Documentos de Académico
Documentos de Profesional
Documentos de Cultura
April 2003
Background to the Design of HRSG Systems and Implications for CCGT Plant
Cycling
Fred Starr, European Technology Development , Surrey, UK
Mr Fred Starr works as Principal Materials and Energy
Consultant with European Technology Development. He is
currently working on projects dealing with the cycling of
power plant, materials for waste incineration, biomass
combustion, and the development of high efficiency micro
cogen systems. He previously worked at ERA Technology on
high temperature corrosion and is a recognised expert in this
field. Prior to this much of his earlier career was spent at
British Gas working on advanced energy conversion systems
including coal gasification processes and indirect fired gas
turbine plant. fstarr@etd1.co.uk
Abstract
The principles which underlie the design of HRSG boilers for CCGT plant are outlined together
with their effect on the susceptibility of HRSGs to plant cycling. The interaction between gas
turbine and HRSG thermodynamics are described. Despite the marked increase in gas turbine inlet
temperatures that has occurred over the past twenty years, there has only been a very slow rise in
flue gas temperatures into HRSGs. The consequences of low flue gas temperatures is that HRSG
plant is large in terms of steam generating ability and have also resulted in the need to generate the
steam from two or more separate evaporators at different pressures. Boiler pinch point temperatures
need to be in the 4-12C range for efficiency reasons, with most of the tubing in the HRSGs being
finned for this reason
It would seem that the preoccupation with heat transfer and HRSG size reduction has caused some
additional problems. These are now well understood by designers and operators. However even
during normal start ups HRSGs will be susceptible to thermal fatigue and flow assisted corrosion. In
older units where condensate drainage was poor, or where restarts after plant trips were excessively
quick, severe thermal fatigue occurred due to blocking of tubes and condensate quenching of
headers.
Although the increase in HRSG steam temperatures has been fairly slow, they are now reached the
point where P91 steels are being used in the superheater sections. The benefits that this class of
alloys confers and its potential shortcoming are briefly described. There are some other materials,
which are coming onto the market, which may be beneficial in construction.
1.
Introduction
Since 1976 no coal fired plant have been built in the UK. Every power plant has been of the CCGT
(Combined Cycle Gas Turbine) type using natural gas as a fuel. To simplify, a CCGT consists of a
gas turbine, which produces about two thirds of the power of the plant. The exhaust from the gas
turbine is used to produce superheated steam, in a HRSG or Heat Recovery Steam Generator, which
then feeds into a steam turbine set, to produce additional electrical power from the plant. On
modern units the steam and gas turbine are linked on the same shaft. Older designs might have one
big steam turbine running off the steam from a number of HRSGs, which were in turn fired from a
number of small gas turbines.
When older power engineers get to hear about the HRSG part of the CCGT system, I am sure many
of them have tended to dismiss this part of the plant as being just another steam boiler, and
concentrated on the more exciting part of the technology; that is the gas turbine. Indeed, apart from
specialised meetings on HRSGs, such as the one being organised by ETD Ltd in November of the
this year (2003), the issues connected with HRSG materials and construction tend to be neglected
by the mainstream conferences on boiler plant and steam turbine design, particularly in Europe. In
consequence the idea for this background article emerged during two recent surveys of plant cycling
by ETD Ltd, one on steam plant and the other on CCGT plant. This paper represents a personal
review of HRSG design, garnered as much from person contacts as from the literature, and for that
reason, a bibliography has been included, rather than sets of references, although where relevant
some individuals are mentioned by name.
Exhaust from
Gas Turbine
1 Inlet Duct
2 Distribution grid
3 HP Superheater 1
4 Burner
5 Split Superheater
6 HP Superheater 2
7 CO Catalyst
8 HP Steam Drum
9 Top Supports
10 SCR Catalyst
11 LP Steam Drum
12 HRSG Casing
13 Deaerator
14 Stack
15 Preheater
16 DA Evaporator
17 HP/IP Economizer
18 IP Evaporator
19 IP Superheater
20 HP Economizer
21 Ammonia Injection Grid
22 HP Evaporator
In an HRSG, it has been flue gas side heat transfer rather than materials, which has governed the
design. A pessimistic view of heat transfer rates could result in a unit, which is too expensive, as
even well designed HRSGs tend to be big in proportion to their output, because of the poor heat
transfer rates and the need to minimise pressure drops along the duct. The positioning and
dimensioning of heat exchanger surfaces is critical, as mistakes could lead to an HRSG working
extremely badly.
This need to concentrate on heat transfer has meant, some would argue, that plant operability and
the consequential materials problems have not been given the prominence they deserve. The
purpose of this article is to indicate the governing factors in HRSG design, and how these have
affected failure mechanisms in HRSG systems, particularly in plant, which is subject to two shift
operation, that is the starting up and shutting down of a plant on a daily basis.
considerations come into play, a relatively low pressure ratio gives a low inlet temperature into the
combustion turbine, so that more fuel can be burnt. Figure 3, based on some simple calculations,
shows how the effect of increased pressure ratio tends to hold down the turbine exhaust
temperature, so that above a pressure ratio of 15 the rise in outlet temperature is very slow. In fact if
pressure ratios were fixed outlet temperature would rise in direct proportion to the inlet temperature,
and gas turbines working at high temperature would be very inefficient.
1600
1400
Exhaust
Temperatures
1200
1000
Inlet
Temperatures
800
600
400
5
10
15
20
25
30
Pressure Ratio
Fig 3: Combustion Turbine Inlet and Exhaust Temperature versus Pressure Ratio
This approach also benefits heat transfer into the HRSG, which is yet another reason for the
differences between industrial and aerospace machines. A low pressure ratio will result in a fairly
high outlet temperature in to the HRSG, which become a vital factor in achieving reasonable rates
of heat transfer when heat transfer coefficients are low. Low pressure ratios are advantageous in
other ways. HRSG size is held down since, because of the high specific work, the volume of
exhaust flow is reduced. HRSG efficiency is also improved due to the reduction in the levels of
excess air going into the gas turbine. Even so, the excess air is about 3 to 4 times that needed for
complete combustion. This mass of excess air, when it leaves the HRSG, carries with it a sizeable
amount of heat energy, even though a typical stack gas temperature is only 100C. In effect, the
stack or chimney of a CCGT plant, which is of an excessive diameter when compared to that of a
steam plant, represents inefficiency in just the same way as does a cooling tower. Any reduction in
excess air will therefore improve efficiencies.
In practice the designers of gas turbine systems have quite a lot of latitude in deciding how much
power to take out of the gas turbine, so that as Figure 4, taken from manufacturers data, shows
exhaust temperatures have tended to group within the 500-600 C band, with just a slight indication
of an upward trend with pressure ratio. The two red points refer to aerospace derivative gas
turbines, which because of the extremely high-pressure ratios have quite low exhaust temperatures.
Such units are best used for standby power generation, the driving of pipeline compressors, or in
niche installations where there is a requirement for relatively low grade steam. The sky blue square,
indicating an exhaust temperature of 640C at a pressure ratio of 30, is showing, perhaps, the shape
of things to come, and refers to the GT24/26 ABB/Alstom reheat design of gas turbine.
Exhaust TemperatureC
700
650
600
550
500
450
400
5
10
15
20
25
30
35
40
Pressure ratio
3. HRSG Geometries
The HRSG evaporators, boilers, superheaters and, where they are fitted, reheaters are located in a
very large rectangular duct. To reduce the velocity of the gas turbine exhaust, the duct opens out in
a vee shaped manner, in both the horizontal and vertical directions, after the gas turbine, otherwise
the pressure drops through the HRSG would be excessive. Swirl vanes are located after the turbine,
as the flue gas will tend to corkscrew up the duct with some force, particularly when the gas turbine
is working off design.
The HRSG section of the ductwork can be laid out parallel to the ground, making the HRSG a
horizontal type model. This type of duct leads into a stack, which will be equipped with silencers as
was shown in Figure 1. The alternative vertical form of HRSG incorporates the boilers and
superheaters within the stack structure. It follows that in a typical horizontal HRSG the heat transfer
tubing is transposed in vertical harp type arrays across the duct. In the vertical form of HRSG the
tubing can be in the form of loops which cross the duct horizontally as in a steam plant tower boiler.
Where harps are used, these too will cross the duct in a basically horizontal manner. It will be clear
that these two arrangements will drain and fill quite differently.
The ducting for HRSGs is big because of the huge amounts of air and combustion products going
through the gas turbine , and is therefore a design and maintenance issue in its own right. Mass
flows for big gas turbines are over 600kg/s, but to put things in more concrete terms, just one of the
big industrial gas turbines has more than enough power to propel a 747 type Jumbo Jet. Horizontal
HRSGs can be up to 25 metres high and are up to 60 metres long. Thermal expansion of the
ductwork is a significant design issue because of the size and the need to preserve the integrity of
the internal insulation. With vertical HRSGs the flue gas has to turn through a right angle up into
the stack soon after it has left the gas turbine. Very little heat will have been lost from the flue gas at
this point and it is an area where thermal expansion problems can become quite difficult.
HP Boiler
and
Superheater
LP Boiler and
Superheater
To
Stack
From Gas
Turbine
HP
Turbine
LP
Turbine
Alternator
This problem is overcome by installing a further set of evaporators, economisers and superheaters in
the HRSG, down stream of the high pressure system. This additional set produces steam at a much
lower pressure, somewhere between 4 and 10 bar, the actual value being that which corresponds to
the exit pressure from the HP turbine, or if the plant is fitted with a reheater, the IP turbine. As the
boiling point of water at this sort of pressure is in the range 140-180 C low temperature heat in the
flue gases can be picked up quite easily. In addition some superheating of the LP steam is done
with the aim of matching the temperature of the cold steam from the HP or IP turbine. It is then
possible to merge the steam from these two different systems and put them to the LP turbine as
shown in Figure 5.
Steam, for deaeration, at an even lower pressure, can also be raised using the last vestige of heat in
the flue gases. To summarise in a modern HRSG steam would be raised at three different pressures
with separate evaporator and pumping circuits for each. This is in complete contrast to a pulverised
fuel steam plant, where all the evaporation takes place at one very high pressure. Another difference
is that feedheaters, of the type used on conventional pulverised fuel plant, using steam extracted
from the turbines are not used on HRSG systems. This would simply raise the water temperature at
the inlet to the economisers, and would reduce the amount of heat, which could be transferred from
the flue gases.
4.2 Pinch and Approach Point Effects
In extracting heat from the flue gas, the ideal is that the flue gas temperature should drop at a
constant rate all the way down the duct. This would be simple to arrange if one was trying to heat
another gas, or a liquid which did not evaporate as it temperature increased. It would then be easy to
keep a constant temperature difference between the flue gas and the fluid being heated at every
point.
This is not too difficult in the economiser or superheater. In the former, the water stays as a liquid.
In the latter, it stays as steam. It is impossible to maintain a constant temperature difference in the
evaporator. Here as the water passes the tubing, the water turns to steam, with the temperature
staying the same all the way through.
There is no great problem with heat transfer at the exit to the evaporator. Assuming that the
evaporator in question is at 100 bar pressure, the bulk water temperature at the exit will be the value
stated previously, 311C, with the tube wall temperate being just above this. The flue gas
temperature at the entrance might be 450C, so that the temperature differential would be around
70C. As the flue gas moves through the evaporator section, it drops steadily in temperature,
although the tube wall temperature in the evaporator would be just above 311C all the way
through. In theory, if the evaporator unit was infinite in size, there would come a point where the
flue gas and water temperatures would be equal. This would give a pinch point of 0C. In practice
to keep the size of the evaporator realistic, the designer will adopt a pinch point of somewhere
between 4and 8C. This is far lower than can be achieved in an HRSG on a refinery and it is one
reason for the size of HRSG systems.
The flue gas leaving the evaporator is likely to enter the economiser. If the same temperature
differential between the flue gas and the water were maintained, as in the evaporator, boiling would
tend to occur at the exit to the economiser. This would result in a steaming economiser with the
possibility of the flow being blocked. It will be apparent that if the water flow were to be become
blocked, all the water in the economiser would begin to turn into steam. Tubes would be subject to a
mild degree of overheating but the main problems would be water hammer and tube-to-tube
differential expansion. Accordingly, the outlet water temperature in the economiser is kept several
degrees below the saturation temperature. This can be done in several ways, for example by cutting
down the economiser heating surfaces and interposing part of the LP superheater system between
the HP evaporator and economiser. The effect of this is to give an approach temperature between
the flue gas and the evaporator outlet water temperature. The approach temperature might be 4C,
so this when added to a pinch point temperature of 8C, would give an overall temperature
differential of 12C, greatly reducing the risks of boiling. See Fig 6.
600
Flue Gas
Pinch
and
Approach
Point
Temperatures
500
400
Superheater
300
Evaporator
Economiser
200
C
Figure 6: Schematic showing pinch and approach points for the HP superheater, evaporator
and economiser tube banks
As might be expected with such tight pinch points a slight change can have massive effect on boiler
costs. Van Manen indicates that in going from an 8C to a 6C pinch point, the heat transfer surface
would need to increase by about 10%. However the effect on steam turbine output, which is only
about a third of the overall power output of the plant, is quite marginal, giving an increase in power
output of 1% or about 1MW for the case described. This may not be the complete story, as
extracting more heat in the HP evaporator would reduce the amount of heat downstream of this unit.
For this reason, that is heat taken out by a tube bank located in the high temperature section will
result in less heat being available in a lower temperature section, an HP economiser needs to be split
into two or more sections. These cooler sections would be located much further along the HRSG,
after the IP and LP evaporators. In this way a good amount of heat is available for heating of IP and
LP steam, with lower grade heat being left for low temperature sections of the economisers. It will
also be apparent that all the water in the system has to be heated, from near ambient temperature,
after it has been returned from the condenser, with separate boiler pumps feeding each part of the
system.
Superheater
Economiser
There will therefore be some conflict in the location of the HP, IP and LP economisers and LP and
IP superheaters. To simplify the construction, a siamised or common header may be used, with one
half of a cross duct exchanger being made up of an economiser and the other half, the superheater.
During startup the superheater tubes are likely to heat up faster, as they will be free from steam at
this point in time, whereas the economiser will always be full of water. The superheater tubes will
be put under a compressive load; the economiser tubes will be strained in tension, as shown
schematically in Figure 7. Figure 8 shows a before and after picture of what happened after the
bottom header was cut through at the point where the two headers had been joined
Flue Gas
Water in
Economiser
Water in
Evaporator
HP Steam
50
500
2500-10000
1000
It follows from this that tube wall temperatures tend to run quite close to the water and steam side
temperatures. Even when temperature differentials are at their highest, the heat transfer rates are
very modest. For example with a 100C difference between the flue gas, as at the beginning of an
HP evaporator, the actual heat transfer rate will be not much more than 5-10 kW.m-2. This compares
with peak heat transfer rates in pulverised boilers which can be well over 200 kW. m-2.
Given such poor flue gas side heat transfer rates, tubes must be of small diameter, with tight
spacings and be of the finned type to provide sufficient heat transfer area. Figure 9 shows a set of
photographs taken during the refurbishment of an evaporator by EH.Wachs Inc, which illustrates
10
just how tight everything is. The only section of the HRSG which might not use finned tubes is the
HP superheater where there might be a possibility of oxidation of the finning.
11
Log Heat
Transfer
Rate
Mode of Boiling
Water Only
Water plus
Steam Bubbles
Steam
Fig 10 : Modes of Flow in Water and Steam- Water Mixtures and Effect on Heat Transfer
The once through boiler is sometime termed the Benson type, wrongly it would seem if one goes
back to Bensons 1930s conception. His ideas were based on the generation of steam under
supercritical conditions, that is at pressures above 221 bar. Most once through designs, particularly
for CCGT plant are subcritical. Here, as the temperature changes, the water in the evaporator
section changes from water, in which steam bubbles are generated at the wall surface, through to
steam going through several different flow geometries in the process. See Fig 10. During base load
operation the steam exiting the evaporator section will have a mild degree of superheat and should
enter the superheater proper in a completely dry condition. During start up, the steam from the
evaporator will not only be saturated but will also contain free water. This needs to be removed in a
separator.
It will be seen that the changes in the modes of flow affect the heat transfer rate, the highest rate
being when steam bubbles are tending to nucleate throughout the tube cross section. This is termed
nucleate boiling. But at some point what is effectively a column of water with some steam must
give way to a tube full of steam, which is on its way to becoming superheated. The heat transfer
designer needs to ensure that steam blanketing does not occur, as shown in the last figure, since heat
transfer rate will drop precipitously and tube temperature will rise. This is termed departure from
nucleate boiling(DNB). Fortunately the extreme tube overheating that does occur in steam plant
evaporators, and which actually promotes this dry out condition is unlikely in HRSGs, as tube
wall temperature and external heat transfer rates can never reach really dangerous levels.
However, during start up, the flow in the IP and LP economisers and evaporators can be of the two
phase water/steam type, in both once through and drum types of HRSG. This results from the fact
that since little HP steam has been generated, the flue gases reaching the back end of the HRSG will
still be quite hot. There is then some potential for damage by water hammer and the development of
tube-to-tube temperature differences.
Water hammer is an obvious phenomenon. It can be heard! Once through boilers are subject to a
more insidious form of damage called Flow Assisted Corrosion (FAC), since they have to use an
All Volatile Treatment (AVT), to avoid deposition of solids in the evaporator. Unfortunately the use
of AVT tends to lead to the dissolution of protective magnetite. Tube surfaces are then susceptible
12
to erosion by two phase, high velocity flows, leading to FAC, as can occur in the low pressure units
of the HRSG. Oddly enough the HP sections of an HRSG are unlikely to be affected by FAC since
the difference in specific volume between steam and water falls rapidly at high pressures. It is this
concern about steaming and the consequential FAC, which has meant that some designers prefer to
stay with a steam drum on the LP units.
13
Fig 11: Flow situation in bottom header. Note that drainage space is highly restricted
This is fine during base load operation, but during plant trips and subsequent purges of the HRSG
duct to remove inflammable gases, condensation of steam will occur in the HP superheater. The
condensate must be drained off before the plant can be properly restarted. Narrow diameter bottom
headers of the type shown are essential unsuitable for this task. Figures 12 and 13 shows the kind of
build up that can occur, in one case with the drainage point in the centre, in the other with the point
at one end. Figure 14 shows that if the drain point is too small, the condensate will back up
uniformly to a near constant depth across the header.
14
Figure 13: Small diameter header with end type drain point
If the condensate is not fully drained off, the subsequent performance of the equipment, during
restart, becomes an interesting exercise in heat transfer and damage to plant! The upper part of those
tubes that are blocked with condensate will tend to overheat, as there is no steam flow. Tube to tube
temperature differences will develop leading to stressing at the tube-to-antler connections. If the
welding of these is of the partial penetration type the life of these could be very short.
15
Eventually the condensate will start to boil and, due to the generation of steam, relatively cold
condensate can be carried over as a mass, quenching the inside of the header. Heat transfer rates in
such situations have been estimated at 8000 kW.m-2. Interligament cracking is a real possibility, and
again the close spacing of the connections and use of headers with a heavy wall thickness will not
help.
Whether or not carry over of condensate is a problem can be ascertained from the changes in steam
temperature during start up. If the increase in temperature is erratic, with sudden drops occurring,
the chances are that condensate is being carried over. A longer time must be allowed to drain off
condensate, and the ramp rate of the gas turbine should be reduced.
16
7. Conclusions
The aim of this paper was to outline the thinking behind the design of HRSG systems and to explain
some of the background to the engineering and materials problems that have been encountered. The
paper has tried to show how the design has been dominated by heat transfer concerns, and that this
has led to units that are difficult to build, inspect, and repair. The preoccupation with heat transfer
has led to clever cost saving design solutions, such as the use of small headers and the neglect of
system drainage. Also, one feels, heat transfer based calculations, may be leading to some over
confidence in the promotion of ramp rate based life assessment programs. Fortunately the problems
with condensate drainage, and the serious effects that this can have on plant, are now well
recognised by both designers and plant operators. The gradual increase in gas turbine exhaust
temperatures should give designers more latitude, so that it should be possible to build in more
flexibility in load following capability.
The paper has also touched on the background to flow assisted corrosion in HRSG boilers during
start up, which again can be traced back to issues associated with heat transfer. FAC is a serious
problem in air-cooled condenser plant, where the use of low chrome steel seems to be the solution.
It may be that this could be an alternative to sophisticated water treatment approaches, which can be
difficult to apply in plant that is two shifting.
The move to higher temperatures and pressures will also imply the need for improved alloys, with
P91 being the material of choice at present. This seems a much better option than the older grades
of steel despite its known and potential shortcomings, but this is something which needs to be
properly evaluated.
Finally the author would like to say, that despite the under current of criticism on the over zealous
application of heat transfer technology, as a long term member of HEXAG, a UK based heat
transfer body, he has the greatest admiration for specialists in this field, especially those working
with compact systems of the HRSG type. If anything, this paper was written with the aim of getting
generating plant personnel and materials specialists to understand the hard work which has gone
into making HRSGs amongst the most efficient recovers of energy that are operating in the world
today.
Acknowledgements
The views in this paper are those of the authors and are not necessarily those of ETD Ltd. However
the author would like to thank Dr Shibli of ETD Ltd for his encouragement in writing this paper,
and Mr A. Fleming of ETD for his helpful comments.
The author would also like to single out Mr M. Pearson of Pearson Associates and Dr D. Bogart of
Innogy for their helpful discussions on HRSG mechanical design and issues connected with HRSG
plant cycling.
Bibliography
D G Wilson The Design of High Efficiency Turbomachinery and Gas Turbines MIT
Publications 1984
17
Anderson R A and Pearson M, Reliability and durability from large HRSGs, pp 21-46, CCGT
Plant Components: Development and Reliability, Professional Engineering Publishing Ltd., 1999.
D Bogart Dealing with the Rigours of Cyclic CCGT Operation: An Operators Perspective pp
33-45 Modern Power Systems Oct 2001
A Van Manen HRSG for Optimum Combined Cycle Performance Paper 94-GT-278 1994
Int. Gas turbine and Aeroengine Congress and Exposition, The Hague, Netherlands, June1994
A Fleming, R V Maskell, L W Buchanan and T Wilson Materials Developments for
Supercritical Boilers and Pipework pp 33-78 Materials for High Temperature Power Generation
and Process Plant Applications, ed Strang Institute of Materials 2000
HEXAG (Heat Exchanger Action Group): This is a UK/Europe based organised comprising
people and organisations involved in compact and two phase heat exchanger design and
construction www.hexag.hw.ac.uk