Árbók VFÍ/TFÍ - 01.06.2000, Qupperneq 331

Ritrýndar vísindagreinar 329
Sensitivity Analysis
The main feature of the 1D casting
process model is not the calculation of
stress values that can be verified in prac-
tice, but rather the construction of a sys-
tem that emulates the behavior of the
physical system correctly. Reasonable
stress values are of course necessary, but
if the model is valid, different values of
design parameters can be tested and com-
parcd with confídcnce although the actual
solution values can not bc verified.
Given that the model is valid, a sensi-
tivity analysis is an effective tool to visu-
alize the effect of the design and boundary
parameters.
Sensitivity analysis is performed to investigate the effect of the independent sub-cycle
duration values and the boundary paraineters on the induced mould stress. The sensitivity of
the solution to the mould initial temperature is also investigated.
The sensitivity analysis is implemented by changing each parameter independently and
simulating the casting process based on thc new input. Each parameter is changed relative
to its original value which results in a sensitivity star graph, where the solution based on the
original parameters is plotted in the core of the star.
Sensitivity analysis is made for both the mould maximum tensile and compression
stress. Simulation runs are therefore made with the maximum tensile and compression
stresses as the output for different sensitivity analysis. The maximum compression stress is
denoted as the minimum total stress.
The resulting sensitivity stars give an indication of what parameters have the most influ-
ence on the solution and hence are the most desirable to control in the process or to study
further.
Sub-Cycles Duration, Maximum Tensile Stress
Figure 7 depicts the sensitivity star for the sub-cycle duration. The maximum tensile stress
is used as an output. The maximum tensile stress is experienced in the mould surface at the
end of sub-cycle 5 (water-based coating). It indicates that a reduction in tensile stress can
generally be obtained by reducing the duration of each sub-cycle, except in the case of the
water cooling sub-cycle. In that case an increase of t2 is more feasible. In general, an
increase of t2 and a reduction of t5 (water-based coating) will give the greatest decrease in
maximum stress. That is reasonable because an increase in t2 will decrease thc heat flux into
the mould, resulting in lower mould temperature at the start of coating, while a decrease in
t5 will shorten the effect of the water used during coating.
To miniinize tensile stress in the surface the reduction of t5 seems to be the most promis-
ing option. As will be seen below, water cooling on the cast and therefore rnould edges, is
also undesirable.
Figure 7. Sensitivity star for mould maximum
tensile stress as a function of relative sub-cycle
duration.