ABSTRACT
Although the major codes for fabrication, generally give clear instructions on the prerequisites
for heat treatment, the implications of carrying out such heat treatment can
be significant, and is easily overlooked by the less experienced contractor.
This paper looks at the most common mechanical problems associated with heat
treatment, and the necessary mechanical solutions to ensure a successful heat
treatment.
Author
Mr. S Chandrashekhar
Silcarb Technical Services (Silcarb)
38, 17th Cross Rd, 6th Block, Malleshwaram West, Bengaluru, Karnataka 560003
THERMAL EXPANSION
The source of many problems associated with heat treatment arises from the
effects of thermal expansion. Although the principal of thermal expansion is readily
understood, problems can arise as a result of thermal expansion, but these are not
always so apparent.
The thermal expansion of carbon/ low allows steels range from 11.5x10-6m/m°C at
ambient, to 14.8x10-6m/m°C at 600°C. An easier number to work with for estimating
purposes is 8mm per meter for carbon steels at 600°C.
Stainless steels have a significantly higher rate of rates of expansion, -approximately
30% greater than that of carbon steel; this can cause problems where stainless steel
components are included in carbon steel heat treatments.
The forces generated by thermal expansion of a constrained component can be
massive, and if constraints are not removed, then both distortion, and/or high
residual stresses are likely following heat treatment.
The following example highlights this:
Pipework – the effects of constrained ends
Many codes make reference to thermal gradients, and the ‘harmful effects’
that may result, and in some cases, minimum heated bands, and insulation
are prescribed to avoid such harmful effects. The harmful effects that are
referred to, are a result of axial temperature gradients.
However, in addition to the radial expansion, the axial expansion also takes place, - ie
the cylinder actually gets longer during the heating cycle. The assumption is
generally made, that the pipe/ cylinder is free to expand/ contract. However,
this may not always be the case, and the effects of constraint can be
significant.
The well-established practice of local heat treatment of piping uses a relatively
a narrow band of heating to achieve the desired soak temperature at the weld.
Radial expansion at soak temperature is only in the order of a few millimeters
at most (for piping), and upon cooling, contraction returns the pipe to normal.
However, if the pipe is constrained, the effects are very different. Even on
relatively small diameter piping, with relatively narrow heated bandwidths,
an axial free expansion is only in the order of 1-3mm. However, if the ends of the
the pipe is constrained, for example between large heavy wall headers, then, as
the expansion cannot be accommodated, the weakest link for the expansion
forces to take is at the heated band, - this results in yielding, and larger radial
expansion whilst at temperature.
• The mechanism is that as the region of higher temperature material has a
lower modulus, and is consequently weaker, therefore, the expanding material
expands into the existing natural bulge like a balloon filling up.
• The bulge creates significant local stresses and yielding, such that even after
cooling, a residual bulge remains, along with high residual stresses.
Radial expansion
larger than normally
seen for no restraint
Permanent bulge
remaining in the pipe after
cool down
HORIZONTAL VESSELS
PERMANENT SUPPORTS
SLIDING SADDLES
Almost all horizontal vessels are based upon two saddle support system; it is also
the standard for saddles to be mounted on concrete plinths, incorporating anchor bolts.
In most cases, one saddle is fixed, and one sliding, but the sliding saddle is designed
to accommodate vessel expansion in service, which may be significantly less than
heat treatment temperatures. Accommodation of the design expansion/contraction
is via slotted bolt holes.
For heat treatment in full of in-situ vessels, it is therefore important that the
expansion can be accommodated by the sliding saddle. If the slot length is
insufficient, options are limited to either extending the slots or jacking up entire
vessel so that saddle base plate is clear of anchor bolts.
Expansion can be significant, - using the above rule of thumb, for a vessel with
saddles spaced at 10m apart, and thermal expansion will be in the order of 80mm.
Whilst thermal expansion between saddles may be readily identified, another effect
associated with heat treatment of welded saddles is not always so clear, and is an
often overlooked complication to heat treatment. Not only do welded saddles
represent a significant heat sink, but they are also a constraint to radial expansion,
except where they are heated fully – in a furnace for example.
Considering initially the saddle alone, and heating only the wrapper plate to 600°C,
and allowing conduction only down the ribs and web plate, typically, the base plate
remains cold.
The net result of the thermal expansion is that the base plate arches up at the
center and high thermal stresses are generated in the main web plate
It has been found that the addition of electrical heating (via ceramic pad heaters)
applied below the wrapper plate significantly reduces the magnitude of stresses in
the saddle, and the level of distortion post heat treatment. This also has the benefit
of compensating for the heat sink effects caused by the saddle, thus guaranteeing
soak temperatures at the saddle to shell welds. (There can be a knock on effect with
this, for saddles in –situ, - if the saddle is low, and mounted on a concrete plinth,
For new build columns, it is most common for heat treatment to be carried out in the
horizontal position. This requires the use of temporary saddles supports. Aside from
the need to ensure that such supports are fit for purpose, - from both saddle
perspective, and vessel perspective, vessel expansion needs to be catered for.
It is not uncommon to see vessels as long as 100m, at 600°C, this equates to
800mm of expansion – a significant amount of movement.
Although as indicated above, the forces of expansion are massive, and the vessel
will expand even if no specific measures are taken to allow for it. However, in doing
so, there is a risk that saddles could snag and potentially tilt, or if internally fired,
saddles could slide on the vessel, dislodging insulation, causing significant risk. It is, therefore, good practice to ensure that saddles are free to move, - ideally utilizing
simple rollers.
This is a simple but very effective method of ensuring the unrestricted expansion of
vessels. However, care must be taken in allowing for significant expansion to ensure
that sufficient rollers are positioned in front of the saddle.
For long vessels that necessitate a number of supporting saddles, a simple measure
that can be taken, is to have the centremost support fixed, and other supports
moving, - this halves the magnitude of movement at any one saddle, but
remembering that expansion from the center is in opposite directions – this is
possibly quite obvious, but if set up wrongly, has the potential for disastrous results.
Fixed -160mm -80mm +80mm +160mm
The typical vessel, 10m between supports
External stiffening Rings
For large diameter thin wall pressure vessels, external stiffening rings may be a part
of the design. Whilst such attachments are not of concern if the vessel is heat
treated in a furnace, they can be a major problem for internally heat treated vessels.
External stiffening rings can be in excess of 300mm deep, usually with an outer
flange ring. Such rings present a risk from a heat sink point of view, and also thermal
expansion. Although code requirements are that the attachment to shell weld
reaches temperature, the effects of a cooler perimeter can be significant.
For large stiffening rings, no matter how much insulation is applied, the effects of
heat loss is such that it is very difficult to achieve even 500C at the outer flange.
For a large diameter vessel, this presents significant differential thermal expansion.
For example, a 10m dia vessel, at a soak temperature of 600°C, radial thermal
expansion is 43mm, whereas, the outer flange of an attached ring at 500°C is 36mm.
This, therefore, represents a constraint to the expansion of the shell, though if the shell is
significantly thicker, typical results are a distortion of the stiffening ring.
Options for mitigating this problem are limited, - external electrical heating can be
applied relatively easily, but for large diameter vessels with a number of stiffening
rings, this can become impractical. One solution is to weld on 50mm short rings to
the vessel shell, and then weld on the main ring after PWHT, - but this would need to
be addressed at the design/ drawing stage.
VERTICAL VESSELS
Heat treatment of vertical vessels in situ (having been in service) present a
a significant challenge in identifying the many factors that need to be considered
before proceeding.
Vessels in service often have a significant amount of attachments of various forms
that can cause restraints against expansion; it is important that a detailed survey of
the vessel is undertaken to identify any attachments that may pose a risk to the heat
treatment. In addition to mechanical constraints, other attachments/ items in close
proximity may also be affected, and require addressing, - such as cable trays,
instrumentation, sight glasses, etc.
MECHANICAL PREPARATION
Restraints on vertical expansion: The height of the vessel section being heat
treated, will dictate the total vertical expansion that must be catered for. If heat
treatment consists of only PWHT of a circ seam, vertical expansion may be
minimal; however if a full vessel requires heat treatment, vertical expansion
may be in the order of 100’s of millimeters. It is important to remember that
everything attached to the vessel above the heated zone will need to move by
the total calculated expansion.
ATTACHED PIPING.
Piping may remain attached providing that there is sufficient flexibility prior to
any restraints. Where there is doubt, the additional loading imposed can be
calculated by simple beam bending theory. If the load cannot be tolerated,
then the pipe will need to be cut.
However, where restraints exist
that cannot be removed, it is
preferable to disconnect flanges if
possible, but the last option is to cut
the line in order to provide the free
expansion to take place.
It is important to look at the pipe
supports to determine where any
cut lines in piping are made, e.g.
below any trunnion supports as
the expansion also lifts the support.
Cut below trunnion
Note that there is an alternative option for cutting pipes, - this is to apply
separate heating on the pipe, purely to ensure that the same thermal
expansion is created. Depending upon the size of heating/ required expansion,
this may/ may not be a viable option.
SERVICE PLATFORMS
Service platforms are generally directly attached to the vessel – the weight of
which must be accounted for in any stability calculations. Any service
platforms that circumvent the vessel by a significant amount, present a
restriction to radial expansion at the point of attachment. It is therefore often
necessary to ensure that all platform attachments are via horizontally slotted
bolt holes.
Note that for larger vessels, radial expansion may cause the vessel shell to
expand beyond the actual platform; - in this case, it may be necessary to
separate the platform into segments to allow additional space for expansion, -
or of course platform removal is an alternative option.
It is not unusual to find a number of service decks adjacent to vessels that
provide access to the vessel and ancillary or adjacent equipment. These
decks often have connections either directly or indirectly to the vessel; any
such a connection is likely to be a constraint;
For example, service deck with stairs down to vessel service platform;
The stairs, attached to the
deck are fixed, but the
service platform needs to
move vertically with the
vessel during expansion.
A section of the stairway
may need removal to
ensure free expansion
VESSEL STABILITY
The major concern for the heat treatment of vertical
In order to assess stability, the primary loads imposed on the vessel at the heated
the area needs to be considered; these are
Vessel self-weight, including all attachments, internals, insulation, etc.
Any additional loads imposed by attached piping for example.
Wind loads
The above loads produce ‘Primary’ stresses in the vessel. It is these stresses that
are of most interest and must be within safe limits.
Also, of concern (but mostly for thinner wall sections) in columns, is the risk of
local shell buckling caused by axial compression in cylindrical sections. This can
be evaluated by hand calculations or analysis.
Thermal Stresses are classified as secondary stresses, and as such, are of far
less concern in terms of risk to stability. However, it is not unusual for thermal
stresses to be quite significant, and well in excess of yield.
WIND LOADING
Depending upon location, wind can produce a significant load on tall vertical
columns, or large diameter storage tanks.
The pressure produced by wind speed is proportional to the square of the
velocity, and taking a simplistic approach, and typical vessel design wind speeds
of 150km/hr, produces a pressure of 1.1kN/m2
. (For reference, API650 uses
1.4KN/m2
in assessing overturning stability of storage tanks).
A number of building codes provide detailed methods for establishing wind loads
for building design; these include detailed methods for establishing a 50yr mean
recurrence wind speeds, based upon specific site locations, which take account of
geographical location, and site-local conditions such as coastal locations;
however, it is not normally necessary to revert to these, as it is reasonable to
assume the design wind speed, which can be considered a pessimistic approach.
Although the safer approach is to consider design wind speeds, this can present
an overly cautious approach, as such wind speeds are based upon the entire
the design life of the vessel, whereas the heat treatment process will be in the order of
24 hrs. Furthermore, it is highly likely that should storm force winds were forecast
and foreseen, and any heat treatment work would be postponed in such
circumstances.
For typical columns, the overall effect of wind loading is that the leading edge of the
wind produces an overall bending moment on the column; this consequently results
in tension at the windward side, which actually counteracts the compressive forces
due to self-weight; however, conversely, on the leeward side, the opposite effect
applies, with the compressive load adding to the magnitude of the self-weight
stresses.
A simplistic approach to wind loading assessment can be made by establishing
the total force on a projected area of the column, and use simple beam theory
calculations to establish tensile and compressive bending stresses.
However, this simplistic approach does not take account for the pressure
distribution around a cylinder. Reference to BS 6399-2:1997 - Loading for
buildings. Code of practice for wind loads. This document provides load factors to
apply to the established wind pressure in order to establish typical pressure
distributions around a cylinder.
As can be seen from above, the pressure distribution around a cylinder varies
significantly, including significant negative areas (suction), though the effects of
this distribution needs to be assessed by analysis to see results in detail.
Whilst the effects of pressure distribution are less pronounced on smaller
diameter, thicker shells, the opposite is true for larger diameter, thin wall vessels
Where heat treatment involves the bottom dished head/ tan line, then the skirt/shell
weld is involved. This will typically require additional electrical heating to compensate
for the heat sink effect of the skirt, and also to prevent steep temperature gradients
at the skirt/shell weld that would place added strain to the skirt to shell weld.
The implications of this are that the top of the skirt sees full heat treatment
temperatures. However as the skirt is not part of the pressure envelope, it is usually
designed with a significantly thinner material (and sometimes lower strength
material). The skirt, therefore, is often the most highly stressed region of the vessel,
as it is supporting the bulk of the overall vessel weight, but also, under wind load, the
skirt is also the most highly stressed region, and this is where calculations/ analysis
usually need to focus on.
VESSEL TEMPORARY SUPPORT
Where calculations indicate that additional support is required to ensure a vessel
stability during heat treatment, options are usually limited to either craneage, or
hydraulic jacks.
Whilst either option can be engineered according to individual cases, with the load
applied being evaluated from calculations and load that is considered safe for the
the vessel, with the balance being taken by the jacks/crane.
However, it must be recognized that as the vessel expands, the jacks/crane must
be constantly monitored and adjusted in order to maintain a constant load,
otherwise, expansion may be sufficient to completely negate the initial applied load.
This requires the use of load cells with hydraulic jacks or inbuilt crane load
monitoring systems.
SPHERICAL STORAGE VESSELS
Spherical storage vessels present a unique but very standardized design, that often
requires heat treatment. The actual heat treatment process is relatively simple,
usually utilizing combustion based internal firing techniques.
Spherical storage tanks are supported on a number of columns that are generally
attached at the equator of the vessel. The implication of this is that during the heat
treatment, supporting columns are pushed out due to the radial expansion of the
vessel. However, typically, the base of the supporting columns are not designed to
accommodate movement, and may even be bolted via grouted bolts.
Thus, during heat treatment, without intervention, the top of the supports move
outwards, but the base is fixed; this will result in bending of the columns creating
bending stresses in the columns and at the crotch connection to the spherical shell,
ultimately risking the integrity of the column.
In order to remove this risk, standard practice is to provide temporary horizontal
jacking mechanisms at the base of each column, and establishing a jacking plan that
matches the movement of the base with that at the top of each column. This is
typically carried at regular temperature intervals during the heat treatment process,
such as every 50C temperature rise, each column is jacked out by the calculated
expansion for each column.
During cooling, the process must be carried out in reverse, with the jacking direction
reversed such that at ambient, the column centers are as original.
ELECTRICAL HEATING OF LARGE SURFACES
It is sometimes necessary to heat treat very large diameter vessels, - circ seams for
example.
Industry standard ceramic pad type heaters have typical power density capabilities
approaching 50kW/m2
. This usually allows the temperature in the order of 700°C to be
reached even without internal insulation. With this capability, and where internal
access may be limited, it may be tempting to carry out the heat treatment without
internal insulation.
However, for large vessel diameters, heat loss should be considered as being to
ambient, and the high levels of heat flux passing through thickness can generate
high-temperature differentials. For 25mm thickness, at 600°C, differentials are in the
order of 25°C, but this doubles to 50°C for 50mm wall thickness.
This assumes worst case heat losses to ambient; - for smaller sizes, heat losses are
reduced by cross radiant heat transfer, and internal bulk temperature build up.
Note that the application of standard thickness of insulation internally, drastically
reduces heat loss, and consequently, temperature differentials are only in the order of
1-2°C.
The heat treatment of tube sheets requires great care, especially if tubes are already
installed and welded. Tube sheets typically are relatively thick, with larger tube sheets
being in excess of 300mm, but the nature of tube sheet forgings is that there is
usually, little material provided to allow welding to the adjacent joining shells on both
gas side, and shell side.
As indicated above, the tube sheet to shell welds are typically very close to the
the main body of the tube sheet. Although it may be feasible to reach the required
the temperature at the weld using only externally applied heaters, and locally
applied insulation, to carry out such a heat treatment would impact hugely
thermal stresses in the tube sheet, and possibly distortion. This is because of the
unheated tube sheet center would remain relatively cool, resulting in a
the significant temperature differential between center and outer, thus creating
large thermal stresses.
Note also that where tubes are attached, not only can the rear face of the
tube sheet be insulated, but the tubes also act as an additional heat sink to the
tube sheet face.
Therefore, in order to overcome this issue, it is important to apply to heat to
the tubesheet face, such that the tube sheet is raised to a similar temperature
to the attachment weld during heat treatment. This ensures that the thermal
expansion is uniform, thus removing the risk of induced stresses and
distortion.
BAFFLE PLATES
On a similar principle to tubesheets, baffle plates can also complicate heat
treatments, if they are close to or in a heated zone in a vessel. If baffle plates are
bolted, then they should be removed, or otherwise, they will need to be heated and
temperatures controlled accordingly.
A perhaps more simple method of visualizing this is to remember that the
diametrical expansion of a cylinder is the same as that for a beam that crosses the
diameter – assuming the same temperature
In this simple example, - if the beam is at the same
temperature as the cylinder, then both will expand the same
amount. Conversely, if the beam is cooler, then it will have
less expansion, and, if attached to the cylinder, will constrain
the expansion
THIN WALL STORAGE TANKS
Due to their often very large diameters, thin wall storage tanks are generally
fabricated and constructed in-situ. Designs are often based upon foundations
including concrete ring walls. Ring walls are usually reinforced with rebar to ensure
hoop loads in service can be supported. Anchor bolts are also commonly grouted
into the ring wall prior to fabrication.
Although not pressure vessels, storage tanks often have caustic or sour service
requirements. Such service has implications for Stress Corrosion Cracking, and in
order to mitigate this risk, heat treatment is usually mandatory.
In-situ heat treatment of storage tanks presents a number of problems:
• Thermal expansion is usually restricted by the presence of anchor bolts
• Concrete will be damaged by exposure to high temperatures
• Any concrete rebar close to the surface of the ringwall will get hot, and
expand causing further damage to the concrete
• Further complications can also arise due to:
Bituminous sand infills
Secondary HDPE membranes close to the tank surface
PVC tubes close to tank underside for CP monitoring
Clearly, to proceed without taking action, would as a minimum be expected to cause
significant damage to the concrete ring wall, and buckling of the tank base adjacent
to anchor bolts. Hence, action must be taken to allow heat treatment to proceed,
though options are generally limited.
There are typically two options available:
Lift tank onto the temporary dry sand base. (For larger tanks, this is usually not an
option, due to the craneage capacity that would be required)
Lift/ jack up the tank, to allow placement of temporary insulated support beams.
These can be major operations and can extend site programs significantly.
The other two main areas of concern for storage tanks are a risk of buckling and roof
stability.
The two most common roof designs in tanks requiring heat treatment are domed,
and cone types. The domed roof design, although less common, is inherently
strong, often not requiring internal bracing; whereas almost all sizeable coned roof
designs incorporate radial rafters/ beams and the compression ring to support the roof –
such designs often require the use of temporary internal support structure to ensure
stability during heat treatment.
Note - that the use of the internal firing technique, and associated combustion air
blowers, generates an internal pressure within the tanks. Although pressures are
generally insignificant from a tank pressure viewpoint, - and typically in the order of
few inches water gauge; this pressure can significantly counteract the weight of large
roof areas. However, caution should be exercised in using the internal pressure as
part of any stability assessment, as an electrical failure at soak temperature could be
catastrophic if pressure is relied upon for roof support.
Finite element analysis is an incredibly powerful software tool that greatly aids the
assessment of most heat treatment engineering problems. Various analysis options
allow the user to establish temperature gradients on both static or transient basis,
thermally induced stresses, static stresses caused by self-weight (Gravity), and wind
loading, as well as buckling analysis, - which are the most relevant analyses in
relation to the subject herein, though the software has much more capability.
Although FEA software is becoming increasingly affordable, with some functionality
even being shipped with higher-end CAD software, including powerful user
interfaces, it is all too easy for engineers to undertake analysis work, without fully
understanding the complexity of the many aspects related to this work. It is
imperative that any analysis has appropriate boundary conditions, elements, mesh
size, assumptions, and applied loads. It is equally important that any results are
checked, and validated before data is relied upon. It is therefore very important that
any company undertaking such work is competent, and experienced in this type of work.
No comments:
Post a Comment