ULTRASONIC MATERIAL PROPERTY DETERMINATIONS

Steven Serabian

University of LowellLowell, Massachusetts 01B54

The use and potential offered by ultrasonic velocity and at-tenuation measurements to determine and/or monitor materialproperties is explored. The basis for such unique measurementsalong with examples of materials from a variety of industries willbe presented.

INTRODUCTION

The ability of ultrasound to measure and/or monitor micro-structure related properties has been established in the laboratory.A number of applications have been implemented for field use,however the full potential of such technologies has not beenrealized. Invariably one makes use of two of the propagationalparameters of ultrasound - i.e., attenuation and velocity. Thelatter are intrinsic to the particular combination of material andprocessing under consideration. Absolute quantitative relationsare usually unavailable, therefore empirical material property -ultrasonic parimeter correlations based upon qualitative reasoningare sought. In effect, a series of specimens containing the rangeof the material property of interest is necessary along with con-ventional destructive material property measurement schemes.Obviously, such correlations are not universal and are only validfor the particular material and attendant processing being investi-gated. Correlation sensitivity is the prime factor that definesthe accuracy and effectiveness of the generated correlation. Oncethe correlation has been established costly and time consumingconventional destructive evaluation procedures can be eliminated.In this paper a number of examples of correlations of significantmaterial properties with ultrasonic parameters will be cited.

I. VELOCITY BASED CORRELATIONS

The original low pressure piping systems used by the utilitiesto transport gas were constructed of cast iron. Mechanical proper-ties and susceptibility to corrosion renders cast iron undesirablefor such an application. Economics precludes the to£al replacement

219

https://ntrs.nasa.gov/search.jsp?R=19860013507 2020-05-13T17:32:45+00:00Z

-2-

by steel wraped pipes, thus one must resort to a gradual, scheduledreplacement based upon a measure of quality. The determination ofthe necessary mechanical properties required the need for an exten-sive length of piping. This presented a costly as well as alimited number of pipe sections that can be evaluated duringnormal installation and maintenance programs. However, it is

possible to trepan a 1-1%" coupon without shutting down the pipingsystem. The readily available large number of coupons would thenpresent a greater sampling rate of the piping system. A programwas initiated which would facilitate the use of the cou_s forthe determination of significant mechanical properties. TM Thelatter along with other parameters such as soil condition, corrosionand road traffic activities are used to formulate pipe quality.

Industry wide practice consists of using the ASTM Talbot stripbeam test as shown in Figure 1 to measure mechanical properties.The maximum deflection of the beam at fracture allows the measure-ment of the moduli of elasticity and rupture. Radiography wasused to aid in the selection of the specimen location within anygiven pipe section to eliminate occasional structural anomaliessuch as voids and inclusions. Each beam specimen was characterizedby velocity and attenuation profiles prior to the destructivephase, see Figure 2. The attenuation proved to be too sensitive tomicrostructure even at low frequencies. However, the velocity

provided a consistent value for each specimen yet sufficient dif-ferences between specimens. It is for these reasons that velocitywas selected as the ultrasonic parameter to be used in the correla-tions with material properties. The velocity measurement schemeinvolves locating the transmitted R-F pulse on a calibrated timescale. The calibration is provided by using specimens with a rangeof thickness and a known velocity.

There is quantitative substantiation of a velocity-mechanicalproperties correlation provided by,

md (i +_)2V L =

(I + 2_) (i -_)(1)

Ed f (%)

Where Ed is the dynamic modulus, _ is the density and_ is Poisson'sratio. By noting that f (_) is essentially unity for a brittlematerial such as cast iron and that there is a direct relationship

between Ed and the secant modulus (Es), therefore equation 1becomes,

2 a EvL s (2)

220

-3-

For the correlation V L and density are taken adjacent to the frac-ture point ; V L from a display typified by Figure 2 and densityby the water displacement technique from a small section ofthe strip.

Figgre 3 indicates the secant modulus. The data spread is± 1 x i0° psi. At the midpoint of the modulus range consideredthis amounts to ± 12% and is within the error associated with

conventional destructive measurement procedures.

The other property of interest was the modulus of rupture (R).In a brittle material,

RE = -- (3)s eT

Where e_ is the strain to failure. Since em can only be measureddestructively it was necessary to seek an i_direct relationshipinvolving a property that is more readily available. One such

property is hardness (/_) or,

= (4)

Where C is a constant. Incorporating Equations 3 or 4 into Equation2 yields,

2

z --PvL " (5)

The hardness was measured adjacent to the fracture point. Theresults of this correlation is shown in Figure 4. The error for amodulus of rupture determination is ± 5,500 psi. Viewing thiserror at the midpoint of the R range translates into ±10% which isa creditable determination.

Once the beam specimen correlations were established all thenecessary measurements of density, hardness and ultrasonic velocitycan be obtained from the parallel faced coupons. The need forthe costly removable of large pipe sections and the associateddestructive evaluations are eliminated. Moreover, the coupons,because of greater availability provides a more realistic overallassessment of the quality of the piping system and can be used asa guide for the pipe replacement program.

Concrete is another material that falls into the brittlecategory and is therefore susceptible to the determination ofstrength related properties by velocity based measurements.(2)

221

-4-

A variety of water to cement ratios as well as curing times wereconsidered. The cylindrical specimens were used for measure-

ment of compressive strength and modulus while step wedgeswere used for the assessment of attenuation. Figures 5 and 6indicate the results of the destructive phase for the compressivestrength and secant modulus, respectively. The influence of thecuring time and the water to cement ratio is as to be expected;both increase the compressive strength and the secant modulus.Figure 7 displays the axial longitudinal velocity as a function ofthe curing time and the water to cement ratio. The similarities

between the data of Figures 5 and 6 with Figure 7 suggests thata correlation exists between these properties and ultrasonicvelocity. Figure 8 shows the compressive strength correlationwhere it is significant to note that notations involving compositionand curing time are lost. The accuracy (± 500 psi) is within thatexpected by conventional destructive means. Much the same remarks

are in order for the secant modulus correlation display as Figure9.

Some interesting observations regarding the state of cure canbe noted by the use of attenuation. The attenuation in concreteis too excessive to measure by noting the decay of a multipleorder back reflection. In this case the slope of an amplitude -thickness display obtained by the step wedge specimen proved tobe a valid technique. A torque wrench was used in a drill pressarranqement to maintain a constant transducer pressure at eachstep._ II) Figure i0 shows a typical determination for a water tocement ratio of 0.5. It can be seen that the attenuation is highduring the initial phase of cure. An attenuation decrease occursas the structure becomes more rigid thus less energy is expendedto propagate the ultrasound.

Another example of characterizing a material by,yelocitymeasurements is graphite of aerospace applications.(_; Figure iiindicates a data display and serves to illustrate the sensitivityfor density; the dynamic modulus was measured by resonant tech-niques. It should be noted that a variety of grades and velocitymeasuring directions are involved. Such a density sensitivitysuggests that transport phenomena are susceptabl4 to velocitycorrelations. Figure 12 supports this notion for the case ofelectrical resistivity.

II. ATTENUATION BASED CORRELATIONS

Let us now turn our attention to the use of attenuation for

the characterization of materials; in particular to the polycrystal-line state. The functional dependency of the average grain size(_) and___frequency (f) upon attenuation is shown in Figure 13

where _K is the direction related average change in elastic prq_,K perties of the grains encountered by the ultrasound._ ;In Rayleigh scattering the whole grain acts as a scattering unit.

222

-5-

For phase scattering one must be cognizant that each portion of the

ultrasonic beam contends with a discrete set of grains, thus pro-

ducing a variety of phases at the detector. The summation of these

individual phases constitutes the phase scattering loss. In dif-

fusion scattering one is essentially concerned with mean free

path considerations - i.e., how many grain boundries are encountered.

The dependency of these loss mechanics upon grain size and frequencycan be shown experimentally. Figure 14 displays typical data for

a steel that has been heat treated to obtain a range of grain size.

The deviation from linearity at the lower frequencies for the less

attenuating specimens is due to transducer beam spreading or dif-

fraction losses. Appropriate corrections would, in effect, reduce

each data point in this region to the linear extrapolated high

frequency data. The slope of each attenuation - frequency display

is the exponential frequency dependency (n). Figure 15 shows the

frequency exponent for a variety of materials with different

grain sizes.(5) The spread in the data is primarily due to the

fact that each specimen can have a range of grain size, thus more

than one type of scattering mechanism is in effect.

By cross plotting the data of Figure 14 one can note the grain

size dependancy (m) of attenuation. See Figure 16. Figure 17shows such data for a number of materials and indicates that the

grain size dependancy can range from zero to three. The data set

is not capable of examining the diffusion type of loss (m = -i).

Of importance for this discussion is the grain size dependency(m). Since for most engineering materials/structures the upper

frequency limitation for propagation restricts scattering to the

Rayleigh (m = 3) and phase (m = 2) loss mechanisms, thus attenua-

tion is quite sensitive to grain size. One obvious use of this

fact is the measurement of grain size. Figure 18 indicates the

observed attenuation in a Timkin Alloy (16 - 26 - 6) for a range

of grain size. If the data is plotted in terms of the ratio of

the average grain size to the dominate wavelength of a narrowband

pulse of ultrasound, a single curve is obtained (Figure 19). This

is in accordance with Roney's generalized scattering theory thatallows one to consider attenuation in terms of a dimensionless

parameter that obviates the need to consider both the grain size

and wavelength individually. By making attenuation measurements

at a variety of frequencies one can obtain an overall grain sizedetermination.

The sensitivity of attenuation to grain size suggests thefeasibility of using attenuation to measure any material property

that is grain size dependent. Fi@_es 20 and 21 indicate thegrain size dependency of hardness _ Jand fatigue limit, (7) respec-

tively. The data of Low(8)in Figure 22 shows the grain size

dependency for strain in failure as well as fracture and yieldstresses. As shown in Figure 23 Klinman (9) has utillzed the

grain size dependency of attenuation to determine transition

temperature for a spectrum of carbon steels.

In addition to grain structure dependency attenuation is also

223

-6-

influenced by other scatters such as porosity and inclusions of

microscopic proportions. Such scattering losses contribute to the

observed attenuation. Figure 24 attests to the sensitivity of

attenuation to porosity in grain refined aluminum. (10) A uniform

grain size throughout the specimen set assures that porosity is

the dominate cause of the attenuation variations. The porosity,

as verified in Figure 25, should effect the ultimate tensile

strength and be measureable by attenuation.

Silicate inclusions are prevalent in 303 stainless steel and

provides an example of microscopic inclusions acting as scatter

sites. The specimens considered contained approximately the same

grain size thus did not contribute to the observed attenuation

variations. Inclusion counts were performed by a quantitative

television microscope and were rated in accordance with length insix levels from 2 to 12 microns in 2 micron intervals. Using a

weighting function of n 2 to bring out the influence of the large_

inclusions a measure of the inclusion severitYhWaS_ taken as _ nZSnwhere Sn is the number of inclusions in the n counting level.

The results are shown in Figure 26, (12) As to be expected the higher

frequencies present a greater sensitivity to the inclusion count.

Such data can be used to quality grade the material.

Another example of inclusion produced attenuation is displayed

in Figure 27. The structure is a large rotor forging used in

large steam turbines. The center is the last to solidify, thus is

the major recipient of the impurities. It is for this reason that asmall borehole is cored out. The data shows the attenuation as a

function of the axial length. The attenuation is measured by the

decay in a multiple order bore refraction display and subsequently

corrected for geometric effects originating from the curved surfacesencountered. It can be seen by the solid line that the initial

measurement indicated an axial region that showed abnormally high

attenuation. This was also verified by an attenuation measurement

of the core material from this region as well as magnetic particle

inspection of the bore surface. It was decided to subject the

rotor to a minimal overboring operation. As evident by the dotted

line display the attenuation decreased appreciably in the regionof high inclusion count while the attenuation in other regions

remained essentially unchanged. The latter lends credance to the

notion that the high inclusion count was localized. While this

example is highly qualitative it does serve to indicate that

attenuation is susceptable to scattering losses due to microscopicinclusions.

SUMMARY AND CONCLUSIONS

Examples of the utilization of ultrasonic velocity and attenua/-

tion for the measurement and monitoring of material properties

have been presented. It is evident that a wide spectrum of activ-

ities in a number of industries are involved. It is significantto note that the measurement schemes and attendant instrumentation

224

-7-

have, in general, minimal requirements of sophistication andcomplexity.

REFERENCES

1. S. Serabian and G. E. Lockyer, "Nondestructive Evaluationof Cast Iron Pipe", Univ. of Lowell Research Foundation,April 1971.

2. S. Serabian and J. Tsererides, "Property Measurements ofConcrete", presented at annual ASNT meeting, 1972.

3. G. E. Lockyer, "Investigation of Nondestructive Methodsfor the Evaluation of Graphite Materials", Air ForceMaterials Laboratory, TR-65-11_ 1965.

4. E. P. Papadakis, "Ultrasonic Attenuation Casued by Scatteringin Polycrystalline Media", in Physical Acoustics (VolumeIVB); W. P. Mason, Editor, 1968.

5. S. Serabian, "Frequency and Grain Size Dependency of Ultra-sonic Attenuation in Polycrystalline Materials", Brit.J. of NDT, 22, 2, 69, 1980.

6. R. W. Armstrong, "The Influence of Polycrystalline GrainSize on Several Mechanical Properties", Met. Trans.ASM, 44, 929, 1952.

7. G. M. Sinclair and W. J. Craig, Trans. ASM, 44, 929, 1952.

8. J. R. Low, "The Relation of Microstructure to BrittleFracture", ASM, 1954.

9. R. Klinman, R. et al, "Ultrasonic Prediction of Grain Size,Strength and Toughness in Plain Carbon Steel", MaterialEvaluation, 38, 26, 1980.

i0. K. Chinnathambi and O. Prabhaker, "Quality Control of AluminumAlloy Castings", Trans. Indian Institute Metals, 33, 4, 296,1980.

ii. S. Serabian, "Influence of Attenuation Upon the FrequencyContent of A Stress Wave Packet in Graphite", J. Acoust.Soc. Am., 42, 5, 1052, 1967.

12. S. Serabian, "Inclusion Count Determinations by UltrasonicAttenuation Measurements", Armco Steel Corp., Middletown,Ohio, Dec. 1976.

225

U)

z__ 3 _

I--

BAR POSITION

(INCHES)

LOAD

)" 0.175-i-

(((_(/i(</</ccc(_ 0.165-Ill

BAR POSITION

Fig 1 Talbot strip test. Fig 2 Velocity/attenuation charact-erization of each specimen.

260 _/ o /

o/Ooo /o 65.0 _--o24.0 . o

Zo / ooo / /

?/ x- /_-- 22.0 50.0x /o o _o o

x oo re 45.0

•_ o o°o__ _ / o,/"" 20.0 ;

> t °1° _==40.0 o_/ o. o35.o ARio _/°/18.0 o :_J 30,0C3 o

025.0

I /o/ 20.016.0 /

5.00 6.00 8.00 I0.00 12.00 20.0 30.0 40.0 psi

SECANT MODULUS ( psi-IO 6) oV2/•

Fig 3 Secant modulus correlation Fig 4 Modulus of rupture correlationin cast iron. in cast iron.

226

_-io !

0 \ o :5 DAYS03 DAYS x 9 % • I WEEK

• I WEEK ._ \_ _ 42 WEEKS

\ A 2 WEEKS ,.,_ 8 X_ • 4 WlEEKS4 ,i,_ • 4 WEEKS

7 i

"'5g4

0.4 0.5 o.s 0.7 < I0

WATER TO CEMENT RATIO Ld0.4 0.5 0.6 0.7

WATER TO CEMENTRATIOFig 5 Concrete compressive strength

vs water to cement ratio Fig 6 Concrete secant modulus vswater to cement ratio

o 5 DAYS 6 /,- o 3 DAYS

• I WEEK x • 1 WEEK /7/_15 _ ,,2 WEEKS "_cL4--_ 2WEEKS--_ J _• 4 WEEKS /r, _ • 4 WEEKS-- I'-x ¢_3,., ', z o o/'_" ._.._14 tu

>- uJ 2 / .T/aI-. h > o "813 _ _.0.j tu /> _. k o.

L 0 o12

,_,""'_- 0 0 /2 3 14 15 160.4 0.5 0.6 0.7 VELOCITY 2(ftlsec x 10"3)2

WATER TO CEMENT RATIO

Fig 7 Compressivestrength/velocity Fig 8 Secantmodulus/velocitycorrelation in concrete, correlation in concrete.

227

I I I iI0 A

'_0 o 3 DAYS 30 c_× 9--* I WEEK I

A 2 WEEKS8 _A 4 WEEKS "_ 20 IC

','"°° 7 ,o_ _o ^/o/ // z 6 oSDAYSOr) -_

4 7 _j ' ' / w • I WEEKio, _',, • I- -- z_ 2 WEEKS --

D /o_£/ ,_ I-C_ <_ 4 • 4 WEEKSo 3

V- 2 .,Z

< / • I I I_ 2m I 0.4 0.60.B I 2 3 4oo _o

I0 15 2o 25 FREQUENCY (MHz)VELOCITY2 (ft/sec x 10-3) 2

Fig 10 Attenuation monitoring of curein concrete with a water to

Fig 9 Secant modulus/velocity cement ratio of 0.5correlation in concrete.

0.110 I ' i I I I I I I I , I

! XI o.100,_, l,

.= \o_ -,o.,{:=':,._t'."A,.10.0- 71 ,0.0. ,.,w cewNO., 0 %_

0.090 • A_INIT_lN _ 0.094 I _w-m'_

ATJ NO, I _ !

O WITHMAIN LiJ !I o inAIIqIT OAAIN )

_ O.00,S _ , ATJNo.I <

O WITHGRAIN t 0.090 \¢ O AGAINGTGRAIN I _

ATJNO. $ %, •

I-- j • AGA.T GRA. ¢_

i 0.080 / m'A_. i Z 0.080 I l,, -_A WITHGRAIN _ , • ,_ ' _b.It AniINIIT GRAIN :=l ' . •RYA NO,I I"= ,

• AOAINiTGRAIN _ io 075 .... _ r _ 7 ....... 0.082 --0.9 1.0 1.1 1,2 1.3 1,4 1.5 1,6 1.8 2.0 O

DYNAMICMODULUS,Ed, 106 psi 0.078900 1000 1100 1200 1300 1400 1500 1600

Fig$l Velocity/modulus correlations in RI:SISTIVITY'#"ohms-ramsgraphite. Densities are 1.72 & 1.85

gm/cm_ for the top and bottom plots, Fig 12 Velocity/resistivity correlationrespectively, in graphite.

228

- AVERAGE i _ --

GRAIN DIAMETER _ 3.6 /53 , )0.46

GRAINDIAMETER MECHANISM ATTENUATION 1.0 1RANGE

-- '¢ HOURS / HOURS I/ HOURS --

_-_ ,,,AsE (_) _,2 __- ,

:_ _ p TREATMENT0,,us,oN C_/"" = / TEM.ERATORE,-_- ,.,\K/ I- r TIME

o.t_ I _, _- ,-- i

- _ _

Fig 13 Ultrasonicscatteringlosses. 2 5 io 2_) 4o 60FREQUENCY, mHz

Fig 14 Attenuation/grainsize effects.

,°°I,, ,

FREQUENCY SLOPE

o / 30mHz 0.39o__/20 .82

:92:.v -- CrMoV 1 [ IA __ NI MoV Steel _ ].0

o---ov t- ___

e-- LOW Alloy Stee I e 1.0 --

3.0 m--Aluminum 5 1.66-- o--41so_ "_= ®® • -- 3140 _ Steel

o • -- Mo_nel& Incoflel -zZ

• -- Nickel - 98.9% IX)z "_-- Magnesium-99.95% "1_

2.0 ¢ Z

a_- --0 /" //o=_ _ _ ,_ O.l / /

v o Z

i • X A A A I-"

0.,o__- (I.01C 2 4 6 8 10 12 14 16 jO

AVERAGE GRAIN DIAMETER ( x 103 inches)/

i I i i L i I l l

1 10 20 40

Fig 15 Frequency dependency as function AVERAGEGRAINSIZE (xlO 3 inches)of grain size.

Fig 16 Cross plot of attenuation/grainsize data typifiedby Fig 14.

229

, _ I I ' I3o- 4.5

_ ,1- 4.0 10 MHz

> _ 3.52.(],, _" UNCORRECTED DATA

z '_z_ 3.0

uJ qO

o / i,/, _ v

uaN ZO 2.5I,-

Z 1.0 A--NIMoV "_ ,¢_E]--NiCrMoV _"Steel _ 2.(v--CrMoV J Z

v/ O-- LowAlloySteel LI.I

m--Nickel"98'9% _ 1,5,<

"/t ] i I J I0 10 20 30 1.0

AVERAGEWAVELENGTHAVERAGE GRAIN DIAMETER Hz

0.5Fig 17 Grain size dependency of

attenuation l [ I J I [ 1 J5.0 _ 2,1 3.0 4.0 5.0 6.0

AVERAGE GRAIN DIAMETER(X103 in)

o 2.25 MHz

4.0 _ Fig 18 Attenuation/grain size for a• 5.0MHz Timkin alloy (6-26-16)

.c_ - r_ 10.0 MHz

,,O 3503.0 --

Z 300

o_- 250 --

:_ 2oo!ZuJ 2.0--t-I-- _ 150

Q 100

50--

1.0 - o_ .[ I I ,I [ [ l I I I25 50 75 100 125 150 175 200 225 250

,._q_• (GRAIN SIZE) "1/2(ram"v2)

0 , I_ t I , I , |,I , I Fig 20 Hardness vs grain size in.01 .02 .04 .06 .I .15 in titanium. After Armstrong (6)

AVERAGE GRAIN DIAMETER.WAVELENGTH

Fig 19 Attenuation as a function of theratio of the average grain sizeto wavelength.

230

20 _ 200 X 103 _ Zl FRACTURE STRESS -"1

A O YIELD STRESS .// <N

-_15- _o-3o,,ASS ,8o-- o STRA,NTOFRACTURE._-- o

lC_ U_ 120 Z_7 u._I-

(/)= _ 80-- -uJ I-

_ 0.605; k-_- - 4o'_,,/ ° a_ o4z

..< /° _0.2 n-

O I I I L I I I I I I o/---Joo>.to--,_°l I o _0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6

(GRMN SIZE) "V=(ram"V=) (GRAIN SIZE)"V=(ram"V=)

Fig 21 Fatiquelimit in 70-30 brass. Fig 22 Influenceof grain size onAfter Sinclairand Craig (7). various propertiesof low

carbon steels. After Low (8)

ATTENUATION (,_)dB/cm1 2 3 4, , , , 0.38 --

• • I .,",::,.H 375 0.28 -- ALUMINUM -- 11.8% SILICON

+220 - ,_,'o_s " " ,.,,G• ,, • ,,, ,F 350 E 0.24 --

+18o_;'_ _'- A _== _ 020, =-° ,,, E

"'== /,_;,/ ....-"'" 325_ ; ---_ +100 hi..._ . .. ,'•" _ n,.'-" _ 0.16g.l= =. _, .-'" 300z =. ;

i , . * ALLOY A rr :i " 0.12

_,_, +40 F :._'" ALLOY FI..- I-"_. • ALLOY G 0.08 --

-'oF'- " oo4_ ,,i , I I• ALLOY,12. I-8o .,.,,,.,._ o.oo0 2 4 6 8 10 0.25 1.0 2.0 3.0 4.0 S.0 6.0

ATTENUATION (,_),dB/inch % POROSITY

Fig 23 Attenuationmonitoringof the Fig 24 Attenuation/porositycorrelat-transsitiontemperatureof ions. After Chinnathambiandcarbon steels. Carbon increases Prabhak (10).the transitiontemperature.After Klinemanet al (9)

231

23 --_-.EE ALUM,NUM--11.8%S,',CON21

19 x. : , . g,5--17 _ _ x x

XXXXX X ::L

x Xx xX zlO

g I I I I II I I I L I:_ 9.0 0 1 2 3 4 5 6 x 1030.01 0.02 0.04 0.06 1.0 2.0 WEIGHTEDDETECTABLEINCLUSIONCOUNT(_n2Sn)

ATTENUATION (db/mm)

Fig 25 Attenuationmonitoringof Fig 26 Attenuationmonitoringofultimatetensile strength.After inclusioncontentin stainlessChinnathambiand Prabhak (10) steel.

1.O I I I I I I I I

-- ATTENUATION FROM _THE CORE SPECIMEN

TOO HIGH TO

MEASURE"_o.0 -

. BEFOREZ OVERBORE0 _ j/ --

2"ZuJ 0.4 - AFTER --

P-k" ALL DATA AT / _- OVERBORE< 5 MHz i "_ -_

- /_...- / _ -

,0.2-- ,,,/t

Fig 27 Monitoringof inclusioncontent in large rotorforgings.

232