Relationship betvveen Fracture Toughness and mechanical Properties of some Structural Steels at Lovv Temperatures
Odvisnost med lomno žilavostjo in mehanskimi lastnostmi nekaterih konstrukcijskih jekel pri nizkih temperaturah
B. Ule, M. Lovrečič-Saražin, Inštitut za kovinske materiale in tehnologije, Ljubljana J. Vojvodič-Gvardjančič, Inštitut za metalne konstrukcije, Ljubljana A. Ažman, A. Lagoja, Acroni Jesenice
The effect of strain-aging on the impact toughness characteristic of Charpy specimens (CVN) and on guasi-static fracture toughness values KIC of some structural steels vvas Investigated in the temperature range of nil-ductility temperatures. Strain-aging provokes shifts of Charpy curves to higher temperatures, but it decreases the nil-ductility temperatures regarding to as purchased steels. The correlation betvveen KIC and conventional mechanical properties valid for lovv temperatures confirms that KIC and probably also Karrest of as strain-aged steels are higher than that of as purchased steels vvith the same Charpy energy.
Key words: fine-grain low-alloy steels, fracture mechanics, fracture toughness, drop-weight test, nil ductility temperature.
Raziskali smo vpliv deformacijskega staranja nekaterih konstrukcijskih jekel na njihovo udarno Charpyjevo žilavost (CVN) ter kvazi-statično lomno žilavost KIC v temperaturnem območju ničelne duktilnosti. Deformacijsko staranje pomakne Charpyjeve krivulje k višjim temperaturam, vendar pa zniža temperature ničelne duktilnosti glede na jekla v dobavnem stanju. Korelacija med KIC in konvencionalnimi mehanskimi lastnostmi, veljavna pri nizkih temperaturah, kaže, da je Ktc in verjetno tudi Karrest vrednost staranih jekel višja kot pri jeklih v dobavnem stanju z enako Charpyjevo energijo. Ključne besede: drobnozrnata malolegirana jekla, mehanika loma, lomna žilavost, test s padajočim bremenom, temperatura ničelne duktilnosti.
1. Introduction
The relationship of microstructure to mechanical properties in low-alloy structural steels has been the subject of considerable research. Such steels vvith increased yield stress are sometimes alloved vvith small additions of various elements so that the characteristics and the properties of such steels are substantialh affected presumablv due to the reduction of the austenite and ferrite grain size and because their yield stress. strength and toughness increase vvhile the ductile/brittle transition temperature decreases vvhich is perhaps one of the most important aspects of microalloying. In most of the previous investigations. fracture behaviour of steels has been evaluated mainlv by means of the Charpy impact test because of its convenience and familiarity. Although the material requirements for a lot of practical applications are based on concepts of fracture mechanics, they are specified in terms of Charpv V-notch impact test results (CVN). Toughness requirements for thick-vvalled nuclear pressure-vessel steels are based on minimum dynamic toughness values, K,d. Hovvever, the actual material-toughness requirements for steels used in these pressure vessels are specified using NDT (nil-ductility transition) values and CVN impact values using lateral expansion measurements. Empirical correlations, engineering judgment and experience are thus used to translate the fraeture-mechanics guidelines or controls into actual material-toughness
specifications". A comprehensive concept for a practical estimation of the dynamic fracture toughness from the CVN impact energy vs. temperature curve vvas proposed by the MPC/PVRC Working Group on Reference Toughness21. It vvas proved that lovver bound curves can be derived from the CVN vs. T-curve for the quasi-static and lovv rate dvnamic fracture toughness (Kk), dynamic and high-rate dvnamic fracture toughness (Kl(l) and crack arrest toughness (K,.,)'1. Besides this. some other correlations betvveen conventional mechanical properties and K„ values for ductile/brittle transition range or for lovver Charpy shelves are also well-known4'51. Hovvever, it is well-known too that strain-aging of several lovv-allov structural steels causes some shifts along the temperature axis vvhich is not the same for both the CVN data and the KIc data. The purpose of the present paper is therefore to determine the more relevant correlation betvveen the conventional mechanical properties and the K„ values for some structural steels in the nil-ductility temperature range.
2. Experimental procedure
Nine non-, micro- and low-alloy structural steels in the form of hot-rolled and heat-treated flats vvere used in this investigation. The chemical composition, the designation of the steels and the thickness of the flats are given in Table 1. These
steels vvith 0.05 to 0.21 vvi/a carbon vvere either non-alloyed or alloyed vvith chromium, nickel, molybdenum, niobium and vanadium in different combinations. The microstructure of the investigated steels vvhich vvas hol-rolled and subsequently cooled at different cooling rates vvas mainly ferritic vvith different shares of perlite (Nioval 47. Č.0562 and Č. 1204) or bainite (Niomol 490 K). Only tvvo types of low-alloyed steels (Nionicral 70 and Nionicral 90) have a microstructure of tempered martensite. The yield stress of the investigated steels varied from 265 MPa for plain carbon steel to 1003 MPa for Nionicral 96 i.e. for submarine steel alloyed vvith chromium, nickel and molvbdenum. Ali the investigated steels vvere tested as purchased i.e. hot-rolled and cooled at different cooling rates but they vvere tested also after strain-aging, i.e. after cold-rolling vvith a reduction in thickness of 10% and additionnally heating for 30 minutes at 250"C.
Table 1: Chemical composition of the investigated steels
(weight %)											
No.	Grade (thickness)	C	Si	Mn	P	s	Cr	Ni	Mo	Nb	V
1	Nioval 47 l2l)iiinil	0.19	0.42	1.49	0.013	0.005	0.13	0.10	0.04	0.05	(1.07
2	Nioval 47 (65 mm)	0.14	0.33	1.53	0.014	0.005	0.16	0.15	0.01	0.04	0.07
3	Nionicral 7(1 l20iiinii	0.11	0.28	0.27	0.009	0.007	1.07	2.81)	0.26		0.06
t	Nionicral 71) (50 mm)	0.11	0.37	0.34	0.009	0.003	1.03	2.63	0.27		(1,08
5	Nionicral 96 (50 mm)	0.14	0.29	0.51	0.017	0.009	1.64	2.76	0.42		
d	Niomol 49(1 k (M) mm)	0.05	0.35	(1.42	0.011	0.004	0.75	0.29	0.33	0.06	
7	Č. (1562 (25 mili)	0.17	0.32	1.28	0.020	0.009	0.21	11.23	(1.115		
8	Č. 0562 (80 mm)	0.18	0.46	1.29	0.036	0.004	0.3(1	0.15	0.03		
9	f. 1204 (30 mm)	0.21	0.25	0.51	0.011	0.025	0.02	0.04	0.01		
Test specimens vvere cut from the plates in transverse orientation and machined to the required dimensions. Besides the standard Charpy V-notch- and Drop-weight test specimens of P3 type (15.9 x 51 x 127 mm), a large number of round-notehed and prefatigue cracked tensile specimens for the low-temperature measurements of quasi-static fracture toughness Kič w;'s made. The drop-vveight test specimens vvere prepared

S

Ud-
i
— D
I
Figure 1: Geometry of a round-notched and precracked tensile specimen
Slika 1: Geometrija nateznega preiskušanca z zarezo in razpoko po obodu
in accordance vvith the ASTM E208-84a vvhere the crack starter bead application is performed by the one bead technique to avoid the undesirable variation of NDT61. The geometry of the round-notched precracked tensile specimens. prepared according Dieter s recommendation7' is shovvn in Figure 1.
At the experiments, it is essential that the fatigue annulus be of a uniform vvidth and concentric vvith the outer diameter of the specimen in order to obtain a state of plain strain at fracture. The fatigue crack grevv to a depth of about 0.2 mm. leaving an unfractured ligament approximately 6.5 mm in diameter.
Figure 2: Experimental set-up vvith crvostat chamber Slika 2: Eksperimentalna ureditev s kriostatsko komoro
An cryostat chamber f i 1 led vvith liquid nitrogen and petroleum ether vvas used during the test to control the specimen temperature range from - 140"C to room temperature and the fracture in the quasi-static test at crosshead speed of 1 mm/min vvas reached by using a universal testing machine (Figure 2). For a round-notched precracked specimen, the stress intensity factor is given by Dieter7' as
K,= ^ (-1.27 + 1.72 D/d)
(1)
vvhere d is the radius of the uncracked ligament after fatiguing, P is the applied fracture load, and D is the outer diameter of the cylindrical specimen. In order to apply linear-elastic fracture mechanic (LEFM) concepts, the size of the plastic zone at the crack tip must be small compared vvith the nominal dimensions of the specimen. The size requirement for a valid KIc test is given by Shen Wei et. al.81 as
B. L le. M. Lovrečič-Saražin, J. Vojvodič-Gvardjančič, A. Ažman. A. Lagoja: Relationship between Fracture Toughness D> 1.5(K„./(t>s)	(2)
where a;s is the initial vield stress of the material obtained at a strain rate comparable to that attained near the root of the noteh in the fracture test. If the specimens did not comply with requirement (2) for valid fracture toughness (Klc) measurements. K(| values were obtained instead of K„, according to E399. However, the concept of the equivalent energy adopted by Wang Chang91 enabled us to determine the virtual fracture load P* instead of load P in equation < 1 > after the transformations of the surface under the parabolic load-displacement curve into the quantitatively equal surface of the triangle as shown on Figure 3.
Figure 5: Charpy V-noteh impact energy versus temperature behaviour
»l as strain-aged steels. Arrows indicate the NDT temperatures Slika 5: Charpyjeve energije v odvisnosti od temperature preiskušanja jekel v staranem stanju. S puščicami so označene temperature ničelne duktilnosti
transition temperatures vvas approximated with linear elastic fracture behaviour.
Displacement
Figu re 3: To the e\planation of the concept ofequivalenl energy Slika 3: K razlagi koncepta ekvivalentne energije
Therefore. the vveak elasto-plastic fracture behaviour of the investigated steels even in the vicinity of the nil-ductility
3. Results
Figure 4 shows the Charpv impact energy of as purchased steels as a function of the testing temperature whereas Figure 5 shows the same relationship for investigated steels as strain-aged. The nil-ductility transition temperatures (NDT) measured at drop-weight test are also indicated in both diagrams. As may be seen, the ductile/brittle transition temperatures of the investigated steels are shifted against higher values due to strain-aging. Hovvcver, the shift of nil-ductilitv transition temperatures nearly in ali the cases shows a slightlv opposite trend vvhich is somevvhat surprising.
The CVN impact energv. the vicld stress cr and the fracture toughness Klr of the investigated steels measured at nil-ductility temperatures are given in Table 2 for both as purchased and as strain-aged eondition. Hovvever, because of
200
	100
	80
E	60
£	50
s	40
LJ	30
20
10
----							! 0 OJ			
	I X«									
__3<D	CB ^						*			
	<									
/	"i									
i		K								
		4								
/ /						• o 1	as purch as strain i	ased -age	d	
1 2 3 i			5 6			8 10 20 30 40				
CVN (J)
100
Figure 4: Charpv V-notch impact energv versus temperature behaviour
of as purchased steels. Arrovvs indicate the NDT temperatures	Figure 6: Relation betvveen K„ and CVN values in the nil-ductilitv
Slika 4: Charpvjeve energije v odvisnosti od temperature preiskušanja	temperatuie ninge
jekel v dobavnem stanju. S puščicami so označene temperature ničelne	Slika 6: Odvisnost med K„ in Charpyjevo udarno žilavostjo
duktilnosti	/ V zarezo v območju temperatur ničelne duktilnosti
-196 -140-120-100-80 - 60-40 -20 0 +20 Temperature (°C)
Grade
Thicknes mm)
Nioval47 NiovaU? Nionicral 70 Ni0ntcral70 Nionicral 96 Niomol490K Č 0562 C 0562 Č. 1204
No	Grade	Thicknes (mm)
1	NiovaU?	20
2	NiovaU?	65
3	Nionicral 70	20
4	Nionicral 70	50
5	Nionicral 96	50
6	NiomoU90K	60
7	C 0562	25
8	C 0562	80
9	C. 1204	30
200-
-140
Temperature (°C)
the lack of diameter of the round-notched specimens for Kl( measurements, only the limited number of the entire data are taken into account, namely onlv data complying with the requirement (2).
Table 2: Mechanical properties of the investigated steels at nil-ductilit.v temperatures (28 valid measurements)
No.	(T,s	CVN	K„	MPa m, :)
	(MPa)	(J)	measured	calculated Eq.(5)
	As purchased			
1	908	13	68.5	75.2
2	900	19	76	80.4
3	524	5	43.5	45.1
4	354	4	31.5	34.2
	717	10	65	62.1
6	1(171	12	85	81.8
7	1(156	17	91.5	86.7
S	1054	21	100.5	90.1
	As strain-aged			
9	681	7	68.5	56.3
10	780	3	53	52.0
11	780	4	63	54.9
12	745	5	73	55.7
13	8(13	8	56.5	63.7
14	791	10	58.2	65.9
15	781	10	65.5	65.9
16	1074	11	106	80.6
17	790	9	59	64.5
IS	791	3	45	51.2
19	675	5	53	52.5
20	665	10	62.3	59.4
21	658	1	38.5	38.1
22	595	2	40	40.9
23	564	3	44.5	42.8
24	933	6.5	68.5	67.0
25	855	7.5	76.5	65.4
26	1343	11	75.5	92.2
27	1308	12	80	92.2
28	1293	14	99	94.3
The relationship betvveen the CVN impact energy and the fracture toughness K,, of the investigated steels is shovvn in the diagram of Figure 6. From the data point distribution it can be concluded that two different correlations betvveen K„ and CVN could be deduced, one for steels as purchased and another one for steels as strain-aged. Mathematical approximation vvith povver function shovvs that the correlation for steels as purchased can be expressed as:
K„ = 15.11 (CVN)"
(3)
vvith a regression coefficient of 0.945, vvhereas the correlation for steels as strain-aged can be represented as:
Ki,
: 35.09 (CVN)0-
(4)
vvith a regression coefficient of 0.819. As strain-aging provokes a considerable increasing of the vield stress of ali the investigated steels, we tried to establish a unique correlation betvveen K„ on one side and both the properties CVN and <r,„ on the other side. The follovving correlation
K„
: 0.776 (Tvs......(CVN)"'
(5)
vvith a regression coefficient of 0.921 vvas deduced from the vvhole set of experimental data given in Table 2.
4. Discussion
Charpy V-notch impact toughness measurements and quasi-static fracture toughness measurements on some none-and low-alloyed structural steels in as purchased and as strain-
286
aged condition respectivelv vvere performed over the temperature range of nil-ductilitv transition temperatures i.e. over the temperature range of-140"C to -40"C. The decrease of the NDT temperatures of steels as strain-aged regarding to the as purchased steels suggests that NDT temperature of steels as strain-aged is a good enough index temperature to represent the quasi-static fracture toughness transition behaviour of such steels, but it is maybe not an enough conservative estimation for the determination of the FTE temperature (NDT + 4()"C). A similar ascertainment, but for dvnamic fracture toughness transition behaviour of some stress-relief heat-treated steels for nuclear reactor pressure vessels. has been previouslv published by Tanaka and covvorkers6. Nevertheless, it seems that the NDT temperatures, measured either in steels as purchased or after strain-aging, correspond to the adequate KarresI value. Beeause the drop-vveight test emplovs a sharp crack. moving rapidly from the notched brittle vveld bead into the test plate. it does not come as a surprise either to find that the NDT temperature defined by this test correlates vvell vvith the beginning of an increase in fracture toughness vvith temperature measured in quantitative, sharp-crack tests"1.
Tvvo different correlations betvveen fracture toughness K[( and CVN values for both groups of steels in the temperature range investigated shovv that steels after strain-aging have a noticeable higher fracture toughness K,< than as purchased steels vvith the same Charpv energv. Hovvever. a verv good correlation betvveen K„ and both properties. CVN and ctvs vvas also deduced from ali the data. The regression coefficient of Equation (5) is relativelv high so that this approach seems to be relevant. Rolfe and Barsom" ascertained that the effects of both the notch acuity and the loading rate should be considered to establish correlations betvveen KIC and CVN test results in the transition-temperature region. Thev found out that K„ values and CVN values in the transition-temperature region can be correlated (a) vvhen the test results for slovv-bend K„ specimens are related to the test results for slovv-bend fatigue-cracked CVN specimens and (b) vvhen the test results for dynamic K„ specimens are related to the test results for dynamic-cracked CVN impact specimens. The correspondence betvveen K„ and the CVN energv-absorption values obtained at a particular test temperature and the same strain-rate for bolh KIC and CVN can be approximated bv1""
K„ = A H (CVN)"
(6)
vvhere A = constant of proportionality, E = Young's modulus. and K1( and CVN are tested at the same temperature and strain rate.
The constant of proportionalitv. A. incorporates - in accordance vv ith Rolfe and Barsom" - the effects of specimen size as vvell as notch acuitv. By changing Ihe value of A in Equation (6) it is then possible to correlate the Kl( data and the CVN energv-absorption values obtained by testing V-notched specimens. This equation suggests that the relationship betvveen slovv-bend Kl( and slovv-bend CVN test results is the same as the relationship betvveen the impact K„ (i.e.K[d) and the impact CVN results. This observation is not unexpected beeause it vvas shovvn" that a particular change in loading rate causes an equal shift along the temperature axis for both the CVN data and ihe K,, data. Both authors also concluded that an engineering estimation of Kk at anv strain-rate can be predicted by using impact CVN data in conjunction vvith Equation (6) and then shifting the curve to lovver temperatures. This approach has been used in investigating the effects of irradiation for steels used in nuclear reactors'". The magnitude of the temperature shift betvveen dynamic and slovv-bend curves vvas given by
Tshm("C)= 118.4 -0.12 a,,
(71
and valid for steels vvith the yield stress <r>s up to a value of 1000 MPa approximately. vvhereas it is diminished at steels vvith higher vield stresses. The general procedure to estimate K]( values in the transition-temperature region from CVN impact results comprises the calculation of KI(I values at each test temperature using Equation (6) vvith follovved shift of K,(1 values at each temperature by the temperature shift calculated vvith Equation (7) to obtain static Kl( values as a funetion of temperature. This procedure vvas adopted from more recent recommendations of Rolfe and Barsom" and it represents a conceptual advantage compared to the previously published methods4'"*'. By comparing our Equation (3) for as purchased steels vvith the Rolfe-Barsom Equation (6) one can scc that the exponents in both equations are relatively close. If the exponent of 0.5 is adopted also in our čase ovving to simplicity and considering some unaccuracy in our calculations (small number of data for relevant statistical analyse), then Equation (3) can be transformed into
Klt = 20 (CVN)"5	(8)
vvhere the calculated constant of 19.97 was rounded up to a value of 20.
It could be assumed that Equation (8) represents the lovver envelope of ali the measured values i.c. it represents the realistic conservative estimation of the fracture toughness of the investigated steels in the temperature range of nil-ductility transition temperatures irrespective by their microstructure or prehistorv. Nevertheless, at very lovv CVN absorbed energies our equation gives higher K„- values compared vvith the values obtained from the previously established Barsom-Rolfe equation for the transition temperature range41. Namely, the mentioned authors41 found that the plane strain fracture toughness KH in the transition region is related to the Charpv energv CVN by
K;IC = 0.22 E (CVN)'"2	(9)
vvhere the Young modulus E is expressed in GPa, Klr is expressed in MPa m1\ and CVN in Joules. The 54 J Charpy energv commonlv used to determine the transition temperature of similar steels'21 corresponds roughly to 150 MPa m"2 vvlicn the relationship (91 is used. Quite a similar value is obtained also vvith Equation (8). vvhich means that both equations could be applied in the transition temperature range. It is not surprising because the loading rate and the noteh acuity do not have a great influence on the fracture-toughness behaviour at slightlv higher toughness values.
Besides the above mentioned equations of Barsom and Rolfe41" /(6),(8|/ there are also some other successful attempts. Namelv. Beglev and Logsdon" suggested that for lovv temperatures vvhere the behaviour is predominantly brittle, the fracture toughness (in MPa m1'2) may be related empirically to the vield stress als (in MPa) alone:
K„. = 0.0717 crys	(10)
Although Equation (10) essentially differs either from the equations of Barsom and Rolfe4'10 or from our equations (3) and (4). it is not in larger disagreement vvith our observations since for lovver Charpy shelves both approaches givc a considerably higher fracture toughness for as strain-aged steels vvith higher vield stress. The empirical correlation (10) is also in good agreement vvith K,(, data'4' so our linking the Eq. (10) vvith the equation of the type K„. = A (CVN)" into a single form (5) vvas relevant. The relatively high regression coefficient for this nevv correlation (5) i.c. a correlation vvhich is compatible vvith the Barsom and Rolfe4'"" approach as vvell as vvith the approach of Beglev and Logsdon51 confirm that Eq. (5) enables the best empirical estimation of the lovv-temperature fracture toughness KI( calculated on the basis of
conventional mechanical properties measured in the temperature range investigated as it is also shovvn in Table 2.
5. Conclusions
1.	The fracture toughness vvas measured in the temperature range of nil-ductility temperatures of nine non- and low-alloy structural steels either in as purchased or as strain-aged condition and it vvas correlated vvith the Charpy V-notch impact energies. Although the strain-aging reduces the Charpy energies i.c. provokes some shifts of Charpv values to higher temperatures, it also decreases the nil-ductility temperatures of such steels.
2.	The fracture toughness Klc of the investigated steels in the temperature range of nil-ductility temperatures can be sucessfully predieted either by the Equation (3) for as purchased steels or by the Equation (4) for steel as strain-aged. The estimation of KI(, vvhich vvould bc conservative enough for both slates of steels can bc given by the Equation (8) vvhich has also a simplc form.
3.	In general, the most suitable and stili plain procedure for obtaining the aetual fracture toughness K„ of structural steels in the temperature range of nil-ductility temperatures, being also compatible vvith various other concepts1'4'5'"", vvould comprise tensile and Charpy testing at lovver temperatures and further application of the generalized correlation (5).
The correlation (5) suggests that K„. and probably also Karnsi of :ls strain-aged steels vvould bc higher than that of as purchased steels vv ith the same Charpv energy because of the inereasing yield stress at strain-aging. Consequently, strain-aged steels have lovver NDT temperatures than those of as purchased steels.
REFERENCES
11 Rolfe, S.T.; Barsom, J.M.: Fracture and Fatigue Control in Structures, Applications of Fracture Mechanics. Englevvood Cliffs, Nevv Jersey, Prentice-Hall 1977. 21 Bamford, W.; Oldfield, W.; Marston, T.: An Improved Reference Fracture Toughness Procedure for Pressure Vessel Steels, 5th ICPVT. San Francisco, 9./14.9.1984, p. 932/65.
31 Kussmaul, K.; Demler, T.: Steel Research 63 (1992) No. 12. p. 545/53.
41 Barsom, J.M.; Rolfe, S.T.: Correlations Betvveen KIC and Charpy V-Notch Test Results in the Transition-Temperature Range, Impact Testing of Metals, ASTM STP 466. American Society for Testing and Materials, Philadelphia, 1970, p. 281/302. 51 Begley, J.A.; Logsdon, W.A.: Correlation of Fracture Toughness and Charpy Properties for Rotor Steels, Scientific Paper 71 -1E7-MSLRF-P1. VVestinghouse Research Laboratories, Pittsburg, July, 1971. 61 Tanaka, Y.; Ivvadate, T.; Suzuki, K.: Int. J. Pres. Ves. &
Piping 31 (1988), p. 221/236. 71 Dieter, G.E.: Mechanical Metallurgy, McGravv-Hill, 1986, p. 358.
81 Shen Wei et al.: Engng. Fracture Mcch. 16 (1982). p. 69/82. " Wanc Chang: Engng. Fracture Mech. 28 (1987). p. 241/250. 101 Barsom. J.M.: Engng. Fracture Mech. 7 (1975). p. 605/18. 1,1 Havvthorne, J.R.; Mager, T.R.: Relationship Betvveen Charpy V and Fracture Mechanics K„ Assessments of A533-B Class 2 Pressure Vessel Steel. ASTM STP 514, American Society for Testing and Materials, Philadelphia, 1972.
121 Faucher, B.; Dogan, B.: Metallurgical Transactions A. 19A (1988), 505/16. "
Langford, W.J.: Canadian Metallurgical Quarterly 19 (1980), p. 13/22. 141 Scarlin, R.B.: Shakeshaft, M.: Metals Technology, Januarv 1981, p. 1/9.