2007, Volume 5, Number 1
Columnar Joints: An Examination of Features, Formation, and Cooling Models
Introduction
The Devil’s Postpile in California and the Giant’s Causeway in Ireland are well known geological tourist attractions. What makes them so attractive is columnar joints, joint systems that intersect at about 120° to form polygonal columns of rock with three to eight sides (Encyclopedia Britannica, 2006) found in a wide variety of rocks such as basalt flows, rhyolite tuffs, and dehydrated opaline quartzarenites (Budkewitsch, 1994). What process formed such an unusual, unnatural looking feature? This paper will provide the answer.
Features of Columnar Joints
Columnar joints are “the only system of natural fractures to approach an optimal hexagonal honeycomb-like pattern” (Gray, 1986) but pentagonal columns are just as abundant (Budkewitsch, 1994). In addition to variations in sides the columns also vary in scale. Sources do not agree on the amount of variation, the largest range being seven centimeters to six meters in diameter (Encyclopedia Britannica, 2006).
It has been observed that the surface expression of columnar joints is not as regular as the pattern expressed at depth within the cooling body (Budkewitsch, 1994). The honeycomb-like pattern at depth was proposed by Mallet in 1875 to be the jointing pattern that would most efficiently release tensile crack energy per unit crack length (Budkewitsch, 1994). However this does not account for the abundant five, three, and eight sided polygons found in columnar joints worldwide. In 1976 a mathematical formula was demonstrated to predict the number of sides per polygon:
n=2(2JT + 3JY + 4JX)/(JT + JY + 2JX)
where n is the number of sides per polygon and JT, JY, and JX are the fractions of T-type triple junctions, Y-type triple junctions, and quadruple junctions, respectively. Experimentation has found that hexagons are more likely to form with a slower cooling rate and pentagons are more likely with a higher cooling rate (Toramaru, 2004).
The faces of the joints show surface markings called striae or plumose structures. Each of these striae is composed of a smooth band and a rough band perpendicular to the direction of joint propagation. Together, these bands represent an increment of joint growth. The striae widths, and therefore single increments of joint growth, range from two to forty centimeters (Grossenbacher, 1995), increasing with column diameter (Budkewitsch, 1994). Since the smooth band always forms before the rough band, the striae can be used to determine the direction of joint propagation (Grossenbacher, 1995), leading to the next topic, columnar joint formation.
Formation of Columnar Joints
According to Budkewitsch (1994) columnar joints are systems of “thermal cracks oriented perpendicular to the direction of maximum tensile stress.” When the edges of a hot body of rock begin to cool and solidify they contract to a greater degree than the still-hot interior. This causes the development of horizontal tensile stresses between the hotter and cooler areas in the rock, in turn causing the rock to crack (Saliba, 2003). These stresses are directed parallel to the isothermal planes within the body, with the cracks perpendicular to the isothermal planes. The direction of crack propagation is also important and from studying the striae, as described above, it has been found that the cracks propagate inward from and perpendicular to the cooling surfaces (Grossenbacher, 1995).
A fracture occurs when thermal contraction stresses exceed the yield strength of the rock mass (Grossenbacher, 1995). Once initiated, the crack propagates inward until it reaches the isotherm corresponding to the glass transition temperature and continues to follow the solidification front (Budkewitsch, 1994). In other words, along a given plane parallel to the cooling surface the body cracks soon after it solidifies. According to Grossenbacher and McDuffie (1995) the “temperature interval (∆Tinit) [for each increment of crack growth] is thought to be a constant, with a measured value of 53C for a Hawaiian basalt.”
This means that each stria, representing one increment of joint propagation, is equivalent to 53C of cooling (for Hawaiian basalt) and that stria width is dependent on the thermal gradient of the cooling body (Grossenbacher, 1995). If the thermal gradient is steep, e.g. a large change in temperature over a short distance, then the stria will be thin (Grossenbacher, 1995). If the thermal gradient is shallow, e.g. a small change in temperature over a long distance, then the stria will be thick.
There has been direct evidence for the thermal cracking hypothesis of columnar joint formation from detailed kinematic analysis of basalt fracturing in columnar structures and through studies on the thermal history of cooling lava lakes (Budkewitsch, 1994). The cooling lava lake studies have also confirmed that the cracks propagate close behind the solidus isotherm (Grossenbacher, 1995). Since the cracks do follow the solidus isotherm they can be used to deduce much about the thermal history of a rock mass. For example, if the upper eighty percent of a rock mass is fractured with downward-propagating joints, then the upper cooling front must have been accelerated with respect to the lower, since usually only the upper sixty percent is cooled through the top (Grossenbacher, 1995).
It has been noted that column diameter is dependent on the cooling rate (∂T/∂t) with narrow columns signifying faster cooling and wide columns signifying slower cooling (Grossenbacher, 1995). A slow cooling rate allows viscous dissipation to act over a wider region resulting in larger columns (Grossenbacher, 1995). The inverse relationship between cooling rate and column width has been confirmed by the starch-water experiments of Toramaru and Matsumoto (2004). It has also been confirmed by higher percentages of mesostasis, and the presence of dendritic iron-titanium oxides (both indicators of a faster cooling rate) in narrow columns (Grossenbacher, 1995). Toramaru and Matsumoto (2004) found that if the cooling rate was accelerated past a critical point no joints would form, and if the rate was slowed past a certain point the crosssectional area of a joint would “approach infinity.” There are also other factors that have an effect, though not as striking, on column diameter. They are elasticity, thermal expansion, crack density ratio at maximum stress, the crack growth coefficient, and glass transition temperature (Toramaru, 2004).
Cooling Models
There is some debate over whether the cooling mechanism for columnar joints is conduction, convection, or a combination of the two (Budkewitsch, 1994). Conduction is the transfer of heat through physical contact, and convection is the transfer of heat through an intermediary fluid. Grossenbacher and McDuffie (1995) proposed a solely conductive model; while Budkewitsch and Robin (1994) proposed a conductive-convective combination where the convecting medium is water, mixed water and vapor, volcanic gas, or air.
The conductive model of Grossenbacher and McDuffie (1995) predicts that the ratio of stria width to column diameter is nearly constant and that column diameter and stria width increase inward from the boundaries of the rock mass. Both predictions are supported by field observations of columnar joint structures such as those exposed in the Makaopuhi Crater at Kilauea Volcano, Hawai’i. The columns have a diameter of one foot at the top and twelve feet at depth (Peck, 1968). Mathematical analysis also supports the conductive cooling model (Grossenbacher, 1995).
The confirmation of the conductive model through field observations and mathematical analysis does not, however, rule out the coupled conductive and convective model because it is “supported by direct field measurements of the cooling Hawaiian lava lakes, data on glass transition, and mathematical cooling models” (Budkewitsch, 1994). With two cooling models, both supported by field observations of lava lakes in Hawai’i and mathematical analysis, at first glance it may seem that a definitive answer is not forthcoming. Further inspection, however, reveals that the coupled model states that heat loss is through (a) conduction toward the cooler joints from inside the flow and (b) convection out of the flow through the cracks when the cooling front is over a certain distance into the interior of the cooling body. Therefore, it is possible that the conduction model is valid close to the edges of the body and the coupled model in the interior. The coupled model allows heat to be removed from the interior of the cooling body through the crack network of the columnar joints. This helps to explain the regularity of the column size by allowing the advance of the cooling front to be a quasi-steady state process (Budkewitsch, 1994). Another feature of the coupled model is that it enables the solidification time to be a linear function of the thickness of the cooling body where in the conductive model the [?]
Conclusion
Columnar joints are systems of joints that formed through thermal contraction from the outside-in as a hot body of rock cooled. Direct observations of cooling lava lakes in Hawai’i have provided much insight into the mechanisms of columnar joint formation. However, even with these observations conflicting ideas on the cooling mechanism remain. Whether the mechanism is agreed upon or not, thickness (Budkewitsch, 1994). columnar joints such as those at Devil’s Postpile will retain the mystery that draws tourists and scientists from around the world.
WORKS CITED
Budkewitsch, Paul and Pierre-Yves Robin. “Modelling the Evolution of Columnar Joints.” Journal of Volcanology and Geothermal Research 59.3 (1994): 219-39.
“joint.” Encyclopedia Britannica. 2006. Encyclopedia Britannica Online. 11 Dec. 2006 .
Gray, Norman H. “Symmetry in a natural fracture pattern: The Origin of Columnar Joint Networks.” Computers & Mathematics with Applications 12.3-4 (1986): 531-45.
Grossenbacher, Kenneth A. and Stephen M. McDuffie. “Conductive Cooling of Lava: Columnar Joint Diameter and Stria Width as Functions of Cooling Rate and Thermal Gradient.” Journal of Volcanology and Geothermal Research 69.1-2 (1995): 95-103.
Peck, Dallas L. and Takeshi Minakami. “The Formation of Columnar Joints in the Upper Part of Kilauean Lava Lakes, Hawai’i.” Geological Society of America Bulletin 79.9 (1968): 1151-65.
Saliba, R. and E. A. Jagla. “Analysis of Columnar Joint Patterns from Three-dimensional Stress Modeling.” Journal of Geophysical Research 108.B10 (2003).
Toramaru, A. and T. Matsumoto. “Columnar Joint Morphology and Cooling Rate; a Starch-water Mixture Experiment.” Journal of Geophysical Research 109.B2 (2004).
This paper was written for Geology 470 (Volcanology) as the final research paper project.