See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301198273

A critical analysis of anchor spacing in refractory lining design

Article · January 2016

CITATION READS

1 5,496

4 authors, including:

Celso Goulart Bruno Teider

São Paulo State University Alcoa Inc
9 PUBLICATIONS   56 CITATIONS    11 PUBLICATIONS   20 CITATIONS   

SEE PROFILE SEE PROFILE

Victor C Pandolfelli
Universidade Federal de São Carlos
650 PUBLICATIONS   7,876 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermomechanical Properties of Refractories of the system Al2O3-3Al2O3.2SiO2-ZrO2 View project

Development of Al2O3-MgO refractory castables with optimized drying behavior View project

All content following this page was uploaded by Victor C Pandolfelli on 10 March 2017.

The user has requested enhancement of the downloaded file.

 PaPer

A Critical Analysis of Anchor Spacing in

Refractory Lining Design

C. A. Goulart, M. A. L. Braulio, B. H. Teider, V. C. Pandolfelli

operation, considering its lack of adhesion

Many of the shortcomings attributed to the refractory lining mater- to the vessel shell, its thickness and its self-
ials may in fact be related to design issues, such as the anchoring supporting inability [2]. The anchoring sys-
one. Key aspects in the engineering of these systems, as the spacing tem must also maintain the lining integrity,
even when it is already cracked, in order to
and position of the anchors, are defined using empirical knowledge inhibit a total structural collapse [3]. There-
in the everyday practice of companies. This approach does not take fore, this issue is of utmost importance be-
into account factors such as thermal behaviour and strength of the cause failures commonly attributed to the
anchors, and their interactions with the lining, highlighting the need refractory component, can be in fact, caused
by shortcomings in the anchoring system.
for a more detailed and technical analysis of the subject. This study Although practical, the anchor space de-
proposes a modification in the calculation model of anchor spacing sign is usually based on rules of thumb or
suggested in the literature and compares the attained results with simplifications that have been developed
standard values presented by few companies and those used in an through trial and error, disregarding import-
ant variables. Despite the wide spread use
actual project of an alumina calciner, aiming to obtain a better mod- of these rules by many industries, various
eling reference and greater reliability in the design of future projects. project particularities are not considered,
which adds to the problems observed in
practice. Failure prevention and an increase
1 Introduction lined with refractories, in order to protect in the lining life are desirable and can be
In many industries, such as metallurgical, and insulate their metallic shells from me- achieved by using engineering principles
cement and glass ones, the processing ves- chanical, chemical and thermal stresses, to such as gravity loads, thermal stress and
sels operate at high temperatures and are control the operational temperatures and strain, and the mechanical and chemical
saving energy. The performance of the re- behaviour of materials at high temperatures
fractory materials directly impacts on safety, when calculating anchor spacing [4]. In this
C. A. Goulart, V. C. Pandolfelli energy consumption and quality of the pro- study, the fundamental features regarding
Federal University of São Carlos (UFSCar) cess. Failures in these materials incurs the anchoring systems are firstly presented. Fol-
Materials Engineering Department potential loss of millions of dollars due to lowing, mathematical models for spacing
São Carlos, SP, 13565-905 the time and labour involved in the cooling calculations are introduced and the values
Brazil and maintenance steps of these structures, obtained theoretically are compared with
besides the lower operational availabil- those proposed by some companies and
M. A. L. Braulio, B. H. Teider ity and consequent decrease in productiv- others used in an industrial project.
Alcoa Alumínio S.A. ity [1]. Thus, it is important for an efficient
Research, Development and Innovation and economically feasible operation that 1.1 Types of anchors
(R&D&I) Department no sudden halts occur due to unexpected Anchors are basically produced by ceramic
Poços de Caldas, MG, 37719-900 failures. or metallic materials and their selection is
Brazil When compared to bricks, the use of mono- based on the equipment operating tem-
lithics as linings presents various advan- perature range, the environmental condi-
Corresponding author: V. C. Pandolfelli tages, because it allows for a diversification tions, as well as the material’s behaviour
E-mail: vicpando@ufscar.br in the geometry of the vessels and in the before thermo-mechanical stresses [5, 6].
installation methods, as well as a reduction Generally metallic anchors are used based
Keywords: refractory lining, design, in the placing time and in the number of on their simpler installation and lower
anchoring system expansion joints, providing savings of time unitary prices, making their use more feas-
and energy. However, the ultimate success ible at wider temperature ranges (up to
Received: 27.10.2015 in the application of monolithic refractory ~1100 °C). The higher costs and the dif-
Accepted: 09.12.2015 intimately depends on the anchoring system ficulties associated with the installation of
selected to keep it in the right place during ceramic anchors restrict their application to

92 refractories WORLDFORUM 8 (2016) [1]

 PaPer

conditions where oxidation, corrosion and Tab. 1 Typical alloys used in metallic anchors and their maximum service temperature
higher temperatures are present. (adapted from [7])
The number and distribution of anchors Anchor Material Operating Temperature [°C]
throughout the vessel impart a direct ef-
Carbon steel 430
fect on the equipment lifetime. Insufficient
number leads to an overloaded system, in- 304 Stainless steel 760
creasing the chances of failure. Conversely, 316 Stainless steel 760
adding more anchors may result in a poor 309 Stainless steel 820
refractory consolidation during installation, 310 Stainless steel 930
as the voids among them are too close and
Inconel 600 1100
may not be completely filled in, resulting in
Incoloy DS 1200
defects, which affect the material’s mechan-  
ical properties [4]. Any problems that may
arise in the anchoring system, either by un-
suitable material selection, design and/or in-
stallation, lead to high costs with refractory
 
lining, maintenance and productivity loss.

1.1.1 Metallic anchors

In conditions where temperatures are lower
than 1100 °C, this type of anchor is usually
the selected one. Steel is the most common
material used and the higher the working
temperature the more particular is the al-
loy [6]. Tab. 1 lists some of the commercially
Fig. 1 Metallic anchor shape used in an alumina calciner refractory lining
available alloys and their maximum operat-
ing temperature.
Metallic anchors are welded directly into
the vessel shell and are covered by the
monolithic refractory so as the anchors’ tip
is embedded in the lining. Approximately
20–35 % of the lining thickness should
cover the anchors’ tip. If the welding is
poorly carried out, it can lead to anchor
(and consequent refractory) failures, due
to uneven load distribution and insufficient
support [7]. Metallic anchors present a wide Fig. 2a–c Usual patterns of anchoring systems: (a) square, (b) diamond, and
range of available shapes and there is no (c) staggered (adapted from [5])
consensus yet on which one provides the
best performance. Many geometries are directions, hindering crack propagation [6]. conditions impose corrosive and/or oxi-
created to solve specific problems and end Anchor features such as shape, length and dizing environments. These anchors possess
up remaining in the market for general pur- angle of legs, and corrugation are not nor- greater holding capacity when compared
poses [2, 3]. In this field, it is usually said malized and their selection is usually car- to metallic ones due to their larger contact
that a successful man, besides planting a ried out based on previous experiences. It is interface and cross-section area, which
tree and writing a book, should also create equally important to keep a regular pattern lowers the overall stress in each anchor [8].
at least one anchor design in his lifetime! in anchor spacing in order to provide an Fig. 3 a presents the characteristic shape of
An example of a metallic anchor shape is evenly load distribution among them. Some ceramic anchors, which are mainly used to
shown in Fig. 1. typical patterns that are commonly used are hold plastic refractory, but can also be ap-
Anchors with larger cross-section areas pro- illustrated in Fig. 2a–c [5]. plied for other types of refractories (such
4   as
 
vide greater holding capacity. Flexible an- castables) [2]. The highest operating condi-
chors, such as a spring-type, unlike rigid and 1.1.2 Ceramic anchors tion for ceramic anchors is around 1700 °C,
straight ones, seem to help reducing cracks Ceramic anchors are refractory blocks pre- depending on the raw materials used in
in the lining [3]. Besides that, there is also senting various grooves along its length their formulations. When possible, it is de-
evidence that anchors featuring legs with which are used to withstand specific cast- sirable that ceramic anchors present similar
different lengths inhibit the development able areas where operating temperatures properties to those of the lining refractory
of preferential shear planes in the same are higher than 1100 °C or when working in order to minimize the thermal expansion

refractories WORLDFORUM 8 (2016) [1] 93

 PaPer
 

considering possible external vibrations

or pressure variations, which can lead the
panel to rotate around a fulcrum. A sketch
of a panel in a vertical wall and the loads
acting on it are presented in Fig. 4, where h
is the height of the panel, t is its thickness,
W the uniform load due to the weight of the
lining (gravity load), R an external load and
  ƒ the lining fulcrum. The horizontal load (R)
is set as equivalent to 30 % of W. Thus, for
1
MR = h R (2) the panel to remain in place without an an-
2
Fig. 3a–bM WCeramic
≥ MR anchor coupled to a C-shaped clip welded
(3) to an external beam (a),
choring system, the moment of the gravity
and diagram illustrating the installation of a ceramic anchor with a U-shape clip directly load (MW ) must be greater than that of the
1
welded to 1the vessel shell (b) external one (MR ), so that the lining does
⌅ t ⌅W ⇤ ⌅ h ⌅ 0.3⌅W ⇥ h 3.3t (4)
2 2 not collapse, as shown in Eq. 4. Thereby, it
for walls (Fig. 3b). Such procedures in- is possible to correlate height and thickness
crease the installation times and associated of the panel, where the former represents
costs [8]. the spacing in which the anchors must be
arranged to sustain the wall.
1.2 Anchor spacing
 
Their spacing is mostly  
determined based 1
(1)
MW = t W (1)
  2
on previous experiences as mentioned be-
fore, which leads to the use of empirical for- 1 R=1 h R (2) (2)
M R = Mh R
12 (2
mulas and rules of thumb. Although these 2M R = 2 h R (2) 6  
  MW ≥ M R (3)
practices are largely used in industries, M W ≥MMW R≥ M R (3) (3) (3
they might ignore specific issues related to 11 1
⌅ t ⌅W⇤⇤1 ⌅ h⌅ ⌅h0.3⌅W
each process and material [4]. A common 1 1 22 ⌅ 0.3⌅W
22 ⌅ t ⌅W ⇥ h⇥ 3.3t
h 3.3t (4) (4) (4)
Fig. 4 Diagram illustrating an external and widespread procedure is to establish ⌅ t ⌅W ⇤ ⌅ h ⌅ 0.3⌅W ⇥ h 3.3t (4
load (R), the weight (W) of the refractory 2 2
the anchor spacing as a multiple of the
lining, and the fulcrum (ƒ) used in the
total refractory lining thickness, regardless 1.2.1 General recommendations
calculation of moment assuring lining
of the materials and of the working condi- According to the company Shinagawa
stability (adapted from [2])
tions. This procedure can be mathematically [5], the spacing for monolithic refractories
coefficient mismatch [9]. Unlike metallic stated [2], according to Eq. 1–4. Given a re- should be determined depending on the
anchors, the ceramic ones are not directly fractory panel located on the wall of a ves- place of installation, type of anchor being
connected to the vessel shell. Instead, they sel, there can be two main loads acting on used and the lining thickness. Tab. 2 reports
are hold by metallic hangers, which in turn it: one due to the panel own weight in the the suggested values based on the practical
are welded to external beams, in the case vertical direction (from its center of grav- experience of the company.
of roofs (Fig. 3a), or directly to the shell ity), and another, perpendicular to the wall, Similarly to Shinagawa, Thermal Ceramics
company [8] also suggests the use of de-
1
MW = t W (1) fault values, as shown in Tab. 3. It is notice-
Tab. 2 Shinagawa’s suggestion
2 for anchor spacing, depending on the lining thickness
able the difference between the spacing
(adapted from [5])
for metallic and ceramic anchors. Whereas
Lining Thickness Anchor Types 6   Shinagawa’s recommendations (Tab. 2) are
Location
  [mm] Metallic [mm] Ceramic [mm] very similar regardless of anchor type and
50–100 – – position, Thermal Ceramics’ ones take into
100–200 230 –
account the higher holding capacity of cer-
amic anchors.
Walls, slopes and cylinders 200–300 380 380
300–400 460 450 1.2.2 Plibrico model
400+ 600 600 Plibrico company [2] stated that the anchor
100–200 300 300 spacing should be based on the features
Roofs and bullnoses
200+ 300 300 of each particular project such as the me-
50–100 230 – chanical properties of the anchor and the
refractory lining as a function of the tem-
Floors 125–230 380 –
perature. Considering these properties a
230+ 500 – mathematical model for the anchor spacing

94 refractories WORLDFORUM 8 (2016) [1]

 PaPer

is proposed, where two methods are adopt- Tab. 3 Thermal Ceramics’ suggestion for anchor spacing, depending on the lining thick-
ed for spacing calculations: (i) based on the ness (adapted from [8])
relationship between anchor strength at Location Lining Thickness Anchor Types
operating temperature and the gravity load [mm]
Metallic [mm] Ceramic [mm]
due to the lining weight, where the thicker
Walls and slopes 50–100 150 –
the lining the greater the weight to be sus-
100–200 230 –
tained and the lower the spacing values; (ii)
based on the refractory strength at operat- 200–300 300 460
ing temperature and its own weight, so the 300–400 – 460
thicker the lining the greater the strength 400+ – 610
and the larger the spacing required. The Roofs and bullnoses 100–200 180 300
lower value obtained considering the two
200+ 250 300
approaches should be adopted for spacing,
Floors 50–100 300 –
as it will be shown in Section 3. Plibrico as-  
sumes L-shaped anchors in the model, as 100–230 460 –
 
this geometry offers the least support pos- 230+ 610 –
sible. It also considers that the loads on  
 
the anchors act entirely in the vertical axis,
ignoring the division of loads in different Weight
directions as in the case of V- or Y-shaped
anchors, which implies in an additional W
safety factor.
For the spacing calculation considering
the anchor strength, it is assumed that the
anchors are equally positioned in all direc- (a) (b)
tions. The gravity load is calculated by using Weight
the dimensions and density of the refractory
M
lining, and the anchor cross-section area.
Then, the tensile strength of the anchor
at the operating temperature is compared
with the calculated stress. Obviously, if the
anchor strength is greater than the applied
(c) (d)  
load there is no failure. Fig. 5a–b illustrate
the variables used in the calculations and  
Fig. 5a–d Lining dimensions and the weight load acting on the system (a), schematic of
the representation of the load due to the the weight
Fig.  5  load acting on one anchor (b), multi-layer lining system (c), and bending
lining weight acting on the anchors. moment   diagram that arises in the lining under the action of its own weight (d)
The load on a single anchor (W ) is given by
 
Eq. 5, where la is the spacing between an- tions. However, from the refractory perspec- the horizontal ones, which are considered
chors, t1 and p1 are the refractory thickness tive, the   action of its own weight generates as 30 % of the weight load, as shown in
and density respectively, and g the gravi- a bending   moment, as it is being hold in Fig. 4 (20 % associated with loads due to
tational acceleration. In order to keep the a fixed   position due to anchoring system an earthquake, according to the Japanese
whole system in place, the anchor tensile (Fig. 5d).   The load in this case is obtained standard for buildings, and 10 %, arbitrar-
strength (σa) should be greater or at least by Eq. 9, where lr is the spacing between ily, to vibrations and thermal expansion).
 
equal to the stress in the system, as indicat- anchors. The bending moment (M ) and the Thus, if the anchors used in the walls.. are
ed by Eq. 6, where A is the cross-sectional modulus   of section lr · t1 (Z ) are given by the same as those for the roof, the spacing.
W = la2 t1 1 g . (5)
area of the anchor. The maximum spacing is Eq. 10–11,
  respectively. For the integrity of isWgiven
= la2 t1by 1Eq.
g 15–16. (5)
then given by Eq. 7 by replacing the variables the system,
  the refractory bending strength W = la2 t1 1 g (5)
of Eq. 1 in Eq. 6 and rearranging them. For (σr) should be greater or at least equal to W =w
la2 t1 1 g (5) (6)
(
  w
the multi-layer lining (Fig. 5c) the spacing is the stress resulted by the bending moment,
a
a
A (6)
Aw
 
calculated similarly to the single-layer case, as indicated in Eq. 12. Thus, the maximum a (6) (6)
Aw
but considering that the load acting on the spacing   allowed is specified by Eq. 13, and a ⇥⇥ a (
la ≤a Aa ⇥⇥ (7)
anchor is given by the sum of the weights in the case of multilayer lining by Eq. 14. la ≤ t1 ⇥ ρ1 ⇥a g (7) (7)
t1 ⇥ ρ1 ⇥ g
of all layers, resulting in Eq. 8. In the case Regarding the walls, due to the anchor hori- a ⇥⇥ a
la ≤ (7)
where spacing is calculated based on the zontal disposition and lack of mobility in t1 ⇥aρ⇥1⇥⇥ g
a ⇤⇥ aa (
refractory strength, it is assumed again that the vertical axis, it is assumed that the only lala⇥≤ a ⇤⇥ (8) (8)
la ⇥ Σ t(1ti⇥⇤ρρ1ia)⇥ ⇤gg (8)
the anchors are equally spaced in all direc- loads able to generate any actual stress are Σ ( ti ⇤ ρ i ) ⇤ g
a ⇤⇥ a 10  
la ⇥ (8)
  Σ ( ti ⇤ ρ i ) ⇤ g 10  
refractories WORLDFORUM 8 (2016) [1]   a ⇤⇥ a 95
la ⇥ (
Σ ( ti ⇤ ρ i ) ⇤ g 10  
 

 
  

PaPer
aa ⋅⋅σσa ⎧⎧ ⎛⎛αα−−15
15⎞⎫⎞⎫
    ll aa ≤
≤ aρ⋅ σa ⋅⋅11.8
.8⋅ ⋅⎨⎨11÷÷⎜⎧⎜60 −⎛15
α⎟⎬−
⎟⎬15 ⎞⎫ (17)
(17)
l a ≤ ΣΣ((ttii ⋅a
⋅ ρii ))⋅⋅agg
σ a) ⋅ g ⋅ 1.⎩8 ⋅ ⎨
⎩ ⎝⎝160 − 15⎠⎭⎠⎭ ⎟⎬
⎧ ÷ ⎜⎝⎛ 60
(17)
( t ⋅ρ α −−15 ⎫
15 ⎠⎞⎟⎭
la ≤ Σ i ⋅ i ⋅ 1.8 ⋅ ⎩ ⎨1 ÷ ⎜ ⎬ (17) (17)
  Σ(t i ⋅ ρ i ) ⋅ g ⎩ ⎝ 60 − 15 ⎠⎭
 
    σσ r ⋅⋅tt122 ⎧ ⎛ α − 15 ⎞⎫
W==lr2lr2 t1t1
W gg ≤ 3(9)
ll rr ≤ (9) r 1 ⋅⋅11.8.8⋅ ⋅⎨⎧11÷÷⎜⎛⎜ α − 15⎟⎬⎞⎟⎫ (18)
(18)
Σ((ttσ i ) ⋅2g
1 1
3 ⋅⋅ Σ i ⋅⋅ ρ
i rρ⋅i t)1⋅ g
⎩⎨ 60 − 15
⎩ ⎝⎧⎝ 60 −
⎬
α⎠⎭⎠−⎭15 ⎞⎫
⎛ 15 (18)
lr ≤ t12 ) ⋅ g ⋅ 1.8 ⋅ ⎨
⎧1 ÷ ⎜⎝⎛ 60 ⎟⎬ ⎫ (18)
⋅ Σσ(t ir ⋅⋅ ρ
3 (9) α −−15
15 ⎠⎞⎟⎭
W = l2 t
W = lr2 t1 1r g1 1 g l r ≤ (9) i ⋅ 1.8 ⋅ ⎩
⎨1 ÷ ⎜ ⎬ (18)
11
M== lrlr WW
3 ⋅ Σ (t
(10)(10)i i ⋅ ρ ) ⋅ g ⎩ ⎝ 60 − 15 ⎠⎭
W = lr2 t1W = 2
1 lg
r t1 g M (9) (9) aa ⋅⋅σσa ⎧
⎪⎧ ⎡⎡ ⎛ α − 60 ⎞⎤ ⎫ ⎪ ⎫
1 22
ll aa ≤ 1 + (1.8 − 1) ⋅ ⎛ α − 60 ⎞⎤ ⎪
⋅⋅⎨⎪ (19)
(19)
1
≤ Σ
a
Σ((ttii ⋅⋅ ρρii ))⋅⋅ gg ⎪
⎨1 +⎢⎣⎢(1.8 − 1) ⎜⎝⋅ ⎜90 − 60 ⎟⎠⎥⎦⎟⎬
⎩⎪ ⎝ 90 − 60 ⎠⎪
⎬
⎭⎥⎦ ⎪
(19)
1 M= l W (10) ⎩ ⎣ ⎭
M = lr W2 r (10) a ⋅ σ a ⎧
⎪ ⎡ ⎛ α − 60 ⎞⎤ ⎫ ⎪
la ≤ ⋅ ⎨ 1 + (1.8 − 1) ⋅ ⎜ ⎟⎥ ⎬ (19)
1
lrMW= 1 lr W 12 (10) (10) Σ(t a ⋅σ
i ⋅ ρ ia) ⋅ g
⎪ ⎢⎣⎡
⎧ α −− 60
⎛ 90 ⎫
60 ⎠⎞⎟⎦⎤ ⎪
⎪
M= ZZ== 1 lrlr t12t12 l a ≤ (11) ⎪
⋅ ⎩ 1 + (1.8 − 1 ) ⋅ ⎝
⎜ ⎥ ⎭ (19)
Σ(11)
2 ⎨ ⎢ ⎬
2 66 (σt i r ⋅⋅ρt12i2 ) ⋅ g ⎧ ⎪ ⎡ ⎣
⎪⎧⎩ ⎛ α − ⎝ 90 ⎪60
60 ⎞−⎤ ⎫ ⎠
⎫ ⎭ ⎦ ⎪
σ ⋅t ⎪1 + ⎢⎡(1.8 − 1) ⋅ ⎜ ⎛ α − 60⎟⎥⎞⎬⎤ ⎪ (20)
1 ll rr ≤
≤ 3 ⋅ Σ(t r ⋅ ρ1 ) ⋅ g ⋅⋅⎨ ⎨1 +⎣⎢(1.8 − 1)⎝⋅ ⎜90 − 60 ⎠⎦⎟⎪⎥ ⎬(20) (20)
1 Z = l t2 3(11)
⋅ Σ(ti ⋅ ρi ) ⋅ g ⎪⎩⎪
⎩ ⎣ ⎝ 90 − 60 ⎠⎭⎦ ⎪
Z = lr t162 r 1 (11) i i ⎭
1 2 1 6M 2
Z = lZ t (11) σ ⋅ t ⎧
⎪ ⎡ ⎛ α − 60 ⎞⎤ ⎫ ⎪
6
Fig. 6a–b Diagram
r = 1 lr t1
for6 upward
2
inclined lining (a),rand
M (11) l r ≤ (12)(12) r
σ ⋅t2
1
⋅ ⎨1 + (1.8 − 1) ⋅ ⎜ ⎟⎥ ⎬ (20)
Zfor downward inclined lining (b) l r ≤ 3 ⋅ Σ(t ri ⋅ ρ1i ) ⋅ g ⋅ ⎪
⎪ ⎢⎣⎡
⎧ α −− 60
⎛ 90 ⎫
60 ⎠⎞⎟⎦⎤ ⎪
⎪ (20)
⎨1 + ⎢(1.8 − 1) ⋅ ⎝⎜
r
Z ⎩ ⎥ ⎭
(adapted from [2]) Mr
M 2 Materials3 ⋅ Σ(t i ⋅ ρ iand
(12)
) ⋅ g techniques
⎪
⎩ ⎣ ⎝ 90 − 60 ⎠⎦ ⎪
⎬
⎭
r Z (12)
M Z
M a ⋅σ a 115 3⎞⎫ (12)
⇥ r ⌅2.1 Proposed calculations
⎧ ⎛ α − 1⌅ lr ⎟3⌅⎬t1 ⌅ ρ1 ⌅ g
r
Z r la ≤ ⋅ 1.8 ⋅ ⎨1 ÷ ⎜ (12)
3⌅ ρ ⌅(17)
l2 ⌅ g t
Z Σ(t i ⋅ ρ i ) ⋅ g ⎩ ⇥ r⎝ 60 ⇤ 2−215⌅ l⎠r⎭⌅ t1 ⌅ ρ1 ⌅ g⇥ ⇥ r ⇤ 3⌅1 ρ1r ⌅ lr2 ⌅⇥ g lr ≤ ⇥1 ⌅ t (13)
⇥ r ⇤ 1 1 ⌅ l1⌅ t 23 ⇥ ⇥ r ⇤ t1 ⇥ lr ≤3⌅ ρ1 ⌅ rg 1 (13)
1 ⌅l1r 2⌅ t1 ⌅ ρ1 ⌅ g 3⌅ tρ1 ⌅l 2 ⌅ g ⇥ rρ⌅1t1⌅ g Thermal stresses
3⌅
⇥ lr ≤ ⇥ r ⌅2.1.1
l 3 ⌅ ⌅tr21lr⌅ ρ
⌅6 ⌅g
1 3 2⇥ r r6⇤ ⌅ t11 ⇥ ⇥3⌅r ⇤ρ1 ⌅ lr2 1⌅ gr t1 (13)
⌅ lr ⌅ t1 ⌅1ρ1 ⌅3g 3⌅ ρ 2⌅ l 2 ⌅ g
⇥ r ⇤
⇥ 1 ⌅ t
1
⌅l ⌅
⇥
t 2 ⇥ r ⇤ t1 ⇥ lr ≤ 3⌅ ρ1 ⌅ g(13)
t1 3⌅ ρ1 ⌅ g
⇥r ⇤ 2 ⌅ lr l⌅rt⇥ ⌅ ρ ⌅ gσ r ⋅ t1 r ⎧ 2 ⎛ α − r15⌅ l1⎞⎫⌅ t (13) 2 r 1
⇥
1≤ 1r ⇤ ⇥
⋅ 1.ρ
3⌅ l ≤
8 ⋅ ⎨⌅r1l ÷ ⌅⎜g ⎟r⎬ 61 ⇥ ⌅ t (18) (13) Thermal stress calculations for refractory lin- 12  12  
⇥1r ⌅⇤lr ⌅2t12 3 ⋅ Σ(t⇥i ⋅ρ⇥ti 1r) ⋅⇤g 1 r ⎝ 60
⎩ 3⌅
⇥ −ρ615
l ⌅⎠≤ g⎭ r 1
⇥1rr ⇤ t12 3⌅ ρ1 ⌅ g    
(13)
6 1 tl1 ⇥ ing systems containing anchors are difficult
⌅ lr ⌅ t12 ⇥ ⇤t 2 (14)
lr ⇥ 3⇤ Σ (ti ⇤r ρi1) ⇤ g 2
r
6 and complex (14) as they involve those arising
⇤i t⇤12ρ⎞⎤i )⎫⎪r ⇤ g1
l⎛r α⇥Σ⇥−(rt60
3⇤ ⇥ ⇤t
a ⋅σ a ⎧
⎪ ⎡ (14)
la ≤ ⋅ ⎨1 + ⎢(1l.r8⇥− 1) ⋅ ⎜ 3⇤ Σ⎟(⎥t⎬⎪i ⇤ ρi ) ⇤ g (19) (14) due (14) to the thermal expansion coefficient 12  
lr ⇥
⇥ r ⇤ t12 Σ(t i ⋅ ρ i ) ⋅ g ⎪ ⎩ ⎣ Σ (−ti 60
3⇤⎝ 90 )⎦⇤⎭g  
⇤ ρi⎠(14) 12  
(
3⇤lΣ ⇥ ti ⇤ ρi ) ⇤ g⇥ r ⇤ t1
2
F 0.3w a ⇧⇥ a mismatch among the anchor, the refractory
⇥a ⌅ ⇤⇥a ⌅ ⇤ la ⇥   (14) ⇧1.8 (15)
3⇤ Σ (ti ⇤ ρi ) ⇤ g Σ (ti ⇧aρ⇧i ⇥ ) ⇧ag ⇧1.8 and the(15) vessel shell, the temperature grad-
r
aF 0.3w
a
⇥ a ⌅ ⇤ ⇥Fa ⌅ ⇤ la ⇥ ⇥
⎪⇤ l ⇥Σa(⇧t⇥ i ⇧ aρi ) ⇧(20)
2
⎪ Fa ⇥
0.3w
⎛ αa− 60 ⎞⎤ ⎫
a ⇧ ag
σ r ⋅ t1 ⎧
a .8 − 1) ⋅ ⎜⇥ a ⌅
⎡ ⌅ ⇤0.3w
( ⇧1.8 (15) ient inherent
(15) to the process, and the differ-
⎦ ⎭ Σ ( t ⇧Σρ( t)i ⇧⇧gρi ) ⇧ g
lr ≤ 1 1
(t ia ⋅ ρ i ) ⋅ ⇥g a ⌅⎪⎩ a ⎣⇤ ⇥ aa ⌅ ⎝ 90 ⇤al⎠a ⎪⇥ ⇧1.8 (15)
⋅ ⎨ + ⎢ ⎟ ⎥ ⎬ a
3a
F 0.3w ⋅ Σ⇧ ⇥
a − 60 ence between the thermal conductivity of
⇥a ⌅ ⇤⇥a ⌅ ⇤ la ⇥ ⇧1.8 (15) 1
i i
a F a 0.3w Σ (ti ⇧ ρi ) ⇧ g a ⇧ ⇥ a ⇧ l ⇧ 0.3⇧ w ⇥ r ⇧ t12
⇥a ⌅ ⇤⇥a ⌅ ⇤ la ⇥ ⇧1.8 M 2 r (15)⇤ l ⇥ metallic
⇧1.8 (16) and ceramic materials [3]. Thus,
a a Σ (ti ⇧ ρi ) ⇧ g⇥ r ⌅ Z ⇤ ⇥ r ⌅ 11⇧ lr1⇧ 0.3⇧ w r 3⇧ Σ (t i⇥⇧ρri ⇧)t⇧122g the use of computational analysis becomes
M
⇤M⇥ r ⌅12⇧6lr⇧ ⇧l2r0.3⇧ ⇧⇧lt1r ⇧ 0.3⇧ ⇤
2
⇥r ⌅ w l ⇥ ⇥ ⇧t ⇧1.8 (16)
w
⇥ r ⌅1 ⇧ l ⇧ t 2 ⇤ lr ⇥ 3⇧⇥Σr ⇧(tt1r2i ⇧ρ1 i )relevant
r
1 M⇥Zr ⌅ ⇤ 2 (16)to define the design parameters of
⇧ g⇧1.8(16)
Fig. 7 DifferentMspacing calculations ⇧ lr ⇧ 0.3⇧ w ⇥ ⌅
⇥ r ⇧ t1 Z
r 2 ⇤ ⇥Z r ⌅
16
1
r 1
2⇧l ⇧ t
⇤2 lr ⇥ 3⇧ Σ ( t ⇧ ρ
3⇧ Σ ( t i ⇧ρii ) ⇧ ig
) ⇧⇧1.8
g
⇥r ⌅ ⇤ ⇥r ⌅ 2 1 ⇤ lr ⇥ ⇧1.8 (16) ⇧ lr ⇧6t1 r 1 anchoring systems and refractory structures.
⇧ l ⇧ t 2 ⇧ lr ⇧ 0.3⇧ w 3⇧ Σ ( t i ⇧ρi ) ⇧ g ⇥ r ⇧ t12
depending on theZ area 1 6
Mof horizontal
cylinders (adapted⇥ r from
⌅ ⇤[2])6⇥ r r⌅ 12 ⇤ lr ⇥ ⇧1.8 (16) (16) However, a simple mathematical model
Z 1
⇧l ⇧ t2 3⇧ Σ ( t i ⇧ρi ) ⇧ g
  6 r 1 [10], which is applied for ceramic matrix
12   composites to calculate the stresses gener-
  For the floors, as the lining is directly with- ated by mismatch in the thermal expansion
stood by the shell/ground, no model is pro- coefficients between the inclusion and the
posed and it is assumed that the anchors matrix, was extrapolated and used in this
are not essential, except in specific situa- work to qualitatively analyse the effect of
tions where there are factors such as vibra- thermal stresses in anchor/refractory sys-
tion and thermal expansion of the lining. For tems. This evaluation also provides insights
  upward inclined linings the configuration about the benefits of anchor caps to reduce
  can be assumed to be in between a floor thermal stresses. The model is based on the
   
 
and a vertical wall (Fig. 6a) and anchor stress balance between a sphere of radius a
spacing is given by Eq. 17–18, which are (representing the inclusion) inside another
  Inclusion/matrix/ valid in the range from 15–60°. Linings with one of radius b (assumed as the matrix),
interface inclinations lower than 15° are regarded as and describes the stress at the inclusion-
  floors and above 60° as walls. The spacing matrix interface (σa ), considering both
 
Fig. 8 Concentric spherical model for downward inclined linings, in which the materials in perfect contact. It is considered
(adapted from [10]) configuration can be considered as in be- that σa acts on the interface between the
W = lr2 t1 1 g tween(9)a roof and a vertical wall (Fig. 6b), spheres in relation to the inner one and it
W = lr2 t1 g (9)
W =l 2
r t1
1

1 g is given by Eq.(9)
19–20, which are applied for can be tensile or compressive, depending
W
M== 1lr2 l t1W 1 g (9) inclinations
(10) greater
(9) than 60°. Below that on the relative thermal expansion coeffi-
r
1
2
M= lr W (<60°) (10)
the spacing is the same as for roofs. cients of the materials. If the inclusion has a
21
M = lr W (10) The spacing for(10)horizontal cylinders follows higher coefficient than the matrix the stress
12
Z=
M 1
= 2 llrr W
t12 the same (10)
(11) principles previously analysed and is tensile and is defined as negative, other-
1
6
Z= 2
lr t1 (11) (11) as shown in Fig. 7. For vertical
is calculated wise it is compressive and considered as
6
1 cylinders the(11)
spacing is obtained with the positive. The surface stress varies as func-
Z = lr t12
16M 2
Z=r l t (12) same(12)
equations
(11) used for vertical walls. tion of the distance (d ) from the inclusion/
6MZr 1
r (12)
Z
M − (α − α m )ΔT
σ96
1 3
⌅ l ⌅ t ⌅ ρ ⌅ gi
2 (h + a ) (12)
(21) refractories WORLDFORUM 8 (2016) [1]
α = M 2 ×
r
2aZ3 (1r − 21υ m 1) + b 3⇥
⇥r r ⇤21 (1 +⇥υrm⇤) 3⌅(1ρ−1 2⌅υlri )⌅ g 3⇥ lrh≤ ⇥ r ⌅ t1 (13) (12)
Z ⌅ l1r32⌅⌅Elt1 ⌅⌅tρ
23 ⌅ g 3
+ ρ ⌅
⇥r ⇤ 2 6 r 1
m b −a
1 (
⇥ ⇥r ⇤
) t
3⌅ ρEi11 ⌅ lr2 ⌅ g
⇥ lr ≤
3⌅ g
⇥1 r ⌅ t1
(13)
1 t1 3⌅ ρ1 ⌅ g
1 3⌅ lr ⌅ t1
2

⌅6lr ⌅ t1 ⌅ ρ1 ⌅ g
)ΔrT⇤ 3⌅ ρ1 ⌅ 2lr ⌅2g(h⇥ ⇥ ⌅t
2
1
⇥ r ⇤ 2⌅ lr3 ⌅ t1 ⌅2 ρ1−⌅(αgi −⇥
α m⇥ a 3 (b 3 − dr 3 ) 1
+ a )lr ≤ (13)
 PaPer
 

matrix interface and can be divided into two

components, one radial [(σa )p] and another
tangential [(σa )θ] (Fig. 8).
As the model was developed for spherical
geometries, a shape factor is applied in
order to extrapolate its use for a cylindric-
al one. This factor consists of an expression
obtained from the division of the cylinder
area/volume ratio by the sphere area/
volume one, where h stands for the height
and a the radius of the cylinder. The model
with the shape factor is illustrated by Eq. 21
and the radial and tangential components Fig. 9 Sketch of the dimensions used for the thermal stresses calculation
by Eq. 22–23. The inner sphere representing
  the inclusion is taken as the anchor, which Tab. 4 Average material properties and conditions used for the thermal stress analysis
 
  has properties such as elastic modulus Ei , Refractory Metallic Anchor Temperature Spacing
Poisson’s ratio υi and thermal expansion αr Er υr αa Ea υa a T l
coefficient �i , and the outer sphere, as the [10–6 °C–1] [GPa] [10–6 °C–1] [GPa] [mm] [°C] [mm]
matrix, represents the refractory lining, with 6,00 20 0,20 18,7 118 0,30 8 1000 300
properties Em, υm and �m, respectively. In
this study, the inner sphere radius (a ) was
set as the radius of a cylindrical anchor, and proposed in order to take into account the system. Any precautions adopted in real pro-
the outer sphere radius (b ) as half the spac- effect of thermal stresses and to improve jects such as anchor caps, expansion joints
ing between anchors (b = 1/2 · l), as illus- the accuracy of the theoretical values cal- and anchor clips, which provide anchor
trated in Fig. 9. The calculations were based culated for anchor spacing. The proposal mobility in relation to the shell, are ignored
on metallic anchors and using average val- consists in calculating these stresses as a and all the stresses are considered as if they
ues for the materials properties, which are function of the lining weight load, following were acting on the anchors as tensile loads.
listed in Tab. 4. the idea suggested in the original Plibrico Therefore, the worst case scenario is adopt-
model (Fig. 4) to determine the magnitude ed, working as an additional safety factor.
2.1.2 Modified Plibrico model of a horizontal one. The model defines this All modifications to the Plibrico origin-
proposed in this work load as representing those related to ex- al model follow the conditions defined
In the original Plibrico model, when it comes ternal vibration and thermal expansion, at- in Eq. 6 and Eq. 12. For roofs, the pro-
to roofs, the only loads considered are those tributing to each factor a percentage of the posed modifications are illustrated by
acting on the anchors due to gravity, which weight load proportional to its effects. Eq. 24–25, in which it is assumed that
is just valid at room temperature. Thermal Firstly, all potential sources of major thermal the overall load acting on the system com-
stresses that may develop during operation stresses were defined as follows: thermal prises the weight load and thermal stress:
due to thermal expansion coefficient mis- expansion mismatch among (i) anchors F = w + T = w + 0,3w = 1,3w, where F in-
match among the anchors, the lining, and and lining, (ii) refractory panels, and (iii) dicates the total load and the thermal stress
the shell, as well as the temperature gra- the shell, the lining and the anchors. Hence, is represented by T. Similarly, modifications
dients, are ignored because of the associ- each of the listed items is arbitrarily defined are proposed for vertical walls, where it is
ated difficulties for calculations. Therefore, as proportional to 10 % of the total weight also assumed the actual loads acting on
the Plibrico model underestimates the total load, similarly to the original model, consid- the system have been underestimated by
load acting on the system and changes are ering an extra load of 30 % acting on the the original Plibrico model. Therefore, the
total load acting on vertical walls is taken
15  
− (α i − α m )ΔT 2 (h + a )
σα = 3 − (α − α )ΔT × 2 (h   + a ) (21) as: F = E + T = 0,3w + 0,3w = 0,6w, where
σ α = 2a 3 (1 − 2υ m− )(α+ iib−33 (α1 +mm υ)ΔmT) (1 − 2υ i ) × 32 (h +h a ) (21)
σ α = 2a 3 (1 − 2υ m ) + b 3 (1 + υ m ) + (1 − 2υ i ) × 3 h (21) (21) E represents the extra load, resulting in the
(( ))
2a (1 2−E2υm mb)33+−ba 33(1 + υ m ) + (1 −Ei
2E m b 3 − a 3 +
2υ i ) 3 h
Ei
(
2E m b − a ) Ei Eq. 26–27.
For downward inclined linings, the same
(σ α )ρ =
− (α i − α m )ΔT
− (α i −3 α m )ΔT
((
2 (h + a ) a 3 b 3 − d 3
× 2 (h + a ) × a 333 b 333 − d 33
)) (22)
ideas for Eq. 15–16 were adopted and
(σ α )ρ
(σ α )ρ
= 2a 33 (1 − 2υ m− )(α
+ ib−3 (α1 +
= 2a 3 (1 − 2υ m ) + b 3 (1 +
)
υ
m mΔ T) (1 − 2υ ) × 2 (h +
3 h a) × da ba −− db
υ ) + (1 − 2υ ii ) × 3 h × d 33 a 33 − b 33
(( )))
(22)
(22) (22) Eq. 28–29 are defined. As previously men-
2a (1 2−E2υ
2E m
b
m 3((
) − a 3 ))
3+ b 3(1 + υ m ) + (1 −Ei
m
+ Ei i
2υ ) 3 h d a −b ( tioned, Plibrico does not suggest a model
m b −a
(
2E m b 3 − a 3 ) Ei
to calculate floor spacing. Yet, following
the same concept used for the other parts
− (α i − α m )ΔT ((
2 (h + a ) a 3 2d 3 + b 3 )
)))
(σ α )θ = − (α i −3 α m )ΔT × 2 (h + a ) × a 33 32d 33 + b 33 (23) (23) and assuming that any extra stress on the
(σ α )θ
(σ α )θ
= 2a 33 (1 − 2υ m− )(α
+ b− (α1 +
mυ
= 2a 3 (1 − 2υ m ) + ib 33 (1 +
)ΔT) (1 − 2υ ) 32 (h +h a ) × a2d 2da +− b
×
υ m ) + (1 − 2υ ii ) × 3 h × 2d 33 a 33 − b 33
( (( (23)
(23)
2a (1 2−E2υ ((
) ))
3+ b 3(1 + υ m ) + (1 −Ei2υ ) 3 h 2d a − b ( ) anchors can be described as a function of
mb 3 − a 3 m
2E m
m b −a
+ Ei i
(
2E m b 3 − a 3 ) Ei the weight load (F = 0,3w ), Eq. 30–31 are

refractories WORLDFORUM 8 (2016) [1] 14   97

14  
  14  
 
 
  
 
 
 
PaPer 1
⋅ 1.3 ⋅ l r3 ⋅ t1 ⋅ ρ1 ⋅ g
σ r ⋅ t12
⋅ 0, 88
M
  σr ≥ → σ r ≥ 12 ⋅ 1.3 ⋅ l r3 ⋅ t1 ⋅ ρ1 ⋅ g → l r ≤ (25)
M Z
→σ r ≥ 2
1
⋅1l r⋅ 1 ⋅ t.132 ⋅ l r3 ⋅ t1→
3 ⋅ σ
Σ (rt i ⋅ ρt12i ) ⋅ g ⋅ 0,88 (25)
  σr ≥ ⋅ ρl1r ⋅≤g 2
g r ⋅ t1
⋅ 0, 88
Z M 16 2 2 3 ⋅ Σ(t ⋅ρ ) ⋅ σ (25)
  σr ≥ → σ r ≥⋅ 1 l r ⋅⋅1t.13 ⋅ l 3 ⋅ t ⋅ ρ ⋅ g → l r ≤i i
Z 1 6 2 1 r 12 1 3 ⋅ Σσ(tri ⋅⋅ρt1i2 ) ⋅ g
⋅ 0, 88
M 3 ⋅ l ⋅ t (25)
  σ ≥ 1.3 l
→σ r ≥ r 1 6 1
⋅ ⋅ ⋅ t ⋅ ρ ⋅r g 1 → lr ≤ 2
M r
→ σ rZ≥ 2 1 1 σ r 3 ⋅ t
⋅ 0, 88
1 ⋅ Σ(t i ⋅ρ i ) ⋅ g (25) proposed, which are valid for inclinations
1, 3w⋅ 1.311,⋅3l r⋅ l a⋅ t⋅1t21⋅6ρ
3 2 l ≤
⋅ ρ⋅131l⋅r⋅gg⋅ t1 r
σr ≥ 2 →
FZ aσ(⋅tσ
0, 88
2
ri ⋅⋅aρt1i ) ⋅ ⋅g
σσaar ≥⇤≥⇥M ⋅ 0,288
  3⋅Σ
la ⇥ a→ →σ⇤1σa ,r≥29 ≥ 2 = ⋅1l r⋅ 1⋅ .t3 1 ⋅ l r ⋅ t1 → ⋅→ ρl1lra⋅ ≤g
≤ (25)
(24)
(26)
(24) up to 15°. In the range from 15–60°, the
a M a a2 Σ(⋅tΣ⋅(tρ ⋅ρ)⋅ g) ⋅σgr ⋅ t1
⋅ 0, 88
6
1
la ⇥ Σ (ati ⇤⇤⇥ρaiZ)σ⇤ gr ⇤1
≥, 29 → σ r ≥ ⋅21l r ⋅⋅1t.13 ⋅ l 3
⋅ t ⋅ ρ ⋅ g
3
→ l ri ≤ i i i (26) (25) equations must be adjusted to account for
⇤ ⇥ 1 3 ⋅ Σσ(t ri ⋅ρ )
Σ l(ti⇥⇤ ρi ) ⇤σgr ≥ →
a Z 6 2 t 2 ⋅g
⋅ 0, 88
M r 1 1
σ,r29
2 ⋅
a
1 ⇤1 ≥3 ⋅ l r ⋅ t1 → lr ≤
i
1 (25)
(26) effects of gravity and Eq. 32–33 are pro-
a
Σ (ta ⇤ ⇥ ρ Z) ⇤ g ⋅ 1.3 ⋅ l r ⋅ t1 ⋅61ρ1⋅ l⋅rg⋅ t12 σ r ⋅ t132 ⋅ Σ(t i ⋅ρ i ) ⋅ g
la ar⇥⇤≥⇥M →i σ ria≥ 2 ⇤1, 29
σ 6 l 0 , 88 (25) (26) (25) posed, whereas when the inclination in-
aΣ ( t ⇤ ρ ) ⇤ g
→ ≤ ⋅
2 i ⇤1, 29
r
la ⇥ Z 1 2 3 ⋅ Σ(t i ⋅ρ i ) ⋅ g (26)
a⇥⇤ ⇥⇤ t ⋅ l r ⋅ t1
lra ⇥ Σ (t⇥i ⇤ rρ⇤ ita)21⇤ g ⇤1,⇤129
i
, 29 6 (27)
(26) creases there is a proportional decrease in
(Σ ( rtρ⇤ρ1) ⇤ )g
a⇤ ⇤ ⇥
lr ⇥ 3⇤
Σ t ⇤
l ⇥i i i i g ⇤1,229 ⇤1, 29
a
(27) (26) the calculated spacing value, up to the point
3⇤laΣ⇥(t i ⇤ρΣi ()t⇤ai⇥g⇤⇤ρ r⇥⇤ t)1⇤ g
⇤1, 29
⇤ g⇤1, 29
lra ⇥ 3⇤ Σ⇥ t ⇤i⇤taρ (27)
(26) (26) that it is the same as the one of a vertical
⇥⇤ ⇥⇤atΣ2 (ti(⇤rρi i )1 i⇤)g ⇤1, 29
2 16  
lr a⇥ (27)
t , ⇤29
  wall. As for angles greater than or equal to
ρi ) ⇤⎧g
Σ (ti⇥⇤ ρ⇤i t)2⇤ ag ⋅(gσ ai⇤1, 29
lla ⇥⇥ r 3⇤ 1 Σ ⇤1 (26)
(27)
lr ⇥ 3⇤l aΣ≤(tr i ⇤Σρ1(ita) ⇤⇥
r ⎪ ⎡ ⎛ α − 60 ⎞⎤ ⎫⎪
⇤1
⇤ ,2g29⋅⎧⎪⎨⎪0,88 +⎡⎢1, 29 − 0,88 ⋅⎛⎜⎝ α90−−60 ⎟⎥ ⎬ (28)
(27) 60°, the spacing should be similar to that of
)
3⇤llaΣ⇥≤(t i ⇤ρi )i ⇤ gri 1 ⋅ ⎨⎩0,88
t 60⎞⎠⎤⎦⎫⎪⎪⎭
⋅ σ ρ ⋅
, ⎢1⎧⎪, 29 − 0,⎡88 ⋅ ⎜ (28)
a ⎣
a ⇤1 +29 ⎟⎥ ⎬
− 60 ⎞⎤ ⎫⎪ (27) (27) vertical walls.
lr ⇥ 2 l a ≤( ri Σ1(i t)ai ⋅ σρ ia⇤1
r Σ3⇤(t iΣ⋅⇥ ρti )⇤⋅ρtg ⇤⎩⎪g a ⎣⋅ 0,88 + 1, 29
2 ⋅ σ ⎝ 90 − 0−,8860⋅ ⎠⎦α
⎛
⎜ ⎭⎪ (28)
)⋅ ,g29⎧⎪⎨⎩ ⎢ ⎟⎥ ⎬
⎛⎝ 90 60 ⎞⎠⎤⎦⎫⎪⎭ (27)
⇥ r ⇤ t3⇤ l aΣ⋅≤σ (tai ⇤Σρ(it ) ⋅⎪⎧⇤ ρg )⋅ g ⎡⋅ ⎨0,88 + ⎢1, 29⎛ −α0−,8860⋅ ⎜⎞⎤90⎪⎫ − 60
⎡⎣ α −−60
⎟⎥ ⎬ (28) 2.2 Applying the proposed
lr ⇥ 1 a
⇤12, 29 (27)
(t ⇤Σρ(it)aiσ⇤⋅⋅rgρσ⋅iat)21⋅ g ⋅ ⎨⎪⎩⎧⋅0⎪⎧⎨,088,88+ +⎢⎣⎡1⎩⎡⎢,129, 29− −0,088, 88⋅ ⎜⎝⎛⋅90 (28)
l ≤ i i ⎪ ⎣ ⎝ ⎠⎦ ⎪⎭
3⇤laΣ 60⎟⎥⎞⎬⎫⎤ ⎪⎫
⎛αα−−−60 models to an alumina calciner
⎜ 60 ⎠⎞⎦⎤⎟⎪ (29)
r ≤ il a ≤ 3 ⋅ Σσ(t ⋅ ρ t
i 1)i⋅)g⋅ a g⋅ ⋅⎨ ⎧0,88 + ⎢1⎡, 29 − 0⎡,88 ⋅ ⎜ ⎛⎝ α ⎭⎥ ⎬
60
90−−6060 ⎟ ⎞⎠⎬⎤ ⎫⎪⎪ (28) (28)
l r ≤ Σ(lt i ⋅r≤ρ ⎪⎩⋅σ⎪⎨⎪⎩0a ,288 +⋅⎣⎧⎪⎢⎣01,,88 29 − 0, 88⎝⋅ 900,−88 ⎛ ⎥α ⎦−⎭ 60 ⎞⎤ ⎫ ⎪ (29) project
  i
3 ⋅ Σla(t≤i ⋅ρ iΣ)(⋅tgσ ⋅ t1 ⎨⎣ ⎧⎪ + ⎢1,⎡29 ⎜⎝−90 − ⋅ ⎜⎠⎦⎟⎛⎪⎭⎥α⎬⎪ − 60⎟⎥⎞⎬⎤ ⎫⎪
60 (28)
⋅ ⎪
rρ
ia ⋅ σi a2⎩ ) ⋅ g ⎪⎩⎪⋅ ⎨0,88 +
⎧ ⎣⎡ ⎢1, 29 − 0, 88⎝⎛⋅90 ⎠⎜α⎦ ⎭− 60 ⎠⎞⎦⎤⎟⎪⎫
⎭⎪⎥ ⎬ (29)
l a ≤ 3 ⋅ Σσ(t i ⋅ρt i )⋅ g⋅ ⎨0⎧⎪⎪⎩,88 + ⎢1⎡⎣, 29 − 0,88 ⋅ ⎜ ⎛⎝ 90
r
α−−−60 60⎟⎥⎞⎠⎬⎤⎦ ⎫⎪⎪⎭
60 (28) Calcination is the last step of the Bayer
laσ
r ⋅≤ σ 2Σ(t i ⎧ ⋅rρ i 1)⋅ g ⎡ ⎪⎩⋅ ⎨0,88 +⎣ ⎢1,⎛29α−−060 , 88⎞⎝⋅⎤⎜90 ⎫ ⎠⎦⎟⎪⎭⎥ ⎬ (29)
⋅ t
r 13 ⋅ Σ⋅(t i0⎪⋅,ρ
⎪ ⎧ ⎡
)+⋅ g+⎢1⎢,1⎪⎩29 α 60 ⎪
⎤ ⎫
, 29− −0,088
a
⋅ ⎜⋅ ⎛⎜ − ⎟⎥⎝⎞⎟⎬⎥90 ⎪ − 60 ⎠⎦ ⎪ (28)
ll ar a≤≤⇤ ⇥ ⎨⋅ 088 ,i88 ,⎣88 ⎭ (29) (29) process to produce alumina, which is char-
la ⇥ a ⋅(tΣiσ
3Σ ⋅)g⋅ g ⎩⎪ ⎨⎪⎩⎧
(⋅trρi ⋅1⋅iρt,)1i283 ⎣ ⎡ ⎝ 90
⎛ 90α−−−6060
60⎠⎦⎠⎞⎭ ⎪⎦⎤⎬⎪⎭⎫ (30)
⋅ ⎨02,88 + ⎣⎢1, 29 − 0, 88 ⋅ ⎝⎜ (29)
la ⇥ (
l ra≤⇤ ⇥ a
)
Σ ti ⇤ ρi3 ⋅ Σg(t i1⋅ρ, 83 σ ⎩⋅ t1
i ) ⋅ g r ⎪ ⎪⎧ ⎡
⎟ ⎥
90 − 60⎛ α − 60 ⎞⎤ ⎪⎫
⋅⎣ 0,88 + ⎢1, 29⎝ − 0, 88 ⋅ ⎜⎠⎦ ⎪⎭
⎬
(30) (29)
acterized by a thermal treatment where alu-
Σ l ti⇥⇤ ρi l rga≤⇤ ⇥3a⋅ Σσ(t ir ⋅1ρ⋅ ,ti183
( ) 2) g ⎨
⋅ ⎪⎩⎧⎪ ⎣⎡ ⎝⎛90
⎟⎥ ⎬
α − 60 ⎠⎞⎦⎤⎪⎭⎫⎪ (30) minum hydroxide [Al(OH)3] is transformed
a
laa⇥⇤ ⇥ σ ⋅ t
Σl rtai≤⇤ ⇥ ( )
ρ2ia3 ⋅ Σg(t ⋅ρ )⋅ g ⋅ ⎨⎪0,88 + ⎢1, 29 − 0, 88 ⋅ ⎜ 90 − 60 ⎟⎥ ⎬⎪
i⎧1, 83 ⎣ ⎝⎤ ⎫ ⎠⎦ ⎭
(29)
(30) (30) into aluminum oxide (Al2O3) at tempera-
( ) g ⋅ ⎪0,88 + ⎢1, 29 − 0, 88 ⋅ ⎛⎜ α − 60 ⎞⎟⎥ ⎪⎬
i ⎡ ⎩
la ⇥ l r ≤ aΣ2 tir ⇤1ρ,183 (29)
(30)
( ) ,Σ ta⇥ ⇤
⇤ r⇥
ρ⇤3at1⋅ Σg(t i ⋅ρ ii )⋅ g ⎨⎪⎩ 90 − 60 ⎠⎦ ⎪⎭ tures in the range 1000–1250 °C. Fig. 10
lra ⇥ i i 2 1 183 , 83 ⎣ ⎝
(31)
(30)
lr ⇥ 3⇤Σl(Σt⇥ ⇥i (⇤rtρi ⇤iρ1) i )g g ⇤1at,283 1, 83
⇤ t a ⇤⇥
(31) (30) shows a schematic drawing of one of the
3⇤laΣ⇥(t i ⇤ρΣi ()tai⇥g⇤⇤ρ r⇥ )1 g
1, 83 (31) (31) calciners currently in operation at Alumar’s
lra ⇥ 3⇤ Σ⇥ t ⇤i⇤taρ g1, 83 (30)
⇥⇤ ⇥⇤atΣ2 (ti(⇤rρi i )1 i )g 1, 83
2
lr a⇥ (31) plant, in São Luis/BR. The regions operating
t , ⇤83
ρi ) ⎧g
Σ (ti⇥⇤ ρ⇤i t)2 ga ⋅(σ ai 1, 83
lla ⇥⇥ r 3⇤ 1 Σ 1 (30)
(31)
lr ⇥ 3⇤l aΣ≤(tr i ⇤Σρ1(ita) ⋅⇥σgρa⇤)1t⋅ ,2g83
r ⎪ ⎡⎛ α − 15 ⎞ ⎤ ⎫⎪ at higher temperatures have their refrac-
⋅ ⎨1, 83 − ⎢⎜ ⎟ ÷ 1, 83⎥⎫⎬ (32)
(31)
−−15
3⇤llaΣ⇥≤(t i ⇤ρi ) g
⎧⎪⎪ 60
⎡⎣⎛⎝ α 15 ⎞⎠ ⎤⎦⎪⎪
⋅ ⎨1, 83 − ⎢⎜⎧ ÷ 1, 83 ⎭ (32) (32) tory held in place with metallic and ceramic
ρti )⇤⋅ρtg2 )a⎪⎩⋅gσ a1, 83
i i ⎩
r 1
⎡⎟⎛ α − 15⎥⎦ ⎬⎪⎞ ⎤ ⎫ (31)
r
lr ⇥ 2
Σ3⇤(tliΣ⋅⇥
a (
≤ ri 1i
Σ(tai ⋅ σ
⎣⋅⎝⎪⎨60
ρ ia)1⋅ g, 83⎧⎪⎩
− 15
1, 83 − ⎢⎠⎜ ⎭⎟ ÷ 1, 83⎥ ⎪⎬ (32)
(31) anchors, depending on the temperature
⎡⎣⎛⎝ 60 15 ⎞⎠
l aΣ⋅≤σ( t i ⇤ρi ) ⎧g
α −−15 ⎤⎦⎫⎪⎭
⇥ r ⇤ t3⇤ 1 a
⋅ ⎨1, 83 − ⎢⎜ ⎟ ÷ 1, 83⎥ ⎬ (32) range and equipment region. For the high
lr ⇥ a1Σ,(83 t ⎪⋅ ρ )⋅ g ⎡⎛⎪α − 15⎣⎞⎟⎝ ÷ 60 − 15⎤ ⎫⎪⎠ (31)
l ≤
3⇤laΣ ( t ⇤ ρ ( )
σ⋅⋅rg
a ⋅ t12 i⋅ ⎨⎧1⎧⎪,i83 − ⎢⎡⎜⎛⎩⎡⎛αα− −
) 15⎞ ⎞ 1, 83⎥⎤⎬⎫⎤ ⎫⎪
15
⎦ ⎪⎭ (32)
r ≤ i
Σ t σ
ρ ⋅ g ⎩⋅ ⎨1, 83 −⎣⎝⎢⎜
⎪
l a ≤ 3 ⋅ Σσ(t ⋅ tρ1 )⋅ g⋅ ⎨1⎧⎪⎪, 83 − ⎢⎧⎜⎡⎛ α
i i a
i2 60 − 15 ⎠ ⎟ ÷ 1, 83 ⎪
⎟ ÷⎠ 1, 83⎥ ⎬⎤ ⎫⎪⎪
⎦ ⎭⎥ ⎬ (33)
(32) (33) temperatures areas, phosphate-bonded
i )i⋅ g a ⋅⎪ ⎝ 60−−− 15
15
⎡⎛⎠ ⎞⎟α
l r ≤ Σ(lt i ⋅r≤ρ i
σ 2 −⎣⎪⎝⎣⎜60
⎩⋅ ⎨⎩⋅1at, 83 83 −−15
⋅ ⎨⎢1⎧⎪,60 ⎜ ÷−1,1583⎦ ⎪ ⎦ ⎤ ⎫
⎭⎥⎞⎟⎬⎭÷⎞ 1, 83⎥ ⎪⎬⎤ ⎫⎪ (33) (32) high-alumina ceramic anchors are used.
3 ⋅ Σla(t≤i ⋅ρ iΣ)(⋅tgσ ⎪ 15
⎢ ⎡⎠⎛ α − 15 ⎪⎭⎟ ÷ 1, 83
a ⋅ ⋅ r
ρ⎩
σ )
1 ⋅ g ⎣
⎪ ⎧
⎪⎝ 1, 83 −
⋅ ⎡
⎣⎝⎛ 60 ⎜α − 15 ⎦
⎠ ⎞ ⎦⎤⎪⎭⎫⎪⎥ ⎬ (33)
lra ≤ 3 ⋅ Σσ(t i ⋅ρta2i )⋅ g⋅⎩⎨1⎧⎪⎨⎪⎩, 83 − ⎢⎜⎡⎢⎣⎛⎝ 60
i i
α−−−15 15⎟ ⎞⎠÷ 1, 83⎥ ⎬⎤⎦ ⎫⎪⎪⎭
15 (32) They present suitable volumetric stability,
laσ ⋅≤σ 2Σ ( t i ⋅
⎧
rρ 1)⋅ g ⎪
i
⎪ ⎪⎧⋅ρ )⋅ g⎡⎛⎡⎛α ⎩⋅ ⎨1, 83 −⎣
− 15 ⎞⎢ ⎝ 60
⎜ ⎫ ⎠
⎤ ⎪⎤ ⎪⎫ ⎟ ÷ 1, 83 ⎦ ⎭⎪⎥ ⎬ (33)
r ⋅ at13 ⋅ Σ⋅(t 1 − −⎢⎜⎢⎜⎪⎩ α − 15⎟⎣÷⎞⎟⎝ 1÷60
r
ll ar ≤
≤ ⎨⋅i⎨,183 ,i83 ,183 ⎥15
,−83 ⎬⎥ ⎬⎠ ⎦ ⎪⎭ (32)
(33) which reduces the stresses originated by
3Σ⋅(tΣiσ(⋅trρi ⋅⋅iρt)1i⋅ )g⋅ g ⎪
2
⎧ ⎝
⎩⋅ ⎪⎩1, 83 −⎣ ⎣⎝⎛⎜ ⎡60 60
α− 15
−
− 15
15 ⎠ ⎠⎞ ÷ 1, 83 ⎪ ⎤
⎦ ⎭⎦ ⎪⎭ ⎫
lr ≤ ⎨ 2 ⎢ ⎟ ⎥ ⎬ (33) the mismatches in thermal expansion coef-
3 ⋅ Σ(t i ⋅ρ i )⋅ σ g r ⎪⎩⋅ t1 ⎪ 60 − 15⎡⎛ α − 15⎦ ⎪⎭⎞
⎧ ⎤ ⎫⎪
lr ≤ ⋅⎣⎝1, 83 − ⎢⎜⎠ ⎟ ÷ 1, 83⎥ ⎬ (33) ficients. In the regions of lower mechanical
3 ⋅ Σσ (t ir ⋅ρ⋅ ti12)⋅ g ⎨⎪⎩⎧⎪ ⎡
⎝
⎣ ⎛ 60
α − 15 ⎠ ⎞ ⎦⎤⎪⎭⎫⎪
lr ≤ ⋅ ⎨1, 83 − ⎢⎜ ⎟ ÷ 1, 83⎥ ⎬ (33)
23 ⋅ Σ(t i ⋅ρ i ) ⋅ g ⎪ ⎣ ⎝ 60 − 15 ⎠ ⎦ ⎪⎭ stresses, Incoloy DS steel anchors are usually
σ r ⋅ t1 ⎧⎪ ⎡⎛⎩α − 15 ⎞ ⎤ ⎫⎪
lr ≤ ⋅ ⎨1, 83 − ⎢⎜ ⎟ ÷ 1, 83⎥ ⎬ (33) selected. Ceramic anchors have the same
3 ⋅ Σ(t i ⋅ρ i )⋅ g ⎪⎩ ⎣⎝ 60 − 15 ⎠ ⎦ ⎭ ⎪
design as those shown in Fig. 2 where their
lengths change as a function of the lining
thickness. The metallic anchors used have
the same morphology as those presented
in Fig. 1. The bottom of the calciner fur-
nace, as indicated in Fig. 10 (Region 1),
was selected for analysis, as it is a region
with different sorts of geometry and higher
operating temperature (1050 °C). Another
area of the calciner, also shown in Fig. 10
(Region 2), was chosen for analysis as it
operates at temperatures around 500 °C
and also comprises floors, walls and roofs.
The analysis carried out compares the aver-
age spacing values used in the calciner de-
sign with those calculated with the original
Plibrico model, the modified one, as well
as the values suggested by few other com-
Fig. 10 Layout of a Mark VII type calciner, with indications of the analysed regions panies, in order to study the mathemati-
(adapted from [1]) cal model validity. Region 1 of the calciner

 
98 refractories WORLDFORUM 8 (2016) [1]
 I2 I1I2 I1
D1 D1 I2 I1I2 D1
I1 D1

PaPer
   

I2 I2 I1 I1 D1D1 I2 I2 I1 I1 D1D1 I3 D2
I3 D2 I3 D2
I3 D2

Fig. 11–d Refractory lining configuration adopted in: (a) Parts 1, 2, 4 and (b) Part 3 of Region 1 for ceramic anchors, (c) Parts 4 and 5 of
Region 1 for metallic anchors, and (d) in all parts of Region 2

Tab. 5 Properties of the refractory Tab. 6 Properties of the anchor materials used in the studied regions of the calciner
materials used in the studied
Anchor Type Tensile Stress Cross-Section Area
regionsI3ofI3theD2
calciner
D2 I3 I3 D2D2
. .
�r(1000 °C) [MPa] a [mm2]
Hot Face Density Bending Metallic     73 50,27
Refractory p [g/cm³] Stress    

�r(1000 °C) [MPa] Ceramic 21 5153

20   20  
D1 2,46 8 two insulation layers  (I1 and   I2) of 50 mm mesh arrangement of anchors. However,
D2 2,65 9 each (Fig. 11a–b). In areas with metallic an- this pattern is inherent to each refractory
Insulating Density chors, the total thickness is 225 mm, lined panel and presents changes according to
Refractory p [g/cm³] only with a dense refractory (D2) (125 mm) the vessel geometry variations, as indicated
I1 1,18 and an insulating material (I3) (100 mm) in Fig. 12b. Therefore, the general analysis
I2 0,56 (Fig. 11c). Region 2 also features cylindrical is based on the average spacing values. As
shapes and parts as roofs, walls and floors, shown in Fig. 12a, Part 1 of Region 1 has a
I3 1,00
hold by metallic anchors. The total lining conical shape with a 60° slope, thus for cal-
. . thickness in all regions is 150 mm, in which culation purpose it should be considered as
presents
   
conical and cylindrical shapes, con- 100 mm is related to a hot face refractory a vertical wall. The average anchor spacing
taining parts of roof, vertical and inclined (D2) and the remaining 50 mm for the in- in this part is 360 mm. Parts 2 and 3 rep-
walls, and floors. The areas in which ceramic sulation (I3) (Fig. 11d).20  Tab.
20   5–6 show the resent vertical walls with ceramic anchors
   
anchors are used present a total thickness of values of the material properties considered and the difference between them relies
  250 mm in one part and 275 mm in other, in the spacing calculations. on the total lining thickness, as previously
both consisting of a hot face refractory (D1) Fig. 12a illustrates in detail the analysed mentioned. The average spacing is 362 mm
of 150 mm and 175 mm respectively, and calciner regions, which show a square for Part 2 and 340 mm for Part 3. Part 4

Region 1 Region 2

(a) (b) (c) (d)

Fig. 12a–d Cross-section of Region 1 and its different parts (a), schematic view of the anchor distribution at the conical part of
Region 1 (b), side view of the top (c), and side view of the bottom part of Region 2 (d)

refractories WORLDFORUM 8 (2016) [1] 99

22  
   
  

PaPer

for all parts of Regions 1 and 2, respec-

Tangential Radial
tively.
200

3 Results and discussion

100
Thermal stress (MPa)

3.1 Thermal stresses

0 The thermal stress analysis, using Eq. 22
–23, and the data presented in Tab. 4, pro-
-100 vided the results shown in Fig. 13. The signs
of the stresses in the graph are adjusted,
so positive indicates tensile and negative
-200
compressive ones, opposite to the reference
adopted in the original model.
 
  -300 Initially, it is possible to notice that the
0 5 10 15 20 25
Distance from anchor / refractory interface (mm)
stresses rapidly decrease towards zero with
the increasing distance from the anchor/re-
Fig. 13 Distribution of tangential and radial stresses from the interface into the matrix fractory interface. However, the calculations
show that null values are only reached at
distances close to half of the anchor spacing
used (150 mm). Thus, considering all simpli-
fications assumed in the evaluation, espe-
cially regarding anchor shape, in which a
simple cylindrical geometry adopted greatly
differs from the complex shapes used in
the project (Fig. 1), the actual stresses may
be even greater in magnitude. So, these
thermal stress profiles may indicate that in
addition to problems in refractory applica-
tion and consolidation during the installa-
tion step, too short anchor spacing can also
lead to an interaction between the thermal
stresses, which may increase the likelihood
Fig. 14a–b (a) Schematic drawing of the bullnoses (Regions 1 and 2), and (b) a bullnose
with metallic anchors
of crack generation. If by one hand the
increase in the distance between anchors
leads to higher stresses due to the greater
weight withstood by each anchor, on the
other, decrease in anchor spacing can also
23  
have harmful effects, since the thermal
  stresses generated at an anchor/refractory
interface may extend along the lining and
interact with those by its nearest neigh-
bours. This scenario may be one of the rea-
sons why areas with complex shapes such
as bullnoses, where there is a great density
of anchors, present so many cracks. Fig. 14a
outlines diagrams of the bullnoses present
in Regions 1 and 2, and Fig. 14b illustrates
  a bullnose with metallic anchors for the
  alumina calciner. Fig. 15 shows a common
Fig. 15 Front view of the lining failure and crack propagation found in a bullnose area
with ceramic anchors in Region 1 failure in the lining for the bullnose contain-
ing ceramic anchors of Region 1.
is a roof with metallic anchors and a small represents the walls connecting Parts 4 The analysis also corroborates the concept of
portion of ceramic ones, with an average and 5, containing ceramic and metallic an- applying degradable coatings to the anchors
spacing of 357 mm for the ceramic anchors. chors, where the spacing for the ceramic in order to reduce thermal stress between
Part 5 is a floor with metallic anchors, and ones is similar to Part 2. The average spac- materials with great mismatch in thermal
Part 6, which is not illustrated in the figure, ing for metallic anchors is 287 and 280 mm expansion coefficients, as the increase in the

100 refractories WORLDFORUM 8 (2016) [1]

 
 
 PaPer

(a) (b)

Fig. 16a–b (a) Incoloy DS metallic anchor

  with polymeric caps, and (b) the corrosion caused by them, after using at high temperatures

interface distance provided by the coating Spacing based on anchor strength

can reduce the initial interaction of the ma- Spacing based on refractory strength
terials and consequently the magnitude of 800

the stresses at the interface and throughout 700

the lining. In this case, the coating thickness
Anchor spacing (mm)

600
should be enough to accommodate the ther-
500
mal stresses but not to the point it affects
400
the interaction between anchor and lining,
resulting in slip bound, what can reduce the 300

holding capability of the anchor to the re- 200 Permissible range

fractory and also lead to failures. However, 100
the nature of the coating used should not in-
0
duce anchor oxidation. Polymers can release 0 100 200 300 400
carbon monoxide during degradation, which Lining thickness (mm)

can oxidize and react even with high-chro- Fig. 17 Relationship between lining thickness and anchor spacing calculated with the
mium content steels [6], such as Incoloy DS modified Plibrico model and using the materials described in Tab. 4–5
alloy. This effect was observed in practice at
Alumar when polymeric caps similar to those Tab. 7 Average spacing in [mm] for ceramic anchors of Region 1, working around 1050 °C
shown in Fig. 16a were used, resulting in Plibrico Model This Paper Shina- Thermal Project
anchor degradation as illustrated in Fig. 16b. Anchor Refractory Anchor Refractory
gawa Ceramics (Alumar)
The presence of chlorine in the coating com- Strength Strength Strength Strength
position is also very harmful and acts on the Roof 546 406 479 356 300 300 357
corrosion of metallic parts. A common instal-
Wall 1 999 744 704 524 380 460 362
lation practice consists of involving the metal
clips used in ceramic anchors with chlorine- Wall 2 938 815 661 575 380 460 340
free tapes, which in addition to relieving Tab. 8 Average spacing in [mm] for metallic anchors of Region 1, working around 1050 °C
stresses do not affect the steel properties.
This is a simple procedure that could be eas- Plibrico Model This Paper Shina- Thermal Project
gawa Ceramics (Alumar)
ily replicated for metallic anchors, replacing Anchor Refractory Anchor Refractory
the polymeric caps. Strength Strength Strength Strength
Roof 295 372 258 326 300 180 287
3.2 Anchor spacing Wall 539 681 380 480 380 300 287
The results for anchor spacing obtained us- Floor – – 539 681 380 460 287
ing the modified and the Plibrico original
model were compared with those proposed Tab. 9 Average spacing in [mm] for metallic anchors of Region 2, working around 500 °C
by few companies and the ones currently Plibrico Model This Paper Shina- Thermal Project
applied in the calciner. Two different cal- Anchor Refractory Anchor Refractory
gawa Ceramics (Alumar)
culations are carried out for the theoretical Strength Strength Strength Strength
models, one with anchor strength as limit- Roof 345 348 302 306 300 180 280
ing factor and another based on the refrac-
Wall 631 637 445 449 230 230 280
tory strength, being the lowest value found
taken as valid, as illustrated in Fig. 17. Floor – – 631 637 380 460 280

refractories WORLDFORUM 8 (2016) [1] 101

 Paper

Tab. 10 Likely reduction for non-critical areas of Region 1 with ceramic anchors time of the lining. Decreasing the number of
Spacing [mm] Number of Anchors Reduction anchors saves costs on materials and on the
Actual This Actual This
[%] installation time. Regarding the anchor mesh
Project Paper Project Paper pattern, there are changes in the regularity
Wall1 Cone 362 524 213 150 30 along the refractory panels due to the transi-
tion of geometry of sections in the vessel. This
Vertical cylinder 1 362 524 81 36 56
scenario may induce irregular load distribu-
Vertical cylinder 2 362 524 131 61 53
tion on the anchors, and could be bypassed
Horizontal cylinder 362 524 62 35 44 through a suitable design of the lining as a
Wall 2 384 575 32 15 53 whole, including the arrangement and posi-
tion of expansion joints, the panel size and
the anchoring system. This study provides
Tab. 11 Likely reduction for non-critical areas of Region 1 with metallic anchors
a preliminary, nevertheless incremental and
Spacing [mm] Number of Anchors Reduction relevant, understanding on the parameters
[%]
Actual This Paper Actual This Paper involved in the design of anchoring systems.
Project Project The mathematical models herein presented
Wall 287 380 56 44 21 offer an interesting basis and a simple and
Floor 287 539 160 40 75 practical way for calculating anchor spacing
using ordinary properties of materials.

Tab. 7 shows the results for the areas
containing ceramic anchors of Region 1,
where the Roof is associated with Part 4
of Fig. 12a, Wall 1 to Parts 1 and 2, and
Wall 2 to Part 3, respectively. The results for
metallic anchors are presented in Tab. 8,
in which the Roof relates to Part 4, the
Wall to Part 6 and the Floor with Part 5.
The results for Region 2 are indicated in
Tab. 9.
For Region 1 the anchor spacing values ob-
tained with Plibrico original model differed
largely from all the others, what indicates
this model overestimates the permissible
spacing, restricting its practical use. There
was a good agreement between the val-
ues of the modified model (this paper)
with those used in the calciner design for
the roofs, which are one of the most critical
­areas concerning anchoring systems, pro-
viding some evidence for the model validity.
In relation to the walls and floors, the val-
ues attained are higher than those actually
used. Thus, considering correct the mod­
ified model, there may be opportunities for
potential changes for the spacing used in
practice and a consequent reduction in the
number of installed anchors, which can lead
to savings. The same trends regarding the
values proximity for roofs and differences
for walls and floors is observed for Region 2.
In addition, the comparison between the
results for Region 1 and Region 2 provides
further and stronger support to the idea of
reducing the number of anchors used in less
critical areas, given that the average oper-
ating temperature of Region 2 is 500 °C,
making this region much less critical, and
that the data used in the calculations refers
to the materials strength at 1000 °C, which
is an additional safety factor. This assess-
ment shows the possibility to decrease an-
chor consumption in less demanding areas
and opens opportunities for conscious and
grounded cost savings. Considering that in
some cases the price of ceramic anchors can
surpass the one of the refractory materials
themselves, the importance for optimiza-
tion related to the number of used anchors
is relevant. Besides that, many failures ob-
served in regions working at moderate tem-
peratures are related to other aspects than
the anchor spacing, for example the refrac-
tory panel size and the lack of expansion
joints. Tab. 10–11 show the possible reduc-
tion in the number of installed anchors in
case the spacing values used were the ones
predicted by the modified model.
4 Conclusions
The values for anchor spacing used in the
actual design of refractory lined vessels can
greatly differ from those attained with the
available theoretical models, especially for
non-critical areas. Generally, there is a trend
of replicating the anchor spacing used for
roofs in walls or even floors, which represent
an opportunity to decrease the number of
applied anchors, without affecting safety.
Their spacing in less critical areas, where the
loads in action are less intense, can be in-
creased without reducing the expected life-
References
[1] Teider, B.: The refractory lining and its effect on
the life time and energy consumption of alumi-
na calciners [in Portuguese]. MSc Dissertation,
Universidade Federal de São Carlos, São Carlos
2012
[2] Plibrico Japan Co. Ltd.: Technology of monolithic
refractories. Tokyo 1999, 314–339
[3] Chen, E.; Dicks, L.; Buyukozturk, O.: Anchor-
lining interaction in a hot-shell refractory lining.
Amer. Ceram. Soc. Bull. 69 (1990) [11] 1813–
1820
[4] Palmer, G.; Tan, K.: Design and failure of mono-
lithic refractory structures – part I. Refractories
Applications and News 14 (2009) [3] 19–26
[5] Shinagawa Refractories: Anchoring of mono-
lithic refractories. Shinagawa’s standard instal-
lation procedures IP-008 anchors (2004). http://
www.shinagawa.biz/index.pl?page=57
[6] Palmer, G.; Smillie, M.: Selecting steel anchors
for monolithic refractory linings Refractories Ap-
plications and News 12 (2007) [5] 22–30
[7] Pessoa, D.S.: Inspection and maintenance of
refractories [in Portuguese]. Course Booklet,
Instituto Santista de Qualidade Industrial, São
José dos Campos 2013
[8] Thermal Ceramics: Anchoring of monolithic re-
fractories, Design & Installation Manual, 2002
http://www.morganthermalceramics.com/re-
sources/installation-manuals
[9] Forni, G.; Recalcati, P.; Nestani, M.: The influ-
ence of thermal bridges in refractory linings La
Metallurgia Italiana 6 (2007) 53–60
[10] Liu, D.; Winn, E.J.: Microstresses in particulate-
reinforced brittle composites J. of Mater. Sci. 36
(2001) 3487–3495

102 refractories WORLDFORUM 8 (2016) [1]

View publication stats