GRINDABILITY OF CLINKER

C E BUCHANAN JR
ROAN INDUSTRIES INC

H J BAYLES
COPLAY CEMENT COMPANY

ABSTRACT

A simple grindability test is presented to quantify Portland Cement Clinker grindability. These data are correlated to KW's/ton based on grinding time and fineness. Alite crystal size and sharpness by Scanning Electron Microscope (SEM) and kiln burning process are compared to grindability.

INTRODUCTION

This work is a continuing project started in 1982 and previously reported in the ICMA proceedings published in 1983. The 1983 report showed that silica modulus best correlated with clinker grindability and that larger, well formed crystals were harder to grind than clinker showing a less prominent crystal structure.

TEST PROCEDURE

The grindability test is done by grinding 47 grams of clinker and 3 grams of gypsum (all previously crushed to minus 1/4 inch) in a Bleuler mill and then measuring Blaine fineness on the sample. Samples are ground for 7.5, 15, 30, 60, and 120 seconds, with the Blaine being run at a porosity of .45 for the 7.5 second grind, .47 for the 15 and 30 second grinds and .50 for the 60 and 120 second grinds.

The SEM work was done on a ETEC Autoscan located at Lehigh University, with the photographs being taken with a Polaroid camera using both positive and negative film. Two clinker pieces were selected at random, one large and one small. The largest crystal found at 1000 magnification was then selected and a picture made. Two additional photographs were then taken at 500x and 200x without movement.

The regression analyses were run on a Hewlett Packard Model 86 Computer using the multiple regression pack supplied by Hewlett Packard which uses the Cholesky square root procedure.

DISCUSSION

This paper is a follow up of work in progress reported on in 1993. It includes the original 33 clinkers, plus 22 additional ones, as well as SEM photographs of all 55. These clinkers represent a total of 36 plants mainly from the US, but with three from France, one from Spain and one from Mexico.

Upon receipt the clinker was passed through a jaw crusher (if this had not previously been done), and the five grinds made, as described above. The data from these grinds can be found in Table 1-RAW DATA. The grinding times listed are the actual time the mill was running, including the amount of time required for the mill to come up to speed. The Blaines were run at the proper porosity, and are reported in square meters per kilogram.

The average curve of grinding time vs. Blaine, and a curve of empirical data gathered from various sources, of KW's vs. Blaine were married so that a new curve was produced showing KW's vs. Blaine. To maintain the best linearity on a log scale, a transformation was made in that 4 was added to the KW's the log taken, and then the log of 4 subtracted. These data are shown in Table 2-REGRESSION COEFFICIENTS, along with the calculated Blaine values at the calculated exact KW's per ton. In addition, Figure 1 depicts one of the clinkers graphically, when compared to the easiest and hardest grinding ones.

Chemical analyses were run on a Philips 1270 X-ray analyzer, using fused beads, with the exception of free CaO and Ignition Loss which were run by normal wet methods. The mineral composition was calculated from these results. The data can be found in Table 3-CHEMICAL ANALYSIS and Table 4-ANALYSIS AND POTENTIAL COMPOSITION. In Table 5-ADDITIONAL DATA, a computation was made of aphthitalite [(3K20)(Na2O)(SO3)], calcium sulfate (CaS04), sodium substituted C3A, [(Na2O)(8CaO)(AI2O3)] and potassium substituted C2S [(K20(23CaO)(12SiO2)]. This was a calculation based on the assumption that the aphthitalite preferentially formed. If there was then excess sulfate calcium sulfate formed, and if there was excess alkali, the alkali substituted C3A or C2S formed.

In Table 5, the crystal size and shape ranking, the grindability ranking (based on Blaine or KW's) and the clinkering process is given.

The clinker size is ranked with the largest clinker being ranked lowest and is the average of several trials. It is estimated that it has a repeatability of 5. The clinker shape is ranked with the sharpest angles having the lowest numbers. It is estimated that it has a repeatability of 8. You can see the differences in the extremes, with the clinkers with large crystals being A307 and A332 while the small ones were A209, A393, and A453. The clinkers showing the sharpest angles were A261 and A311 while those with the least sharp edges were A228 and A351.

The Blaine rankings were based on the 54.4 KW's per ton Blaine value, with the lowest Blaine having the lowest rank. The KW's were based on the KW's required to grind from 50 to 200 sq. meters per kg surface area, with the highest KW's having the lowest rank. The clinkering process was divided into wet, dry, and pre-heater, with wet being assigned 1, dry being assigned 2, and pre-heater being assigned 3.

Table 6-shows the simple correlation between the major variables. Some interesting comments are:

1. Crystal size shows a negative correlation with calcium and a positive one with potassium, kiln type, sulfur and aphthitalite.

2. Kiln Type shows a positive correlation with sulfur and size, and a negative one with C3S.

3. KW's had a positive correlation with Silica and Kiln Type.

4. Blaine had a positive correlation with ignition loss.

In Table 7, the major items are shown from a step-wise correlation made using the Blaine obtained at 54.4 KW's per ton energy input as the dependent variable. This regression used all the clinkers in the study. It can be noted that the best variable was the ignition loss, with CaS04 being second, which is surprising, as less than half the clinkers contained any calculated CaS04. The next three variables entered were then, Si02, Shape and Size.

In Table 8 the same data is presented with the exception that the KW's needed to grind from 50 Blaine points to 200 Blaine points was calculated and this used as the dependent variable. The S. M was entered first with it then being deleted at step 4 due to the fact that silica correlated better at this point. The final model then ended up with CaS04, Si02, Shape and Size.

Due to the fact that a few clinkers had been exposed to the atmosphere and consequently had high ignition loss, which inflated the Blaine values, an arbitrary cut off point was made at one percent loss, with all observations above this (10) being deleted. The regressions were then run with 45 observations and these data are reported in Tables 9 and 10.

It is interesting to note the clinkers of high ignition loss, especially numbers A262, A350, A351, and A471. You can see the distinct hair like growth. This structure is possibly .1 to .5 microns in diameter and 1 to 3 microns long. In all probability, this accounts for the discrepancies between the Blaine and the Wagner when grinding outside clinker; the Blaine which measures air flow through a cell would count the surface area of the hydrated products while the Wagner does not differentiate between a .1 micron particle and a 7.4 micron one. They both contribute the same amount of surface area in this test on a weight basis whereas in actuality, .1 micron particles possess 50 times as much surface area as does a 5 micron particle containing the same weight.

If additional people desire to participate in this program, they are requested to send a 5 Kg sample of fresh uncrushed clinker to:

Mr. Jim Bayles
Coplay Cement Company
Nazareth, FA 18064
U. S. A.

CONCLUSIONS

1. A simple test is presented which can differentiate between different clinkers as to their grindability.

2. Ignition loss plays an important part in Blaine surface area. The most probable cause of the increase in Blaine is the hydration products which are definitely shown in the SEM Photographs.

3. Crystal size and shape definitely affect grindability. However, they are extremely hard to define as an assignable cause because of being able to see a very limited number.

4. There is a definite relationship between the excess sulfate available, above that required to satisfy the alkalies. Further work needs to be done on this to see if through some minor chemical changes clinker grindability can be improved.

FIG. 1. Relative Grindability of Sample A315

Table 1-RAW DATA

Table 2-REGRESSION COEFFICIENTS

Table 3-CEMICAL ANALYSIS

Table 4-ANALYSIS AND POTENTIAL COMPOSITION