Damage Control: Brittle Fracture Detection

Editor’s Note: This regular column offers practical insights into various damage mechanisms affecting equipment in the O&G, petrochemical, chemical, power generation, and related industries. Readers are encouraged to send us suggestions for future topics, comments on the current article, and raise issues of concern. All submissions will be reviewed and used to pick topics and guide the direction of this column. We will treat all submissions as strictly confidential. Only Inspectioneering and the author will know the names and identities of those who submit. Please send your inputs to the author at damagecontrol@inspectioneering.com.

Introduction

Brittle fracture is characterized as the sudden rapid fracture of a component subject to stress (either residual or applied), where the material exhibits little-to-no evidence of ductility or measurable plastic deformation [1,2]. A running fracture can occur at extremely high speeds (as high as 7,000 feet/second in some steels) and is usually distinguished by a flat cleavage fracture surface with little-to-no shear lips. Furthermore, rapid fracture can occur at average stress levels (well below the expected yield strength of the material) [3]. While historically, the overall number of documented cases of structures that have failed due to brittle fracture is relatively low compared to other failure modes such as fatigue, gross overload/yielding, or buckling, brittle fracture failures often have catastrophic consequences and pose a serious health and safety risk given their sudden nature. In the past, brittle fractures have occurred in all types of engineering structures, such as storage tanks, pressure vessels, ships, bridges, aircraft structures, and even buildings [3]. Even today, despite the widespread use and acceptance of modern fracture mechanics principles and more damage-tolerant design and fabrication practices, engineering structures and, in particular, pressure-retaining equipment may still be at risk for brittle fracture failure (or possibly ductile tearing). While it is neither practical nor feasible to detect a brittle fracture itself before it occurs, identifying crack-like flaws, particularly at/near welds or other geometric discontinuities, represents the primary line of defense in quantifying potential failure risk since unstable fractures initiate at smaller defect locations. To this end, as highlighted herein, understanding the importance of reliably and accurately detecting and characterizing crack-like flaws in pressure equipment is paramount.

In general, materials are categorized as behaving in either a brittle or ductile manner. Engineering materials and steels that are considered ductile are often characterized as being governed by yielding (plasticity or large permanent deformations). Contrarily, brittle materials are typically limited by the propensity for fracture. Additionally, brittle materials will not generally display any well-defined yielding behavior during a uniaxial tension test (e.g., the absence of necking in the test specimen). It is remarked that normally, brittle materials can exhibit notable ductility under loading conditions that produce an elevated hydrostatic component or when subject to applied stress that is highly compressive [4]. Furthermore, empirical observations indicate that customarily ductile materials tend to fail as ductility increases for compressive hydrostatic stress states, and the opposite correlation holds true; that is, as ductility decreases in a component subject to tensile loading, the material is usually more prone to failure [4]. In carbon and low-alloy steels commonly used to construct pressure equipment, ductility (e.g., fracture toughness) is distinctly a function of metal temperature. In fact, most ferritic materials exhibit a ductile-to-brittle transition as described herein, where brittle behavior is generally observed at colder metal temperatures.

This article is Part 1 of a three-part Damage Control series on brittle fracture. Specifically, this article focuses on the fundamental concepts of brittle fracture and the typical damage morphology associated with the unstable rupture of pressure equipment. A brief commentary on ductile tearing characteristics of austenitic stainless steels is also provided. Additionally, critical variables that influence fracture susceptibility (including material properties, weld characteristics, component geometry, original heat treatment, etc.) and affected process units/equipment are also outlined. Special considerations for heavy-walled pressure equipment (e.g., in hydroprocessing units) and a concise summary of low fracture toughness issues in carbon steel forgings (e.g., flanges) and fittings are rendered in this article. Lastly, a brief overview of how industry’s collective understanding of brittle fracture concepts has evolved over the years is offered herein.

Historical Overview of Brittle Fracture

In the late 1800s, members of the British Iron and Steel Institute reported perplexing cases of cracking in steel that manifested in a brittle manner [5,6]. Specifically, in 1886, a 250-foot-tall standpipe in Gravesend, Long Island, failed by brittle fracture during its hydrostatic pressure acceptance test. During this same period, other brittle failures of riveted structures such as gas holders, water tanks, and oil tanks were reported, even though the materials used in these structures had met all existing tensile and ductility requirements of the day [3]. One of the most infamous tank failures likely attributed to brittle fracture was of a large molasses tank (2,300,000 gallon capacity) in Boston, which failed in January 1919. This catastrophic failure resulted in twelve fatalities and many more injuries [3]. Prior to World War II, several welded Vierendeel truss bridges in Europe failed shortly after construction. In these instances, all the bridges were lightly loaded, the temperatures were low, the failures were sudden, and the fractures were brittle in nature. Results of a subsequent comprehensive failure investigation indicated that...