A review of experimental investigations on stress development during the bl
ockage of a water-filled pipe by freezing was undertaken with the parallel
development of an effective finite element thermal stress model. A wide spr
ead of measured stress values was noted as well as a degree of uncertainty
in the cases when the gauge output did not return to zero at the end of the
freezing cycle. A methodical examination of stress- and temperature-time h
istories showed that it is possible to divide a freeze into three stages: f
illing, constant wall temperature and thawing. Since each stage produces qu
antitatively and qualitatively different stress states, it needs to be exam
ined separately. The filling stage causes stresses through the pipe wall, w
hich vary from tensile on the outside surface to compressive on the inside.
These stresses can be significant but are also short lived and their magni
tude may be greatly affected in practice by the way that the coolant is app
lied. During the constant-temperature phase, when the pipe wall temperature
is maintained at the coolant temperature, the stresses are mainly compress
ive, their variation is small within the freezing jacket and they appear to
depend on the diameter-thickness ratio. A significant difference between t
he behaviour within the jacket and at the end of the jacket is sometimes ob
served. Finally, tensile stresses arise during the reverse process of thawi
ng. Comparison of experimental data with numerical predictions confirmed th
e above observations regarding the magnitude, distribution and nature of th
e developing stresses. There are important quantitative and qualitative dif
ferences between measured and predicted values, which can be explained by t
he uncertainty and scarcity of data as well as the simplicity of the adopte
d material model for ice behaviour.