CORRELATION BETWEEN CHARPY-V AN D SUB-SIZE CHARPY

TESTS RESULTS FOR AN UN -IRRAD IATED

LOW ALLOY RPV FERRITIC STEEL

1 2 1
R. Schill , P. Forget and C. Sainte Catherine
1
CEA Saclay, DRN -DMT, F-91191 GIF-SUR-YVETTE
2
CEA Saclay, DTA-DECM, F-91191 GIF-SUR-YVETTE

ABSTRACT

Conventional Charp y-V is largely u sed for the surveillance program of RPV (Reactor Pressu re
Vessel) em brittlem ent by neutron irrad iation. Prolongation of service life of nu clear p ow er plants
and m ore stringent safety requirem ents are increasing the request for sm all test specim ens su ch as
su b-size Charp y. Furtherm ore, em pirical correlation form ulas betw een conventional Charpy-V
and fractu re tou ghness are largely used in practice. So, before using red uced specim en size, w e
have to look for possible relations betw een the tw o test geom etries.
The p resent investigation is p erform ed on a nuclear RPV typical low alloy ferritic steel 16MN D5
(eq. A508 Cl3). The batch used here w as extracted from a nozzle opening in the RPV. Instrum ented
im pact tests w ere perform ed on conventional Charp y-V and su b-size Charp y for a large
tem p eratu re range (-196°C up to 300°C). The fam ous hyperbolic tangent curve from W. Old field
[1] is fitted to the absorbed im pact energy results. This ad justm ent is then used in ord er to get
som e m eaning values for this test, such as for exam ple Upper Shelf Energy (USE), Ductile to Brittle
Transition Tem perature (DBTT) or tem perature for a given level of energy (TK). The uncertainty
on these p aram eters d eterm ination for a given confid ence level w ill be also given.
Based on these resu lts, d ifferent transferability correlation form ulas such as norm alisation factors,
VTT correlation and ORN L [2] m ethod are tried and com m ented . Furtherm ore, som e d iscu ssion
w ill be ad d ed on the fractu re m od e of the d ifferent specim ens and in particular for the su b-size
geom etry.

IN TROD UCTION
The possibility of u sing Su b-Size sp ecim ens to m onitor the m echanical properties of RPV steels is
receiving increasing attention. The p otential results of using such sm all sam ples are as follow s :
- the red uction of activated m aterials to u se for a given resu lt,
- the p ossibility of testing Su b-Size Charpy specim ens m anufactured from broken halves
Charp y-V p reviou sly tested su rveillance specim ens,
- the increase of tested specim ens for experim ental irrad iations.
If this size red u ction is nearly w ithou t any problem for som e specim ens such as tensile tests, it is
not the sam e for Charp y tests. The resu lts obtained on red uced specim ens m ust be com parable to
those on stand ard size and com pact tension (CT) specim ens. In fracture m echanics, it is called a
transferability p roblem and is alw ays a m atter of d iscussions and research even for m ore
conventional geom etry.
The Su b-Size Charpy geom etry (3x4x27 m m ) d eveloped in the fram e of ESIS TC 5 [3] is of high
interest. But, of course, before using another test than the w ell know conventional Charp y-V test, it
requires a large valid ation for the new test and it also need to establish strong connections w ith the
usual ones (Charp y-V and CT 25). This article w ill present the m aterial used for this valid ation, the
test d evices and the results in term s of absorbed transition curves. Then, correlations betw een Sub-
Size Charp y and Charp y-V alread y p rop osed in the literature w ill be tested and com m ented .
This article is linked to another one [4] d evoted to finite elem ent sim u lations.

MATERIAL AN D TEST D EVICES

Mat erial
The m aterial is a low alloy m anganese ferritic steel w ith nickel and m olybd enum ad d itions. It is
referenced 16MN D5 in French d esignation and is very near from A508 Cl.3. It is a nozzle opening
of a real nu clear pressure vessel m anufactured by Creusot Loire Ind ustry. It is p erfectly
rep resentative of 1300 MW PWR. This block has alread y been largely used in preced ing stu d ies
and in p articu lar in [5]. The chem ical com position is given in the follow ing table :
Table 1 : Chemical composition for PW R 16M N D5 steel.
C Mn Si N i Cr Mo Cu S P Al V
Mes. 3/ 4 Thickness 0.159 1.37 0.24 0.70 0.17 0.50 0.06 0.008 0.005 0.023 <0.01
Im p osed Mini 1.15 0.10 0.50 0.43
RCC-M Maxi 0.22 1.60 0.30 0.80 0.25 0.57 0.20 0.015 0.02 0.04 0.03

Therm al treatm ent of this steel includ ed three d ifferent steps :

- Qu enching : after austenitisation at 865/ 895°C d uring 4h40, a w ater quenching by prod uct
im m ersion w as achieved . This step w as repeated tw o tim es.
- Recovering : 630/ 645°C d uring 7h30 follow ed by a cooling in free air.
- Stress relief : w as perform ed at 610°C for 8h00.
The resu lting m icrostru ctu re is bainitic. The sam pling w as perform ed at the 3/ 4 thickness and the
d irection is axial (T) for tensile tests and axial-rad ial (T-S) for Charpy and toughness tests.

Conv ent ional Charpy -V

Span : 40 mm
45°

R = 0.25 10

55 10

Figure 1 : M ain dimensions of Charpy-V specimen, from EN -10045.

The test d evice u sed for conventional Charpy-V tests is an instrum ented 350 J Tinius-Olsen Im p act
m achine. It is in conform ity w ith EN -10045 for the m achine and also for specim en geom etry.

Sub-Siz e Charpy
The test d evice used here is a 50 J Zw ick Mini Charp y w ith an au tom atic p ositioning of the
specim en. The striker is instru m ented w ith stain gauges and an optical sensor is installed on the
p end u lu m m ass for d isplacem ent m easurem ent. Specim ens, striker and anvils geom etries are in
accord ance w ith ESIS TC5 Draft [3].
Span : 22 mm
60

R = 0.1 4

27 3
Figure 2 : M ain dimensions of Sub-Size Charpy specimen, from [3].

Table 2 : M ain dimensions for Charpy and Sub-Size Charpy, from EN -10045 and [3].
Dim ensions (m m ) Charp y-V Su b-Size Charp y
Specim en Length 55 27
Specim en H eight 10 4
Specim en Wid th 10 3
Specim en N otch Depth 2 1
Specim en N otch Angle 45° 60°
Specim en N otch Tip Rad ius 0.25 0.1
Distance betw een Anvils (Span) 40 22
Anvil Rad iu s 1 1
Angle of Taper of Anvils 11° 11°
Angle of Taper of Striker 30° 30°
Rad ius at tip of Striker 2 2
Maxim u m Wid th of Striker - 18

Specim ens and im p act tools d im ensions are com pared in the follow ing table 2. Load calibration
has been p erform ed und er static cond ition w ith a reference piezo-electric sensor. For this, a sp ecial
d evice w as d esigned in ord er to press the tw o load cells one against the other one. The evaluation
of tem p eratu re loss d uring the transfer from the furnace to the support until the im pact has alw ays
been a m ajor concern for experim entalists. Size red uction of specim ens increases this problem . So,
a carefu l calibration of this tem p eratu re loss has been perform ed w ith an instrum ented Su b-Size
Charp y sp ecim en. A 0.5 m m K therm ocouple w as introd uced in the specim en by one of its
extrem ity and the top of the therm ocouple is located ju st u nd er the notch. The transfer tim e from
the fu rnace u ntil the im pact has been m easured and is equal to 1.85 s. Then, series of tests w ith the
instrum ented sp ecim en has been perform ed at d ifferent tem peratures and thus give tem p eratu re
loss after 1.85 s. A tem perature loss graph has been d one from -150°C up to +300°C. The correction
to ap p ly varies respectively from +3°C d ow n to -10°C. This curve is not perfectly linear in
p articu lar for p ositive tem peratures.
TEST RESULTS

Absorbed Energy Transit ion Curv e

For Charp y-V, m ore than 100 specim ens w ere used for the transition curve. This gives a reliable
shape and w ell id entified characteristic p oints and it w ill also be useful for a scatter evaluation.
180.0
16MN D 5
160.0 Charpy-V Tests
in T-S orientation
140.0

120.0
Absorbed Energy (J)

100.0

80.0
68 J

60.0
E (J) = A + B *tgh((T(°C) - To)/C)
A = 86.2 LSE = 3.8
40.0
B = 82.4 USE = 168.6
To= -4.4
20.0 C = 43.3
TK68 (°C) = -14.2
0.0
-200 -150 -100 -50 0 50 100 150 200 250 300
Tem p eratu re (°C)

Figu re 3 : A bsorbed energy transition curve of 16M N D5 for Charpy-V specimen.

9.00
16MN D 5
8.00 Mini Charpy Tests
in T-S orientation
7.00

6.00
Absorbed Energy (J)

5.00

4.00
3.1 J
3.00 E (J) = A + B *tgh((T(°C) - To)/C)
A = 3.85 LSE = 0.61
2.00 B = 3.24 USE = 7.09
To= -71.13
C = 23.39
1.00 TK3.1 (°C) = -76.6

0.00
-200 -150 -100 -50 0 50 100 150
Tem p eratu re (°C)

Figu re 4 : A bsorbed energy transition curve of 16M N D5 for Sub-Size Charpy specimen.

For Sub-Size geom etry, m ore than 60 specim ens have been broken giving also a good d efinition of
the transition curve. Then, characteristic values have been d eterm ined by the Old field [1]
hyp erbolic tangent ad ju stm ent w hich is given by the follow ing equation :
  ( T − T0 ) 
E( J ) = A + B ⋅ tanh   (1)
 C 
w here A, B, C and T0 are param eters and
are graphically d efined in figure 5.
Relation w ith conventional characteristic
points can be established , such as :
USE = A + B
LSE = A − B
DBTT = T0
Transition Onset = T0 − C
Slope = B / C
Where USE is the Upper Shelf Energy,
LSE the Low er Shelf Energy and DBTT the
Ductile Brittle Transition Tem perature.
The next table gives these param eters
values for the tw o specim en geom etry.
Figu re 5 : Hyperbolic tangent adjustment as defined by [1].
Table 3 : Characteristic parameters of the
hyperbolic tangent adjustment for full and sub-size.
E (J) A (J) B (J) To (°C) C (°C) LSE (J) USE (J) TK (°C)
Charp y-V 86.2 ± 4.3 82.4 ± 5.0 -4.4 ± 4.0 43.3 ± 6.4 3.8 ± 9.3 168.6 ± 9.3 -14.2
Mini Charp y 3.8 ± 0.46 3.2 ± 0.63 -71.1 ± 7.4 23.4 ± 15.0 0.6 ± 1.1 7.1 ± 1.1 -76.6

In this table, w e also ind icate the tem peratu re ind ex corresp ond ing to a p articu lar energy level
w hich is com m only equal to 68 J for Charpy-V and 3.1 J for Sub-Size Charpy. For each param eters,
confid ence intervals on the estim ates of each param eter are given for a confid ence level of 95% (±
2σ). These values w ere obtained via the use of a conjugated grad ient for the fitting and a
sensitivity analysis. The correspond ing upper and low er confid ence bound s for the p red icted
energy level are given on the graphs only for the transition range (d ue to statistical inference, the
valu es given for low er and upper shelves are too high [1]). For temperature ind exes, the p red icted
95% confid ence interval are not ind icated in table 3 because they are non sym m etric. For Charpy-V
and Su b-Size Charpy, they are respectively equal to {-2.8°C,-26.5°C } and {-108.5°C,-58.5°C }. These
resu lts show that scatter associated to Sub-Size Charpy seem s to be greeter than the one observed
for Charp y-V, bu t this requires further confirm ation.
From table 3, w e also noticed that the DBTT shift (∆DBTT) is equal to 66.7°C and the tem p eratu re
ind ex shift (∆TK) is equal to 62.5°C. The ratio betw een upper shelves energies is 23.8 and only 6.3
betw een low er shelves energies. Instrum ented curves giving load versus tim e or d eflection are
given and u sed in the com panion article [4].

CORRELATION BETWEEN SUB-SIZE AN D FULL SIZE CHARPY

Upper Shelf Energy (USE)

A lot of w ork has alread y been d one in ord er to correlate USE betw een fu ll size and su b-size (e.g.
Corw in [6], Schu bert [7], Loud en [8] and so on). Lucon [9] gave a synthetic overview of these
ap p roaches. Proposed form ulas are often based on a N orm alisation Factor (N F) of energy for the
USE. The d efinition of this N F is as follow s :
  USE Charpy− V 
NF =   (2)
 USESubSize Charpy 
The valu e of N F can be established either d irectly from experim ents, either through quite sim ple
m echanical based relations. The m ost com m only u sed exp ressions give valu es inclu d ed betw een
8.9 and 26.5. The lack of these approaches is that they are based only on elastic consid erations and
are not able to take into account the yield ing, the d u ctile tearing nor constraint change. The
exp erim ental based approach gives values w ithin 21.6 ± 2.6 [9] for m ore than 10 d ifferent m aterials
(m ainly p ressu re vessel steels).
For the steel consid ered in the present investigation, the experim ental N F value is equal to 23.8
and so, is in accord ance w ith the value rep orted by other investigators su ch as [9] and [2].
Another proced u re proposed for the evalu ation of N F is the VTT ap p roach w hich is based on a
p ow er law fit of the correspond ing J-integral versus d uctile stable crack grow th ( J = C ⋅ ∆a m ) [9].
This form ulation has been established on an exp erim ental basis. The p articu larity of this equation
is that it is transcend ent as m is d ep end ant on the USE Charpy− V . Z and n w ere found to be
resp ectively equ al to 184 J and 0.34 [9].
1+ m n
10  8   USE Charpy− V 
NF = ⋅   with m=  (3)
B  b  Z 
The next figu re is giving a graphical representation of these results. The experim ental reported N F
valu e for the steel und er consid eration (23.8) is in very good agreem ent w ith the interval
(21.6 ± 2.6) based on other experim ental results or w ith the VTT correlation.
225.0

200.0

N Fexp = 23.8
175.0
USE Full Size (J)

150.0
N F = 18.9

125.0
N F = 24.2
100.0
VTT Correlation
Z = 184.0 J
75.0 n = 0.34

50.0
3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
USE Su b-Size (J)

Figu re 6 : Experimental and empirical values for N ormalisation Factor between USE.

Duct ile Brit t le Transit ion Temperat ure (DBTT) Shift

The transition tem perature shift (∆DBTT) is d efined by :
∆DBTT = DBTTCharpy− V − DBTTSub−Size (4)
Of course, this relation can be applied to DBTT bu t also to tem p eratu re ind exes (TK). We
exp erim entally find 66.7°C and 62.5°C respectively. This is in very good agreem ent w ith the valu e
(65 ± 30°C) rep orted by E. Lucon et al. [9].

Low er Shelf Energy (LES)

For LSE, the N F ratio can also be ap p lied . The value experim entally obtained on 16MN D5 is equ al
to 6.3 w hich is low er than the value (8.9) reported by Solokov et al. [2]. N ote that this value is not
often rep orted and that the fitting proced ure has a high influence on the precision of LSE
d eterm ination.

O RNL Correlat ion

Solokov et al. [2] have p rop osed a norm alisation proced ure applicable to the w hole transition
cu rve. The Su b-Size Charpy absorbed energy is first re-scaled and then a tem perature shift is
ap p lied . This gives a transition cu rve w hich is com p arable to the Charp y-V cu rve. These tw o step s
are d etailed as follow s :
1- Absorbed energy is re-scaled by the follow ing relation :

NF = (1 − SFA( T)) ⋅ NFLSE + SFA( T) ⋅ NFUSE (5)

w here SFA(T) is the Shear Fracture Appearance and can be m easured either from fracture
surface observation (i.e. surface ratio of d u ctile tearing and cleavage p artsor via em pirical
form u las [3]). The obtained valu e varies from 0 (purely brittle) to 1 (only d uctile) w ith
increasing tem perature.
2- Sub-Size tem peratures are shifted w ith the follow ing relation :
TCharpy− V = TSub −Size + ∆DBTT (6)

225.0
16MN D 5 En (J) = A + B *tgh((T(°C) - To)/C)
200.0 in T-S orientation

175.0 Ch arp y-V Tests

Ch arp y-V H yp erbolic Fit
150.0 Su b-Size Charpy Tests
Charpy V (10x10x55 mm) :
Su b-Size Charpy H yp erbolic Fit
125.0 A = 86.2 LSE = 3.8
En (J)

B = 82.4 USE = 168.6

100.0 To= -4.4
C = 43.3
75.0 Mini Charpy (3x4x27 mm) :
A = 90.6 LSE = 3.3
50.0 B = 87.3 USE = 177.9
To= 8.0
25.0 C = 23.9

0.0
-200 -150 -100 -50 0 50 100 150 200 250 300
Tem perature (°C)

Figu re 7 : ORN L methodology applied to Sub-Size Charpy results and compared to Charpy-V .

This m ethod has been ap p lied here w ith calibrated coefficients (i.e. N F factors and ∆DBTT are
those m easured on 16MN D5) to Sub-Size exp erim ental p oints in ord er to p u t them in the Charp y-
V absorbed energy graph. Then, the hyperbolic tangent fit w as perform ed by consid ering those
transform ed p oints. So, the correspond ence betw een the tw o curves is quite good , in particular for
the u p p er and low er shelves energies. DBTT shift is a little too big giving thus a d ifference of abou t
12°C betw een the tw o curves. But an interesting feature is observable on these curves : the slop es
B/ C are not the sam e. C rem ains approxim ately constant d u ring the transform ation and B is
increased in ord er to get the correct u p p er shelf energy. This m eans that the second step of the
m ethod ology is p robably not so sim p le as a sim ple tem perature shift. A slope correction m ust also
be ap plied in Eqn. 6. This m eans that a coefficient equ al to the ratio C Charpy− V CSub −Size (1.9 in our
case) should be applied to the tem peratu re scale. With su ch a transform ation, the tw o cu rves
exhibit a very good agreem ent, but this m od ification need s fu rther valid ation.

CON CLUSION S
Zw ick 50 J Sub-Size pend ulum is able to test ESIS TC 5 [3] sp ecim en typ es. Before the tests, a first
p hase w as d evoted to calibration of the load cell and the op tical sensor. Attention w as also p aid to
tem p eratu re loss d uring the specim en transfer from the furnace to the anvils until im pact occurs.
Then, transition curves have been established w ith Charpy-V (10x10x55 m m ) and Sub-Size Charpy
(3x4x27 m m ) sp ecim ens for un-irrad iated 16MN D5 RPV low alloy ferritic steel (eq. A508 Cl.3).
Characteristic valu es have been d eterm ined by Old field hyperbolic tangent ad justm ent [1].
Confid ence intervals on param eters and on p red icted energy show that scatter associated to Su b-
Size Charp y seem s greater than for Charpy-V bu t this need fu rther confirm ation.
The ratio betw een upper shelves experimentally reported is equal to 23.8. This valu e is in
accord ance w ith the values reported in the literature for sim ilar steels [9] and is w ell pred icted by
the VTT correlation. It is only 6.3 for low er shelves. The transition tem perature shift (∆DBTT)
betw een the tw o specim en geom etry is equal to 66.7°C w hich is also in good agreem ent w ith the
values given by Lu con [9]. A m ethod ology d eveloped by Solokov [2] at ORN L w as app lied in
ord er to transform the Sub-Size Charpy transition cu rve in the Charp y-V scales. This w as d one
w ith ad ap ted coefficients and w e noticed that a sim p le tem p eratu re shift is p robably insufficient.
A slop e correction shou ld also be ap p lied to the tem perature relation.

REFEREN CES
1. OLDFIELD W., "Curve Fitting Impact Test Data : A Statistical Procedure", ASTM Stand ard ization
N ew s, N ovem ber, (1975), pp. 24-29.
2. SOKOLOV M.A. and ALEXAN DER D.J., "A n Improved Correlation Procedure for Subsize and
Full-Size Charpy Impact Specimen Data", N UREG/ CR-6379, ORN L-6888, (1997).
3. ESIS TC 5, "Proposed Standard M ethod for Instrumented Impact Testing of Sub-Size Charpy-V N otch
Specimens of Steels", Draft 8, European Structural Integrity Society (ESIS), TC 5, May, (1999).
4. SAIN TE CATH ERIN E C., H OURDEQUIN N ., GALON P. and FORGET P., "Finite Element
Simulations of Charpy-V and Sub-Size Charpy Tests for a Low A lloy RPV Ferritic Steel", ECF 13,
This volu m e.
5. ROSSOLL A., "Fracture Toughness Determination of a Low A lloy Steel by Instrumented Charpy
Impact Test", Ph. D. Material Science, Ecole Centrale d e Paris, N ovem ber, (1998).
6. CORWIN W.R. and H OUGH LAN D A.M., "Effect of Specimen Size and M aterial Condition on the
Charpy Impact Properties of 9Cr1M oV N b Steel", ASTM, STP 888, (1986), pp. 325-338.
7. SCH UBERT L.E., KUMAR A.S., ROSIN SKI S.T. and H AMILTON M.L., "Effect of Specimen Size
on the Impact Properties of N eutron Irradiated A 533B Steel", J. N ucl. Mat., 225, (1995), pp. 231-237.
8. LOUDEN B.S., KUMAR A.S., GARN ER F.A., H AMILTON M.L. and H U W.L., "The Influence of
Specimen Size on Charpy Impact Testing", J. N ucl. Mat., 155-157, (1988), pp. 662-667.
9. LUCON E., CH AOUADI R., FABRY A., PUZZOLAN TE J-L. and Van WALLE E.,
"Characterising M aterial Properties by the Use of Full-Size and Sub-Size Charpy Tests : A n Overview
of Different Correlation Procedures", ASTM, STP 1380, (1999) .