Schaeffler Diagram

Related terms:

Austenitic Stainless Steel, Chromium, Austenite, Ferrite, Martensite, Filler Metal,

Dilution, Base Material, Filler Material

View all Topics

Stress corrosion cracking (SCC) in stain-

less steels
V. Kain, in Stress Corrosion Cracking, 2011

5.1.2 Classification of stainless steels

As the stabilizing effect of each element for a given phase is known, it is possible to
make use of simple diagrams like the Schaeffler diagram to predict which phase is
the stable phase for a given composition of stainless steel (Pickering 1976, Schneider
1960). The Schaeffler diagram (see Fig. 5.1) is prepared by establishing the stable
phase (at room temperature) for different compositions of stainless steels after
heating an alloy to 1050°C for 30 minutes and then quenching it in water.

5.1. Schaeffler diagram showing stable phases for a given alloy composition after
heating to 1050°C for 30 minutes and quenching in water. Here % Ni equivalent =
%Ni + %Co + 30(%C) + 25(%N) + 0.5(% Mn) + 0.3(%Cu) and %Cr equivalent = %Cr
+ 2%Si + 1.5(%Mo) + 5(%V) + 5.5(%Al) + 1.75(%Nb) + 1.5(%Ti) + 0.75 (%W).

From Pickering 1976 and Schneider 1960.

Figure 5.1 shows that four main categories of stainless steels (austenitic, ferritic,
martensitic and duplex) can be produced by varying the chemical composition. In
addition a fifth category – precipitation hardenable stainless steels – is also a popular
category.

> Read full chapter

The weldability of steel

Klas Weman, in Welding Processes Handbook (Second Edition), 2012

Welding of stainless steels to dissimilar metals

There is often a need, when fabricating stainless steel structures and/or when
incorporating them in other plant, to weld stainless steel to plain carbon or low-alloy
steels. In addition, different types of stainless steels often have to be welded to each
other. In order to obtain the best performance of such joints, particular attention
must be paid to the metallurgy of the weld metal. For this purpose, the Schaeffler
diagram is a good help. In general, the following methods of working can be
employed when deciding on the appropriate choice of filler material.

The position of the material in the Schaeffler diagram is determined from its
chromium and nickel equivalents, and a straight line is then drawn between the
points. Similarly, the position of the proposed filler material is plotted. When welding
symmetrical joints, it can be assumed that the joint surfaces of the base material will
be more or less equally melted. A straight line can therefore be drawn between the
position of the filler material and the centre of the line between the base materials.
In general, about 20–40 % of the filler material will be 'diluted' by the molten base
material, with the result that the structure of the weld metal will be as indicated by a
point about 20–40 % of the way along the line from the position of the filler material.
If this structure is suitable, then the proposed filler material can be used. If it is not
suitable, repeat the above procedure for a filler having a different composition.

The Schaeffler diagram in Figure 19.6 shows an example of the above for the case of
a low-alloy steel EN 1.0401, (A), being welded to stainless steel 1.4301, (B), using 23
12 2 L filler (D). This gives a weld metal structure as shown at point E, i.e. austenite
with about 10 % ferrite.
The filler material to be used for joining the two materials must also be selected as
described above when welding compound materials consisting of plain carbon and
high- alloy materials. The same applies for selection of the filler material for the first
layer of stainless material to plain carbon or low-alloy steel.

> Read full chapter

Plant Piping Systems

Alireza Bahadori PhD, CEng, MIChemE, CPEng, MIEAust, RPEQ, in Oil and Gas
Pipelines and Piping Systems, 2017

10.5 Welding
End preparation, alignment, and fit-up of pipe pieces to be welded, preheating,
welded, postheating, and inspected after postweld heat treatment, should conform
to standards.

Austenitic stainless steel weld deposits should have a ferrite content of 3–10%.
One deposited weld metal sample should be taken for every 30 linear meter of
welding and should be checked for carbon, chromium, nickel, silicon, molybdenum,
manganese, and columbium content. These analyses should be used to determine
the ferrite content by the Schaeffler Diagram. When approved by the engineer, the
ferrite scope may be used as an alternative method to verify ferrite content. The
ferritescope should be used prior to weld heat treatment.

Branch and nonpressure, part-attachment welds should not cross longitudinal

seams or circumferential butt welds and should be subject to the toe-to-toe sep-
aration distance specified for circumferential butt welds.

Where such intersections are unavoidable the main weld should be subject to
nondestructive examination prior to making the attachment weld. The extent of
examination should be at least twice the diameter of the branch pipe measured from
the center line of the branch.

Joints involving the intersection of more than two welds should be avoided.

Joints to be seal-welded should be made up clean and without the use of tape
or any compound. Welding should be performed in accordance with the qualified
procedure by a qualified welder. All exposed threads should be covered by the seal
weld.

> Read full chapter

Austenitic steel grades for boilers in ul-
tra-supercritical power plants
P. Barnard, in Materials for Ultra-Supercritical and Advanced Ultra-Supercritical
Power Plants, 2017

4.2 Overview of austenitic steels

There is a range of austenite-forming matrices; most notably in the context of
boiler alloys are the Fe-, Ni-, and Co-based alloys. In this chapter, when we refer
to austenitics, we are only discussing those based upon Fe. The austenitic steels
generally have compositions highlighted by the Schaeffler diagram, see Fig. 4.1
[4]. Although specifically for weld metal, this diagram gives an indication of
the austenite stabilizing elements, on the vertical axis, and the ferrite stabilizing
elements on the horizontal axis, which is relevant for wrought product as well.

Figure 4.1. Schaeffler diagram for weld metal [4].

A wide range of austenitic steels have been used for many years in boiler applications
for tubing, ie, heat transfer surfaces, and for piping, ie, transport of fluid without
heat transfer. A comparison of the creep performance of ferritic and austenitic steels
and Ni-based alloys is given in Fig. 4.2. Steam-side oxidation resistance is often
attributed to Cr content, as per Fig. 4.3.
Figure 4.2. Envelopes of 100,000 h creep rupture for the three main alloy classes.

Compiled from literature and internal Doosan data.

Figure 4.3. Generic plot of oxidation versus the Cr content of the alloy.

Compiled from literature and internal Doosan data.

Generally, austenitic steels are characterized by having good ductility, a low yield
stress, and relatively high ultimate tensile strength compared to carbon steel. Other
favorable characteristics are that welding has little effect on mechanical properties
and normally post-weld heat treatment (PWHT) is not necessary, except where
stress corrosion cracking may be an issue. Austenitic steels are not susceptible to
hydrogen embrittlement, therefore in most cases, pre- and post-weld heating is also
unnecessary.
Other than the cost of austenitic alloys, the main disadvantage for their use is the
high coefficient of expansion and low thermal diffusivity, ie, thermal fatigue
and high residual stresses in dissimilar metal welds (DMW). Other considerations
include alloy sensitization, stress corrosion cracking, stress relief cracking, secondary
hardening, and strain-induced precipitation hardening (SIPH). For the rest of this
chapter, we will examine some of these disadvantages to see how they manifest
themselves in service or during component manufacture.

> Read full chapter

INVESTIGATION OF FAILED ACTUA-

TOR PISTON RODS
T.F. RÜTTI, E.J. WENTZEL†, in Failure Analysis Case Studies II, 2001

4 DISCUSSION
The absence of evidence of fatigue and the short operational life of the actuators in-
dicated that the components entered service flawed or contained such high residual
stresses in the weld that operational loading resulted in cracking and fracture.

Table 2 shows the nickel and chromium equivalents of the various steels used. The
exact extent of dilution of the weld metal was not determined, but the combination
of the parent metals and E312 filler metal places the resultant weld metal composi-
tion into the A + F or A + M + F zone on the Schaeffler diagram. Since the welds were
very wide and a large degree of dilution had taken place, the former composition
was more likely. Indeed, the microstructure of the E312 weld consisted mainly of
austenite and ferrite (Fig. 3), but a large degree of grain boundary and intragranular
carbide precipitation, probably chromium carbides, was also observed. The carbide
precipitation probably took place during nitriding, resulting in the high hardness of
the weld and weakening of the grain boundaries.

Table 2. Ni and Cr equivalents calculated on average nominal compositions

Steel grade Ni equivalent Cr equivalent

En9 16.8 0.5
En32 (uncarburised) 4.8 0.5
E312 13.8 30.0
ER 70S-6 5.2 1.5

The high residual stresses (estimated at 300–500 MPa) due to the different thermal
expansion coefficients of the duplex weld metal and the low alloy parent metal
(11–13 × 10–6/K and 16–18 × 10–6/K for En9 and high Cr/high Ni steels respec-
tively), coupled with the highly constrained joint configuration and the weakened
microstructure, probably resulted in cracking of the weld on cooling after nitriding.
This conclusion is supported by the fact that the actuator piston rods that were phos-
phated did not fail—there was no thermal cycle to allow for carbide precipitation.
Unfortunately, no phosphated piston rods were available for evaluation, so this could
not be confirmed.

The absence of preheat resulted in a very high cooling rate which had a twofold
effect. Firstly, the HAZ was very hard and susceptible to cold cracking which was
confirmed by the martensitic structure and some small cracks found in the HAZ.
Although the nitriding subsequently tempered this microstructure to 200 Hv, the
delay between welding and nitriding was often several days which was sufficient for
damage to occur. Secondly, the weld microstructure comprised long, narrow grains
with no equiaxed grains near the centreline. The result was increased weld centreline
segregation and reduced toughness.

The weld bead was very wide and relatively shallow (Fig. 8), and in some cases
insufficient penetration resulted in a circumferential stress concentrator in line
with the centreline of the weld. Ideally, where two components are only welded
along a part of their contact surface, the un-fused plane should be off-centre and
not parallel to the weld centreline.

Fig. 8. Schematic drawings of the original and recommended weld preparations.

> Read full chapter

Materials, properties and selection

Mechanical Engineer's Reference Book (Twelfth Edition), 1994
7.3.17 Weldability of steels
Steels may be welded by almost all varieties of electric arc welding methods, in-
cluding gas-shielded MIG and TIG with and without filler, flux shielded manual
metal arc, submerged arc, electro slag, spot, projection and flash butt. Other fu-
sion methods include the more recently developed electron beam and laser, and
relatively old-fashioned gas welding. Solid-phase methods, forge, diffusion, friction
and explosive welding may also be used. Many of these procedures are concerned
with relatively thin sections or special design and applications.

7.3.17.1 Weldability of non-stainless steels

‘Weldability’ of steels usually implies the ability to make long runs in fairly large
sections either by manual metal or submerged arc and is governed in ferritic steels
by the ‘Carbon Equivalent’:

(Boron is not taken account of in this equation but has a great influence on harden-
ability and therefore on voidability.) Steels with carbon equivalent below 0.14% are
readily welded without special precautions in a wide range of thicknesses.

Table 7.11. Mechanical properties of the rotor shown in Figure 7.34 after quality heat
treatment*

Sample No. Position and Tensile test Charpy impact

orientation of test (notch: 2mm
sample V)

S0.2(kg Su (kg e (%) AR (%) As re- De-em- Embrit- Δf.a.t.t.- Δf.a.t.t.-

mm−2) mm−2) ceived brittled† tled‡ 1 (°C) 2 (°C)

Energy f.a.t.t. Energy f.a.t.t. Energy f.a.t.t. AsRe-WQ STC-WQ

( J) (°C)
( J) (°C) ( J) (°C) ( J) (°C)

Desired properties
— Surface R 82.4 84.5 15 45 4.1 ≤ + 15.5
— Centre of L 81.0 84.5 15 45 — —
core bar T — — — — 4.1 ≤ + 15.5
Actual properties
S—1 Surface O 84.0 95.3 19.1 69.6 out. <−75
11.0
radial I in. 12.1 −65
S—2 O 83.8 96.0 19.1 65.6 out. 12.7 −63
I in. 12.1 −29
S—3 O 83.7 94.8 20.5 68.3 out.12.7 <−75
I in.12.5 −37
S—4 O 92.1 102.8 18.0 62.8 out. 9.3 <−75
I in. 11.0 <−75
S—5 O 90.9 101.6 19.1 68.8 out. 10.8 <−75
I in. 11.5 <−75

O-1 Ends L 90.9 101.9 20.5 67.4

O-2 92.8 103.6 20.5 67.9

C-1 Cen- L 84.3 98.4 19.8 57.9 4.6 + 15 10.5 + 1 4.1 +38 + 14 +37
tre
of
core bar
T 83.8 98.7 17.7 51.9 5.9 + 12 6.9 + 13 4.1 +43 −1 +30
C-2 L 85.0 98.5 18.3 60.5 6.7 0 12.5 −18 7.7 +6 + 18 +24
T 85.0 99.0 16.9 53.9 7.3 0 13.0 −20 9.2 +5 +20 +25
C-3 L 84.0 97.1 19.9 59.5 8.7 −8
T 83.8 96.4 18.2 57.8 8.7 −10
C-4 L 83.4 96.0 20.8 62.6 8.1 + 11 14.0 −4.5 7.1 +20 + +24.5
15.5
T 83.4 95.7 18.7 59.9 10.6 + 10 9.0 +8 5.8 +22 +2 + 14

R, radical; T, transverse; L, longitudinal; I, inner; O; outer.Δf.a.t.t., increase in f.a.t.t.;

AsRe-WQ, s re-water quenched; STC-WQ, step cooled and water quenched.

* Reproduced by courtesy of Japan Steelworks.

† De-embrittled: 590°C − 1 h – WQ.

‡ Embrittled: step cooled.

Figure 7.34. Dimensions of typical large rotor forging showing location of test
specimens in Table 7.11

Steels with carbon equivalent between 0.14% and 0.45% require the following
precautions, depending on the value of carbon equivalent and section size to prevent
the formation of austempered martensite cracking aggravated by hydrogen.

Specification of low-hydrogen electrodesThis is always desirable but requires operator

skill to compensate for the more sluggish metal and slag flows compared with other
electrodes.
Use of preheat before welding The preheat temperature required depends on the CE
and the metal thickness; for a carbon equivalent of 0.2, 40°C and 110°C are advisable
for respective metal thicknesses of 25 and 225 mm, while for a carbon equivalent of
0.45, 170°C and 260°C are advisable for the same thicknesses.

Control of heat input Other things being equal, a higher heat input gives less risk
of formation of austempered martensite than a lower heat input but care must be
taken to limit distortion and the introduction of stress.

Use of post-heat after welding This is seldom required for CEs below about 0.35%,
but high-duty components with restrained welds should be post-weld heat treated
at between 600°C and 650°C for one hour per 25 mm thickness. Besides preventing
immediate cracking (or making it obvious during inspection) post-weld heating
improves dimensional stability. It is essential when welding thick and complex
structures to post-heat-treat one weld before commencing to weld a cross seam.

Steels with carbon equivalent above 0.45 These present very severe problems in
welding. Very high preheats ranging up to 340°C for carbon equivalent 0.6 and 225
mm thickness, low-hydrogen electrodes (preferably lower in carbon equivalent than
the parent material) and immediate post-heat-treatment at temperatures around
800°C are essential. Sample test welds are advisable.

Maraging steels With carbon contents around 0.03% these have a soft martensite
matrix and are highly weldable with no risk of decarburization, distortion or cracking.
They should be used where very high strength combined with weldability is required.

7.3.17.2 Welding of stainless steels

Welding is the normal method of fabricating stainless steel vessels. The heat-affected
zones are raised to incipient fusion temperature but time at temperature varies with
different welding processes. Argon arc and spot welding are most satisfactory in
heating for minimum time; metal arc welding, inert-gas metal arc and submerged
arc are less so in that order from this point of view.

The problems associated with welding stainless steels fall into two categories. The
first, associated with carbide precipitation, includes ‘Weld Decay’ and ‘Knife Line At-
tack’ and affects mainly corrosion behaviour. The second includes those phenomena
which may be assessed by means of the Schaeffler Diagram (see Figure 7.33).

7.3.17.3 Carbide solution and precipitation

The solubility of chromium carbide in austenite decreases with decreasing tempera-
ture and increasing nickel content (Figure 7.35). At room temperature the solubility
in 18% Cr 8% austenite (solid line) is approximately 0.03%. If an 18% Cr 8% Ni
alloy containing, say, 0.06% C is annealed at 1050–1100°C all chromium carbide is
in solution and remains in unstable solution after quenching to room temperature.
If the alloy is heated to an intermediate temperature excess carbide is precipitated.

Figure 7.35. Variation of solubility of carbon in 18% chromium steels with temper-
ature and nickel content

The mode of precipitation of carbide is dependent upon whether or not the austenite
has been worked. If the quenched but unworked alloy is heated in the temperature
range 450–750°C chromium carbide is precipitated at the grain boundaries; the
lower the temperature, the longer the time required. Thus at around 450°C the time
taken for precipitation can be about two years whereas at 700°C it is a matter of
minutes.

Precipitation is effected by diffusion of carbon atoms to the grain boundaries where

they each combine with approximately four times the number of chromium atoms.
Diffusion of carbon is relatively fast at these temperatures but that of chromium
extremely slow. Consequently, the chromium atoms are almost entirely supplied by
the grain boundaries so that the grain boundary chromium content is substantially
lowered. This local depletion of chromium causes loss of passivity in acid corrodants
with consequent attack along grain boundaries.

7.3.17.4 Weld decay

The resultant ‘Intergranular Penetration’ in a casting or, if the heating has been
caused by welding, ‘Weld Decay’ (see Figure 7.36) can completely disintegrate the
material. Precipitation in cold-worked material takes place along slip planes as well as
grain boundaries, consequently the distance that the chromium atoms must diffuse
is small. Hence, although the same amount of chromium is removed as carbide, the
depletion is more uniformly distributed with a consequential lowering of general
corrosion resistance but a lower tendency to intergranular failure.
Figure 7.36. Example of ‘weld decay’ in an austenitic steel

There are two alternative approaches to the problem of preventing intergranular

corrosion. Either the carbon content of the steel is limited, by using an AISI ‘L’ or BS
970 Code ‘11’ steel, to 0.03% at which precipitation of carbide in sufficient quantity
to cause trouble is impossible, or an element such as titanium or niobium, which
has a stronger affinity for carbon than chromium, is added to form the appropriate
carbide by using an AISI ‘Nb’ or ‘Ti’ or BS 970 Code ‘40’ steel. The theoretical amounts
required to ensure that all carbon in excess of 0.02% is combined are titanium = 4 ×
excess carbon, niobium = 8 × excess carbon. In practice, allowance must be made for
nitrogen combining with the added element (particularly Ti) and for the efficiency
of combination - carbon levels below 0.06% requiring a higher titanium or niobium
to carbon ratio for complete combination than those above 0.08%.

7.3.17.5 Knife-line attack

Furthermore, when a stabilized (Ti or Nb treated) steel is heated to successively high-
er temperatures above 950°C up to 1250°C the carbide enters solution and is broken
down into its constituent elements to an increasing extent so that above 1100°C a
relatively small amount of carbon remains combined. The free carbon is then avail-
able to form chromium carbide on subsequent re-heating in the 450–750°C range.
Combination of carbon with titanium occurs in the 850–950°C range given adequate
time. The whole question of stabilization is concerned with time, temperature and
amount of free carbon.

Time at temperature affects the extent of re-solution of the titanium and niobium
carbides present in stabilized steels; titanium carbide dissolves more rapidly than
niobium carbide. Re-solution takes place at temperatures in excess of approximately
1200°C under welding conditions. The extent depends on carbide particle size as
well as time at temperature. If reheated within the sensitization temperature range
(around 650°C) this narrow zone immediately adjacent to the weld metal precipitates
intergranular chromium carbide, because combination of Ti or Nb with C cannot
occur at this temperature.

Thus, although the stabilized steels will not precipitate chromium carbide in the
region of the heat-affected zone raised to 650°C by welding, there is the possibility, in
conditions where the edge of the weld metal is reheated to 650°C, that intergranular
attack can occur. The existence of such conditions depends on the welding practice
but in most fabricated articles, as distinct from samples with single-run

welds, positions must arise at weld junctions where these conditions will obtain;
welded samples should therefore always have crossed welds.

This particular type of intergranular attack at the weld metal edges is known as
‘knife-line’ attack (see Figure 7.37). It is most likely to be seen in boiling dilute nitric
acid solutions. The composition of the steel affects its incidence; steels with lower
nickel-to-chromium ratio, which produce a greater amount of delta ferrite in the
knife-line zone, are less susceptible. Fully austenitic Ti stabilized grades appear to
be more susceptible than fully austenitic Nb stabilized.

Figure 7.37. Example of ‘knife-line’ attack on stainless steel weld

Where corrosion conditions are known to offer a knife-line hazard, treatment of the
fabrication at 870°C will promote precipitation of the carbon as titanium of niobium
carbide with consequent resistance to attack. The unstabilized 431 and 434 grades
are susceptible to intergranular attack after welding. This can be prevented by heat
treating for 2 hours at 600–800°C which coalesces the carbide films.

7.3.17.6 Weld problems which may be assessed by means of the Scha-

effler Diagram
Fully austenitic weld metal tends to crack on solidification because of inherent
weaknesses at the boundaries of columnar grains. The composition of the filler metal
is therefore adjusted to ensure that all the molten zone contains a small proportion
of ferrite (that is, it lies on the A + F side of the A/A + F line in Figure 7.33) and also
to ensure that its strength is not excessive compared with the parent metal.

As long as the weld metal composition is maintained within the zone in which ferrite
and austenite co-exist austenitic steels have excellent weldability. Ferritic stainless
steels are weldable but suffer from the brittleness and grain growth problems de-
scribed in Section 7.3.12 Martensitic stainless steels suffer from the same brittleness
problems as carbon and quenched and tempered steels unless the carbon content
is below 0.12%.
Both precipitation hardening and duplex stainless steels (the compositions of which
can be roughly estimated from Figure 7.28) are fully weldable without preheat, and
the precipitation-hardening steels may be hardened by precipitation after welding.
Brief summary notes on corrosion and welding aspects of stainless steels are given
in Table 7.12This is also referred to in Section 7.9 where for the convenience of the
reader some of the figures in Sections 7.3.13 and 7.3.17 have been repeated.

Table 7.12. Corrosion resistance and weldability of stainless steels

AISI No. Nominal composition Corrosion notes

C Cr Ni Mo Other
410 0.08−0.4 12−14 – – – Rust
resisting.
Higher–car-
bon grades
for
engineering
applications,
turbine
blades,
cutlery, etc.
465 0.08 max. 13 – – Al Weldable
grade.
430 0.1 max. 16.5 – – – Resists mild
acids.
Special
feature is
resistance to
nitric acid.
May require
heat
treatment
after
welding
(600–800°C)
to avoid
intergranu-
lar attack.
Forming of
sheets up to
3 mm at
room
tempera-
ture; greater
thickness at
200–350°C.
430Ti 0.1 max. 17.5 – – Ti Weldable
grade not
requiring
heat
treatment.
Argon arc
(gives
minimum
grain
growth)
preferred.
Both grades,
if welded,
should not
be applied
under
conditions
 of shock
loading or
vibration.
304 0.08 max. 18 10 – – Rust and
acid resis-
tant. Suitable
for welding
in certain
applications.
304L 0.03 max. 18 10 – – Extra low
carbon. Very
resistant to
intergranu-
lar
corrosion.
Weldable for
practically all
applications.
309Cb 321 0.1 max. 18 10.5 – Nb or Ti Not
susceptible
to
intergranu-
lar attack
(but see
reference to
knife–line
attack in
text).
Applicable
above
300°C.
Weldable.
316 0.07 max. 17.5 11 2.2 – Resistance to
chemical at-
tack better
than 18/8
(e.g. se-
vere acid at-
tack). Re-
sists inter-
granular at-
tack up to
6 mm thick-
ness. Ap-
plicable be-
low 300°C.
Weldable for
most appli-
cations.
316L 0.03 max. 17.5 11 2.2 – Superior
resistance to
intergranu-
lar
corrosion,
suitable for
thicknesses
greater than
6 mm.
316Cb 316Ti 0.1 max. 17.5 11.5 2.2 Nb or Ti Not suscep-
tible to inter-
granular at-
tack (but
see knife–-
line at-
tack). Applic-
able above
300°C. Suit-
able for
strong acids
at elevat-
ed tempera-
 tures. Weld-
able.
317 0.07 max. 17.5 12 2.8 – Resists
intergranu-
lar attack up
to 6 mm
thickness.
Applicable
below
300°C.
Corrosion
resistance
superior to
2% Mo
alloys.
Weldable for
most
applications
317Cb 317Ti 0.1 max 17.5 12.5 2.8 Nb or Ti For strong
acids at high
tempera-
tures.
Applicable
above
300°C.
Weldable.
317LM 0.03 max. 17 13.5 4.5 – Resistance
to strong
organic
acids at
elevated
tempera-
tures.
Increased
resistance to
pitting.
Applicable
below
300°C.
Resists
intergranu-
lar attack.
Weldable for
most
applications.

> Read full chapter

ScienceDirect is Elsevier’s leading information solution for researchers.

Copyright © 2018 Elsevier B.V. or its licensors or contributors. ScienceDirect ® is a registered trademark of Elsevier B.V. Terms and conditions apply.