design using SiC side blocks operated at a higher current density and corresponding higher
superheat prior to the shutdown. That average cooling rate is very similar to those presented in
figure 10 of [reference 5]. Figure 3 presents the cell temperature obtained after 24 hours of cooling
from the full quarter cell model while figure 4 presents only the cathode panel temperature section.
It can be noticed that the temperature on the cathode panel surface is lower than the one directly
below at the collector bars level.
Cathode cooling cracks:
That cell cooling is sufficient to produce cooling cracks on the cathode
surface mostly in the transverse direction of the cell as the one presented in figure 1. A level of
tension stress of at least 8 MPa in the longitudinal direction of the cell is required to generate those
cracks according to the cathode block properties presented in [Reference 6].
It was not possible to obtained that level of longitudinal tension stress in the previous study
[Reference 4]. In that previous study, the cathode panel was prevented to deflect down but was free
to contract in both horizontal directions. By using this limited type of displacement constrains, the
level of tension stress obtained was only around 2 MPa or about 4 times less that was is required to
generate cooling cracks.
Yet, already in that previous study, the level of longitudinal tension stress obtained by solving the
2D thermal stress model in plain strain mode was sufficient to generated cooling cracks. Figure 6
presents that longitudinal stress component obtained using the 2D thermal stress model in plain
strain mode with the thermal gradient after 24 hours presented in figure 5 itself the results the new
transient analysis model producing the new faster cooling rate. Like in the previous study [reference
4], the thermal gradient used to carry-up the thermal stress analysis is obtained by subtracting the
initial steady state temperature to the temperature obtained after 24 hours of cooling.
When assuming plain strain, the 2D model do predict longitudinal tension stress level high enough
to expect cracking problems as it did in the initial study, but those results were then considered
unrealistic as they are based on the assumption that the cathode is restrained from shrinking
longitudinally. After discussing the issue with Dr. Morten Sorlie the authors reconsided the
situation, according to Dr. Sorlie, the collector bars who are anchor by the pier are restraining the
cathode panel to shrink freely in the longitudinal direction. Figure 7 is presenting the longitudinal
stress component obtained using the 3D quarter cathode panel model assuming that the collector
bars are preventing the vertical carbon faces in the slots to move longitudinally. As it can be seen in
figure 7, with this type of motion restrains, there is enough longitudinal tension stress to generate
transverlase cracks so it is safe to assume that as Dr. Sorlie is proposing, collector bars do prevent
the cathode panel to move freely in the longitudinal direction of the cell.