51

A design;guide for the refractory linings

of vesselpnd transfer lines
P K Richards,* BSc, CEng, FIMechE, FTWeldE, MInstFuel
F h o r Daniel BV, Haarlem, The Netherlands

The paper is concerned with the design of refractories where there are significant pressure and temperature loadings on the structure.
?'he factors that aflect the performance of linings are discussed, including the properties of the material as manufactured. Design
methods are proposed which enable a suitable lining and pressure vessel combination to be determined for those cases where direct
experience is not available. Examples are given to illustrate the design methods.

NOTATION significant impact. There have been circumstances

where a particular refractory selected for a project has
D diameter (m) proved itself good or bad, but the reason has not been
E modulus of elasticity (GPa) fully understood.
F force (N) There has therefore been relatively little theoretical
1 thickness, length (mm)
design and investigative work and most designs evolve
P internal design pressure (N/m2)
from trial and error and experience. Reliable design
Sl shrinkage at hot face (inside) (mm/mm) methods are available for simple structures as are nor-
so shrinkage at cold face (outside) (mm/mm)
mally met in furnaces and boilers based on practical
Ti temperature at inside ("C)
experience. This paper is concerned with a rational and
T* initial mean temperature ("C)
T mean temperature ("C) reliable guide for the lining of vessels and transfer lines
a coefficient of expansion (mm/mm°C) which need a more sophisticated approach because the
E strain shell is subjected to significant pressure and tem-
Poisson ratio perature loadings.
P
Cr stress (N/m2) For design purposes (and safety) it is always assumed
that the pressure vessel carries the full pressure load and
the refractory carries none. This is very nearly correct in
Subscripts practice because the refractory is porous to varying
c circumferential degrees and can contain cracks of various sizes. The
1 longitudinal expansion difference between the inner and outer faces
p pressure of the refractory and between the refractory and shell of
r refractory course produces stresses in each which can be consider-
s shell able.
S shrinkage It should be noted that for this application the refrac-
T temperature tory is attached to the steel shell at the cold face so that
both refractory and shell are at a similar temperature,
which is relatively low. This situation does not apply to
Examples an internal steel part in a vessel that is lined on both
E , ~strain
~ in shell, due to pressure, circumferential sides, and after a short time all reach substantially the
direction same temperature level, including the steel. The refrac-
~ r ~stress
~ , in refractory by shell, longitudinal direction
tory in this case is of course not applied to keep the
steel part cool but to protect it from erosion or some-
times to reduce thermal stresses due to transient tem-
1 INTRODUCTION peratures or at transitions where the internal part is
Refractories have developed and improved as a result of connected to the outer shell.
the demands of the end user so that currently with
much improved quality control, manufacturers have 2 REFRACTORY DESIGN
assembled an enormous amount of experience and
knowledge. Installation contractors have a largely prac- A refractory is installed inside vessels and transfer lines
tical background, though sometimes with a limited to maintain the wall temperature at a reasonable tem-
theoretical ability. Refractory materials themselves perature or to protect from erosion/corrosion or a com-
cannot be refined to the same consistency as metal, and bination of these.
the skill and care of the installing operator can have a Obviously it must not fall off, and so the physical
properties must be suitable for the environment it is in
The M S was received on 5 October 1989 and was accepted fur publication on 23 contact with; in addition it must show a low thermal
Januory 1990. conductivity but sufficient strength to withstand expan-
* Author is deceased: all cummunicotiuns should be directed to the Editor. sion stresses and its own weight. It is the combination
E01489 0 lMechE 1990 0954 4089y90 $2.00 + .05 Proc lnstn Mech Engrs Vol 204
52 P K RICHARDS

of these properties including its own thermal expansion/ be calculated and T3 estimated (based on the relative
dry-out shrinkage that is important rather than any one l / A for each layer).
individual property. The results can be used to estimate the mean tem-
Should the wrong combination of properties be perature for each layer of refractory to establish the
chosen then the refractory will fail. It must be remem- mean conductivity of each. These do not vary too much
bered that even with the right material in the correct with temperature so that provided the estimated tem-
thickness problems can occur due to too fast heating up peratures are not grossly in error sufficiently accurate
during dry-out or initial operation and that tem- figures can be obtained.
perature cycling or fast cooling are also injurious. The total heat loss and the corresponding individual
temperatures T2, T3 and T4 are the calculated and com-
pared with the estimate. The calculation is repeated if
3 HEAT TRANSFER
necessary to obtain more accurate results.
The calculation methods for flat walls are well known
but for convenience the basic formulae are given below, 3.2 Cylindrical vessels (see Fig. 2)
as well as those for circular vessels and transfer lines.
The flat wall formula is reasonably accurate for circular The same basic method applies but the formula has
vessels where the refractory thickness is less than 10 per been corrected to allow for the fact that the circum-
cent of the internal diameter of the lining. ference through which the heat flows increases as the
heat flows through the refractory towards the outside
surface. Sufficient accuracy is normally obtained for a
3.1 Flat walls with two layers of different refractory (see transfer line to assume that the inside face temperature
Fig. 1 ) is the same as the bulk gas temperature and the tem-
perature drop through the shell is zero; thus TI = T2
Heat loss Q = Tl - To and T3 = T4.
l/a, + 11/%1 + 12/22 + l/ao
Heat loss Q per m2 of outside surface
-
- TI - To
+ (l/A)d&
~ / E O log d J d ,
A simplified approach is possible which is suffciently
accurate for normal purposes (less than 4 per cent error
for a range of diameters up to d , / d , = 2; see Fig. 2) as
follows: m m

where 1 = refractory thickness.

Typically a1 may be taken as 25 kcal/m2 h "C
(5 Btu/ftz h O F ) for inside refractory line vessels, but for 4 CALCULATION OF EXTERNAL
radiant walls or transfer lines T2 can be taken as equal HEAT-TRANSFER COEFFICIENT
7',( a 1 = a).
Typically a. may be taken as 10 kcal/m2 h "C The heat loss from the outside surface of a vessel or
(2 Btu/ft2 h OF) for normal furnace casings in still air but furnace casing is affected by a number of factors:
should be calculated for abnormal conditions or more (a) surface emissivity,
accurate results (see Section 4). These approximations (b) position, that is horizontal (top or bottom) or verti-
produce a maximum error of only a few degrees in the cal,
shell metal temperature but inaccurate estimates of (c) position relative to other hot bodies (for example
external heat loss produce errors that are not acceptable two pipes close together),
(see Appendix 4). As I , and lo are known, T, and T4 can (d) wind velocity and wind direction,
(e) surface temperature above ambient.

t- - 1- - *I
* L j
Fig. 1 Fig. 2
Parl E: Journal of Process Mechanical Engineering
0 IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 53

Items (d) and (e) have the major influence but item (c) Pressure strain :
can occasionally have a dramatic effect and must be
considered. Circumferential : eSpc
For comparison purposes and various standards it is
normal to specify still air at the maximum expected
ambient temperature that gives a conservative higher Longitudinal: eSpl= % ! (1 - 2ps)
skin temperature; frequently a heat-loss figure of 10 Es
kcal/m2 h “C (2 Btu/ft2 h O F ) is used for casings of fur- Mean thermal expansion :
naces with skin temperatures of the order of 50°C (90°F)
above ambient. Approximate estimates of factors (a)
and (b) are sufficient for furnace casings [(c) is normally
ignored unless there is some specific problem], because
Circumferential : (T), (y)T = = %(T, - T:)

these normally are only 4CL5O”C above ambient and Longitudinal :

even quite approximate estimates only make a few
degrees of difference. However, the external surfaces of
Note. In the circumferential direction expansion refers
lined vessels can be significantly above ambient and
to circumferential dimensions (that is diameter of
accurate calculations are required (see Fig. 3).
thickness), whereas in the longitudinal direction expan-
It is normal engineering practice to make conserva- sion refers to longitudinal direction (height).
tive calculations including safety contingencies. It must
be clear, however, why a calculation is being done and Total shell elongation
what is conservative. Overestimating surface tem- without refractory stress:
perature gives a ‘safe’ figure for heat loss but an ‘unsafe’
view of refractory stresses. Installation of a thicker ref-
ractory than necessary can again be in the ‘unsafe’
Circumferential : CES, = Espc + (y) T
direction because stresses due to the high differential
temperature between hot and cold faces can cause
failure. Therefore, the ‘worst case’ can be either to over-
Longitudinal: 1 E , ~= + (y) T
or underestimate the surface temperature depending
upon what is under consideration.
5.2 Refractory (unrestrained expansion)
5 CALCULATION OF REFRACTORY STRENGTH Mean thermal expansion :
IN VESSELS AND TRANSFER LINES
5.1 Shell (unrestrained expansion) Circumferential : (%)T = (T)T = %(T, - T:)
Pressure stress:

Circumferential: ,a,,
PD
=-
Longitudinal: (y)* = %(T, - 7-3
24 Mean shrinkage :
PD
Longitudinal: aspl= - Circumferential :
41s

50 -
Longitudinal : (y) =; S
(S, + So)
Nett mean unrestrained expansion :

P
c
40 -
CircumferentiaI: 1 E,,
= (y)T (I:s -

-.
“E
z
Longitudinal : CE,I = (y)T (y) - S
ru:
Nett differential expansion:
Circumferential: AS, = E,, - 1 E,,

Longitudinal: AE, = 1 E,, - 1

01 1 I I I I I
5.2. I Stress calculations
0 SO 100 150 200 250 300 For clarification reference is made to Appendix 4. The
Temperature_difference
_ - surface tu air mean refractory stress due to the restraint imposed by
“C the shell is:
Fig. 3 The external heat-transfer coefficient [derived from
The American Society of Heating and Refrigerating E, 4
Circumferential: orsc= AE, E , (la)
Engineers Association, Handbook 191 Es 1s + Er lr
@ IMechE 1990 Proc Jnstn Mech Engrs Vol 204
54 P K RICHARDS

tent results. Tensile strengths are typically 30-50 per

cent of compressive strengths so obviously refractories
are much weaker in tension and may have severe crack-
ing on the cold face (usually under tension) when the
hot face (usually under compression) has not.
5.2.2 Stress differential (hot to cold face)
Stress differential from mean at hot and cold faces
6.2 Modulus of rupture (refer to ASTM C133)
caused by thermal expansion gradients :
In the opinion of the author, this test is considered as a
E,ar(T,l - T:) more important criterion and is more representative of
Ort = (4
2(1 - Pr) the structural strength as required in practice.
Due to shrinkage gradient:
6.3 Permanent linear change
(3) A low shrinkage is often considered to be a good thing
because it is thought to reduce cracking. However, it
Above the stress differential relations are valid in the must be remembered that the refractory expands when
circumferential as well as in the longitudinal direction. it gets hot and in any situation of restraint the shrink-
Stress in refractory at hot face: age can compensate for the thermal expansion so that
the compressive stresses under operating conditions are
Equations (1) - (2) + (3) reduced. Cracks will appear when the refractory cools,
but these will close up again when reheated. Any com-
Stress in refractory at cold face:
pressive stresses will of course be reduced by shrinkage.
Equations (1) + (2) - (3) Not all refractory materials show a negative permanent
linear change (shrinkage), as some expand
Mean circumferential stress in shell due to refractory: (permanently) on heating and some may shrink initially
1 and permanently expand at higher temperatures.
oSrc
= - [equation (la)] 2 The permanent linear change test itself is normally
1, done by measuring at room temperature the length of a
Mean longitudinal stress in shell due to refractory : specimen: by example 50 x 50 x 230 mm in the 230
mm direction first after drying at 105°C and again (at
Dr 1,
G,,,= -[equation (1b)] - room temperature) after heating to 8 15°C. Permanent
Ds 1, linear change is the percentage of linear change based
There are few data available on the properties of refrac- on the length measured after drying. These tests ignore
the fact that the original sample has shrunk (or
tory material and what is published is often contra-
occasionally expanded) from the as-cast stage to the
dictory. Section 6 is therefore devoted to this subject.
dried stage (l05T) and so the total shrinkage in an
actual application is greater than (usually) that indi-
6 PHYSICAL PROPERTIES cated by the test.
(of refractory material in general) No reliable data are available on these phenomena
but for hydraulic setting refractories less dense than
(For reference to various ASTM standard methods for about 2000 kg/m3 the permanent linear change from
testing see Appendix 5.) cast to 100°C can be taken as 0.1 per cent.
There are various standard methods for testing refrac- It is again stressed how important it is that tests are
tories and it is essential t s use consistent methods or carried out at a condition truly representing the oper-
different results will be obtained. It is equally essential ating condition and that standard test results as quoted
to ensure that the basis of the test is applicable to the by the manufacturers should show the initial linear
case in mind. For example, it is useless to compare change, which is a once-only phenomenon, as well as
erosion test results when one test has been carried out the permanent linear change.
with the jet at 90" to the sample and another has been
at 45". Similarly, it is useless to use results from a
sample that has been prebaked to 500°C when consider- 6.4 Thermal conductivity
ing the cold face of a refractory lining that never in its Here again there are various different tests performed in
life exceeds 100°C. different parts of the world and different laboratories.
Tests on refractories both for preshipment qualifi- Some types of tests are slow (for example ASTM C201
cation and to verify 'as-installed' specimens have his- takes about 2 weeks) but all give different results. The
torically been carried out to ASTM C704 Standard so hot wire (parallel wires) test gives results about 20 per
that proper comparisons can be made. To test linings cent higher than ASTM C201. Correlation is possible
for specific purposes requires tests of a different type. between different types of test and different labor-
atories; however, this has to be done in a consistent
6.1 Cold compressive strength (refer to ASTM C133) way to obtain reasonable results. For calculation pur-
poses and consistency with actual as-built results the
This is generally used as a quality measurement rather ASTM C201 results appear to be the most reliable. All
than as a basis for assessing the strength under load. calculations in this paper have been based on ASTM
Tensile strength is rarely measured as it gives inconsis- C201 results.
Part E: Journal of Process Mechanical Engineenng 0 IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 55

Conductivity varies with temperature and conse- are met when the shrinkage equals about half the
quently differs through the refractory lining thickness. reversible thermal expansion at the hot face. However,
Normally mean values will be applied. both these factors are properties of the material that
Conductivity is also very high during initial dry-out cannot be changed and the only control is by choosing
because of the water content and may easily be 20 per a material with suitable properties overall.
cent higher than expected, causing concern if surface
temperatures are measured during dry-out. 6.6 Modulus of elasticity ( E )
Conductivity also varies with the type of atmosphere
in the vessel. Most common gases have similar thermal This is defined as the ratio of stress to strain for normal
conductivities but for hydrogen the thermal conductivi- compressive or tensile stresses. Very little data are avail-
ty is extremely high and thus when a porous refractory able for refractory materials and frequently this is con-
absorbs gas containing more than about 10 per cent flicting (usually because of different test bases). Tests are
hydrogen the apparent conductivity of the refractory is normally done under compression because tensile data
measurably increased (see Fig. 4). are erratic and unreliable. The value is also dramatically
Thermal conductivity of refractory mass in gaseous affected by the temperature to which the test samples
atmosphere = (1 - P2/3)4+ Pli3Ag have been heated and to a lesser (but significant)
amount by the time after manufacture of the samples
where and the actual temperature during the test.
P = apparent porosity E is a function of composition, structure, water
I , = conductivity of refractory content, temperature, load history etc., but for a given
A, = conductivity of gas mixture set of conditions has been shown to be reasonably con-
Note : This correction for the increased thermal conduc- sistent and within the ranges of permitted stresses and
tivity due to hydrogen atmospheres has been found to temperatures to show elastic behaviour. Note particu-
give a reasonable estimate even for ceramic fibres with larly that water tensile properties (with consequently a
their extremely high porosity. lower elastic limit) and the effect of creep at higher tem-
peratures.
There are two basic methods used for the test which
6.5 Reversible thermal expansion give different results: the sonic method (ASTM C215)
Limited data are available but for common refractories and the static method (ASTM C845). These ASTM
values for thermal expansion normally lie between : numbers are quoted as examples only as there is at the
present time no International Standard for measuring
6x per "C (3.3 x per O F ) the modulus of elasticity, although there are a number
9x per "C (5 x per O F )
of National Standards and University and Research
Centre methods which differ in detail and produce a
for densities between 2300 kg/m3 and 1300 kg/m3 scatter of results. This paper is based on results
respectively. Figures may be taken pro rata (on a linear obtained by the static method generally in accordance
scale) if specific data are not available for the refractory with ASTM C845. Correlation between results from dif-
under consideration. ferent methods can only be made if samples of the same
Reference should be made to the previous section on material have been tested by each method.
linear shrinkage as one can clearly compensate (partly) For testing the strength design of a refractory the
for the other. As a general guide, optimum conditions static method of testing is preferred because it more

0.5 t Hydrogen in air (moi a)

0.3 -

0.2 -

0.1 -

O L I I 1 I I I
2 4 6 8 10 12 14
Average temperature
x 100°C
Fig. 4 Thermal conductivity of air, hydrogen and mixtures of the two
(Lindsay-Bromley expression)
0 TMechE 1990 Proc Instn Mech Engrs Vol 204
56 P K RICHARDS

nearly resembles the actual installed condition. 0.3

However, the sonic test is preferred by laboratories
because it is easier to perform. This method gives results Refractories with densltles up to 1600 kglm'
up to 30 per cent higher than the static method depend-
ing upon the type of refractory under test. If only sonic
figures are available a reasonable (and generally
0.2
! ,Densities above 1600 kgim'
conservative) estimate may be taken as 85 per cent of
this value in place of static figures.
Fortunately, test results have shown that there is a 0. 1
reasonable relationship for unbaked samples of concrete
Ternperaturc
between the modulus of elasticity at room temperature
"C
( E ) and cold compressive strength (f): 3X 260 s3x 816
I I I I
E = CJf x d e n ~ i t y ' . ~ 0 I 1 I 1
100 500 1000 1500
where C is a reliable constant for a particular strength Temperature
refractory but is itself dependent upon the strength "F
being somewhat lower for the stronger materials (see Fig. 6 Poisson ratio for typical refractories (with or without
Fig. 5). added fibres)
This relationship holds reasonably well for many ref-
ractories at high temperature using the crushing
strength at the appropriate temperature in the formula. addition or not of fibres appears to have little, if any,
However, some materials diverge widely as the elastic effect on the Poisson ratio (see Fig. 6).
modulus falls far more rapidly than would be antici-
pated. A lower than anticipated modulus of elasticity
leads to lower stresses with the same strain in restrained 7 STEEL FIBRES
conditions in lined vessels. It is, however, suspect for the
weight loaded condition, as in a self-supporting wall, Fibres of various types and materials can be added to
where the same load (stress) produces a bigger deflec- concrete as reinforcement. (They have little or no effect
tion (strain). on the erosion properties.) Naturally the material must
be suitable for the operating temperature and environ-
ment, so refractory reinforcement is usually by heat-
6.7 Poisson ratio resisting steels, frequently some type of stainless steel. A
This is defined as the ratio of transverse (lateral) strain typical fibre would be type 302 or 304 stainless steel,
to the corresponding axial (longitudinal) strain resulting about 25 mm long and 4 mm in diameter.
from uniformly distributed axial stress. This varies lin- Fibres are manufactured by two basic processes :
early for refractories of the ranges under consideration (a) chopped wire,
from 0.12 for lighter weights (density 1300 kg/m3) to (b) melt extraction process.
0.17 for the heavier weights (density 2000 kg/m3). The
denser types are affected by temperature and approach Chopped wire is normally used for cast refractory appli-
similar values to the lighter types at temperatures of cations and the melt extract type, which is more flexible,
loOo"C, while the lighter types are hardly affected. The is normally used for gunned applications, but it may
also be used for cast. The addition of fibres has been
demonstrated to increase the flexural strength and
general toughness as well as increasing the crack and
spa11 resistance. Obviously the fibres make no direct
contribution in areas where the temperature.exceeds the
oxidation resistance and the fibres corrode away.
30 -
However, the refractory generally has only the hot face
at high temperatures and for the majority of the refrac-
tory thickness the fibres do not deteriorate.
Furthermore, when a refractory is raised to its oper-
ating temperature the hydraulic bond is replaced by the
ceramic bond formed by the reaction of the com-
ponents. Both the hydraulic and ceramic bonds nor-
mally have sufficient strength, but a weak zone exists
where the temperature is high enough to weaken the
- hydculic bond but not high enough to form the
ceramic bond. Fibres can give significant benefit in this
zone.
01 I I 1 I I The amount of fibres added is small, between 1.5 and
1100 1400 1600 1800 2000 2200
2.5 per cent by weight, but steel has a conductivity of up
Density to 150 times that of a refractory and an elasticity of
kgirri ' about 30 times that of a refractory.
Fig. 5 Relationship between modulus of elasticity with It is normal practice with reinforced concrete to
density for unbaked samples of concrete assume lower values of elasticity and strength in design-
Part E: Journal of Proms Mechanrcal Engmeering @ IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 51

ing, because this is 'safer'. The opposite is the situation Care must therefore be taken in interpreting the results
in this case as this underestimates the stresses and loads of such calculations.
in vessels and transfer lines when considering the
stiffness of the refractory-lined item and the effects of 7.1 Calculation of thermal conductivity with
expansion/bending. added fibres
A number of commentators have stated that the addi-
tion of fibres has a negligible effect on conductivity, but If the fibres are fully randomly orientated in three
this is contradicted by others who state that it has a planes, it can be shown that the combined conductivity
substantial effect on the lightweight insulating types. of the composite mass for small fibre percentages is
The situation is further confused because conductivities
are normally quoted for 'prebaked' samples which will
+ sin 45" (A, + $is)
+ (1 sin 45")A,
-

cause a reduction in conductivity of the fibre-reinforced where

refractory which does not correspond to the situation in t j = % by volume of fibres
the usual case where the temperature will never exceed

(1,+ A)+
a much lower figure (maybe only 2W3OO"C at the
outer edge of the refractory lining, resulting in a higher = 0.50 ~ 0.30JL,(approx.)
conductivity). Calculations confirm that there is only a
marginal effect for fibres on denser-type refractories (10 II
per cent for 2250 kg/m3), but it can be substantial for = 0.8A, + 0.5 100
lighter types (nearly 50 per cent for 1000 kg/m3) (see
Fig. 7). for 1% fibres by volume (see Fig. 7)
Calculations of elasticity E for a refractory without Clearly much more effect is gained from adding fibres to
fibres give figures that are reasonably consistent with less dense refractory (insulating refractory) material
the limited test data available for that material. The than from others. If there is doubt about the strength of
effect of fibres on elasticity is considerable, being of the an insulating refractory for a particular duty, it is pref-
order of 40-50 per cent depending upon the type of ref- erable to use a slightly stronger material and accept its
ractory and the temperature. The calculation is based slightly inferior insulating properties rather than add
on treating the mixture as reinforced concrete. Again fibres which are expensive and which themselves reduce
'prebaking' of samples will produce low values, which the insulating properties. If preferential orientation
may not be representative, as the refractory suffers a occurs then the value varies from zero if all the fibres
permanent reduction in elasticity after heating. are perpendicular to the direction of heat flow to (A,
The data available and the calculation methods are
not precise and there are various factors that can
+ A,/lW) if all are parallel.
increase the vessel and transfer line stresses, which have
to be taken into account, for example 7.2 Calculation of modulus of elasticity with
added fibres
(a) variation in the physical properties of a refractory,
(b) heat conducted through refractory anchors, Consider a cube of refractory (without fibres) under
(c) erosion causing thinning (particularly in transfer direct normal compressive stress 0 on all faces. The
lines), strain of each side is
(d) heating effect from other hot surfaces (particularly IT
E = - (shortened) -
0
2 p - (lengthened)
lines between vessels). E E

Conductivity ot refractory AIt

kcal/m2 h "C
Fig. 7 Conductivity calculated for new material heated only to operating
temperature (not prebaked)
8 IMechE 1990 Proc Instn Mech Engrs Vol 204
58 P K RICHARDS

where p = the Poisson ratio. Hence, negating effects of fibre pull-out and fibre burn-out (at
high temperatures). None of this work is conclusive and
U much is even contradictory. However, it is clear that the
& = - (1 - 2p)
E composite strength is only slightly increased in tension
and somewhat more in compression, as influenced by
Volumetric strain:
the above characteristics and depending upon the
strength (density) of the refractory under consideration.
&,,I = (E + & + &) = 3 -UE (1 - 2p) For design purposes it is recommended to take the
increased tensile strength due to 1 per cent added fibres
-_ U as nil and for compression as plus 10 per cent.
-
K Permissible compressive stresses both with and
without fibres may be taken as 80 per cent of the corre-
where K = bulk modulus. Therefore, sponding cold crushing strength; permissible tensile
E = 3K (1 - 2 p ) stresses without fibres may be taken as 40 per cent of
the corresponding cold crushing strength and with
If the bulk modulus due to addition of fibres is fibres as 60 per cent to allow for the crack repressing
increased by, say, 10 per cent it is obvious that if the properties. Designs have been built to these limits using
Poisson ratio is not increased then the modulus of elas- the calculation methods in this paper with satisfactory
ticity is also increased by 10 per cent. This is, however, results. Where stresses higher than these are calculated
too sweeping an assumption as tests indicate that the it does not mean that the refractory will necessarily fail,
Poisson ratio also increases by a marginal amount. It is but higher maintenance can certainly be expected.
reasonable to assume that the Poisson ratio in all main
directions for the cube will be increased by 1.5 per cent.
Hence, 8 PRACTICAL LIMITATIONS

E with fibres = 3 ( 1 . K ) { 1 - 2(1.015p)} 8.1 Basic material

Taking for refractory p = 0.15, Mechanical engineers sometimes think about a refrac-
tory in the same way as they do about steel, forgetting
E = 2.3K that these types of materials (bricks, stone, cement etc.)
compared with 2.1K without fibres. Thus, increase in have been the basis of buildings since Roman times and
E = 2.3/2.1 = 9.5 per cent (that is almost equal to the 10 before. Refractories do contain defects, cracks, cavities,
per cent increase in bulk modulus). Clearly the change porosity to a far greater degree than steel, and it is not
in the Poisson ratio has a negligible effect. practical to X-ray or ultrasonic test a refractory-lined
It is frequently assumed that fibres randomly orien- vessel containing hundreds of tons of refractory. Much
tated in three planes have a much reduced effect on the therefore depends upon the quality and consistency of
elastic modulus, but this is clearly not true although it the refractory material and its application. Catalogue
certainly does apply when considering the effect on data cannot be accepted because of gradual changes in
increasing the direct strength. Clearly, as with heat con- the raw material used by the manufacturer and there
duction, preferential orientation of fibres perpendicular may be a complete change of source.
to the load have a zero effect on the strength (see Fig. 8).
Hence, 8.2 Testing
E with fibres = $Es + (100 - $)E, All the major properties should therefore be tested to
where ensure that the material type is what is required, and
$ = % by volume of fibres then a limited number of representative properties
should be tested regularly to check that the material is
E, = modulus of elasticity of steel
E , = modulus of elasticity of refractory
consistent and properly and consistently applied. It is
vital that tests carried out by different suppliers, techni-
Note the variation in shape of the curves for different cal laboratories etc. are on the same basis (or corre-
materials. ‘Weaker’ materials (insulating refractories) lation test performed to obtain comparative results) and
are less affected by heating whereas some ‘strong’ also that test samples truly represent the condition that
materials are considerably affected. Fibres have a pro- you are interested in. For example, a ‘standard’ test may
portionately greater effect on ‘weak‘ materials but this be performed at 1000°C which is in no way representa-
tends to be overestimated because of the greater pro- tive of the material at the cold face.
pensity for pull-out under tension. Refractory ages and property changes occur in air-
dried (not baked) samples with time, and so it is impor-
tant when establishing properties for design purposes
7.3 Calculation of refractory strength with fibres
that samples be left for 28 days in the air-dried condi-
It is clear that fibres orientated perpendicular to the tion before testing or before baking for further testing.
direction of direct tensile stresses have little or no effect This is not necessary during production testing where
on the tensile strength of the composite. However, for testing methods must be uniform in order to check con-
direct compressive stresses, fibres orientated other than sistency of the final installed product.
in the stress direction do have an effect. Much has been Hydrotesting of refractory-lined pressure vessels is
written about the effectiveness of randomly orientated possible without damage to the refractory provided that
fibres, their contribution to crack repression and the the pressure is relieved slowly after test. Otherwise the
Part E . Journal of Process Mechanical Engineering Q IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 59

Density 1900 kgiin '

CCS (unbaked) 85 MPa
With fibres
After baking to 800°C 24.6 MPa

i b
Without fibres With tibrcs ( 1 76 by vol)

10 - 1C
Denvty 2250 kg/rn3
CCS (unbaked) 100 MPa
After baking to 800°C 90 MPa

I 1 I 1 1 I
200
1
400
I
600
1
xo0
0 200 400 600 800
Temperature
~~~

"C

- -

Dcnvty 1400 kglm'

CCS (unbaked) 14 MPa
20 - After haking to 800°C I 3 MPa
20 -
Uenvi) 1200 k g / m l
CCS (unhakcd) 5 MPa
After baking to X00"C 5 MPa

With tihre\
10 - 10 - With fibres ( I % by bol)

Without fibre\
d
1 1 1 I
f Without fibre\
1 1
0 200 400 600 xoo 0 200 400 600 800
Temperature
"C
Fig. 8 Effect of temperature on modulus of elasticity (with and without
fibres)

refractory can be shattered, if porous. This is normally 8.3 Quality control

only a problem on small vessels which can be almost As inspection after installation is, to say the least,
instantaneously relieved of pressure if care is not taken. imprecise: control at every stage is essential. The full
Hydrotesting of new lined vessels and transfer lines is properties on which the design was based may never
normally only accepted for welds which are not covered fully develop or may be destroyed if, for example,
by refractory, as in site closing welds which are later
refractory lined. Particular note must be taken of 1. Too much water was used for easier application.
certain chemical bonding refractories, which must be 2. Rebound was buried under fresh refractory.
fully baked before being exposed to water, which can 3. It was installed during hot weather and dried too
destroy them. A check should always be made with the quickly.
manufacturer with all refractories if a hydrotest is 4. It was installed during cold weather and froze.
expected. Some or all of these incorrect installation practices are
0 IMechE 1990 Proc Instn Mech Enprs Vol 204
60 P K RICHARDS

not detectable during later inspection, especially if they dards, testing and design. National Physical Laboratory Conference
occur towards the outer layer of the refractory against Proceedings, April 1974.
BS CPllO Part 1: 1972 The structural use of concrete, 1972 (British
the steel shell. Standards Institution, London).
Cbesters, J. H. R$ractories production and properties, 1973 (Iron and
8.4 Dry-out Steel Institute, London).
Dixon, J. and Mayfield, B. Concrete reinforced with fibrous wire.
Finally, all the efforts outlined above to choose and Concrete, March 1971, 73-76.
install a suitable refractory can be spoilt by not adher- Edgington, J., Hannant, D. J. and Williams, R. I. T. Steetfibre rein-
ing closely to a suitable dry-out, heat-up and cool-down forced concrete. Current paper C P 69/74, July 1974, Building
programme. The rate of cool-down can be just as Research Establishment.
Hackman, L. E. Steel fibres reinforcement of refractories, May 1976
important as the rate of heat-up, and this applies also to (American Ceramic Society, Pittsburgh).
operational start-up and shut-down. Hackman, L. E. and Baker, R. Ribtec steel fibre in refractories. Tech-
nical paper presented at Bedford Springs, Pa., October 1977
(American Ceramic Society, Pittsburgh).
Handbook of concrete engineering, 1974 (Van Nostrand Reinhold, New
9 SUMMARY York).
Hannant, D. J. Steel fibres and lightweight beams. Concrete, August
Although the data and calculation methods are not 1972.
precise, different types of refractory can be compared Hannant, D. J. Fiber cements andfiher concretes, 1978 (John Wiley
and the most favourable chosen for a particular duty. and Sons, New York).
Appendix 1 shows a typical calculation for different Holiday, L Composite materials, 1966 (Elsevier, Amsterdam).
Holiiter, G . S. and Thomas, C. Fibre-reinforced materials, 1966
types for the same duty. It can clearly be shown which (Elsevier, Amsterdam).
designs are unacceptable and which are borderline cases Kelly, A. Strong solids, 1973 (Clarenden Press, Oxford).
and require further consideration. The lightest refrac- Kelly, A. Some scientific points concerning the mechanics of fibrous
tory (type C) is well outside the acceptable stress level composites. National Physical Laboratory Conference on
(both with or without fibres) and is liable to heavy Composites-standards, testing, and desiyn, April 1974.
Johnston, C. D. and Coleman, R. A. Strength and deformation of
maintenance. On the other hand, the strongest refrac- steel-fibre-reinforced mortar in uniaxial tension. In Fibre-reinforced
tory (type A) would probably be accepted both with concrete. American Concrete lnstitute Publication SP-44, 1974.
and without fibres (hardly for tensile stresses without Johnston, C. D. Steel-fibre-reinforcedmortar and concrete: a review of
fibres), but it has to be so thick that the increased mechanical properties. In Fibre-reinforced concrete. American Con-
crete Institute Publication SP-44, 1974.
weight would make it uneconomic. The middle range Krenchel, H Fibre reinforcement, 1964 (Akademisk Forlag,
(type B) has clearly the most favourable properties, even Copen hagen).
without added fibres. In Appendix 6 the listing (in Krenchel, H. Fibre spacing and specific fibre surface. RILEM Sympo-
BASIC) is shown of the computer program used for ref- sium, on Fibre-reinforced cement and concrete, 197.5, Vol. 2, 1976
ractory calculations. (Construction Press).
Lankard, D. R. and Sheets, H. D. Use of steel wires in refractory
The importance of using design data based on the castables. Bull. Am. Ceram. Soc., May 1971,50(5).
actual operating conditions cannot be overstressed. Lankard, D. R. Flexural strength predictions. Construction Engineer-
Data based on standard tests at standard conditions ing Research Laboratory, Conference Proceedings M-28, December
can be completely misleading, particularly when con- 1972, Champaign, 111.
Laws, V. The efficiency of fibrous reinforcement of brittle matrices. J .
sidering the refractory strength at the outer (cold) walls. PhyS. D : Appl. Phys., 1971,4,1737-1746.
The common practice of considering the refractory Mogha, N. C. Effective stiffness of concrete coated lined pipe. ASME
weak under tension and providing no stiffness under Publication 71-Pet-26, 1971.
bending on the tension side due to cracking can often Palin, F. T. and Padget, G . C. Thermomechanical behaviour of refrac-
tory castable linings. British Ceramic Research Association 320,
be completely erroneous and refractory-lined transfer March 1981.
lines may be considerably stiffer than has been assumed. Peterson, J. R. and Vaughan, F. H. Metal fiber reinforced refractory
In order to assess the sensitivity to errors in certain of for petroleum refinery applications. National Association of Corro-
the basic data, calculations are shown in Appendix 3, sion Engineers, 1981.
with variations of up to 50 per cent in the internal and Shah, S. P. and Rangan, 6 . V. Effects of reinforcement on ductility of
concrete. Proc. Am. SOC. Civ. Engs: J . Struct. Div., June 1970.
external heat-transfer coefficient and in thermal conduc- Shah, S. P. and Rangan, B. V. Fibre-reinforced concrete properties. J .
tivity of the refractory. These clearly show that errors in Am. Concrete Inst., February 1971, -14.
the estimate of the external coefficient are extremely Snyder, J. M. and Lankard, D. R. Factors effecting the flexural
important. Appendix 7 gives the listing (in BASIC) of strength of steel fibrous concrete. J . Am. Concrete Inst., February
the computer program used for sensitivity calculations. 1972,6!3-9.
Swamy, R. N. The technology of steel-fibre-reinforced concrete for
practical applications. Proc. Inst. Civ. Engrs., paper 7694, 1974.
Swnmy, R. N., Mangat, P. S. and Rao, C. V. S. K. The mechanics of
BIBLIOGRAPHY fibre reinforcement of cement matrices. In Fibre-reinforced concrete.
American Concrete Institute Publication SP-44, 1974.
ACT 318; 1977 Building code requirements, American Concrete Insti- Technology of monolithic refractories, 1979 (Phybrico Japan
tute, 1977. Company).
Aveston, J., Mercer, R. A. and Sillwood, J. M. Fibre-reinforced Wygant, J. F. and Crowley, M. S. Designing monolithic refractory
cements-scientific foundations for specifications. Composites, stan- vessel linings, May 1963 (American Ceramic Society, Pittsburgh),

Part E : Journal of Process Mechanical Engineering @ IMechE 1990

 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 61

APPENDIX 1
Calculation results for different refractories for same duty (typical examples)
(a) Basic
Vessel application Gasifier
Vessel diameter 16 m
Vessel thickness (steel) 30 mm
Circumferential pressure stress 90 MPa
Inside gas temperature 980°C
Vessel metal temperature 125°C
Ambient temperature 20°C
Wind speed 2 m/s

(b) Dutu
A = 2250 B = 1900 C = 1400
Refractory (gunned) density (kg/m3)
1% by volume added fibres Without With Without With Without With
Cold crushing strength at 20°C (MPa) 100.0 101.0 85.0 87.0 14.0 15.4
CCS baked to 815°C (MPa) 90.0 91.0 25.0 26.5 13.0 14.3
Thermal conductivity at 540°C (kcal/m2 h "C) 0.96 0.98 0.68 0.74 0.39 0.49
Thermal expansion (mmjmm "C x to6) 6.5 6.5 6.3 6.3 8.75 8.75
Permanent linear charge ('%)
Cold face 0.1 0.1 0.1 0.1 0.1 0.1
Hot face 0.3 0.3 0.4 0.4 0.4 0.4
Young modulus
AT 125°C (GPa) 29.5 31.5 19.1 21.1 6.0 8.0
At 92WC (GPa) 20.5 22.5 5.3 7.3 4.0 6.0
Poisson ratio 0.14 0.14 0.14 0.14 0.10 0.10
Permissible stress*?
Compressive (hot face) (MPa) -72.0 -73.0 -20.0 -21.2 -10.4 -11.4
-
Tensile (cold face) (MPa) f40.0 +60.6 +34.0 +52.2 + 5.6 + 9.2
* Permissible stress based on 80 per cent of cold crushing strength for compression in refractory with and without fibres, and 40 per cent of
cold crushing strength for tension in refractory without fibres and 60 per cent with fibres.
t Tensile stress denoted positive +.
Compressive stresses denoted negative -.

(c) Results
A = 2250 B = 1900 C = 1400
Refractory (gunned) density (kg/m3)
1% by volume added fibres Without With Without With Without With
__
Circumferential pressure
strain (mm x lo3) 21.5 21.5 21.5 21.5 21.5 21.5
Thermal expansion of shell (mm x lo3) 3 1.95 31.95 31.95 31.95 3 1.95 31.95
Total shell expansion (mm x lo3) 41.80 41.80 41.80 41.80 41.80 41.80
Mean expansion of refractory (mm x lo3) 82.5 82.5 80.2 80.2 112.3 112.3
Mean shrinkage (mm x lo3) -51.0 -51.0 - 63.5 -63.5 -63.5 -65.3
Net mean unrestrained
expansion of refractory (mm x lo3) 31.5 31.5 16.7 16.7 48.8 48.8
Net differential circular
expansion (mm x lo3) 10.3 10.3 25.1 25.1 - 7.0 - 7.0
Required refractory thickness (mm) 500.0 520.0 355.0 390.0 200.0 250.0
Mean circular refractory stress (MPa) + 3.0 + 3.1 +5.1 + 5.8 - 1.1 -- 1.4
Stress differential from mean
due to thermal expansion (MPa) f66.0 72.2 f22.1 k28.1 17.4 f25.5
Stress differential from mean
due to shrinkage (MPa) f25.6 28.0 f13.2 +
- 16.8 3.6 f5.2
Circular stress at hot face (MPd) - 31.4 -41.1 - 3.8 - 5.5 - 14.9 -21.7
Circular stress at cold face (MPa) +43.4 + 47.3 + 14.0 + 17.1 + 12.7 + 18.9
Circular stress in shell due
to refractory (MPa) - 50.0 - 54.0 - 60.0 - 75.0 + 7.3 + 11.7
Actual stress
Allowable stress at hot face 0.52 0.56 0.19 0.26 1.43 1.9
Allowable stress at cold face 1.09 0.78 0.41 0.33 2.27 2.05

(Q IMechE 1990 Proc Instn Mech Engrs Vol 204

62 P K RICHARDS

APPENDIX 2
Calculation of vessel internal with refractory on both sides
Temperature 700°C

. . . .. . . . . . . . . .
. . .
. .. . .. .
. . . . .. .
. . . . . .
, .
. .
. . .
. ..@
I
Refractory type C (see Appendix 1)

100 mm thick each side on 12.5 mm thick steel plate, supported by 10 mm diameter anchors at 250 mm square pitch.
Assume slow heating so all at substantially same temperature and fixed attachment of refractory and steel at the
anchors.
Due to temperature:

12.24 x (700 - 20) = 8.32 x

Expansion of steel = __ mm/mm
lo6
6.48
Expansion of refractory = - x (700 - 20) = 4.41 x 10 ~ mm/mm
lo6
Differential train = 3.91 x rnmjmm
Total load in refractory = total load in steel (equilibrium condition)
Stress in refractory x area (ref.) = stress in steel x area (steel)
x
CT, A , = r ~ ,x A ,
Take a strip containing one line of anchors (250 mm wide)
12.5 x 250
0,= 0 , x = Q.0625aS
200 x 250
0s CT-
Differential strain = - (compression) + (tension) = 3.91 x
Es Er

=-
rJS
+ 0.0625 'T. = 3.91 x
E S Er
3.91 x lop6 - 3.91 x
6,= = 189MPa
l/Es + 0.0625/Er - (1/196 + 0.0625/4) x l o p 9
6,= 0.06250, = 11.8 MPa (tension)
It will readily be seen that both the refractory and the steel (depending on type) are overstressed, even disregarding
peak stresses at anchors (permissible tensile stress refractory 5.6 MPa, see Appendix 1).
Let us assume the refractory does not crack nor the steel buckle. Then:
Load on one anchor x
= CT, A, = 11.8 x 250 x 100 MPa
= 295000 N

'II
Cross-section area =- x lo2 = 78 mm2
4
295 O00
Shear stress on anchor = ~ = 3782 MPa
78
which is obviously impossible.
Long before the system has reached full temperature any refractory bond with the steel plate will have broken, the
refractory will have cracked (commencing from the anchor positions) and the anchors will have severed, depositing
the refractory in pieces upon the floor.
This situation of gross overloading (with a superficially reasonable design) can easily be overcome, for example by
using a stainless steel plate (with approximately half the coeficient of expansion), a thinner layer of refractory (only 25
rnm is necessary for erosion resistance), combined with a greater number of anchors of higher tensile strength. Even
so, some expansion gaps must be left between sections of the refractory and around the anchors themselves to prevent
the refractory from being broken up due to expansion.
Part E Journal of Process Mechanical Engineering @ IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 63

APPENDIX 3
Sensivity to errors
Gas temperature 588°C Gas temperature 540°C
Ambient temperature 20°C Ambient temperature 20°C
Shell 0.d. 1.1684 m Shell 0.d. 0.4648 m
Refractory 0.d. 1.143 m Refractory 0.d. 0.4658 m
Refractory i.d. 0.889 m Refractory i.d. 0.6858 m
Internal coefficient 50 kcal/m2 h "C Internal coefficient 50 kcal/m2 h "C
Refractory condition 0.72 kcal/m2 h "C Refractory condition 0.72 kcal/m2 h "C
External coefficient 20 kcal/m2 h "C External coefficient 20 kcal/m2 h "C
Heat loss 1711 kcal/m2 h "C Heat loss 1770 kcal/m2 h "C
Hot face 455°C Hot face 49PC
Cold face 106°C Cold face 108°C
Metal temperature 105°C Metal temperature 108°C

(a) Variation in internal coefflcient

-
M1 -Q 5 t3
- t4
-
%I % Yo % %
~

-50.0 -8.5 -8.1 -6.9 -6.9 -50.0 -0.7 -8.3 -7.1 -7.1
-40.0 -5.8 -5.6 -4.7 -4.7 -40.0 -6.0 -5.7 -4.9 -4.8
-30.0 -3.8 -3.6 -3.1 -3.1 -30.0 -3.9 -3.7 -3.2 -3.2
-20.0 -2.2 -2.1 -1.8 -1.8 -20.8 -2.3 -2.2 -1.9 -1.9
-10.0 -1.0 -0.9 -0.8 -0.8 -10.0 -1.0 - 1.0 -0.8 -0.8
+o.o +o.o +o.o +o.o +o.o +o.o +o.o +o.o +o.o +o.o
+10.0 +0.8 +0.8 +0.6 +0.6 +10.0 +0.8 +0.8 f0.7 +0.7
+20.0 +1.5 +1.5 +1.2 +1.2 +20.0 +1.6 +1.5 +1.3 f1.3
+30.0
+40.0
+2.2
+2.7
+2.1
+2.6
+1.7
+2.2
+1.7
+2.2
+30.0 +2.2 +2.1 + 1.8 + 1.8
+40.0 +2.8 +2.6 +2.2 +2.2
+50.0 +3.0 +3.0 +2.6 +2.6 +50.0 +3.2 +3.1 +2.6 +2.6

(b) Variation in refractory condition

-
1 -
Q -t 2 -
t3 -
t4 0
-
YO Yo Yo % ?4J
~

- 50.0 -42.0 -4.1 -34.1 -34.1 - 50.0 -42.3 - 4.2 -34.5 -34.5
-40.0 -32.6 -3.2 -26.4 -26.4 - 40.0 - 32.8 -3.3 -26.8 -26.7
-30.0 -23.7 -2.3 - 19.2 -19.2 - 30.0 - 23.9 -2.4 - 19.5 -19.5
-20.0 -15.3 -1.5 - 12.4 -12.4 - 20.0 - 15.4 - 1.5 -12.6 -12.6
- 10.0 -7.4 -0.7 -6.0 -6.0 - 10.0 - 7.5 - 8.7 -6.1 -6.1
+ 0.0 +o.o +o.o +o.o +o.o +0.0 + 0.0 +0.0 +0.0 +o.o
+ 10.0 +7.8 -8.6 +5.7 +5.7 + 10.0 + 7.1 -8.7 +5.8 +5.8
+20.0 +13.7 -1.3 + 11.1 +11.1 + 20.0 + 13.9 - 1.4 +11.3 +11.3
+ 30.0 +20.1 -1.9 +16.3 + 16.3 + 30.0 + 28.3 - 2.0 + 16.6 +16.6
+40.0 +26.2 -2.5 +21.2 +21.2 +40.0 + 26.5 - 2.6 +21.6 -21.6
+ 50.0 +31.9 -3.1 +25.9 +25.9 + 50.0 + 32.3 - 3.2 -26.4 -26.3

(c) Vuriation in external coefjcient

t4
- -
t3 -
t4
% YO Yo
~

- 50.0 -15.1 +1.4 + 56.2 + 56.5 -50.0 -14.5 + 1.4 + 57.6 + 57.8
-40.0 -10.6 +1.0 + 39.4 - 39.6 -40.0 -10.1 +1.0 +48.3 +48.5
-30.0 -7.0 f0.7 + 26.3 +26.5 -30.0 -6.8 +0.6 + 26.9 + 27.0
- 20.0 -4.0 t0.4 + 15.0 - 15.9 -20.0 -4.0 f0.4 + 16.1 - 16.2
- 10.0 -1.9 f0.1 + 7.2 + 7.2 -10.0 -1.0 +0.1 + 7.3 + 7.3
+ 0.0 fO.0 +o.o + 0.0 +0.0 +o.o +o.o +o.o + 0.0 +0.0
+ 10.0 +1.6 -0.1 -6.1 -0.1 +10.0 +1.5 -0.1 -6.2 - 6.2
t 20.0 +3.0 -0.3 11.3 - 11.4 +20.0 +2.9 -0.2 11.5 - 11.6
+
~

-
30.0 +4.2 -0.4 - 15.9 - 16.0 +30.0 +4.0 -0.4 - 16.1
+ 40.0 +6.3 -0.6 - 23.4 -23.6 +40.0 f5.1 -0.5 -20.2
- 16.2
- 20.3
+ 50.0 +6.3 -0.6 -23.4 -23.6 +5O.O +6.0 -0.6 - 23.8 -23.9

8 IMechE 1990
Proc lnstn Mech Engrs Vol 204
64 P K RICHARDS

APPENDIX 4
Calculation of refractory stress due to restraint imposed by shell
Consider the shell with refractory attached.
(a) Assumptions
1. Internal pressure is taken by shell ( E , $ Er);refractory does not contribute.
2. Unrestained shell (that is without fixed connection with refractory) tends to expand due to AT and AP.
3. Unrestrained refractory (that is without fixed connection with shell) tends to expand due to AT minus shrinkage.
4. Due to assumed fixed connection between shell and refractory, the dominant tendency to expand of either material
with respect to the other shall impose interactive forces of same magnitude and reverse sense in each material, for
example tension in shell and compression in refractory if unrestrained expansion of refractory exceeds unrestrained
shell expansion or vice versa.
Circumferential Long i t ud inil1

Shell
Refractor.

f
1
I I

fr,i =

(See figures above) Circumferential Longitudinal

Unrestrained shell&elongation : C E S C = Espc + (3,

-

Unrestrained refractory-elongation:
C E r c = ($)T - ()’ S

Nett differential expansion :

Total restrained shell elogation:
Total restrained refractory elongation :

With following stress-strain relations :

Dsrc Dsrl Dsrl *src
Esrc = -- Ps - Esrl = -- P s -
Es E S Es ES

with
Fsrc
Gsrc =-
21,1

with

Note: 0 positive (tension) c* E positive

D negative (compression)c-,I: negative
Part E : Journal of Process Mechanical Engineering 0 IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS O F VESSEL AND TRANSFER LINES 65

(b) Strain compatibility

CE,*,= CE:, and X E ; =I&$
Hence,
A&, = C E S , - CE,, = E,,, - E,,, and A&,= X E , 1 - C&,1= &,,I - &,,I

Hence, in circumferential direction for refractory,

= AE,

(c) Equilibrium of forces

F,,, = - P ISC and Fsrl = - FIS1

Hence,
E , I, + E , 1, Ds Es 1s + ~s Dr Er Ir
Frsc( ) “”(
2E, 1, E , I, 1 -
PI

nD,E , I , D, E , I,
= A&,

21,l
with F,,, = orsc and Frsl= ors,nDrI, . Hence,

Similarly in longitudinal direction for refractory,

(d) Remarks
The terms containing p, and ,us account for the Poisson effect in case longitudinal as well as circumferential
constraints exist.
Formulae as stated in Section 5 do not account for this Poisson effect and therefore correspond to circumferential
refractory stress in the case of free longitudinal expansion or, respectively, longitudinal refractory stress in the case
of free circumferential expansion.
Using formulae as stated in Section 5, the case of existing longitudinal as well as circumferential constraints leads
to slightly underestimating the stiffnessand hence results in slightly lower stress values.

APPENDIX 5
List of various ASTM Standards
A580 Stainless and heat-resisting steel wire
C92 Sieve analysis of refractory materials
C113 Reheat change of refractory brick (PLC)
c133 Cold crushing strength (2 in or 50 mm cube specimens) and modulus of rupture (9 in x 43 in x 3 in or
230 mm x 115 mm x 50 mm) of refractory brick and shapes
C134 Size and bulk density of refractory brick
c201 Thermal conductivity of refractories
C704 Abrasion resistance of refractory material at room temperature
C860 Determining consistency of refractory concretes
C862 Preparing refractory concrete specimens by casting
C865 Firing refractory concrete specimens

APPENDIX 6
Listing of computer program used for refractory calcudations
1OOO:‘S’; REM REFRACTORY STRESS CPL CS. 1010:PAUSE “SHELL PROPERTIES.”
1002:REM INPUT REQUIRED FOR SHELL, 1020:INPUT “1NT.DIA. ? ”; 0
REFRACT. AND GENERAL PROPERTTES. 1030:INPUT “WALL TH. ? ”; TS
1004:REM CONSISTENT UNITS ARE REQUIRED. 1WO:INPUT “M.O.E. ? ”; ES
1006:REM OUTPUT IS SELF EXPLANATORY. 1050:INPUT “R.T.E. ? ”; CS
1008:USING ‘‘+****.*” 1060:INPUT “T.C. ? ”; LS
@ IMechE 1990 Proc Instn Mech Engrs Vol 204
66 P K RICHARDS

1070:INPUT “P.R. ? ”; US 1365:PRINT “MEAN SHELL TEMP.: ”; MS

1080:PAUSE “REFRACTORY PROPERTIES.” 1370:FP= P*D/2/TS
1000:INPUT “REFR. TH. ? ”; TR 1380:PRINT “C.PR.STRESS : ”; F P
1100:INPUT “M.O.E. ? ”; ER 1390:XP=FP/ES*(l -US/2)
1120:INPUT “R.T.E. ? ”; CR +
1400:XL = FP/ES*(-US . 5)
1130:INPUT “T.C. ? ”; LR 1410:INPUT “CIRC.STRESSES 7 ”; 0%
1140:INPUT “P.R. ? ”; UR 1420:IF Q$=“WTHEN LET X P = x L
1150:INPUT “COLD SHRK. ? ”; SS 1430:XT =CS*(MS -TA)
1160:INPUT “HOT SHRK. ? ”; SR +
1440:EX = XP XT
1170:PAUSE “GENERAL PROPERTIES. ” 1450:PRINT “PR.STRAIN : ”; XP*JE6
1180:INPUT “T. AMBIENT ? ”; TA 1460:PRINT “TH.EXP.SHELL : ”; XT*JE6
1190:INPUT “T.OPER. ? ”; I0 1470:PRINT “TOT.SH.EL. ”;EX*JE6
1200:INPUT “0PER.PRESS. ? ”; P 1480:YT= CR*(MR-TA)
1210:INPUT “INLFILM COEFF .? ”HI 1490:YM =(SS + SR)/2
1220:INPUT “EXT.FILM COEFF. ? ”; HE 1500:EY=YT-YM: E=EX-EY
1230:REM CALCULATIONS. 1510:PRINT “MEAN REFR.EXP.: ”; YT*JE6
1235:INPUT “FLAT WALL ? ”; C$ 1520:PRINT “MEAN SHRINKAGE : ”; YM*IE 6
1237:IF C $ N T H E N GOTO 1242 1530:PRINT “NET MEAN UNRES. EXP.: ”; EY*lE6
1240:KC= l/(l/HI+TR/LR + IS/LS+ J/HE) 1540:PRINT “NET DIFF.EXP.: ”; E*IE6
1241 :GOTO 1245 1550:FC =ER*E*(ES*TS/(ER*TR+ES*TS))
1242:OD = D + 2*TS: ID = D - 2*IR 1560:FT = ER*CR*(T)-T2)/2/(1- UR)
1243:KC= l/(OD/HI/ID +OD/2/LR*LN (D/I 0)$ 1570:FS=ER*(SR-SS)/2/(1-UR)
OD/2/LS*LN (OD/D)+ 1/HE) 1580:HF = FC-FT +FS: CF = FC +FT - FS
1245:0 =(TO-TA)*KC 1590:HS= -FC*TRflS
1250:T3=TA+Q/HE 1600:PRINT “MEAN RFRSTR.: ”; FC
1290:T2=T3 +Q*TS/LS 1610:PRINT “DIFF.STR.TEMP.: ”; FT
1300:TJ =T2+Q*TR/LR 1620:PRINT “DIFF.STR.SHRK.: ”; FS
1310:PRINT “HEAT LOSS : ”; 0 1630:PRINT “RFR.STR. H.F.: ”; HF
1330:PRINT “HOT FACE TEMP.: ”; TI 1640:PRINT “RFR.STR. C.F.: ”; CF
1340:MS=(T2+T3)/2 1650:PRINT “SH.-RFR. STR.: ”; HS
1350:MR = (T2+ T1)/2 1660:END
1360:PRINT “MEAN REFR. TEMP.: ”; MR

APPENDIX 7
Listing of computer program used for sensitivity calculations
2000: “H”:REM INFLUENCES ON HEAT LOSSES. 2200: LPRINT “ VARIATION IN IN T.COEFF.”
2001: DIM HT(10):DIM CT(10):DIM MT(1O) 2210:LF 1
2005 :USING 2220:B$ =“HI”:COSUB “J”
2010:INPUT “ITEM ? ”; I$ 2235:LF 1
2020:INPUT “GAS TEMP. ? ”; T6 2240:USING “+ ****.*”
2030:INPUT “AMB TEMP. ? ”; TA 2250:FOR I = -5T0 5STEP 1
2040:INPUT “SHELL TH. ? ”; TS 2260:HI =IH*(1/10+ 1)
2050:lNPUT “REFR. TH. ‘? ”; TR 2270:HO=OH:KR=RK
2060:INPUT “REFR. ID. ? ”; ID 228O:COSUB “Q”
2070:INPUT “COND. STEEL ? ”; KS 2290:COSUB “0”
2080:INPUT “COND. REFR. ? ”; RK 2300:HT(I+5)=TH
2090:INPUT “INT. COEFF. ? ”; IH 2302 :CT(1+ 5)=TC
2100:INPUT “EXT. COEFF. ? ”; OM 2304:MT(I+5)=TM
2105:CSIZE 1 2310:NEXT I
21 10:LPRINT “ITEM : ”; I$ 2315:COSUB “I”
2115:GRAPH :LINE (0, 5)-(80, 5), 0 : TEXT :LF ] 2317:LF 2
2117:CSIZE 1 2320:COSUB “C”
+
21 18 : D = 1D 2*TR*IE - 3 :OD = D + 2*TS*lE-3 2325:COSUB “I”
2120:LPRINT “GAS TEMP. : ”; TG 2330:LF 2
2130:LPRINT “AMB TEMP. : ”; TA 2335 :LPRINT ‘‘ VARIATION IN RE FR. COND.”
2140:LPRINT “SHELL OD. : ”; OD 2340:LF 1
2150:LPRINT “REFR. OD. : ”; D 2350:H$ =“KR:COSUB “J”
2160:LPRINT “REFR. ID. : ”; ID 2365:LF 1
2162:LPRINT “INT.COEFF, : ”; 1H 2370:FOR 1= -5T0 5STEP 1
2164:LPRINT “REFRCOND. : ”; RK 2380: KR = RK*( 1/10 + 1)
2166:LPRINT “EXT.COEFF. : ”; OM 2385:HO = 0H:HI = IH
2170:HI =IH:HO = 0H:KR =RK 2390:COSUB “Q’
2180:COSUB “ Q 2395 :COSUB “ 0 ’
2185:QO=Q: T1 =TM: T2=TC: T3=TH 2400: HT(1+ 5)= TH
2190:LPRINT “HEAT LOSS : ”; INT (QS) 2402: CT(1+ 5) = TC
2192:LPRINT “HOT FACE : ”; INT (TH)
2193:LPRINT “COLD FACE : ”; INT (TC)
+
2404 : MT(1 5) = TM
2410:NEXT I
2194:LPRINT “METAL TEMP: ”; INT (IM) 2415:COSUB “1”:LF 2
2195:CSIZE 1:LF 2 2420: COSUB “ G
Part E: Journal of Process Mechanical Engineering @ IMechE 1990
 A DESIGN GUIDE FOR THE REFRACTORY LININGS OF VESSEL AND TRANSFER LINES 67

2425:GOSUB “T”:LF 2 2610:LPRINT ‘‘ (‘X) (%) (%) (Yo) (%)”

2430: LPRINT I‘ VARIATION IN EX T. COEFF.” 2620 :RETURN
2435:LF 1 3000:“I”: GRAPH :RLINE -(216, 153), 0, 0, 0
2440:B$ =“HO:GOSUB “J” RLINE -(216, 125), 0, 0, B
2455:LF 1 3010:RLINE (43, 0)-(0, 153), 0
2460:FOR I = -5T0 S T E P 1 3012:RLINE (43, 0)-(0, -153), 0
2470:HO =OH*(1/10+ 1) 3014:RLINE (43, 0)-(0, 153), 0
2480: HI = IH :KR = RK 3016:RLINE (43, 0)-(0, - 153), 0
2485:GOSUB “ Q 3020:EXT :CSIZE 1
2490:GOSUB “0” 3030:RETURN
2492:HT(I+5)=TH 3100: “G”:REM GRAPHICS
+
2494 :CT(I 5 ) =TC 3110:GRAPH :Y =TG-TA:YI =Y/2 + 20:X = 30+
+
2496: MT(1 5) =TM +
IR/2 TS/2
2500:NEXT I 3120:RLINE (30, 0)-(0, -Yl), 0:RLTNE
2505:GOSUB “I”:LF 2 TR/2, 0)-(0, YI), 0
2510:GOSUB “ G 3130:RLINE (TS/2, 0)-(0, YI), 0
2515:GOSUB “I”: LF 2:CSIZE 2 3140:RLINE (0, 0)-(-X, YI-lo), 0
2520: END 3160: RETURN
2525:“Q:REM SUBROUTINE HEAT LOSS 3170:“ T :REM
2530:Q=(T6-TA)/(OD/HI/ID+ LN (D/TD)* +
3 180:X = X 30
+
OD/2/KR/LN (OD/D)*OD/2/KS IjHO: 3190:FOR I=OTO l0STEP 5
2531 :TM = TA + Q/HO 3200 :YO =TG/2 :YI = HT(1)/2 :Y2 = CT(I)/2:
2532:TC=TM +Q*OD*LN (OD/D)/2/KS Y3 = MT(I)/2:Y4 = TA/2
2533 :TH = TG - Q*OD/ID/HI 3210:RLINE (0, 0)-(25, 0)-(5, (-YO+Yl) )-(TR/2,
2535 :RETURN + + +
(- Y 1 Y2))-(TS/2, (- Y2 Y 3)) -(5, (- Y 3 Y4)), 0
2540:“O”: DQ=(Q-QO)/QO*lOO 3220:RLINE (0, 0)-(25, 0), 0
2542:DM = CTM - Tl)/T1*100 3230:RLINE (O,O)-(-X, (YO-Y4)), 0
2544: DC =(TC -T2)/T2*100 3240: NEXT I
2546:DH=(TH-T3)/13*100 3250:RLINE (0, 0)-(0, -(Y0+20)), 0
2550:LPRINT TAB 0; !*lo; TAB 7; DQ; TAB 3255:RLINE (0, 0)-(216, 0), 0
14; DH; TAB 21; DC; TAB 28; DM 3260:TEXT :CSIZE 1
2560:RETURN 3270:RETURN
2600: “J”:LPRINT “ ”:B$; “ : H TC IM”

@ IMechE 1990 Proc Instn Mech Engrs Vol 204