Background of the invention W086/01755 (incorporated by reference) discloses a method for producing com- pacted graphite cast iron by using thermal analysis. A sample is taken from a bath of molten cast iron and this sample is permitted to solidify during 0.5 to 10 minutes.

The temperature is recorded simultaneously by two temperature-responsive means, one of which is arranged in the centre of the sample and the other in the immediate vicinity of the vessel wall. So-called cooling curves representing temperature of the iron sample as a function of time are recorded for each of the two temperature- responsive means. According to this document, it is then possible to determine the necessary amount of structure-modifying agents that must be added to the melt in order to obtain the desired microstructure. However, the cooling curves disclosed in this document are rather uniform and no variations are disclosed.

In order to accurately determine the graphite microstructure in cast iron specimens, conventional thermal analysis techniques, such as the one disclosed in WO 86/01755, require cooling curves where the first thermal arrest caused by austenite formation is distinctly separated from heat release caused by the onset of eutectic solidification. However, sometimes cooling curves are obtained without such a dis- tinctly separated thermal arrest. This is the case when the molten cast iron solidifies as eutectic or hyper-eutectic iron. Until now, it has not been possible to use cooling curves corresponding to near eutectic cast iron for monitoring graphite growth.

W093/20965 teaches that hyper-eutectic melts, where graphite nucleates before iron, do not provide a distinct plateau as the temperature crosses the liquidus line.

This is correctly attributed to the fact that graphite crystallisation has a lower latent heat release than iron.

By placing a small amount of low-carbon iron in the melt, the low-carbon iron par- tially dissolves locally while the sample is still molten. As the sample cools, the relatively pure iron surrounding the remaining solid portion of the nail begins to so- lidify because of its lower carbon equivalent (CE). Ultimately, as the sample vol- ume cools below the liquidus line, the remaining solid volume of nail, the sur- rounding low CE and the melted portion of the nail begin to solidify and"trigger" the solidification in an otherwise hyper-eutectic melt. The net result is that an aus- tenite arrest plateau appears on the cooling curve.

W093/20965 also states that the temperature difference (AT) between Tg and Tc max can be used to establish a correlation with the carbon equivalent. However, W093/20965 refers to hyper-eutectic melts, i. e. melts where the carbon equivalent is so high that the release of heat from the primary solidification does not coincide with the minima of the cooling curve (in the hatched region of Fig. 2 (a) of W093/20965). Accordingly, the primary solidification and the eutectic solidifica- tion are separate.

None of the above cited references discuss anything about carrying out thermal analysis on cast iron melts in order to determine the carbon equivalent of melts which are near-eutectic. Moreover, W093/20965 suggests measurements on melts having a carbon equivalent of up to 4.7%. It is disadvantageous to reach such high values because of graphite flotation and the degeneration of graphite shape. Like- wise, the method of W093/20965 is disadvantageous in that it requires an extra ad- dition of low-carbon steel or iron to the sampling vessel, and accordingly lead to higher costs and a more laborious method.

Summary of the invention The present invention provides a possibility to evaluate cooling curves recorded in near-eutectic cast iron melts. The curves are evaluated by determining the net amount of heat generated in the centre of the melt sample as a function of time. This information is then used to identify the part of the centrally recorded cooling curve that can be used as a basis for determining the amount of structure-modifying agent that must be added to produce compacted graphite cast iron, and/or spheroidal graphite cast iron, and to identify the part of said curve that is associated with for- mation of primary austenite.

Definitions The term"cooling curve"as utilise herein refers to graphs representing tempera- ture as a function of time, which graphs have been recorded in the manner disclosed in W086/01755, W092/06809.

The term"heat generation curve"as utilise herein relates to a graph showing the heat that is generated in a certain zone of a molten cast iron. For the purposes of the present invention, all heat generation curves herein are determined for a zone lo- cated in the centre of a sample of molten cast iron. This zone is generally referred to as the"A zone".

The term"sample vessel"as disclosed herein, refers to a small sample container which, when used for thermal analysis, is filled with a sample of molten metal. The temperature is then recorded during solidification in a suitable way. preferably the sample vessel is designed in the manner disclosed in W086/01755, W092/06809, W091/13176 (incorporated by reference) and W096/23206 (incorporated by refer- ence).

The term"sampling device"as disclosed herein, refers to a device comprising a sample vessel equipped with at least two temperature responsive means for thermal analysis, said means being intended to be immersed in the solidifying metal sample during analysis, and a means for filling the sample with molten metal. The sample vessel is preferably equipped with said sensors in the manner disclosed in W096/23206.

The term"structure-modifying agent"as disclosed herein, relates to compounds af- fecting the morphology of graphite present in the molten cast iron. Suitable com- pounds can be chosen from the group of magnesium and rare earth metals such as cerium, or mixtures of these compounds. The relationship between the concentra- tion of structure-modifying agents in molten cast irons and the graphite morphology of solidifie cast irons have already been discussed in the above cited documents W092/06809 and W086/01755.

The term"CGI"as utilise herein refers to compacted graphite cast iron.

The term"SGI"as utilise herein refers to spheroidal graphite cast iron.

Figures The invention is described with reference to the accompanying figures, in which: Figs. 1A-4A disclose cooling curves. A continuos line relates to the temperature in the centre of the melt and a dashed line relates to the temperature close to the wall of the sample vessel. Figs. 1B-4B show A zone heat generation curves correspond- ing to the cooling curves in figs. 1A-4A. Fig. 1 relates to normal hypo-eutectic cast iron and figs 2-3 show curves for near-eutectic cast irons comprising increased val- ues of the carbon equivalent. Fig. 4 discloses curves relating to hypo-eutectic cast iron.

Fig. 5 is a schematic presentation of an apparatus for controlling production of compacted graphite or spheroidal graphite cast iron according to the present inven- tion.

Fig. 6 is a schematic representation of a sample vessel in which the heat is approxi- mately uniformly transported in all directions. During measurements the vessel is filled with molten cast iron. This molten cast iron can be considered as a freezing sphere. The A zone relates to a sphere in the centre of the molten sample and the B zone relates to the molten iron surrounding the A zone. The radii ri and r2 relate to the mean radii of the A zone and the B zone, respectively.

Detailed description of the invention As already mentioned above, the present invention relates to an improved method for interpreting cooling curves in near-eutectic cast iron melts. One of the important aspects of a cooling curve is the maximum slope of the recalescence peak of the centrally recorded cooling curve. This point is referred to as a in figures 1A, 2A and 3A. In cooling curves corresponding to near-eutectic cast iron melts, the inflexion point (referred to in the figures as"tp") is located much closer to a and it can be very difficult to determine the slope correctly. Examples of such curves can be found in figure 2A and 3A. However, it has now turned out that the determination of the maximum slope at a can be considerably simplified if a heat generation curve corresponding to the central zone (the A zone) of the molten cast iron is calculated.

Such a heat generation curve renders it possible to determine the location oftp in the centrally recorded cooling curve and to check whether it affects the slope of the re- calescence peak or not.

The thermal balance of any uniform element can be described by the relation:

QstoredQgenerated'Qin*<out) where Qstored is the amount of heat stored by the heat capacity of the material, Qgenerated is the amount of heat generated by the volume of material, Qin is the heat transferred into the material from its surroundings and Qout is the heat transferred out of the material to its surroundings.

When carrying out the present invention, it is advantageous to use a sample vessel as disclosed in SE 9704411-9. In such a sample vessel, the heat transport in a sam- ple contained in the vessel is approximately the same in all directions. From now on the heat transport between the centre (Fig. 6, zone A) and the more peripheral (Fig. 6, zone B) part of a molten cast iron contained in a sample vessel is described.

As the A zone is situated in the centre of the molten cast iron sample, and as that zone can be considered to be a freezing sphere, no heat is transported into the zone <BR> <BR> <BR> and Q. n is therefore equal to zero. Appropriate substitutions in relation (1) above give the following equation: CpmA dTA/dt = QgellA where Cp is the heat capacity per unit mass, mA is the mass of the A zone, dTA/dt is the temperature change of the A zone per unit time, QgenA is the heat generated in zone A, ke is the effective heat transfer coefficient of the material, and (1/r,-I/r2)-' is the mean distance for heat transport. The radii ri and rz are both defined in fig. 6.

TA and TB are the temperatures in the A zone and B zone respectively.

From equation (2) we can isolate the heat generation term and calculate the average bulk heat generation in the A zone: QgenA = CpmAdTA/dt + 4#ke [(TA-TB)/(l/rl-l/r2)] (3)

In equation (3) all variables are constants except dTA/dt and (TA-T, 3). Accordingly, equation (3) can be simplified to: Qg = k, dTA/dt + k2(TA-Te)(4) where ksi and k2 are constants. Hence, a heat release curve can be calculated from a set of cooling curves recorded in the centre and the periphery of a sample of molten cast iron.

As already mentioned, these calculations are based on a situation where heat is uni- formly transported in all directions. The skilled person can of course also calculate other equations corresponding to other heat transport conditions.

As mentioned earlier, the carbon equivalent of a certain cast iron melt affects the appearance of its corresponding cooling curves and heat release curves. Such cool- ing curves and heat generation curves can, depending on the carbon equivalent, be classified into one of four model types: Type 1, which is disclosed in fig. 1, relates to hypo-eutectic cast iron. The cooling curve recorded in the centre of the sample is characterised by a first inflexion point associated with formation of austenite, followed by a local minimum. Interpretation of such curves does not present any particular difficulties and can be carried out in accordance with the principles outlined in W086/01755 and W092/06809. It should be noted that there is a peak (tp) on the zone A heat generation curve that corresponds to the inflexion point (and accordingly with the onset of austenite for- mation) on the centrally recorded cooling curve.

Type 2, which is disclosed in fig. 2, relates to near-eutectic cast iron and is charac- terised by a very flat local minimum on the cooling curve recorded in the centre of the sample. An inflexion point cannot be seen. However, it is much easier to find a peak corresponding to the austenite formation point tp on the heat generation curve.

In this case, it turns out that the austenite formation point tp is located in connection to the local minimum resulting in a very flat minimum. From figures 2A and 2B it can be seen that tp does not affect the slope at a.

Type 3, which is disclosed in fig. 3, also relates to near-eutectic cast iron, and is- characterised by a detectable inflexion point after the local minimum in the cooling curve recorded in the centre of the sample. After determining the heat generation curve it is easy to establish the position of the austenite formation point tp to a loca- tion immediately after the local minimum. However, in this situation the recales- cence, and accordingly a is affected by the austenite formation. In order to be able to disregard the effects of the"moved austenite formation point"and to find a cor- rect a located at t, (where ta > tp), the part of the centrally recorded cooling curve immediately adjacent to tp, and having time values t such that t-t,, are larger than a predetermined threshold value tv, are searched for a maximum value a ; of the first time derivative. This value (xl is then used as a correct a in the method. A suitable threshold value tv can easily be determined by the skilled person, but typical exam- ples of such values are 1-5 seconds.

Type 4, which is disclosed in fig. 4A, relates to hyper-eutectic cast iron. This type is characterised in that no evident inflexion points or wide minima can be seen. The heat generation curve of the A zone does not comprise any detectable peaks. The present invention is not applicable in this situation. lt is preferred to carry out the prediction method by using a computer-controlled system, especially when a large number of measurements must be carried out. In this case, the same kind of sampling device 22 that has been described above is used. Such a computer-controlled system is outlined in fig. 6. During the measure- ment of a particular sample the two temperature-responsive means 10,12 send sig- nals to a computer 14 comprising a ROM unit 16 and a RAM unit 15 in order to generate the cooling curves. The computer device 14 is coupled to a memory means

19 in which a program routine is stored. When, in the following, it is described that the computer operates, it is to be understood that the computer is directed by the program in memory 19. The memory 19 may be a ROM circuit or a hard disc. The program may be supplied to the memory 19 from a non-volatile memory such as a CD-ROM or a diskette, e. g. via a databus (not shown). The computer has access to calibration data in a ROM unit 16 and calculates the amount of structure-modifying agents that must be added to the melt. This amount is signalled to a means 18 for administrating structure-modifying agent to the melt 20 to be corrected, whereby the melt is supplied with an appropriate amount of such agents.