When a component fails, the fracture surface of the broken component can tell a great deal regarding the mechanism of failure. It can often provide details or suggestions as to why a component has failed. In this short column, I will attempt to describe the four basic types of fracture and indicate possible causes. Often, multiple fracture mechanisms will be present on the failed component, and can lead the investigator to determine why a component failed.
Examination of the fracture surface and metallography are used to determine the cause of failure. First, it is necessary to determine the fracture mode. Unfortunately, there is no clear or logical classification of fracture. Generally, classification is based on the crack growth mechanism. We will discuss four types of fracture mechanisms: ductile failure, brittle fracture, intergranular fracture, and fatigue.
Ductile failure is a very common mechanism of failure. In this mechanism, the material is loaded beyond its ultimate tensile strength. Loading beyond the tensile strength could be due to the designer failing to apply a suitable factor of safety; or service conditions beyond design criteria.
On a macroscopic scale, a ductile fracture is accompanied by a relatively large amount of plastic deformation before the part fails. After failure, the cross-section is reduced or distorted. Shear lips are observed at the latter part of the fracture and indicate the final failure of the part. The fracture surface is dull, with a fibrous appearance. Examination with an SEM shows that the fracture is by micro-void coalescence  (Figure 1).
In this failure mechanism, the creation of a free surface from a small particle such as an inclusion occurs first. The free surface around the particle creates a void. The void grows by plastic strain and hydrostatic stress. Finally, the voids grow to a size that they join or coalesce with adjacent voids. These voids coalesce to form a central crack, perpendicular to the applied tensile stress. Depending on the applied stresses, the shape and configuration of the dimple shape can be changed (Figure 2). This fact is useful in determining the type of loading in a failure analysis investigation .
Ductile failure can occur with any of the types of inclusions. This is true whether it is the brittle alumina type inclusion or the more ductile sulfide type inclusions. Inclusions generally initiate ductile cracking above a critical size. Coarser inclusion sizes tend to have a larger local stress concentration factor, which can cause local decohesion and micro-crack formation.
In a similar fashion to that of inclusions, the distribution of carbides can also influence the toughness and ductility of the steel. The strain needed for void formation decreases with increasing carbide volume fraction. Spheroidal carbides will not crack at small strains and exhibits decohesion. Spherodized steel is much more ductile than similar steel of the same hardness containing only ferrite and pearlite. Pearlite has a lower critical strain for void formation. In addition, once a crack or void forms in a pearlitic matrix, it will tend to run along the length of a pearlite lamella. Examining this type of fracture under the SEM, the base of the dimples will contain fractured pearlite lamella.
Very little plastic deformation and a shiny fracture surface characterize brittle fractures. Often, chevron patterns point back to the origin of failure  (Figure 3). It can occur at low stress and propagate with rapidity — often at speeds approaching the speed of sound in the failed material. Analysis of fracture surfaces of brittle failures indicated that brittle failures would initiate at a notch or stress concentration, and propagated with little plastic deformation. These notches were of three types: design features, fabrication details, or material flaws.
Design features were notches such as tight radii or structural members that were rigidly joined at angles less than 90 degrees and welded. Fabrication details result from the production of notches during the manufacture of the component. Welding strikes, deep gouges, and similar machining marks create mechanical notches. Metallurgical notches arise from abrupt changes in microstructure or porosity from welding or casting. These flaws can also be related to mill practice and can be large inclusions, internal oxidation, or porosity.
In brittle fractures, limited energy is absorbed by the fracture. Energy is absorbed through regions of small plastic deformation. Individual grains separate by cleavage along specific crystallographic planes (Figure 4).
Visually, little or no plastic deformation or distortion of the shape of the part characterizes brittle fractures. The fracture is usually flat and perpendicular to the stress axis. The fracture surface is shiny, with a grainy appearance. Failure occurs rapidly, often with a loud report. Because the brittle cleavage is crystallographic in nature, the fracture appearance is faceted (Figure 4).
There are three basic factors that contribute to brittle cleavage type of fracture in steels: a triaxial stress state, low temperature, and a high strain rate or rapid loading rate. These three factors do not have to be present for cleavage-type fracture to occur. Most brittle, cleavage-type fractures occur when there is a triaxial stress-state and low temperature. This is actuated by a high rate of loading.
The notch toughness of low- and medium-strength steels is highly dependent upon temperature. There is a transition from ductile fracture to brittle fracture as the temperature decreases. One criterion for the transition temperature is the Nil Ductility Temperature (NDT). The Nil Ductility Temperature is the temperature where fracture becomes 100 percent cleavage, and there is essentially no plastic deformation.
Changes in the NDT can be produced by changes in microstructure and chemistry. The largest change can be affected by changes in the amount of carbon and manganese. The NDT is lowered by about 10°F for every 0.1 percent increase in the Mn concentration. Increasing the carbon content also lowers the NDT. The Mn/C ration should be approximately 3:1 for good notch toughness.
Nickel is beneficial for increasing ductility. Up to 2 percent Ni is effective in lowering the Nil-Ductility Temperature. Increasing concentrations of silicon have the effect of increasing the NDT. Chromium has nearly no effect, while molybdenum is extremely effective in increasing the ductility of steels, and drastically decreases the NDT. Oxygen strongly decreases the ductility. It can also cause increased propensity to intergranular fracture, by creating brittle oxides at the grain boundaries. Decreasing the grain size has a strong effect on increasing the ductility and notch toughness.
Section thickness can also influence ductile and brittle behavior . Investigations  showed that there was considerable variation of toughness with the thickness of the specimen. Further, at large thickness, the toughness appeared to reach a constant value  (Figure 5). Within this curve, there are three regions apparent: First, there is the region where maximum toughness is obtained (thin sections). Second, there is the region of intermediate toughness, with the final region, a region with relatively constant toughness (thick sections).
In the first region, the fracture appears to consist entirely of a shear lip, or in other words, the fracture surface is inclined at angle of approximately 45˚ to the tensile axis. In this situation, the stress in direction of the thickness of the specimen tends toward zero, and a state of plane stress is achieved.
In the intermediate range, the fracture behavior is complicated. The fracture does not consist of entirely “slant” type fracture, nor does it contain entirely a “flat” plane strain-type fracture. Instead the regions of “flat” and “slant” fracture are approximately equal. It has been found that the amount of “flat” fracture depends only on the thickness of the test specimen and was independent of crack length.
In the third region, the fracture consists of predominantly “flat” fracture. Some evidence of very small shear lips may be present at the later part of fracture. Fracture is catastrophic and rapid. No plastic deformation is evident. In this third region, any increase in the thickness of the test piece causes no further decrease in the toughness.
One famous failure involving brittle fracture was the “Great Boston Molasses Disaster” . In this failure, the United States Alcohol Company fabricated a large cast iron molasses tank in Boston in December 1915. This tank was 90 feet wide and 58 feet tall, with a head of 49.5 feet of molasses. It was fabricated of cast iron plates, riveted together. It held 2.3 million gallons of molasses, ostensibly used for the fermentation of ethanol used for liquor. The man who oversaw construction could not read blueprints, nor had any technical training. No engineers or architects were consulted to ensure that the tank was constructed safely. On January 15, 1919, the tank exploded and molasses flooded the streets of Boston with waves eight to 15 feet tall (Figure 6). This great wall of molasses was reported to have moved at speeds up to 35 miles per hour, and it devastated a large section of Boston.
Half-inch steel plates were torn apart, and these plates were thrown with enough force to cut girders of the elevated railway. This explosion, and the subsequent wave of molasses, resulted in 21 people killed, 150 people injured, many buildings destroyed, and an entire area devastated.
Investigation many years later indicated that the probable cause was brittle fracture of the tank at rivets, with the temperature below the ductile to brittle transition temperature. One interesting result of this disaster was that Massachusetts and many other states created laws to certify engineers and to regulate construction. It also required stamped drawings certifying that an engineer had reviewed the drawings. It was this failure that was the origin of the Professional Engineer’s License and stamp as we know it today. As a side note, the 18th Amendment was ratified and Prohibition was signed into law on January 16, 1919.
In this short article, two types of fracture were described. Ductile failure is the result of overload and exceeding the ultimate tensile strength of a material. Extensive plastic deformation is observed. In brittle fracture, catastrophic failure is observed, with little or no warning. No plastic deformation is observed. Triaxial stress states, a stress concentration, or rapid loading are usually required for brittle fracture to occur.
In the next article, intergranular fracture and fatigue will be discussed. Should you have any questions regarding this, or have any suggestions for further articles, please contact the editor or myself.
- B. Bogner, G. Rorvik and L. Marken, “Bolt Failures – Case Histories from the Norwegian Petroleum Industry,” Microsc. Microanal., vol. 11 (Suppl 2), pp. 1604-1605, 2005.
- D. S. MacKenzie, “Overview of the Mechanisms of Failure in Heat Treated Components,” in Failure Analysis of Heat Treated Steel Components, L. Canale, R. A. Mesquita and G. E. Totten, Eds., Materials Park, OH: ASM International, 2008, pp. 43-86.
- Transportation Safety Board of Canada, “Derailment and Collision Canadian National Train No. U-783-21-30 and Train No. M-306-31-30 Mile 50.84, Saint-Hyacinthe Subdivision Mont-Saint-Hilaire, Quebec 30 December 1999”.
- D. S. MacKenzie, “Failure Analysis of a Naval Arresting Hookpoint,” in Proc. 2nd Mediterranean Conference on Heat Treatment and Surface Engineering, 11-14 June, Dubrovnik, Croatia, 2013.
- J. M. Kraft, A. M. Sullivan and R. W. Boyle, “Effect of dimensions on the fast fracture instability of notched sheets,” in Proc of the Crack Propagation Symposium, Cranfield, 1961.
- J. E. Strawley and W. F. Brown, “Fracture Tougness Testing,” in ASTM STP 381, 1965.
- A. S. Tetelman and A. J. McEvily, Fracture of Stuctural Materials, New York, New York: Wiley & Sons, 1967.
- J. F. Knott, Fundamentals of Fracture Mechanics, New York, New York: J. Wiley & Sons, 1973.
- J. S. Colton, Great Boston Molasses Disaster, Jan 15, 1919, Georgia Institute of Technology.