P auerkari-mom paper1.doc

Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003
Factors influencing the performance and reliability of tubular boiler components
Pertti Auerkari, VTT Industrial Systems, Espoo, Finland.
Olavi Lehtinen, Fortum Service, Naantali, Finland
Abstract

Pressure and thermal loads as well as wall thinning due to erosion, corrosion and oxidation inflict a
gradually accumulating damage to boiler tubes, which therefore have an inevitably limited
technical life. Design attempts to take this into account to provide for the assumed minimum life
expectancy, impact in a way that can guarantee low maintenance costs. However, in addition to
occasional deficiencies in design or manufacturing, the actual service conditions can differ from
the initially assumed conditions or change in time, introducing damage that was not accounted for
in design. In-service inspections, assessment and maintenance are the main tools to update the
information on the condition and predicted technical life. Power plant boilers include a large
number of tubes exposed to a variable combustion environment, and reliable updating is
particularly required when approaching the expected end of technical life. By that time any
manufacturer's guarantees have been expired, and much of the technical challenge rests on
maintenance. The practical means to manage boiler life are based partly on the measurement of the
wall thickness of tubular components. The results from these measurements as wall thinning rates
can include considerable uncertainty, with inevitable consequences to predicted life and
maintenance planning. The uncertainty tends to be highest for relatively new plant, hiding useful
indications of potential improvement in the boiler operation.
Critical aspects in condition and life assessment of boilers

In addition to routine inspections and maintenance, assessment of boiler condition and life could
be motivated by anticipated need for design modifications, changes in the way the boiler is
operated (e.g. increasing cycling or start-up rates), observed defects from fabrication, repairs or
service, tube failures or operation far beyond the design life. In particular, changes in operational
conditions cannot be fully known during the design stage, and at least some unexpected changes
are likely to occur during the relatively long lifetime of a typical boiler plant [1, 2].
Boiler tubes are affected by parallel life-limiting damage mechanisms from the combustion
environment, corrosive effects of the water/steam side, and mechanical and thermal loading.
Therefore, boiler tubes are subject to:
- fir eside erosion, corrosion and deposit formation
- water/steam side corrosion (also during shutdown periods)
- fatigue and corrosion fatigue driven by variable mechanical and thermal loads; and
- creep and creep fatigue in areas of high material temperatur e.
Nearly all failure mechanisms are accelerated by increasing temperature, and therefore
temperature control is one of the main issues in maintaining simultaneously high efficiency and
availability for the foreseen lifetime of the boiler. However, to be able to set the operating
conditions to an ideal level, the operator should have an indication of the consequences of his/her
action of control. As quality of fuels can differ in time, boiler surfaces can show different levels of
deposit formation and the tubes accumulate increasing internal oxide thickness, the thermal loads
Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 do not remain constant even if firing does. As generally there is incentive to extract the same power, the thermal balance within the boiler may gradually shift to increase the temperatures at some locations, shortening life of these areas. Because of the severe conditions inside the boiler, the continuous monitoring is mainly done from the fuel and water feed, flame control and water/steam exit. For the purposes of prediction, however, much of the boiler condition monitoring still relies on off -line inspections, which are largely thickness measurements from the boiler tubes. The boiler life can exceed the career span of at least one generation of the plant and support service personnel within in the tasks related to maintenance and life of a given boiler. Therefore, it is not self-evident that there would be experienced personnel to cover the past O&M history of the boiler for well justified predictions and decisions for future. Considering risk (i.e. both probability and consequences) of unplanned plant shutdown, boilers firing solid fuels are among the worst components or subsystems of any power plant (Fig 1). SYSTEM FAILURE RATE (1/h)
MWh LOST PER FAILURE AND 1 MWe OF CAPACITY
Fig 1. Mean risk of major components to cause unplanned lost production in power plants, estimated from data of [1]. The descending lines correspond to constant risk; in comparison, coal fired boilers represent highest risk of lost power. Boilers using solid fuels may of course also differ in many aspects. For example, in relatively large coal fired plants used for base load power generation or CHP, the boiler internals can have a required and achievable life of 10 years or more. In contrast, in waste incineration even yearly replacement of some superheater sections can sometimes make economical sense and be therefore acceptable. However, in most cases it is of interest to determine the initial condition, the rates of degradation and the control functions to optimise the operation and maintenance plans for the foreseen economical targets. As there are several sources of uncertainty (Table 1), the assessment Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 can only provide useful results when the uncertainty does not excessively hide the true underlying trends. Table 1. Sources of uncertainty and their characteristics in condition and life assessment. Sources of uncertainty
Characteristics
Load and temperature history, firing, fuels Detection, sizing, interpretation, false calls Note that human factors enter many "technical" sources of uncertainty, supporting the rather unsurprising truism that personnel and training are important for success. Common types and locations for boiler damage are shown in Table 2. Table 2. Common types of boiler damage. Principle of finding
locations
maximum damage
Erosive/corrosive wall thinning High temp/gradients/flow The present paper will particularly consider boiler tube wall thinning and the uncertainty in its measurement in condition and life assessment of boilers. Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 Wall thickness and estimates of wall thinning rate

The simplest common estimate of the wall thinning rate is taken from
where s and s0 are the current and initial wall thickness values at corresponding points of time t and t0. In reality this rate is not necessarily constant in time because of shifts in e.g. fuel, firing and deposit formation. However, for any reasonably comparable sets of measurements, the estimates from the equation (1) are clearly affected by the uncertainties in measuring the wall thickness. The time to failure for a boiler tube under internal pressure can be predicted by solving for tr from where h is wall thickness, r the stress exponent of the inverse Norton law tro = Aσ-r and tro the corresponding time to creep rupture under similar conditions without wall thinning. The resulting time to failure depends on tube size, material and operating conditions (pressure and temperature history), but wall thickness and wall thinning rate are among the most influential factors to the predicted life. Uncertainty in C is likely to be highest when the tubes are relatively new and ∆s relatively small, because then the uncertainties both in measured (current) wall thickness and in the initial wall thickness can be large in comparison with ∆s. When the tube is as-new and delivered to comply with the standard dimensional tolerances (Table 3), the wall thickness should generally not be less than 9 to 12,5% below the nominal wall (depending on tube size and chosen standard), and not more than about 9 to 15% above the nominal wall [3]. The corresponding tolerance for tube diameter is of the order of ± 1% (min ± 0.5 mm), but this has much less influence on life than wall thickness. Table 3. Standard tolerances for wall thickness of boiler tubes (DO ≤ 130 mm, DIN 17175) Outer dia
DO ≤ 2⋅sn
2 sn < D O ≤ 4 sn
DO > 4 sn
For example, assume that a particular section of a boiler could show wall thinning from the initial nominal wall of 6.0 mm to a specified minimum allowed level of 3.5 mm in an expected life of 100000 hours. This translates to C = (6-3.5)/100000 = 2.5⋅10-5 mm/h on average, or about 0.2 mm wall loss per year assuming 8000 service hours per year (base load operation). If there is no other information of the actual initial wall thickness at the location of interest than the nominal wall thickness, ± 9-12,5% tolerance can cover a wall thickness range from 5.46 to 6.75 mm. In addition, if after a few years of service the wall thickness is measured by ultrasonic testing (UT), the measurement itself adds a component of uncertainty. Because of inherent uncertainty in the technique, geometric scatter from sensor and tube alignment and possible material loss when cleaning the tube surface for the measurement, the resulting measured wall thickness will be Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 known to within some ± 0.2 mm even if no pitting or similar localised discontinuities occur. It is then clear that without some knowledge of the initial (measured) wall thickness, it would take many years before the actual wall thinning rate can be discerned. However, even the initial wall thickness values of new tubes are often measured (if they are measured and recorded) with the same technique. This narrows down the uncertainty in the initial wall thickness to about ± 0.2 mm, assuming that the whole range (i.e. extreme value statistics) is used. Again, at least a couple of years would be needed to see the actual wall thinning rate to predict meaningful safe life. More accurate wall thickness measurement is possible by destructive tube sampling, but this is limited in scope. In general, wall thickness measurements are useful in a relatively new or recently replaced boiler section only when much faster than expected wall thinning occurs, typically of the order of 0.5 mm/year or more. Then it may make sense to inspect the wall thickness, to see possible strong deviations from the expected trend. The situation is usually easier in older plants or when using corrosive fuels, because the measured wall thickness is then likely to be much less than in the as-new state, and safe side assumptions on the initial wall can be used with better confidence. The results are generally useful if the resulting predicted safe life is provided as multiples of the periods between possible maintenance campaigns (usually at least about one year). However, correct sampling for measurements remains important (Fig 2), because boiler surfaces are very large and wall thinning (and other factors) can vary considerably within the boiler (Fig 3). Fig 2. Schematic arrangement of boiler tubes in a 135 MWe coal fired plant. Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 The result from life assessment point of view is that the predicted life involves less relative uncertainty in an old (or fast thinning) boiler, but again it would be of practical advantage if measurements are timed so that at least one year to the next maintenance campaign will be predicted as the minimum safe life. These features are not affected by e.g. the relatively new European boiler assessment standa rd EN 12952-4. To be able to use all the available measurements to their full potential, it is naturally important that data storage and management can handle this over relatively long periods of time and changes in personnel, repairs and replacements. P (cumulative)
Wall thickness (mm)
Wall thickness (mm)
Fig 3. a) Measured initial wall thickness distribution for a boiler section with a nominal wall thickness of 8.8 mm (mean 8.74 mm, median 8.7 mm, SD 0.2 mm); b) wall thickness variation in the waterwall after more than 100000 h of service (not the same boilers). In conclusion, it can be harmful to use wall thickness measurements from a relatively new boiler unless the factors of uncertainty are properly compared with the results. Old or relatively fast thinning walls are easier to handle but then it is important not to measure too late. As wall thickness measurements alone are coarse, they have poor control (monitoring) value particularly for new plant where this would be most valuable.
Summary

Both for new boiler plant and maintenance investment, the return is partly measured by the degree
of achieving or exceeding the plant performance and reliability targets. The excellent ideas can in
boiler practice degenerate into hard work on dirty tubes, and often a significant fraction of this is
related to performing and interpreting wall thickness measurements. Particularly challenging
aspects may be involved in wall thickness measurements from a relatively new boilers unless the
factors of uncertainty can be properly managed. Old or relatively fast thinning walls are easier to
handle, if assessment is done in time. Wall thickness measurements alone can have poor control
Performance, reliability of tubular boiler components OMMI (Vol. 2, Issue 2) Aug. 2003 (monitoring) value for plant operation and must be complemented with other more indirect indications of the boiler condition and life expiration rate.
References

[1] Balkey et al, 1994. R isk-based inspection - development and guidelines. Vol 3. Fossil-fired
electric power generating station guidelines. ASME Research Report CTRD-Vol 20-3, ASME, New York. 177 p. [2] Jovanovic, A., Auerkari, P., Brear, J. & Lehtinen, O., 2001. Risk-related issues in life management of power plant components: inspection, monitoring, code -based analysis. Proceedings of Baltica V Conference on Condition and Life Management for Power Plants. June 6-8, Porvoo, Finland, Vol 2, p. 427-448. [3] DIN 17175, 1979. Nahtlose Rohre aus warmfesten Stählen (seamless tubes of heat resistant steels). Verlag Stahleisen, Düsseldorf.

Source: http://www.ommi.co.uk/PDF/Articles/44.pdf