A STUDY OF THE THERMAL SHOCK RESISTANCE OF COKE-OVEN DINAS UNDER LABORATORY CONDITIONS UDC 666.762.2.017:536.495

E. K. Aksel'rod and A. I. Portnova

The thermal shock resistance (stability) of ceramic materials, characterizing their fracture resistance or resistance to strength reduction under the action of thermal stresses, depends on the characteristics of the material itself (porosity, strength, homogeneity, and the coefficient of linear thermal expansion), and also, on the operational factors (the size and shape of the products, heating and cooling rates, service temperature, the presence of mechanical loads, reactions with the gaseous atmosphere, the effect of dust, slag, etc.). Thus, the thermal shock resistance is not a physical property and it can not be unambiguously described by any quantitative coefficient or parameter that is valid under all conditions. Of practical value are only the comparative data evaluating, in one way or the other, the resistance of a specific product or specimen to thermal shocks under the given test conditions. The test procedure for evaluating the thermal shock resistance must consist of two successive or parallel stages: thermal cycling of the selected products or specimens under predetermined regimes (temperature, dwell, heating and cooling rates, temperature gradients in the volume of the material, etc.) for creating thermal stresses; and evaluating the stability of the objects against abrupt temperature gradients using direct (based on the extent of failure) or indirect (based on the change in any physical property) methods. In spite of the simplicity of the direct methods in conducting and in interpreting the obtained results, they are often unsuitable for the refractories lacking thermal shock resistance: for example, the dinas of magnesite refractories fail within 1-2 thermal cycles when tested according to GOST 7875--83. The indirect methods usually require more time-consuming experiments. However, their advantage lies in the fact that during the tests, as a rule, the specimens are not cycled up to fracture. This permits one to conduct multiple (repeat) tests on the same specimen. It is known that dinas brick has a considerably high thermal shock resistance at the temperatures exceeding 600~ However, at lower temperatures its thermal shock resistance rapidly falls because of the phase transformations (modifications) of silica leading to volumetric changes of the phases. In coke ovens dinas work mainly at temperatures exceeding 1000~ but at the moment of loading the coal charge, the temperature of the oven walls abruptly decreases (within a period of 5 min) up to 200-250~ [I]. A study of the thez~al shock resistance of dinas under the conditions close to the service conditions forms an important practical problem. In this paper we present the results of investigations on the thermal shock resistance of the dinas specimens having different values of porosity (P = 15-21%), density (Q = 2.342.40 g/cm3), and refractory clay (additive) content (~cl = 0-I0%). The specimens were cut out from the wall components of a batch manufactured at the Pervoural'sk Dinas Factory on a trial basis. In our studies we employed the direct method of evaluating the thermal shock resistance (based on the number of thermal cycles withstood by the specimen up to fracture in the medium of air) as well as the indirect method (based on the change in the resonance frequency of vibrations in the specimens during the thermal cycling process). The standard DIN 51068--80 [2] (slightly modified to suit the specific objectives of the present investigation) formed the basis of the method used to determine the thermal shock resistance according to the number of thermal cycles in air. According to this standard, the specimens measuring I15 x 65 x 65 mm (i/4th of the normal brick) are introduced into a furnace heated up to 950~ After holding for 45 min, they are taken out from the furnace and the faces measuring 115 x 65 mm are subjected to compressed air blasting (forced cooling) at a pressure of 0.1 MPa. Thereafter, the specimens are subjected to three-point bending under Ukrainian Scientific-Research No. ii, pp. 5-9, November, 1985.

Institute of Refractories.

0034-3102/85/1112-0587509.50

Translated from Ogneupory,

9 1986 Plenum Publishing Corporation

587

TABLE I. Values of the Parameters of Thermal Shock Resistance 21 fn/f~ and K Obtained at Different Densities of Dinas Brick and Averaged in the Porosity Ranges AP %

,, g/er~ 2,34

1 1

2,35 0

2,36

2,37 1

2,38

2 2 l 1

2,39 2,40

2 4 4 l 4

•P, %

fn/f~

K,therrna]

15--17 17--19 15--17 17--19 19--21 15--17 19--21 15--17 17--19 19--21 15--17 17--19 19--21 15-r-17 19--21 15--17 19--21 15--17 19--21 15--17 17--19 17--19 17--19 17--19 17--19 17--19

0,87

6,0

0~'90 0,94 0,86

14,0

cycles 4,5

0"~4 0,87 0,95 0]-82

l 1,9 10,0 16,0 5,7 9,2 23,0

--

9,3

0"~2

25,0__ 16,0 7,0 13,3

79 0"~0

0,87 0,91 0~1 0,85

16,0__ 6,5 7,0 4,0 11,4

a stress of 0.3 MPa applied on to the air blasted face. The thermal shock resistance K is determined on the basis of the thermal cycles withstood by the specimens up to fracture. In contrast to the procedure given in the standard DIN 51068--80, the test specimens were held in the furnace under unilateral (one-sided) heating conditions at a hot-face temperature of 950~ The temperature gradient in the 65 mm thick layer of the material amounted to 300-350~ which corresponds to the temperature regimes maintained during the service of dinas in coke ovens [i]. When blasting the hot face of the specimen, its temperature decreased by 200-250~ within a period of 5 min. This simulates the cooling process of the oven walls at the moment of loading the oven with the coal charge. Such a procedure permits one to evaluate the relative characteristic of shock resistance (stability) of dinas specimens exhibiting different properties to the action of sharp temperature gradients and mechanical loads. The test regimes used in this study are close to the actual service conditions under which the refractory lining is subjected to mechanical effects (pressure, vibrations, impacts, etc.) besides the thermal effects. The relationship existing between structure weakening of the specimen and the changes occurring in its mechanical properties during thermal cycling forms the basis for the indirect method of evaluation. The microcracks developing under thermal shock decrease the strength and the modulus of elasticity of the refractory and, thereby, affect its thermal shock resistance. The method involves measurement of the normal modulus of elasticity E or shear modulus G of the specimens before and after a predetermined number of thermal cycles in the given test-regimes. The ratio of the values of E or G after thermal shock to their initial values: En/Eo or Gn/Go (n is the number of thermal cycles) is considered as the index of thermal shock resistance [3]. The measurements were carried out* at room temperature using the apparatus developed earlier [4]. In view of the fact that the process of determining E and G reduces to the measurement of the frequency of flexural (transverse) (fflx) and torsional (ftr) vibrations of 2 the specimen, respectively, and since E ~ f~Ix and G ~ ftr, the thermal shock resistance can be evaluated by comparing the ratio of the squares of the corresponding frequencies. This *A. S. Bobkovaya and V. I. Pechenezhskii participated in the work. 588

TI x

x

x x

B

/,0

~

0

O0

x Q

o,s I

o,s 1,0

t

0,,8 o,8 l

x x

I

~

l

I

I

x

I

I

UoI

x

I

I

I

Fig. I. Dependence of the index of thermal 2 2 shock resistance fn/fo on the porosity P of the dinas specimens after 1 (a), 6 (b), 21 (c) thermal cycles. Weight content of clay in the specimens: 0 (o), 1 (• 2 ( 0 ) , 4 (A), and 10% (Q). Vertical segments indicate s c a t t e r i n the exper, imental data.

co I 0"61S

16

17

18

19

20 P,%

significantly improves the accuracy of measurement because the error in determining f2 amounts to approximately 1%, whereas, the calculated error in the values of E and G is equal to 14-16%. In view of the fact that the measurement of ftr is characterized by better reproducibility than that of fflx, for determining the thermal shock resistance we used the ratio of the squares of the respective frequencies of torsional vibrations after and before thermal cycling fn/fo. 2 = Thermal cycling was carried out in the following way. Specimens measuring 200 • 30 • 30 mm were arranged in the form of a wall in the chamber of a laboratory furnace that permits one to conduct programmed unilateral (one-sided) heating of the specimens and holding at temperatures up to 1400~ After heating up at a rate of 2-3~ and holding for i h, a temperature gradient close to that obtained under the service conditions is created in the dinas specimens. The external surface of the wall, showing a temperature of 950-I000~ is then subjected to forced cooling using compressed air and water spray so that its temperature is brought down to 200-250~ within a period of 5 min. Thereafter, the furnace door is closed, and within a period of 1 h the specimen temperature is increased again up to the previous value. The process of temperature variation of the specimens during thermal cycling is registered on the chart of a recording instrument. A study of the relationship between the thermal shock resistance of dinas and its important technological properties such as the density, porosity, and phase composition is quite complex. In view of the fact that these characteristics are, as a rule, related to each other, we encounter the problem of identifying the role of each of these characteristics in the behavior of the material during its thermal loading. For this purpose, it is necessary to have a possibility of successively varying one of the aforementioned properties without changing the other two properties. Unfortunately, the production technology of dinas bricks does not always permit one to obtain products having the characteristics within the present limits. Besides this, it is often found that a change in one of the properties involves a change in other properties (for example, increasing the clay content leads, as a rule, to an increase in the product density). Therefore, it is not always possible to separate out a fairly representative region of the variation of one of the characteristics of dinas bricks at fixed values of the other characteristics. These facts determined the ranges of the density and porosity values studied in this work. The results of investigations showed that the thermal shock resistance of dinas brick is not a function of its density within the rms scatter of the experimental data, viz., • 34% (for the method based on the standard DIN 51068-80) and • 11% (for the method based on the measurement of resonance frequencies). This conclusion is valid only for the ranges of variation of density, porosity, and the clay additive content investigated in this work (Table i).

589

TABLE 2. Porosity Variation in the Dinas Specimens during Thermal Cycling Product number %1%

P ~'%

n

i

20,0

1

20,6

0,6

16,7 19,2 18,4 16,6 19,8 20,6 15,7 15,8

I.

16,9

0,2

6

20,2

1,0

3 6 8 45 12 2

1 1 0 0 ! 0 1 90

6 6 21 21

19,3 16,9 20,9 21,6

0,9 0,3 1,1 1,0

21

17,0

" 1,3

21

16,5

0,7

E

i

7 II

8

P~,% (Prl~oPo),

/o 9

0

016

x

QO X

I

I

I

I

17

18

19

20

P, %

Fig. 2. Dependences of the index of thermal shock resistance K on the porosity P of the dinas specimens. Symbols are the same as in Fig. i. A study of the relationship between the porosity of dinas and its thermal shock resistance yielded the following results. During thermal cycling of the specimens with subsequent determination of the fn/fo = 2 parameter after one thermal cycle all the experimental points show a single dependence on porosity (Fig. la) independent of the density and the clay content (except the point for the specimen with acl = 10%) (Fig. la). This curve has a parabolic form with a minimum appearing in the region of porosity values P ~ 17-19%, and the branches of the parabola exceed the minimum value by 35-40% at an 11% rms scatter of the experimental data. Such a trend is also observed after six thermal cycles (Fig. ib) although the curve is slightly flattened and the scatter of points is increased. Finally, after 21 thermal cycles 2 2 (Fig. le) the fn/fo = F(P) curve separates out according to the ~cl values (which is noticeable even after 6 cycles), and the single parabolic relationship is converted into a number of linear relationships. Here, flattening of the parabola occurs due mainly to the decrease 2 2 in the parameter fn/fo in the region of P ffi 15-16%. The decrease in the resonance frequency of vibrations in the dinas specimens indicates structure weakening of the material. This occurs because of the thermal shock-induced structural discontinuities or because of the development of the already existing defects. Confirming the well-known fact that the strength of dinas increases with decreasing porosity, in the present study the maximum strength is exhibited by the specimens having P = 15-16%~ i.e., crack nucleation in these specimens requires the maximum energy. Apparently, thermal cycling of these specimens up to one or even six cycles does not lead to crack formation capable of causing significant structure weakening. Therefore, their thermal shock resistance 2 2 (characterized by the parameter fn/fo) is fairly high. When the porosity of the specimens exceeds 17%~ such cracks are developed immediately after one thermal cycle. However, they lead to a reduction in the thermal shock resistance only in the range P ~ 17-19%. At P > 19%, the growth of microcracks is inhibited owing to thermal stress relaxation in the pores and the concentration of microcracks increases~ thereby facilitating improved thermal shock resistance [5]. The effect of these microcracks on the resonance frequency of vibrations with the background of increased total porosity of the material is small and, therefore, at P > 19% the magnitude of fn/fo 2 a remains higher than that of P ~ 17-19%. This leads to the parabolic dependences shown in Fig. la, b.

590

On increasing the severity of the test regimes by increasing the number of thermal cycles, crack growth is intensified in the less porous specimens also and, consequently, in 2 2 the dinas specimens containing 15-16% porosity, the parameter fn/fo decreases and the parabolic relationship transforms into a linear relationship (see Fig. ic). In this case, the maximum thermal shock resistance is exhibited by the specimens containing 1 and 4% clay, and the minimum value is shown by the specimens made without clay additions. The given interpretation of the obtained results is confirmed by the porosity measurements conducted on the specimens before (Po) and after (Pn) thermal cycling (Table 2). When the initial porosity of dinas brick is less than 17%, one and six thermal cycles do not lead to an increase in the porosity (fissuring develops to a less extent), whereas, after 21 thermal cycles, there is a significant increase of Pn indicating development of cracks in the specimens. In the dinas specimens having Po ~ 18%, the porosity increase due to crack formation is noticeable immediately after one thermal cycle. The determination of the thermal shock resistance of dinas specimens according to the method based on the DIN 51068--80 standard also showed a parabolic dependence of the index of thermal shock resistance K on the porosity in the 16-21% range at all values of Q and acl (Fig. 2). In this case, the maximum increase in the branches (arms) of the parabola above its minimum value amounts to 300% at a • rms scatter of the experimental data. In view of the fact that the tests conducted according to the aforementioned method involve application of bending loads on the specimen after each cycle, the method must be very sensitive to the changes in the strength characteristics of the material. This circumstance and the fact that identical results are obtained in our studies on the thermal shock resistance of dinas specimens using two fundamentally different methods indicate the correctness of our interpretation of the experimental relationships and the reliability of the obtained results. It was previously mentioned [6] that the use of fairly strong materials is recommended for the situations involving moderate thermal shocks, and that under severe thermal loading conditions subjecting the product to destructive loads, the materials having a low initial strength are recommended. In view of the fact that the strength of the dinas specimens varies even within a relatively narrow range (15-21%) of porosity (by more than two times according to our data), the aforementioned concepts may be fully applied to dinas. This is confirmed by the data obtained in the present study: dinas having the maximum porosity (minimum strength) is thermal shock resistant under the most severe thermal cycling conditions chosen, whereas, dinas having a low initial porosity shows thermal shock resistance only under relatively mild test regimes. CONCLUSIONS Under the selected test regimes, no relationship is found between the thermal shock resistance of dinas brick and its density in the ranges of variation Q = 2.34-2.40 g/cm 3, P = 15-21%, and ~cl = 0-10%. The effect of the porosity of dinas on its thermal shock resistance is determined by the conditions of thermal loading of the specimens. The moderate test regimes (small number of thermal cycles, air cooling of specimens) lead to a parabolic dependence of thermal shock resistance on the porosity. In this case, the minimum thermal shock resistance is shown by the specimens having 17-19% porosity. Clay additions up to 4% do not have a significant effect on the thermal shock resistance of dinas; on the other hand, addition of 10% clay decreases it. Increasing the severity of the test conditions (increasing the number of thermal cycles, water-spray cooling of the specimens) alters the nature of the relationship between thermal shock resistance and porosity: the parabolic relationship changes into a linear relationship. The low-porosity (15-16%) specimens show a decreased thermal shock resistance because of the development of detrimental (destructive) cracks in them. The highest thermal shock resistance is exhibited by the dinas specimens having 19-21% porosity. Increasing the clay content in the specimens up to 4% helps in improving the thermal shock resistance. The identical nature of the porosity dependences of the thermal shock resistance of dinas observed during different methods indicates the reliability of the test results and 2 2 the applicability of the parameter fn/fo as an empirical criterion of the thermal shock resistance of refractory materials. 591

LITERATURE CITED i. 2.

3. 4. 5. 6.

V . T . Krivoshein, V. !. Aleshin, and V. D. Ignatov, Koks Khim., No. li D 1.7-20 (1976) o DIN 51068--80 (FRG). Testing Ceramic Raw Materials and Products. Part 2. Determination of the Resistance to Abrupt Temperature Changes Based on the Method of Fast Air Cooling [in German]. R. Iskander, Sprechsall Keram., Glas, Email, Silikate., 105, No. 7, 296 (1972). A . I . Kovalev and I. I. Vishnevskii, Zavod. Lab., No. 9, 1109-1111 (1959). R . D . Smith, H. V. Anderson, and R. E. Moore, Am. Ceram. Soc. Bull., 55, No. ii, 979982 (1976). K . K . Strelov, Structure and Properties of Refractories [in Russian], Metal!urgiya, Moscow (1982).

STRUCTURE AND CREEP OF MULLITE--CORUNDUM REFRACTORIES I. I. Vishnevskii, A. V. Kushchenko, L. Do Smirnova, and V. S. Shapovalov

UDC 666.762.11o017:620.!72.251~2

The modern concepts of the deformation mechanisms of heterophase refractory materials are primarily based on establishing an interrelationship between the structure and the mechanical properties~ Creep is one of the structure sensitive properties of the refractories that significantly depends on the amount and the distribution of the phases, the degree of coherence (connectivity) in the crystalline framework, and on the composition and the density of the binding agent. According to the published data [i, 2], sliding of the weakly bound (adhering) regions of the structural framework along the binder layer causes the high-temperature plastic deformation of most of the types of refractories produced according to the conventional chamotte technology~ In aluminoscilicate refractories the binder contains a glasscrystalline composition, and their creep is determined by the concentration and the degree of localization of the glass phase, the structure and the density of the binder, and its adhesive (interface) strength with the chamotte grains. In this paper we present the results of creep measurements of various mullite--corundum based refractories and analyze its magnitude as a function of the structural features of the material. Four types of refractories were studied: the industrial grades (trade marks) MKS of the Semiluksk Refractories Plant (TU 14-8-140--75); the experimental grades MKG with fused mullite produced by the same plant according to the technology developed by the Ukrainian Scientific-Research Institute of Refractories (UkrNllO); the laboratory grades MKV made at the UkrNllO based on the charge used at the Semiluksk plant for the MKV-72 products and fired at 1600~ and the MPS products [3].* Tables 1 and 2 show the composition and the properties of the specimens. All the materials under study have a mullite--corundum structure. The MKS fractories have higher porosity and greater aluminum oxide content in the form The glass phase content in the MKS, MKG and MKV refractory specimens is nearly the MPS specimens it is twice higher. Besides this, the MPS material contains The strength of the MKS specimens is considerably lower than that of the other refractories, and there is a significant scatter in their strength values.

and MKG reof corundum. the same; in cristobaiite. experimental

Creep was measured under axial compression in air in the temperature range 1300-1600~ and in the stress range 0.1-8 MPa using the furnaces operating with molybdenum disilicide heating elements. The measurements were carried out on cylindrical specimens measuring 36 to 50 mm in diameter and 50 mm in height. Their axis coincided with the pressing direction *The MKS and MKG specimens were prepared by T. A. Ansimova, pared by T. V. Ivashchenko. Ukrainian Scientific-Research No. Ii, pp. 9-13, November, 1985.

592

and the MPS specimens were pre-

Institute of Refractories.

0034-3102/85/1112-0592509.50

Translated from Ogneupory,

9 1986 Plenum Publishing Corporation