The stress corrosion cracking of austenitic stainless steels in H3BO3 + NaCl solutions between 100°C and 200°C

The stress corrosion cracking of austenitic stainless steels in H3BO3 + NaCl solutions between 100°C and 200°C

Corrosion Science, Vol. 35, Nos 1-4, pp. 44,'~455, 1993 Printed in Great Britain. THE STRESS STAINLESS 0010-938X/93 $6.00 + 0.00 © 1993 Pergamon Pre...

802KB Sizes 0 Downloads 25 Views

Corrosion Science, Vol. 35, Nos 1-4, pp. 44,'~455, 1993 Printed in Great Britain.

THE STRESS STAINLESS

0010-938X/93 $6.00 + 0.00 © 1993 Pergamon Press Ltd

CORROSION CRACKING OF AUSTENITIC STEELS IN H3BO 3 + NaC1 SOLUTIONS BETWEEN 100°C AND 200°C

R. KASRI, J. M. OLIVE, M. PUIGGALI and D. DESJARDINS Laboratoire de Mdcanique-Physique, Universitd Bordeaux I - - U R A CNRS 867,351, cours de la Liberation, 33405 Talence Cedex, France

A b s t r a c t - - T h e stress corrosion cracking of 316L austenitic stainless steel in H3BO 3 + NaCI aqueous media between 100 and 2(1()°C has been studied. The results obtained from slow strain rate and constant load tests, the polarization curves, the rapid tensile straining tests and the fractographic investigation of the fracture surface by scanning electron microscopy, permit the examination of the nature of mechanisms at work. The experimental crack propagation velocities and those calculated from Faraday's law have been compared. The analysis of the electrochemical behaviour in the absence of stress has revealed the potential ranges of m a x i m u m stress corrosion cracking susceptibility. Etch-pitting and stereographic observations are employed in order to determine the cracking crystallography. The transgranular fracture surfaces show crystallographic morphology and though the cracking is cleavage-like and is discontinuous in some cases, the cracks propagate mainly under the control of localized anodic dissolution. This process seems to be the limiting factor of the p h e n o m e n o n whatever the proposed mechanisms may be. In fact. the crack propagation velocity seems to be clearly consistent with a film rupture-dissolution-repassivation process, although it proves insufficient to quantify the p h e n o m e n o n . The influence of the various environment parameters (temperature, concentration of chemical species, oxygen) which have been studied, has confirmed the importance of the localized anodic dissolution on the cracks propagation.

INTRODUCTION

IN THE NUCLEARindustry, austenitic stainless steels are used as the structural material for the pressurized water reactors primary circuit (PWR). In the auxiliary circuits of these reactors, stainless steels are subjected to various cases of stress corrosion cracking (SCC) in boric acid solution. 1 Cracks result essentially from the presence of chloride and oxygen which are the main polluting agents in these circuits. In atmospheric pressure, studies on some physico-chemical properties of boric acid and chloride solutions have been undertaken to specify electrolyte behaviour. These studies have underlined a different structure of solutions which depends on the concentration of chemical species. A small variation of physico-chemical parameters is sufficient to modify completely boric acid solution behaviour. 2 The present work has used various studying methods of SCC to define the conditions required to obtain SCC of 316L austenitic stainless steel in boric acid and chloride solutions between 100 and 200°C. The influence of various elements, such as boric acid concentration, chloride levels and oxygen, has been studied in order to determine the mechanism at work during initiation and propagation of cracks. EXPERIMENTAL

METHOD

The chemical composition (wt%) of 316L austenitic stainless steel used in atmospheric pressure study was: C: (I.(/15, Cr: 17.35, Ni: 12.35. Mo: 2.2, Si: 0.51, Mn: 1.02, S: (/.(116, P: 0.3. The specimens were 36(I m m gauge length and 1.5 m m diameter. The composition of steel used at 20(1°C was: C: 0.002, Cr: 17.(/1, Ni: 13.8, Mo: 2.21, Cu: (/.(132, Si: 0.38, Mn: 1.63, S: 0.013, P: 0.018, N: 0.019, Co: 0.21. The 443

444

R. KASRIet al.

samples were 85 mm gauge length and 4 mm diameter. All specimens are annealed and water quenched. Before each experiment, the electrodes received an identical surface treatment, i.e. mechanical polishing followed by an acid etching. In atmospheric pressure tests, the electrolyte was boiling concentrated solutions of boric acid (270 g 1- l H3BO3) containing chloride (2 g I J NaCI)3 and for 200°C tests, it was a solution with various concentrations of boric acid and chloride. Rapid strain rate experiments were conducted at a strain rate of 3 × 10 2 s i. The constant load tests were conducted at free corrosion potential and the applied stress was 450 MPa. All potentials were measured with respect to the saturated sulphate electrode (SSE). Slow strain rate tests were performed in an Hastelloy autoclave at 200°C, 12 bar and strain rate of 2 × 10 7 S 1. After stress corrosion tests, specimens were observed by scanning electron microscopy.

SCC A N D E L E C T R O C H E M I C A L B E H A V I O U R AT

100°C

E l e c t r o c h e m i c a l results

T h e e l e c t r o c h e m i c a l b e h a v i o u r o f 316L steel has b e e n s t u d i e d in o r d e r to i n v e s t i g a t e the r e s i s t a n c e o f the b u i l t - u p passive film. T w o d i f f e r e n t m e t h o d s w e r e used: cyclic v o l t a m m o g r a m s w e r e s t u d i e d to d e t e c t the electrical b r e a k d o w n a n d h e a l i n g o f the passive film, a n d r a p i d strain r a t e tests w e r e u n d e r t a k e n at fixed p o t e n t i a l n e a r the c o r r o s i o n p o t e n t i a l in o r d e r to define the m e c h a n i c a l b r e a k d o w n o f t h e passive film a n d to d e t e r m i n e crack p r o p a g a t i o n velocity c a l c u l a t e d f r o m F a r a d a y ' s law. F i g u r e 1 shows a n o d i c p o l a r i z a t i o n curves o b t a i n e d by a p p l y i n g t h e p o t e n t i a l s w e e p of 20 m V m n -1. A hysteresis p h e n o m e n o n was o b s e r v e d a n d i n d i c a t e s a p o o r passive l a y e r p o t e n t i a l z o n e w h e r e l o c a l i z e d c o r r o s i o n is f e a r e d . W i t h o x y g e n a d i s p l a c e m e n t o f the hysteresis l o o p to m o r e a n o d i c p o t e n t i a l s is o b s e r v e d . R a p i d strain rate tests h a v e b e e n c o n d u c t e d at strain r a t e o f 3 × 10 -2 s -1 a n d total d e f o r m a t i o n o f 2 % . T h o s e tests a t t e m p t to r e p r o d u c e the passive film r u p t u r e d i s s o l u t i o n - r e p a s s i v a t i o n p h e n o m e n o n i n t e r v e n i n g in cracking. B e s i d e s , t h e y also p e r m i t t e d the a p p r e c i a t i o n o f b o t h d e p a s s i v a t i o n a n d r e p a s s i v a t i o n kinetics of steels. F i g u r e 2 shows the c u r r e n t - t i m e a n d e l o n g a t i o n - t i m e r e c o r d e d with b o t h the p r e s e n c e a n d a b s e n c e o f o x y g e n . T h e c u r r e n t s o b t a i n e d f r o m the steel d e p a s s i v a t i o n show curves w h e r e t h e slope a n d m a x i m u m c u r r e n t at d e f o r m a t i o n are m o r e i m p o r t a n t with o x y g e n . T h e r e p a s s i v a t i o n kinetics has b e e n e x p r e s s e d by: 4

4 --

Withouto x y ~

3 -2--

with o x y g ~

0 --

~ ~ f f r e e corrosionvotential' A: 224 rnV(Without02) B: -140mY(with02)

-I --2--

B

-3--

I

-lOOO

800

600

400 200 E (mV/E.S.S)

o

200

400

Fl~. 1. Cyclic polarization curves (20 m V rain-l): 270 g I-l H 3 B O 3 + 2 g 1-l NaCI.

445

SCC of steels in H3BO 3 + NaCI

o

- - - o - - I (withoutoxygen) A I (withoxygen)

L(mm)

10 --

8

--

0.5

--

0.4

6 --

-- 0.3

4 --

-- 0.2

2 --

-- 0.1

0

tttt,rtt

~ t l t t t t t t : ~ : l ~ t r t t t : t

500 Fro. 2.

--

t Cms)

1000

0

1500

Anodic current and elongation evolutions during 316L steel fast straining: 270 g I i H3BO 3 + 2 g l iNaCi.

i/io = (t/f1)-" where i0 is the durrent at time to when repassivation occurs. The repassivation rate decreases in the presence of oxygen (Fig. 3). Then, oxygen increases the dissolution current density and slows down repassivation.

Stress corrosion cracking results Constant load test results have been given in terms of initiation time and crack propagation velocity, and then correlated with the electrochemical study. The propagation time of cracking (tp) is determined from logarithmic creep vs time curves. 5 Experimental crack propagation rate is calculated by: 6

0 --

-0.4

"° ~_h~,¢.~ y~ a ~ ' x~ ~/ x .

i/i0 = (t/to)-~

-0.8

,...a

-1.2 without

oxyg~en ~

-1.6

n= 2 -2 0.1 FIG. 3.

''''1''''1''''1''''1'"'1''''1 0.2

0.3

0.4 Log(tA~

0.5

Repassivationcurvesof316Lsteel:270gl

0.6

0.7

IH3BO3+2gl

INaCl.

446

R. KASR!et al. TABLE 1.

RESULTSOFCONSTANTLOADTESTSIN270 g I-l H3BO3 + 2 g I l NaCI BO1LIN~SOLUTION(450 MPa)

Steel

Oxygen

316L

Without With

tr

ta

Vp

(mV(SSE))

E

(rain)

(min)

(pm h -1)

-224 - 140

13 639 1582

12 745 1340

3 8

Vp = Lf/tp

where Lf is the deepest crack length observed on rupture surface by SEM, and tp is the propagation time. Table 1 gives the SCC results: initial corrosion potential, rupture and initiation times and crack propagation rates. It shows a reduction in rupture time and an increase in crack propagation rate in presence of oxygen. Then, oxygen influence is shown by an accelerating effect on both initiation and propagation of cracks. When they are placed on the polarization curves (Fig. 1), the initial corrosion potentials are situated in passive film instability range. The potential with oxygen is more anodic and is situated near the transpassive range. Thus, the corrosion potentials are situated in a frontier zone where localized dissolution can occur. The experimental crack propagation velocities have been correlated with those calculated from Faraday's law. 7 The values of the latter are estimated by supposing that crack propagation is continuous and results only from anodic dissolution current. The predicted maximum crack propagation rate has been given by: Vp = i a ( M / z p F )

where ia is an anodic dissolution current density which is evaluated during the metal depassivation. By supposing that the passive film has a brittle mechanical behaviour, the bare area has been given by: 6 A• = ~0{[(AL + Lo)/Lo] 1/2 - 1} where ~0 is the initial surface, L0 the initial length in contact with electrolyte and A~ the bare surface corresponding to AL elongation. 8 The predicted crack propagation velocities are 7 . 3 p m h -1 in the absence of oxygen and 13/~m h -1 with oxygen. These values show clearly the oxygen influence on the increase of anodic dissolution. In fact, the propagation rate is proportional to current density which is more important in presence of oxygen (Fig. 2). This occurs because corrosion potentials become more anodic with oxygen (Fig. 1). These results have confirmed those obtained by constant load tests which have showed that oxygen accelerates cracking. The results obtained in atmospheric pressure show that oxygen destabilizes the passive film and increases anodic current density. The predicted crack propagation velocities that are the upper limit in estimation of this rate are more important than those measured experimentally and have confirmed the oxygen effect. Rupture surfaces, observed after constant load tests, have shown a crystallographic facet feature of fracture in which a cleavage process intervenes. 9 Thus, to quantify this phenomenon, the anodic dissolution model is not necessarily the only one to be taken into account.

SCC of steels in H3BO~ + NaCI

STRESS

CORROSION

CRACKING

447

AT 200°C

This part of the work has permitted the study of the influence of the main polluting agents (oxygen and chloride) which are found in the auxiliary circuits of pressurized water reactors. This study has been undertaken by using a slow strain rate test at a strain rate of 2 x 10 - 7 s - l . The SCC sensitivity has been studied by comparing tensile curves obtained in corrosive media and in inert media at the same temperature and strain rate. The parameters used to quantify SCC sensitivity are the ultimate tensile stress and the rupture strain.

Influence of boric acid concentration Boric acid influence has been studied by fixing the chloride level at 2 g l- ~NaCl. True tensile curves show a lowering of both ultimate tensile stress and rupture strain of steel in corrosive media for various levels of boric acid (0, 100 and 270 g 1-J H3BO3), with respect to inert media (Fig. 4). Thus, the SCC resistance is higher in 270 g 1-1 H3BO 3 at the strain rate used of 2 x 10 -7 s- 1 than 100 g 1-1 concentration which gives maximum sensitivity of SCC. After slow strain rate tests, the surface of gauge length and rupture surface of steel have been observed by SEM. A diameter plane section from the specimen has undergone a metallographic examination, and the crack length and number of cracks were measured in order to study the crack distribution. The SEM observation showed that crack morphology is transgranular in the cases where there is not boric acid and in 270 g 1 ' HsBO3 solution. In 100 g 1-1 H3BO 3 solution, the cracking is transgranular in the beginning and then is intergranular. The crack number evolution vs crack length (Fig. 5) has showed that cracks are numerous and very short (crack depth <50 pm) in 270 g 1-1 H3BO3. In 100 g 1-1 H3BO 3 solution, there are few deep cracks (depth varying between 250 and 650/xm). In absence of boric acid, cracks are numerous and deeper than in 270 g 1-1 H3BO3. This result has showed the localized and fast aspect of SCC phenomenon in 100 g 1-1 HsBO 3 + 2 g 1-1 NaCI solution. The actual results have led to two possible interpretations of the boric acid role. In the first case, the addition of boric acid has led to the formation of a protective layer constituted by neutral compounds obtained from the combination of borate 800-70O inert r~dia

6O0 500 t~

400

300 -

so~ + 2 s.l 1 ~cl

200 ::

g.l" 1 NaCI

lOO o 5

tO

15

20

25

30

35

40

(%) FIG. 4.

True tensile curves (2 × 10 7 s - i ) ; T = 200°C. Influence of boric acid concentration.

448

R. KAsru et al.

t~ ÷

I

'

I

'

I

'

I

'

I

'

@

r-

3

r~

I ....



I ....

I ....

I ....

I ....



o

i ....

Z. u

o

l ' ' ' ' l ' ' ' ' l ' ' ' ' l ' ' ' o

,=

~.

S C C of steels in H s B O 3 + N a C I

449

anions and sodium cations. At 100 g 1-1 H3BO3, this layer has played a barrier role which, considering the important size of formed compounds, has slowed down the CI- ion displacement. Yet this has not occurred everywhere and has only led to lowering of initiation sites, and then to a localization of cracking and an increase in the dissolution kinetics. Consequently, the crack time has been shortened. At high boric acid concentration (270 g l - l ) , the number of formed compounds is very important and the protective layer becomes thick. This has reduced the chloride effect by stopping the C1- ion crossing. In fact, a study with the absence of chloride voluntary addition has showed that SCC behaviour is not modified, t° The plastic strain level reached in this case has led to a destruction of passive film and then to the creation of a lot of anodic sites leading to a dense cracking of steel which occurred at the end of the test. In the second case, we can imagine that boric acid has an activating role which has prevented the localization of corrosion in the situation where imposed strain rates were too important. In order to confirm this suggestion, it is necessary to carry out slow strain rate tests with varying strain rates to define the depassivation critical rates, l~ which can lead to the understanding of the boric acid role for different boric acid concentrations.

Influence of chloride levels The solution containing 270 g 1-1 H3BO 3 has been chosen to study the influence of chlorides. True tensile curves obtained in inert media and in corrosive media show, for all cases of sodium chloride levels (0, 2, 5 and 10 g I l NaC1), a lowering of both ultimate tensile stress and rupture strain of steel (Fig. 6). The sensitivity is a maximum in 5 g 1 t NaCI solution and intermediate in 10 g 1 l NaCI. Addition of 2 g I i NaCI has not modified the (o, c) curve. The SEM observations have showed numerous and short transgranular cracks in 0 and 2 g !- 1NaC1. In 5 g 1- l NaCI, cracks are few, wide and deep and are transgranular in the beginning and then intergranular. In 10 g I t NaC1, the rupture surface has presented transgranular cracking and cracks are less wide and deep than in 5 g I I NaC1. m These results have showed that SCC phenomenon is more localized in 5 g 1 NaCI solution, where there are few deep cracks, than in 2 g I 1NaCI where there are 800-7.

700

inert media

4

60O 500

~

400

.

0g'l'l

30O 20O

1

1oo 21 0

.

llllllllilllllllllll4111rllllllllill'll S 5

10

15

20

25

30

35

40

(%)

FIG. 6.

T r u e tensile c u r v e s (2 × 10 7

t): T = 21)(I°C. I n f l u e n c e of c h l o r i d e levels.

450

R. KASR! et al.

numerous short cracks. This has showed that the increase in chloride level has led to the creation of anodic zones which result from the local rupture of passive film and where dissolution kinetics is important. In 10 g l - 1NaC1, a delocalization of SCC has occurred and slowed down the cracking propagation process. This case is situated in the limit of general corrosion as it has also been observed on 304 steel in H N O 3 solutions for sufficient chloride levels. 12

Influence of oxygen Oxygen influence has been studied in 100 g 1-1 H3BO 3 + 2 g 1-1 NaC1 solution which has given a maximum sensitivity of 316L steel to SCC for the strain rate used of 2 × 10 -7 s -1. The tests have been carried out by adding oxygen in solution on the one hand, and on the other hand excluding oxygen by purging with argon before the strain rate test. Figure 7 shows that suppression of oxygen has increased the SCC resistance of 316L steel and that its presence has accelerated the cracking phenomenon. The metallographic examination has showed that cracks are numerous and very short in de-aerated solution (crack depth <30/~m). With oxygen, cracks are few and very deep (Fig. 8). Oxygen can play the role of anodic polarization which brings the steel near the transpassive range with increasing dissolution kinetics and decreasing repassivation rate. MICROGRAPHIC

STUDIES

OF RUPTURE

SURFACES

The fracture surface inferred from the slow strain rate tests has been examined by SEM in order to determine the feature aspect of the rupture surface which would identify the cracking mechanisms. The SEM observations have showed as a rule that crack initiation occurs according to the transgranular mode. In some cases, when stress intensity factor becomes important 13 and when electrochemical conditions vary at the crack tip, ]4 transition from transgranular to intergranular mode can occur. Figure 9(a) shows a scanning electron micrograph of the specimen brittle region obtained in 2 g l - 1NaC1 at 200°C and strain rate of 2 × 10 -7 s- 1. The cracking mode appears to be transgranular and the cracks have crystallographic character-

800 --: 70O

-

inert media

: s00

D

400 300

--

:

1.001g.l "1 H3BO3 + 2g.l" l NsCI+ Argon

~.~P"

0 T''"I''"I''"I""I""I""I'"'I""I 0 FIG. 7.

5

10

15

20 (~)

25

30

35

40

T r u e tensile curves (2 x 10 7 s 1); T = 200°C; 100 g 1- I H3BO 3 + 2 g 1-I NaCl. Influence of oxygen.

Fro. 9. Scanning electron micrographs of specimen brittle region of 316L steel tested at strainrateof2× l(I 7s I:(a)2gl tNaCI. T=200°C;(b)270gl iH~BO~+2gl i N a C l + oxygen, T = 200°C, (c) Micrograph of zone three.

451

SCC of steels in H3BO 3 + NaC1

453

Z

-t-

II I , , , i , , , ~ , , , i , ~ , l , , ,

18 o

0

+

z 2 -

~

+

o6

Z

ol ÷

d ~

!11" I

~

I

,

I

,

i

,

l"<~qlnrilI :lio~.ll ~'~llli~

i

;

454

R. KAsmet al.

istics consisting of primary facets separated by ligaments which indicate propagation direction. Striation-like markings, which are perpendicular to the propagation direction, have been observed. They seem to interact with ligaments and the distance between these lines is approximately 1 ~tm. The same markings have been observed in some cases and are produced by load pulsing. 15 Then, these lines correspond to crack-arrest markings which indicate that the propagation of transgranular SCC occurs by discontinuous cleavage. Figure 9(b) shows a transition from an intergranular mode to a transgranular one obtained in a solution containing 100 g 1-1 H3BO 3 + 2 g 1-1 NaC1 + oxygen at 200°C and strain rate of 2 x 10 7 s - l , The transgranular crack has presented three appearances and two netting of slip traces which have showed that crack is globally propagated according to the (100) direction. In the first zone, near the transgranular crack initiation, long river lines have been observed and serrations have interacted with the slip system. The second zone, where the river lines disappear, looks like a transition zone between the first and third zones. At the scale observation used, this zone is planar and the two nettings of slip traces form an angle of near 90 ° , which indicate that propagation plane is {100} which has been confirmed by the etch-pitting method. 16 The third zone presents similar facets to those observed in the first, but another netting of parallel lines, perpendicular to river lines, appears in this case (Fig. 9c). This last netting does not coincide with slip system previously described. Spacings between markings are about 1 /xm. This indicates that the propagation of cracks can occur by discontinuous cleavage. As a conclusion to this study, micrographic observations have showed that the propagation mechanism can take place with a discontinuous cleavage process with crack advance distance about 1 ktm in certain favourable situations with or without the presence of boric acid at 200°C. at the scale observation used, no existence of dissolution has been observed, but the dissolution-repassivation process is certainly the limiting parameter of the propagation. CONCLUSION The results of the constant load tests and the electrochemical techniques have both shown the part of anodic dissolution in the SCC phenomenon. Oxygen presents a negative effect: it produces an increase in dissolution and an acceleration on both initiation and propagation of cracks. In 200°C and high pressure (12 bar), the results of slow strain rate tests (2 x ]0 -7 s -1) have underlined critical concentrations leading to a high sensitivity to SCC of 316L steel, as well as the negative effect of oxygen. Cracking analysis (number, depth, fracture surface) shows in which conditions dissolution is more confined and damage more important. The observation of fracture surface has suggested that in some cases a phenomenon of corrosion-plasticity interaction can participate in the fracture mode of samples. Thus, micrographic studies have underlined the crystallographic characteristics of transgranular fracture surface. So we may suppose that crack propagation sometimes takes place with a microcleavage. In some cases, the observations of crack arrest markings have showed the discontinuous feature of the phenomenon. Acknowledgements--We

port.

thank Electricit6 de France (EDF--Les Renardi6res) for their financial sup-

SCC of steels in H3BO3 + NaC1

455

REFERENCES 1. J. P. BERGE, Corrosion sous contrainte: ph(nomdnologie et mdcanismes (eds D. DESJARDINSet R. OLTRA). Editions de Physique (1992). 2. R. KASRI,Thesis, University Bordeaux I (1991). 3. D. NOEL, B. PRIEUX,J. ECONOMOUand M. DA CUNHABELO, Colloque Internalional Fontevraud II, (Sepembre 1990). 4. F. P. FORD, Corrosion sous contrainte: ph~nomOnologie et m~canismes (eds D. DESJARDINSet R. OLa'RA). Editions de Physique (1992). 5. M. PUIGGALI,D. DESJARDINSand L. AJANA, Corros. Sci. 28, 587 (1987). 6. D. DESJARDINS,Mdtaux-Corrosion-Industrie (Mars 1982). 7. R. N. PARKINS,Br. Corros. J. 14, 5 (1979). 8. D. DESJARDINS,M. PUIG~ALIand M. C. PETIT,M~m. Et. Sci. Revue de Metallurgie, 233 (1981). 9. R. KASRI,D. DESJARDINS,M. PU1G~ALIand C. CLEMErCr,Mdm. Et. Sci. Revue de Metallurgie (June 1992). 10. R. KASRI,D. DESJARDINSand M. PUIGGALI,Mdm. Et. Sci. Revue de Metallurgie (February 1993). 11. R. OLTRA,A. DESESTRET,E. MIRABALand J. C. COLSON,M6m. Et. Sci. Revue de Metallurgie, 353 (1987). 12. M. C. PETIT, D. DESJARDINS,M. PUIGGALI,A. EL KrtELOUIand C. CLEMENT,Corros. Sci. 32, 1315 (1991). 13. P. BALLADON,J. FREVCENONand J. HERmER, Am. Soc. Testing and Mat., Philadelphia (1981). 14. EL HASNIand D. NOEL, Congres CORROSION 89, New Orleans, Louisiana (17-2l April 1989). 15. E. N. PUGH, Corrosion N A C E 41,517 (1985). 16. J. M. OLIVE, C. SARRAZlNand R. KASRI,Int. Conf.: Corrosion-Deformation Interactions, Fontainebleau, France (5-7 October 1992).