15
(r=72%)
where F = feed rate
(Note CAL 600 run neglected as mill drying capacity was exceeded). 2.6. Interactions With the exception of coal abrasiveness none of the single factors correlated well with wear. However, some combinations of these fared better. If the as received moisture content and free silica are combined to form a single factor then the correlation improves significantly over the individual correlations to: War = O.O80(Mar X SiO,) + 1.99
( r = 78%)
and W, = O.O3(M, X SiO,) + 3.48
(r = 68%)
A further possibility lay in AI and HGI forming a combined factor. This gave an equation which was not significant: War = O.O008l(AI x HGI) + 3.49
(r = 46%)
3. Pulverizer effects The prime factor in the effect of mill mechanics on component wear is the roll grinding pressure followed by the rate of recirculation within the mill which is controlled by the vane setting in the cyclone at the mill outlet. This induces various levels of recirculation of oversize coal in the mill. About 5% of the feed coal is reduced
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
575
Table4 Roll wear rates at different recirculation rate (as received coal) Bay 5.68
Cal 4.76
Con 5.10
Muj
0
10.42
Tar 11.87
Wal _
7.5 IS 22.5
5.30 5.92 7.28
6.3 1 5.93 7.96
3.17 4.20 6.74
13.32 11.86 11.06
8.93 11.22 10.63
0 -
Recirculation vane angle (“)
to product sized material per pass under the roll (Sligar, 1976). A high level of recirculation is therefore necessary in this type of mill to achieve reasonable output of desired product. 3.1. Roll grinding pressure From Table 1 a correlation between wear of mill components and roll grinding pressure was developed: War = 1.24 GP + 0.77
( r = 82%)
where GP = roll grinding pressure (MPa) 3.2. Recirculation TablIe 4 shows the roll wear rate as a function of recirculation within the mill at 60% load. Recirculation tests were carried out at this reduced rate because the range of vane angle for efficient mill performance near full load is very limited. Taking all results a correlation was obtained: War = 0.0059R + 0.094
(r = 33%)
where R = vane angle (“) was obtained. This was repeated using only vane angles 7.5 to 22.5 as the zero vane angle would never be used in practice. This gave: War = 0.019R + 0.76
(r = 56%)
This was repeated with vane angles 15, and 22.5 giving a similar correlation: War=0.03R+0.58
(r=51%)
Attempts were made to link classifier vane angle with real recirculation rate within the mil’l. This proved to be extremely difficult and also appeared to be specific to each type and capacity of mill. Classifier vane angle is adjustable and is usually set as part of the commissioning procedure with a mill. At full load other mill performance criteria limit the range of vane angle adjustment severely.
J. Sligar/Int.
576 Table 5 Mill air/coal
J. Miner. Process. 44-45 (1996) 569-581
ratio tests
Air: coal ratio
Classifier vane position (“)
3.0: 3.0: 3.0: 3.0:
1 I I I
0 7.5 15.0 22.5
2.5 : 1 2.5: 1 2.5: 1 2.5: 1
0 7.5 15.0 22.5
3.3. Air/coal
ratio
Tests were carried out on one coal using two air/coal ratios of 2.5 : 1 and 3 : 1 at 60% mill capacity. The results are shown in Table 5. From these results it can be seen that the air/coal ratio in combination with classifier vane angle has a significant effect on mill component wear. This occurs because at the lower air/coal ratio the upward velocity in the air gap surrounding the rotating table is reduced. This, in turn, reduces the particle size of coal or mineral matter which can be lifted into suspension. This effectively results in a bed of finer material as the coarser material cannot be lifted into suspension until it is ground to a smaller particle size. The effect is magnified with higher density minerals such as quartz which would result in the increased proportion of quartz on the mill table under reduced air flow. There were insufficient data collected to separate the vane angle and air/coal ratio effects and predict a relationship. 3.4. Interactions All the above factors, whether coal properties or mill adjustments, separately show varying degrees of correlation. Some combinations of these show superior correlations. Considering the basic approach of trying to concentrate the coal effects in one factor or area and the mill effects in another. The most important factor for coal is coal abrasiveness and for the mill is roll grinding pressure which give a very good combined correlation. Using values in Table 6 the equation representing these results is of the form: War = O.O17(AI x GP) + 1.8
(P = 97%)
Previous work on full scale mills (Sligar, 1986) showed a linear relationship between mill component wear and coal abrasiveness. The grinding pressure effect would presumably have been hidden in the relevant equation constant as the full scale mills would have been operating somewhere near their optimum grinding pressure and level of recirculation at the time.
J. Sligar/Inr. Table 6 Determination Coal
Bay
Cal
Con
of optimum
grinding
Moisture (W)
J. Miner. Process. 44-45 (1996) 569-581
pressure AI
GP (MPa)
P
54 55 _
41 49 -
4 3.5 _
8.66 10.1 _
4.0 4.1 -
80 _ -
7 -
2.2 2 2
7.3 6.2 7.5
3.6 2.2 2.7
82 82 -
25 66 _
2 2.02 2.05
7.69 7.64 7.65
4 4.77 7.37
HGI
4.0 (ad) 6.3 car) 9.3 (+ 3%) 9.8 (ad) 12.5 car) 15.5(+3%)
I .2 (ad) 3.6 (ar) 6.6 (+ 3%)
577
(kWh/ton)
W (pm/ton)
Muj
18.0 (ad) 29.1 car) 32.1 (+3%)
50 60 _
29 68 -
7 5.4 3.7
12.59 11.2 8.6
8.8 8.0 4.8
Tar
5.6 (ad) 7.1 car) 10.1(+3%)
50 47 -
45 44 -
3.8 6.5 4.9
10.3 11.2 11.3
6.5 7.0 6.72
The other factors obviously provide a minor correction to this basic equation. As an alternative using the other coal properties available the best of these was a combination of moisture and free silica. Thus the following correlation was possible where abrasiveness was not measured and the other factors were: War = O.O12(Mx In addition,
X
an equation
SiO,)GP
+ 2.15
was developed
(r = 72%) linking
grinding
pressure with HGI:
1 GP = (0.0091 HGI - 0.265) Com.bining
(‘=
94%)
this with the above, correlations 0.017AI
’[
War =
(0.0091 HGI - 0.265)
or:
+ 1’8
of the form were developed:
1
O.Ol(Mar X SiO,)
4. Mill wearing
component
alloys
As part of this program it was possible to compare a limited range of metal alloys for wear resistance. The pilot scale mill has a number of table segments and these were manufactured out of different alloys and subjected to identical coal wear conditions.
J. Sligar/lnt.
578
J. Miner. Process. 44-45 (1996) 569-581
Table 7 Table segment alloy properties Element (%o)
NiCrl
Cr2
CrMol5
Fe C Mn Si S P Cr Ni Mb CU Al V
balance 3.30 0.54 0.5 1 0.015 0.029 2.13 3.73 0.01 0.02 0.01 -
balance 2.5 1.3 0.8 0.02 0.03 28.0 0.05 1.30 0.03 -
balance 2.77 0.24 1.2 0.02 0.03 13.9 0.05 2.43 0.05 0.01
3
balance 3.36 0.29 1.09 0.02 0.03 19.9 0.07 1.10 0.03 0.085 -
balance 2.0 0.17 1.5 0.22 0.03 4.5 0.07 4.0 0.03 0.09 4.0
Hardness(BHN)
627
512
682
712
461
_
CrMo20
1
CrMoV 5 5 5
While the performance of individual alloys with specific coals could not easily be separated the relative wear of the alloys after being treated with all coals could be determined. 4.1. Table segment wear 4.1 .I. Alloy selection
A range of alloys was chosen which included the nominal industry standard, Nihard 1 (AS2027- 19851, a number of more controversial high chrome alloys and an experimental alloy. The high chrome alloys were chosen because of their anomalous results in industrial coal pulverizing mills. The alloys, together with their properties are set out in Table 7. 4.1.2. Sample wear
At the start and conclusion of the test the sample table segments were weighed and a weight loss per ton of coal throughput determined. The results are set out in Table 8. As a first comparison the wear results were compared with the surface hardness (BHN) of
Table 8 Table segment wear Alloy
Weight loss (g/t
NiCrl-550 Cr27 CrMol5 3 GM020 CrMoV 5 5 5
0.37 0.17 0.21 0.19 0.42
coal)
J. Sligar / In!. J. Miner. Process. 44-45 (1996) 569-581
579
the original alloys. This suggested a linear relationship between alloy hardness and resulting wear rate. A linear regression of the results suggested an equation of the form: Wau = - 0.000493 BHN + 0.57
( I = 90%)
BHN = hardness (BHN) A further comparison was made with the alloy constituents and a relation was found with chromium. A further linear analysis was made resulting in an equation of the form: War = - 0.0099 Cr + 0.406
( r = 92%)
Cr = chromium % in alloy Using the relationship BHN X Cr: War= -O.O000174(BHNX
Cr) +0.42
(r= 96%)
From these results it would appear that higher surface hardness alloys are more wear resistant in a relatively predictable manner. This obviously depends on the rate of change :in hardness with depth in the sample. Alloys with higher proportions of chromium can be metallurgically treated to be harder and hence provide superior wear performance. 4.2. Roll wear All the tests in Sections 2 and 3 were for roll wear rather than table segment wear as in Section 4. This was limited to a replaceable roll wearing ammlus of a proprietary alloy “Welten 360C” with a Brinnell Hardness of 405 BHN. This is not as hard as the conventional Nihard roll cast material and would be expected to wear at a greater rate as suggested from the equation developed for predicting wear from table segment material BHN. 4.3. Industrial mill perjtormance The results of this work on specific alloys is of great interest. However, it must be remembered that the test samples were small and capable of very precise metallurgical treatment to develop desired properties. The relative depth to which consistent high hardness exists in the samples would be greater than in industrial mills. Industrial mills require castings of up to 9 tons for mill rolls and up to 100 kg for table segments. These components cannot be held to such fine metallurgical treatment tolerances over the whole casting. Because of the above, industrial scale wear levels will not be as good as pilot scale results. In fact the higher chromium alloys have had a clouded introduction to the power industry, Rolls cast by some manufacturers have had an enviable reputation for low wear rates while rolls manufactured in the same alloy provided by other manufacturers have performed poorly. It should be noted that as an alloy becomes harder there is a greater tendency for brittle failure. The correlation of wear vs. hardness for small samples obviously has an upper limit where brittle failure becomes the dominant mode of failure rather than the
580
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
component wearing out. Brittle failure implies unplanned forced outage of the mill rather than a planned outage resulting from steady wear. A balance must be maintained between metallurgical factors to minimize forced outages but maximize life. From the results of this project a relatively reliable indication of likely wear in roll and table segments due to material used can be obtained from the mean hardness (BHN) of the wearing layer of material on the roll or table segment and the Chromium content of the alloy.
5. Mill wear prediction
A broad range of experimental results have been obtained linking pilot mill component wear with a number of variables. Correlation coefficients for every relationship give an indication of their reliability. Most are linear relationships where the slope of the line provides a measure of relative importance. The most effective relationships were found to be a function of coal properties and of machine adjustments. Interactions between these were not additive but multiplicative. From previous work with mills with fixed settings the wear rate was found to be proportional to the abrasion index of the coal with the mill at full load and near optimum grinding pressure. The present project is a logical extension of this leading to a predictive tool using coal properties and mill adjustments and including the properties of the component wearing alloys. An equation of the following form is appropriate: war = D(coa1) X
(mW1LWw)l
[f( std. alloy)]
I[
In addition the relationship linking grinding pressure with coal HGI may be used: 0.017 AI war=
0.0091 HGI - 0.265
or war=
O.O12(Mar X SiO,) 0.0091 HGI - 0.265
0.000049 BHN + 0.57
+ 1.8
0.55
I
- 0.000049 BHN + 0.57
+ 2.15 I[
0.55
1
These relationships may be used together to get separate estimates of likely wear depending upon what basic data is available on the coal and on the grinding machine. Two other factors were investigated at mill part load. This was necessary because one was the feed rate to the mill and was obviously at a range of loads. The other was the degree of recirculation in the mill. At full load the possible adjustment of this in an industrial mill is very limited, more flexibility being available at lower load. These two factors cannot be used directly in the prediction of wear as they were not measured at full load. However, if a mill is to run at part load for a considerable period they can be used to predict variation in wear with load and to determine the best vane settings for recirculation providing the full scale mill has the necessary flexibility.
J. Sligar/lnt.
J. Miner. Process. 44-45 (1996) 569-581
581
In this case: W,ar = -0.0023F
+ 2.15
where F = feed in kg/h
or war = 1.15c + 2.15 where C = mill capacity ratio In addition, where major vane angle changes can be made: war = 0.0059R + 0.94 where R = vane angle (“) The feed rate function would be expected to be common for similar mills. The vane angle, however, is a much more variable factor, depending on the internal mechanics of the particular cyclone used within the mill body.
Referertces Sligar J., 1976. Simulation of a medium speed mill. Ph.D. thesis, University of Newcastle. Sligar J., 1986. Mill component wear rising from size reduction of coal. In: World Congress Particle Technology (6th European Symposium Comminution), Nuremberg, Part II, pp. 109- 119.