|
The experiment based on
measuring the matrix properties of ductile iron (GJS) castings
of the same chemical composition but of differing shapes and
thus of different rates of heat removal has shown that it is
not possible to find a reliable regression function of the
speed of the passage of ultrasonic waves with respect to
structural or mechanical properties. A result of this fact is
that before beginning to measure the speed of ultrasound it is
not possible to determine what values of hardness, strength,
ductility or other matrix properties of the casting should
correspond to this speed. Before introducing this control
measuring method into operation it is therefore necessary to
carry out direct measurement of structural and mechanical
properties on a statistically significant number of specimens,
to set again control limits and only then can this method be
regarded as reliable and conclusive in establishing the
quality of castings. The direct measurements must be repeated
periodically in order to check the trend in properties.
The correlation relations established on one type of GJS
casting cannot, however, be applied to another type of GJS
casting and even within one casting such relations holding for
one area cannot be applied to another area of the casting
since we obtain different statistical sets. Of particular
consequence can be regarded the finding that the statistical
sets measured do not mostly come from normal distribution,
which points to some non-recognized parameters that affect the
measurement, even in laboratory conditions.
It has been shown that the values established for the surface
layer of specimens (HB. chemical composition, etc.) have a
large scatter since they describe in the specimens only a
layer whose depth is in the order of mm. As a result, such
measurements are sensitive to local differences in properties.
Measurements through the whole wall of the casting (speed of
ultrasound) include the effect of the whole wall, due to which
the measurement of the speed of ultrasound is less sensitive
to local inhomogeneities.
The temperature values obtained with the aid of thermocouples
cannot always be regarded as correct and accurately reflecting
the reality. In addition to errors of thermocouples and
measuring sets, which can be foreseen and corrected, the
greatest problem occurring during measurement are the
inaccuracies caused by the temperature gradient along the
thermocouple length. The most accurate measurement can be
obtained if the temperature gradient along the thermocouple
length is as low as possible; this means that it is convenient
if the thermocouple tip is at least a few centimetres in the
wall of the metal being poured. In practice, however, this
requirement is often impossible to meet. The proposed method
of data correction starts from a comparison of the temperature
course measured and its characteristic temperature points with
tabulated data or with the temperature courses in a casting of
the same alloy, where the requirement for a limited
temperature gradient along the thermocouple length has been
fulfilled.
It has been shown that the only effective way of increasing
the accuracy of computer simulations of pouring and cooling
castings is using more accurate thermophysical data. I have
therefore proposed an experiment that serves to make
thermophysical data more accurate for the simulation needs -
which I refer to as reverse simulation.
For practical application, I have set up an integral database
of thermophysical values (for GJS in particular) that are used
in simulation computations and I have compared values reported
by different authors.
An evaluation and mutual comparison of thermophysical
parameters is always necessary with respect to the problem
under solution since comparing the difference in numbers alone
does not give a clear idea of what these differences mean when
used in the calculations in simulation programs. One
possibility how to compare these data is to perform the
calculation using data for several alloys.
The comparison of input values for the computation of the
structure of castings in simulation programs does not give a
sufficiently clear idea of the percentage difference of
structure after computation. When using data obtained for
alloys of different chemical composition (be it for one
materials group - as ductile irons in our case) we obtain a
difference in the proportion of individual components in the
order of as much as tens of per cent points. Such computation
is unusable in practice. Therefore it is necessary to use here
as accurate data as possible from TTT diagrams. For this
purpose I have included in the work an integral database of
input TTT diagrams for GJS and the corresponding values
designed for use in simulation programs.
It has been shown that combining the computer simulation of
temperature field and the matrix evaluation via picture
analysis can bring a much more complex view of a casting. It
has also been found that computer simulation is able to
capture differences in the temperature course both at
individual points on the casting and along the wall section.
Moreover, the course of cooling curves was in good agreement
with theoretical assumptions about graphite formation, again
both along the section of yoke casting and through its wall.
Stáhnout
PDF soubor s kap 15 - 21 (0,3 MB)
Předcházející kapitola: Kap.
17: Závěry
Následující kapitola: Kap.
19: Schlussfolgerungen
Upozornění: Pokud
použijete část z moji disertační práce, dodržujte
Autorský zákon a dbejte na správnost citací |
