TECHNOLOGY Procedure improves line pipe Charpy test interpretation

Michael J. Rosenfeld
Kiefner & Associates Inc.
Worthington, Ohio

These effects are readily accounted for by a simple mathematical procedure, offered here, which enables extrapolation of the full-scale transition curve from as little as a single subsize specimen test.

This procedure is useful when the toughness transition curve is incomplete or nonexistent. Toughness data may be incomplete because the API 5L toughness test establishes minimum performance at a single temperature, which does not reveal the full transition curve.

Toughness data may be nonexistent because the first requirements for toughness testing of line pipe appeared in the 16th Edition of API 5LX in 1969, and those requirements remain at the option of the purchaser today.

Consequently, much of the pipe in service today may never have been tested for toughness properties. Although material can be removed for testing from pipe in service by making a small hot tap, the coupon provides insufficient material to obtain the full toughness transition curve.

The full toughness transition curve, when available, can contribute to engineering satisfactory solutions to many pipeline problems, including the following:

• Development of in-service relocation and maintenance procedures
• Evaluation of ground movement problems
• Fitness-for-service assessments
• Determination of the maximum flaw size that can survive hydrostatic testing
• Risk assessment.

Transition behavior

Plain carbon and low-alloy steels exhibit a dramatic transition in toughness with temperature, being brittle at "low" temperatures and ductile at "high" temperatures. The lower part of Fig. 1 [18713 bytes] illustrates the increase in absorbed impact energy as the fracture mode changes from brittle to ductile.

The ductile impact energy level is sometimes referred to as the "upper-shelf toughness." The impact energy is a measure of resistance to fracture propagation from a defect, but it is not an intrinsic measure of the ductility of the material.

Ductility involves plastic deformation and is indicated in the CVN specimen by lateral expansion and by the proportion of the fracture surface consisting of shear. The upper part of Fig. 1 illustrates the increase in fracture-surface shear area (as a percentage of the total fracture surface area) as the fracture mode changes from brittle to ductile.

Brittle fracture takes place at nominally elastic stress levels without appreciable energy absorption occurring through deformation. Consequently, a brittle material may fail catastrophically at low levels of stress applied in the presence of relatively small defects.

By contrast, ductile fracture involves significant energy absorption through plastic material deformation which inherently requires high levels of applied stress to initiate. Consequently, ductile materials have high defect tolerance compared to brittle materials.

The differences between these two modes is illustrated in the fractured CVN specimens shown in Fig. 2 [34294 bytes]. The upper specimen, being ductile, exhibits deformation near the fracture. The lower specimen exhibits no such deformation, being brittle.

Fig. 3 [45961 bytes] shows the fracture surfaces of full-size and half-size CVN specimens arranged in order of increasing ductility (and test temperature) from fully brittle to fully ductile. The brittle areas are bright and crystaline in appearance, while the ductile shear areas are dull and fibrous in appearance.

Fully ductile behavior occurs at temperatures greater than the "transition temperature." Many definitions for the transition temperature have been used, but the pipeline industry generally regards the transition temperature as that at which 85% of the CVN fracture surface area consists of shear.

This is referred to as the shear-area transition temperature (SATT).

CVN impact test

The CVN test is performed by impacting a notched bar specimen with a weighted pendulum having calibrated kinetic energy at the point of impact. The standard size CVN specimen is a rectangular bar with cross section measurements of 10 mm x 10 mm (0.394 in. x 0.394 in.) and length of 55 mm (2.165 in.).¹

A V-notch 2 mm deep is machined across the specimen face opposite the impact point. When the pipe W.T. is less than 10 mm, "subsize" specimens must be used. Only the specimen width is reduced, while the depth relative to the notch is maintained.

One might expect absorbed impact energy at a given test temperature to be proportional to the specimen width. This may be the case at extreme positions (the upper or lower shelves) on the transition curve. In the transition area, however, the situation is complicated by the fact that the reduced width of the subsize specimen reduces the constraint on crack-tip plasticity.

Consequently, a subsize specimen may fracture with greater ductility than a full-size specimen at the same temperature. The problem this creates is that a smaller specimen will indicate the brittle-ductile transition at a lower temperature than will a larger specimen.

Likewise, test results from a specimen that is significantly narrower than the pipe-wall thickness may lead one to underestimate the transition temperature unless the specimen size effect is accounted for.

Drop-weight tear test

The drop-weight tear test (DWTT) addresses the specimen size effect by using plate-like flattened specimens of full wall thickness. Hence, it reflects the full-scale temperature and thickness-dependent resistance to crack propagation.

The shear-area transition temperature obtained from the drop-weight tear test (TD) is taken as the true fracture-propagation transition temperature of the full-scale pipe wall material. The standard DWTT specimen is 12 in. long by 3 in. deep and contains a pressed-in notch 0.20 in. deep opposite the impact point.²

The major disadvantages of the DWTT are: much greater kinetic energy is required to break the specimens compared to the CVN test necessitating larger test frames; few independent materials-testing laboratories are equipped to perform the test; and the specimen is too large to be obtained from coupons in-service.

Adjusting transition temperature

Fortunately, it is possible to avoid the difficulties of the drop-weight tear test by making suitable adjustments to CVN test results. Maxey³ developed an empirical graphical procedure to estimate the DWTT transition temperature (TD) from the SATT obtained from CVN tests (TC; Fig. 4 [18698 bytes]).

The graphical procedure is emulated by the relationships shown in Equations 1 and 2 in the accompanying equations box in which tW is the pipe W.T. and tC is the width of the CVN specimen (units = inches; degrees = °F.).

Agreement between Equations 1 and 2 and Maxey's graphical procedure is shown in Fig. 5 [13331 bytes].

If we assume that TD is an inherent property of a given material of thickness (tW), we can predict the SATT of the material in any CVN specimen size (b) knowing the SATT in a specimen size (a). This is accomplished by equating TD in Equation 1 written in terms of specimen size ta to TD written in terms of specimen size (tb), giving Equation 3.

Adjusting Charpy impact energy

Maxey demonstrated a relationship between the CVN impact energy and the shear area (SA) at temperatures less than the upper shelf regime, as shown in Equation 4 in which CV¢ = the upper shelf impact energy, CV = the impact energy at some temperature below the upper shelf regime, and SA = the shear area as a fraction of the net specimen fracture surface area.

Inverting Equation 4 yields the shear area expressed as a function of the impact energy (Equation 5).

Until now, it has still been necessary to perform 6-12 impact tests in order to develop the transition curve. Observing that most simple brittle-ductile transitions have similar profiles makes it possible to develop general expressions that might be used to forecast the entire transition curve from a single point.

It turns out that the transition curve of many plain carbon and low alloy steels can be adequately characterized with a Sigmoidal transition expressed as a function of X = T - TC (Equation 6 in which A and B are specimen-size dependent).

We can readily invert Equation 6 to obtain the SATT from the shear area obtained at some other temperature (T) and obtain Equation 7.

The Sigmoidal curve fit is demonstrated in Fig. 6 [15960 bytes] with data from McNicol.⁴ These data were used to develop A and B for 1/4, 1/2, and full-size CVN specimen sizes, while data from Maxey3 were used to determine A and B for 2/3-size specimens.

A and B are as given in Fig. 7 [12397 bytes]. The values given for A and B work in Equations 6 and 7 with temperature in °F.

The drop-weight tear test transition curve is seen to be more abrupt than is the case with CVN transition curves. Effective values for the DWTT curve is A = 19 and B = 11 over a range of actual material thickness encompassing most pipeline applications.

Generating complete curve

We can now develop a complete impact energy and shear-area transition curve in a given specimen size using Equations 4 through 7 starting with only two of the following four quantities: CV at a test temperature (T); SA at T; CV'; and TC.

If only one of the four quantities is known, CV at T, for example, with no information as to shear area, SATT, or upper shelf impact energy, the transition curve cannot be developed.

Using Equations 1-3, one can estimate the full-size or full-scale transition temperature from sub-size data.

Equations 4-7 can then be used to forecast the full-scale transition curve. They may also be used to generate the full-size (10-mm wide specimen) CVN transition curve for use with full-size acceptance criteria.

Experimental verification

The methodology described here was verified against data from several sets of CVN impact tests of line pipe steels.⁴⁵ The correlation between the actual and estimated CVN impact properties is shown in Fig. 8 [82367 bytes].

The figures reflect that some scatter is to be expected even in the most complete set of CVN test data. Typical synthesized transition curves are compared to actual test data from Specimen C15T of Eiber's Fig. 1.5 The curves were synthesized with only TC and CV'.

The relationships given here are applicable to plain carbon or low-alloy steels having a simple transition behavior. They are not validated for certain contemporary high-toughness steels having a "rising upper shelf."

In the U.S., however, there are hundreds of thousands of miles of line pipe in the ground and more still being purchased for new construction which exhibit the typical simple transition behavior and which can be reliably modeled as shown.

Although there is great utility in being able to perform only a single test, CVN tests are always susceptible to scatter. A confirmation test at the same or different temperature may be worthwhile where material availability permits.

Load-rate temperature shift

Notch toughness as measured by the CVN test describes the ability of material to absorb energy dynamically and resist fracture propagation.

Fracture toughness (Kc) measured under conditions of slow loading and nominally elastic behavior describes the ability of a material to resist initiating a crack at a notch under static loading.

Fracture toughness undergoes a temperature-dependent transition in carbon steel similar to that for CVN impact energy, at a temperature that may be significantly lower than TD, indicating a load-rate effect, as shown in Fig. 9 [12016 bytes].

Data in Rolfe and Barsom⁶ indicate that the static transition temperature (TK) can be estimated from the full-scale dynamic transition temperature (TD) as shown in Equation 8 (temperatures = °F.; yield strength = ksi, in the range 30-160 ksi [207-1,100 Mpa]).

The fact that lower strength grades may be greatly overstrength should be considered when estimating DTrate. Maxey⁷ reports observed DTrate typically around -60° F. (-33° C.) for line-pipe steels, which is a smaller shift than that given in Equation ⁸.

DTrate may be used to verify ductile fracture initiation, which is a necessary precondition for the valid use of the ln-sec expression for longitudinal defects7 and subsequent acceptance criteria such as ASME B31G.

Ductile fracture initiation is verified by showing that the operating temperature exceeds TC, corrected for size and rate effects (Equation 9).

It is important to recognize that a failure that initiates in a ductile manner may propagate in a brittle manner. This could be expected to occur where the operating temperature is greater than the static fracture initiation toughness transition temperature but below the full-scale fracture propagation transition temperature, or TK

Examples

The following two examples illustrate the use of the methods described here and are adapted from actual problems worked out for operating pipeline companies.

Problem No. 1

Consider a company purchase specification that pipe must meet or exceed the CVN impact energy of 25 ft-lb and shear area of 60% in a full-size specimen at a temperature of 50° F. The pipe to be purchased has a 0.281 in. W.T., which is less than the width of a full-size CVN specimen (0.394 in.).

What are acceptable test temperatures and test results for a subsize specimen that will be equivalent to meeting the full-size specification?

First, estimate the shear-area transition temperature that would be given by a full-size specimen just meeting the specification. Using Equation 7 and A = 55 and B = 32 from Fig. 7 yields Tc = 50 - 32 ln (0.60/[1 - 0.60]) + 55 = 92° F.

Equation 4 yields the full-size upper shelf energy: Cv' = 25/([0.9][0.60] + 0.1) = 39.1 ft-lb.

The largest standard specimen that can be obtained from this material is a 2/3-size flattened specimen with a width of 0.263 in. The shear-area transition temperature of a 2/3-size specimen according to Equation 3 would be TC = 92 + [(66 x 0.281)0.55] [(0.394)-0.7-(0.263)-0.7] = 71° F.

The shear area at various test temperatures would be calculated from Equation 6, using TC = 71° F., and A = 47 and B = 27 from Fig. 7. The 2/3-size specimen upper shelf energy would be CV' = (2/3)(39.1) = 26.1 ft-lb.

The impact energies at various test temperatures would be calculated from Equation 4, using the shear areas calculated previously and CV' = 26.1 ft-lb. The resulting criteria curves are shown in Fig. 10 [79743 bytes].

These represent minimum test or target values which would ensure at least the same toughness in the actual pipe as would be assured by meeting the full-size specification. It is not necessary to test at a specific temperature except that it should lie within the transition range shown for the 2/3-size specimen.

The next smaller standard specimen is a 1/2-size, with a width of 0.197 in. The same procedure could be used to develop target test curves for that size, also shown in Fig. 10.

Problem No. 2

Consider a pipe in service with 30 in. OD x 0.312 in. W.T.; the material is API 5L X-52 but with an actual strength of 54 ksi and operating at 55° F.

The goal is fully to characterize toughness properties of the material, but sufficient material can be obtained from a small hot-tap coupon for a CVN test at a single temperature with one additional test for verification.

Two 2/3-size specimens 0.263 in. wide are tested at 32° F., producing average impact energy of 9 ft-lb and shear area of 40%.

First, extrapolate the upper shelf energy and transition temperature for the 2/3-size CVN properties. From Equation 4, Cv' = 14.5/([0.9 x 0.4] + 0.1) = 31.5 ft-lb.

From Fig. 7, A = 47 and B = 27 for 2/3-size specimens, and from Equation 7, Tc = 32 - 27 ln (0.40/1 - 0.40) + 47 = 90° F.

One can now develop the entire 2/3-size transition curve using Equation 6 and then Equation 4.

Next, estimate the full-scale static and dynamic transition temperatures. From Equation 2, (Tsize = ([66] [0.312]0.55/0.2630.7 - 100 = -11° F. so that TD = 90 - 11 = 79° F. This indicates that the pipe operates below the fracture propagation transition temperature.

From Equation 10, DTrate = (1.33)(54) - 187 = -115° F., so that TK = 79 - 115 = -36° F. or using Maxey's DTrate = -60° F., obtain TK = 15° F.