- Email: [email protected]
Down-time corrosion in boilers
FUPROC-04745; No of Pages 9 Fuel Processing Technology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Down-time corrosion in boilers Frida Jones ⁎, Daniel Ryde, Anders Hjörnhede SP Technical Research Institute of Sweden, Energy and Bioeconomy, Box 857, SE-501 15 Borås, Sweden
a r t i c l e
i n f o
Article history: Received 22 March 2015 Received in revised form 24 September 2015 Accepted 3 October 2015 Available online xxxx Keywords: Down-time corrosion, Online-corrosion probe, Deposits
a b s t r a c t Down-time corrosion can occur on boiler surfaces, e.g. furnace walls, superheaters, or economizers that are covered with hygroscopic deposits, when the temperature drops under 100 °C. This scenario takes place when a boiler is shut-down for cleaning, maintenance, or other reasons, such as unplanned shut-downs. Initially, the dry deposits will absorb moisture from the surrounding air, potentially creating a corrosive environment. After this, corrosive acids can form in the deposits. In this study modified online-corrosion probes were used in combination with deposits taken from 6 different boilers at various locations (for example, from the furnace, the superheater, and the economizer), where the fuels have been waste, demolition wood or biomass. The deposits were ground and dried in an oven at 160 °C for several hours before exposed to a moist environment (RH 65%) during online measuring of the corrosion rate and the pitting activity. Four types of alloys were tested: low-alloy ST45.8-steel, 9% Cr ferritic P91-steel, austenitic stainless steel 304L, and Ni-based super Alloy 625. The results for ST45.8 show that in biomass boilers a corrosion rate from negligible values up to 0.7 mm/year can be reached within a week, while waste-fired boilers can have rates as high as 1.8 mm/year. Furthermore, for some samples from waste-fired boilers show a high pitting activity already after 24 h. The tests with the P91-steel show values up to 0.16 mm/year, for samples from different locations in the boiler. For 304L and Alloy 625 the downtime corrosion was negligible even after a two-week exposure. The ability to follow the down-time corrosion online has provided data that show that even though thought to be negligible, the risk of down-time corrosion is of significance, especially if the fuel is waste. Also, even if the initial corrosion rate is low, it increases during the first 24 h due to the exposure to moist environment, motivating immediate cleaning of the boilers after shut-down, especially on surfaces of lower steel quality. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Down-time corrosion can occur on boiler surfaces, e.g. furnace walls, superheaters, or economizers that are covered with hygroscopic deposits when the temperature drops under 100 °C. This scenario takes place when a boiler is shut down for cleaning, maintenance, or other reasons, such as unplanned shut downs. Initially, the dry deposits will absorb moisture from the surrounding air, creating a potentially corrosive environment. Furthermore, depending on the chemical composition of the deposits, it is possible that corrosive substances are formed in the deposits when in contact with moisture. Down-time corrosion can also occur if improper procedures are used during boiler shutdown; this is especially true for down-time corrosion on the economizer. To reduce the effects of down-time corrosion it is common to clean the boilers as soon as possible after shut down, either by blasting, brushing, washing or scraping the surfaces. Sometimes the flue gas is diluted during the shut-down procedure, or soot blowing takes place ⁎ Corresponding author. E-mail address: [email protected] (F. Jones).
simultaneously, to minimize the effects. Generally, down-time corrosion is considered negligible compared to the corrosion taking place during the boiler process. However, the field of down-time corrosion still needs investigation and the process need to be further studied as several observations suggest that the corrosion rate could be substantial. If this type of corrosion can be minimized or avoided by simple measures, material investment costs could be decreased and boiler lifetime improved. This study focuses on uniform corrosion by measuring the corrosion rate, and pitting by measuring a parameter denoted pitting corrosion index. These two corrosion mechanisms are thought to be the most important when it comes to down-time corrosion. When the protective oxide layer of steel is broken down and corrosion appears more or less over a whole surface it is considered uniform corrosion. When the corrosion is limited to smaller areas it is local corrosion, also called pitting. The formed pits are often very small but the corrosion rate can be high (several mm/year is not unusual) and results in a small but detrimental puncture of the material. This kind of corrosion is common in acidic environments with access to chloride ions. Although there is a general consent that boilers operating on chloride-rich fuel, such as straw, have problems with down-time
http://dx.doi.org/10.1016/j.fuproc.2015.10.009 0378-3820/© 2015 Elsevier B.V. All rights reserved.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
2
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
corrosion there are only a few available studies. This is mostly due to the difficulties performing measurements on this type of corrosion. Some observations of down-time corrosion include a study of a coal power plant where the corrosion rate was measured with real-time corrosion probes during a three month period. During shut-down, as the temperature decreased, the corrosion went from immeasurable levels to as high as 0.4 mm/year. The corrosion occurred when moisture from the air was absorbed into the deposits forming acids. The corrosion subsided and after twelve weeks it had stopped. The study concluded that the acids were consumed and/or neutralized after this time [1,2]. Another study shows that coatings of austenitic stainless steel can be sensitive to down-time corrosion. If the coating has tension due to residues and is exposed to acid environment it can crack owing to stress corrosion [3]. In a study of a boiler in Haag, The Netherlands a high corrosion rate was measured (7–8 mm/year) on the primary superheater. It was concluded that the corrosion was mainly caused by an adjustment of the primary and secondary air but also by down-time corrosion [4,5]. A previous study has shown that online-corrosion probes are useful when studying down-time corrosion [6]. Since this study a new generation of probes is available, with improved technique and up to fifty times faster measurement cycle. They also have a very high accuracy; at a corrosion rate of 0.5 mm/year the deviation is only 0.1%, although the accuracy is lower at very low corrosion rates. By performing online studies of corrosion rates more qualified information about the corrosion process is gained. The corrosion rate is seldom constant and it is plausible that the most damaging corrosion takes place during 10% of the time causing 90% of the material loss. By measuring online the most unfavorable corrosion conditions can be identified. This cannot be done when assuming that the material loss is divided by the exposure time, which is the conventional way of estimate corrosion rates. This type of online measurements can also be used to characterize different deposits owing to how corrosive they are, or study corrosion rates on different alloys to choose the most suitable for the wanted application. Some examples of techniques [7,8] to measure the corrosion rate online: Electrical Resistance (ER) Zero Resistance Amperemetry (ZRA) or Galvanic Current (GC) Linear Polarization Resistance (LPR)/Electrochemical Impedance Spectroscopy (EIS) Electrochemical Noise (EN) Out of these techniques only Electrochemical Noise can measure pitting, while the others only measure uniform corrosion.
The aim of this study is to characterize the down-time corrosion rate for different alloys with different fuels, and see if it is necessary to take action to minimize the down-time corrosion. Traditionally down-time corrosion is thought to be negligible compared to the corrosion occurring during operation. In addition, boiler shut-down time is limited; from that the material temperature reaches 100 °C during the shut-down to cleaning it is normally not more than a few days, maximum a few weeks. However, the corrosion rate could be high during this time period and cannot be considered completely negligible. Furthermore, different fuels and alloys might show different corrosion rates, hence different material losses (if any) during boiler shut-down. It is also of interest to study how the corrosion rate changes with time. It can be assumed that moisture can be absorbed from the surrounding air when the deposit temperature reaches 100 °C, which will then increase the corrosion rate. By the absorption of moisture it is also possible that acids form in the deposits and depending on the deposit composition different acids with different strength might form at different times. It is therefore of interest to see when the maximum corrosion rate is reached, when it decreases and when it can be considered negligible. If the results would show a significant contribution to the total corrosion at some positions in the boiler, it is easy to prioritize cleaning of these surfaces at boiler shut-down. Since deposits have different composition at different positions in the boiler, such as boiler wall, superheater, and economizer, deposits from all these positions are used in this study. Deposits will also have different compositions owing to the fuel; therefore the study includes several different boilers operating on different fuels.
Table 1 Plant, sample position, fuel and alloy used for the tests.
Table 2 XRF-analysis results for chosen elements in the deposit samples.
2. Materials and methods 2.1. Deposit samples To be able to perform a test that resembles the true conditions as much as possible, the deposit samples were taken from various positions in six different boilers as soon as possible after boiler shut-down. They were directly taken to the laboratory and placed in desiccators. The sample positions varied between the different boilers depending on the availability to certain positions. Table 1 shows from which boiler and position the samples were taken. The sample preparation started with grinding of the deposits, either by hand in a mortar or in an electrical grinder depending on the hardness of the deposit. After this de-ionized water was added to the
Plant Sample position
Fuel
ST45.8 P91 304L Alloy 625
Plant, position, fuel
Element (%)
Cl
1 1 1 2a 2a
Waste Waste Waste Waste Waste
X X X X X
1 1 1 2a 2a
Waste Waste Waste Waste Waste
2.2 4.6 2.9 2.8 4.8 3.5 0.2 2.4 2.8 2.3
Waste Demolition wood Demolition wood Biomass Demolition wood Waste Demolition wood Demolition wood Demolition wood
X X
X
X
X
X X
X
X X
X X
X
X
X
X
2a 2b
Primary superheater Furnace top Economizer Inside center tube Secondary superheater Economizer Furnace
3
Fly ash (superheater)
4 5a
Furnace Loop seal
5b 6
Furnace Superheater (1)
6
Superheater (2)
6
Superheater (3)
X X X X
X
X
X
Primary superheater Furnace top Economizer Inside center tube Secondary superheater 2a Economizer 2b Furnace
Waste Demolition wood 3 Fly ash (superheater) Demolition wood 4 Furnace Biomass 5a Loop seal Demolition wood 5b Furnace Waste 6 Superheater (1) Demolition wood 6 Superheater (2) Demolition wood 6 Superheater (3) Demolition wood
S
K
Pb
1.0 0.3 1.6 0.4 1.1 0.2 0.7 b 0.1 1.0 0.2
Zn
Ca
1.2 22 0.7 20 0.8 21 2.6 3.0 1.0 2.5
2.6 0.4
2.7 8.0
1.7 9.0
1.2 1.0 4.3 1.4 9.2 17
3.4
3.7
2.0
0.9 2.0 16
2.1 0.2
3.5 4.6
8.7 3.1
0.1 2.3 11 9.0 1.7 13
0.1 b0.1 b0.1 10
0.2 8.1
0.1 0.5 21 3.9 6.2 7.4
b 0.1
8.1
3.4 3.7
8.3
1.3 2.4
8.5
0.1
9.5 10
11
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
material until it was firm clay like consistence. This clay was then mounted on the electrodes of the probe, made of one of the virgin materials presented in Section 2.2, and directly put in a 180 °C oven for 2 h. When the samples were dry they were placed in a closed container with a relative humidity of 65%. The corrosion process was monitored from the moment the samples were placed in the container. 2.2. Test materials Four different alloys of varying quality were used for the corrosion tests: ST45.8 (EN 1.0406) ST45.8 is a commonly used steel in boiler walls and economizers in power plants. In this study it also represents the unalloyed and/or low-alloyed steels used. Chemical composition: Fe(0.1–0.35)Si(0.4–1.2)Mn
304L (EN1.4307) The austenitic stainless steel 304L is the most commonly used steel of this quality. It is not normally used in power plants but in this study it represents stainless austenitic steel of the type “18–10”. The corrosion resistance is markedly better than P91. Chemical composition: Fe19Cr10Ni b 2Mn b 0.75Si
P91 (EN 1.4903) The ferritic 9% steel P91 is often used for superheaters in power plants. This steel is far more corrosion resistant than ST45.8. Chemical composition: Fe9Cr0.45Ni0.35Si0.2 V0.1Nb
Alloy 625 (EN 2.4856) Alloy 625 is a nickel-based super alloy that is frequently used in power plants as a protection against corrosion. It is often used for coating of low-alloy. This is the most corrosion resistant steel used in this study. Chemical composition: Ni21.5Cr9Mo3.75Nb1Co0.4Al0.4Ti
3
accuracy of the probes used is very good, and at least 2 μm, in many cases better. The LPR technique is developed based on the Stern–Geary theory [12]. Here the corrosion current is measured at regular intervals over a period of time, and from this it is possible to calculate the total mass of steel lost. The second technique used (EN) consists of recording current and potential fluctuations spontaneously generated by corrosion reactions, so non-intrusiveness is one of the main advantages of this technique [13]. Studies have shown that EN can be used to distinguish between different corrosion types, such as metastable and stable pitting [14], crevice corrosion [15], uniform corrosion [16], and stress corrosion cracking [17]. The virgin alloy samples were exposed to the deposits for at least a few days, some of them for several weeks. During the exposure the corrosion probes measured online and logged data in 30 s intervals. The reason for the extreme measurement time of several weeks was to study long-term effects. However, it is very unlikely that a boiler will be shut-down and not cleaned for this amount of time. 2.5. Experimental schedule Table 1 shows which plant (1–6), sample position, fuel and alloy that was used for each corrosion test. In plants number 2 and 5 samples have been taken from two different boilers, presented as 2a, 2b, 5a, and 5b. In plant 6 three samples were taken from the superheater, presented as superheater (1), (2), and (3), respectively. 3. Results 3.1. Chemical analysis with handheld XRF-instrument
The allyos were in mint condition and had not been used or exposed to higher temperature prior to the experiments. All electrodes were of the same size and shape and polished clean before the samples were mounted on them.
The results of the chemical analysis (of the chosen elements Cl, S, K, Pb, Zn, Ca) of the deposit samples are presented in Table 2. There is no evident correlation between fuel type and the measured concentrations.
2.3. Deposit composition
3.2. Corrosion rate and pitting
Prior to the corrosion test the deposits were chemically analyzed with a handheld XRF-instrument (X-Ray Fluorescence). This is an analysis method measuring radiation from a material excited with x-rays [9]. It gives a fast analysis of the chemical composition of the deposit without consuming or destroying the sample. The measurement technique primarily used in handheld XRF-instruments is Energy Dispersive X-Ray Fluorescence (EDXRF) [10]. A clear limitation with this instrument is that no elements lighter than magnesium (Mg, atom number 12) can be detected. Unfortunately this means that sodium (Na) cannot be measured, and Na is an important element in corrosion, and often found in significant concentration, especially in waste-fired boilers. Furthermore, there are other interference issues that must be accounted for. The XRF is limited when it comes to measure amorphous elements. A non-crystalline structure with mixtures of a variety of chemical elements can affect the behavior of photons in highly complex ways. These effects on light translate directly to complexities in interpreting the fluorescence radiation that is detected in the XRF [11]. The XRFanalysis in this study enabled a quick, inexpensive overview of the sample composition. However, the analysis has the limitation for elements lighter than Mg and possible interference, meaning amorphous elements will not show up in the analysis.
The corrosion rates were measured for all deposits for at least six days. The pitting activity was measured simultaneously but the measurement is not reliable when the corrosion rate is low, hence it is not presented for all samples and steels. Please note the difference in the y-axis of the following figures. In each section Tables 3–8 refers to the samples for the specific boiler, respectively.
2.4. Corrosion tests The down-time corrosion was measured with online probes using Linear Polarization Resistance (LPR) and Electrochemical Noise (EN) calculating the corrosion rate and pitting index internally. For the experiments in this study the maximum pitting activity is set to one “1”. The
3.2.1. Plant 1–3 samples Fig. 1 and Fig. 2 show the corrosion rate and pitting activities for the ST45.8-steel and the P91-steel, respectively, while exposed to the deposit from the superheater region in plant 1. For the ST45.8-steel the Table 3 Samples in plant 1. Plant
Sample position
Fuel
ST45.8
P91
Alloy 625
1 1 1
Primary superheater Furnace top Economizer
Waste Waste Waste
X X X
X X X
X
Table 4 Samples in plant 2. Plant
Sample position
Fuel
ST45.8
P91
304L
2a 2a 2a 2b
Inside center tube Secondary superheater Economizer Furnace
Waste Waste Waste Demolition wood
No corrosion X X No corrosion
X X
X
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
4
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Table 5 Sample in plant 3. Plant
Sample position
Fuel
ST45.8
P91
3
Fly ash (superheater)
Demolition wood
X
X
Table 6 Sample in plant 4. Plant
Sample position
Fuel
ST45.8
P91
304L
4
Furnace
Biomass
X
X
X
Table 7 Samples in plant 5. Plant
Sample position
Fuel
ST45.8
P91
5a 5b
Loop seal Furnace
Demolition wood Waste
X No corrosion
No corrosion
Fig. 2. The corrosion rate for superheater deposits from plant 1 on P91-steel. The corrosion rate exceeded 0.01 mm/year after four days.
Table 8 Samples in plant 6. Plant
Sample position
Fuel
ST45.8
P91
6 6 6
Superheater (1) Superheater (2) Superheater (3)
Demolition wood Demolition wood Demolition wood
X X X
X X No corrosion
corrosion rate increases fastest during the first 48 h and reaches a maximum of 0.25 mm/year while the P91-steel reaches about 0.01 mm/year. Vast pitting occurs three times during the experimental time. For the deposit samples from the furnace (Fig. 3–Fig. 5) the ST45.8-steel shows a corrosion rate of 1 mm/year after 24 h and after approximately 4 days it reaches 1.6–1.8 mm/year before plateauing. The pitting activity is high the first 48 h and then stops. For the P91-steel the maximum corrosion rate is 20 μm/year and the Alloy 625 reaches 0.2 μm/year. Hence, the corrosion rate of Alloy 625 is two magnitudes lower than P91, which in turn is two magnitudes lower than ST45.8. When exposed to deposits from the economizer (Fig. 6 and Fig. 7) the corrosion rate for the ST45.8-steel reaches 1.6 mm/year and pitting activity can be seen the first two days. For the P91-steel the corrosion rate reaches 0.08 mm/year and the pitting is vast during the first few days.
Fig. 1. The corrosion rate and pitting activity for superheater deposits from plant 1 on ST45.8-steel. The maximum corrosion is about 0.25 mm/year.
Fig. 3. The corrosion rate and pitting activity for furnace deposits from plant 1 on ST45.8steel. The corrosion rate reaches about 1.75 mm/year after four days.
3.2.2. Plant 2–4 samples The sample taken from the inside center tube from plant 2 showed no corrosion. The deposit from the superheater was tested with three types of steel and gave a corrosion rate of 0.4 mm/year for the ST45.8steel. The corrosion rate for the P91-steel and the 304L-steel was 0.12 mm/year and 0.035 mm/year, respectively. Pitting activity was initially measured for the ST45.8-steel but not for the two other steels. This is shown in Fig. 8–Fig. 10.
Fig. 4. The corrosion rate for furnace deposits from plant 1 on P91-steel. The corrosion rate almost reaches 0.2 mm/year after three days.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 5. The corrosion rate for furnace deposits from plant 1 on Alloy 625. The corrosion is negligible.
Fig. 6. The corrosion rate and pitting activity for economizer deposits from plant 1 on ST45.8-steel. The corrosion rate is about 1.6 mm/year after three days.
5
Fig. 8. The corrosion rate and pitting activity for superheater deposits from plant 2 on ST45.8-steel. The corrosion rate reached 0.4 mm/year after approximately 30 h and then increased to 0.5 after 75 h.
Fig. 9. The corrosion rate and pitting activity for superheater deposits from plant 2 on P91steel. The corrosion rate reached 0.12 mm/year after five days.
The economizer deposit sample gave a corrosion rate of 0.64 mm/year for the ST45.8-steel and 0.06 mm/year for the P91-steel, see Fig. 11 and Fig. 12. Only limited pitting occurred for the ST45.8-steel and for the P91-steel the pitting is most likely an artifact as the corrosion rate is relatively low. The furnace sample showed no corrosion.
3.2.3. Plant 3–1 sample From plant 3 a fly ash sample was taken from the superheater region. Fig. 13 shows the exposure on ST45.8-steel. A maximum of 0.55 mm/year was reached after approximately 3 days, no pitting activity was registered. For the P91-steel 0.16 mm/year was reached within 24 h and then decreased to approximately 0.8 mm/year after 50 h. During the first 50 h pitting occurred, see Fig. 14.
Fig. 7. The corrosion rate and pitting activity for economizer deposits from plant 1 on P91steel. The corrosion rate reaches about 0.08 mm/year.
Fig. 10. The corrosion rate for superheater deposits from plant 2 on 304L-steel. The corrosion rate reaches 0.035 mm/year after about one day and sustains that level.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
6
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 11. The corrosion rate and pitting activity for economizer deposits from plant 2 on ST45.8-steel. The corrosion rate is steadily increasing during seven days.
Fig. 14. The corrosion rate and pitting activity for fly ash from plant 3 on P91-steel. The maximum corrosion rate 0.16 mm/year is reached after one day.
Fig. 12. The corrosion rate and pitting activity for economizer deposits from plant 2 on P91-steel. The maximum corrosion rate is about 0.06 mm/year The pitting activity is over exaggerated.
Fig. 15. The corrosion rate for furnace deposits from plant 4 on ST45.8-steel. The maximum corrosion rate is 0.7 mm/year and is reached after approximately six days.
3.2.4. Plant 4–1 sample The deposit sample from the only biomass boiler included in the experiments was exposed to three different kinds of steel, ST45.8-, P91-, and 304L-steel, the results shown individually in Fig. 15–Fig. 17, and all together in Fig. 18. For the ST45.8-steel the corrosion rate continued increasing during seven days, and at the experiment end, it had reached 0.7 mm/year. The P91-steel exposure was shorter and was also
increasing slowly, reaching a maximum of 0.12 mm/year before the experiment was stopped. The corrosion rate for the 304L-steel is practically zero. 3.2.5. Plant 5–2 samples For sample 5a the corrosion rate reached 0.12 mm/year on ST45.8steel after approximately two days and then decreased, see Fig. 19. Sample 5b was sintered deposit samples, which was prepared like all other samples, but the corrosion rate was not measurable. 3.2.6. Plant 6–3 samples From plant 6 three different deposit samples from the superheater region was used. For sample (1) the corrosion rate was slowly increasing over approximately five days to 0.37 mm/year on the ST45.8-steel, see Fig. 20. On P91-steel it does not have much effect, only reaching 0.01 mm/year after eight days of exposure shown in Fig. 21. Samples (2) from plant 6 had a lower corrosion rate on the ST45.8steel only reaching 0.15 mm/year after three days and then decrease while the corrosion rate steadily increased for the P91-steel, but only to 0.04 mm/year, see Fig. 22 and Fig. 23. Fig. 24 shows deposit sample (3) from plant 6. The corrosion rate reaches only 0.09 mm/year after approximately three days on ST45.8steel, while the P91-steel did not show any corrosion. 4. Discussion
Fig. 13. The corrosion rate for fly ash from plant 3 on ST45.8-steel. The corrosion rate reaches 0.55 mm/year after three days.
Down-time corrosion starts from no corrosion rate at all and reaches rates of a few hundredths mm/year to a few mm/year. The time it takes
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
7
Fig. 16. The corrosion rate for furnace deposits from plant 4 on P91-steel. The maximum corrosion rate is 0.12 mm/year and is reached after approximately six days.
Fig. 19. The corrosion rate and pitting activity for loop seal deposits from plant 5 on ST45.8-steel. The corrosion rate reaches 0.12 mm/year two days and then decreases.
Fig. 17. The corrosion rate for furnace deposits from plant 4 on 304L-steel. The corrosion is negligible.
Fig. 20. The corrosion rate for superheater deposits (1) from plant 6 on ST45.8-steel. The maximum corrosion rate, 0.037 mm/year is reached after five days and the falls back.
to reach the maximum corrosion rates varies from approximately one day up to a week. In some cases, such as plant 2 (waste boiler, Fig. 11), the corrosion rate is still increasing after several days while others plateaus and some decreases after reaching maximum. The corrosion rate is at its highest the first few days. Some of the measured corrosion rates are considered high. For example, the furnace deposit sample from plant 1 (waste boiler) on ST45.8-steel (Fig. 3) shows corrosion rates between 1.5–1.8
mm/year. In addition the economizer deposit from the same boiler on ST45.8-steel (Fig. 6) shows a corrosion rate of 1.6 mm/year. Also the furnace deposit sample from the biomass boiler (plant 4, Fig. 15) has a corrosion rate of 0.7 mm/year. With corrosion rates of this magnitude it is possible that the down-time corrosion could be worse than the corrosion during operation. In these cases it is of great importance to clean the boiler surfaces without too much delay after boiler shut-down. It is not the material wear that
Fig. 18. A comparison between all three steels, ST45.8, P91, and 304L, in furnace deposits from plant 4.
Fig. 21. The corrosion rate for superheater deposits (1) from plant 6 on P91-steel. The corrosion rate reached 0.01 mm/year.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
8
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 22. The corrosion rate and pitting activity for superheater deposits (2) from plant 6 on ST45.8-steel. The corrosion rate reaches 0.15 mm/year after three days.
Fig. 24. The corrosion rate for superheater deposits (3) from plant 6 on ST45.8-steel. The corrosions rate maximum is 0.09 mm/year.
is the main problem, but the possibility for the down-time corrosion to destroy any protective oxide films added to prevent corrosion during operation, especially when pitting occurs simultaneously. Although the corrosion rate is low and in some cases negligible, it is concluded that the corrosion starts during the first 24 h after the boiler has been shut-down. The results suggest that the initiation of corrosion is dependent on the moisture content but the corrosion rate is not. However, it needs to be taken into consideration that all steel test materials used for the experiments were of mint condition and not exposed to high temperature corrosion inside a boiler.
corrosion rate decreases as the S-concentration increases. For the P91-steel there are no clear correlations. The same correlation analyses were made for Ca, see Fig. 27, showing a similar, but weaker, trend as for Cl for both ST45.8-, and P91-steel. Out of all compared elements it is Cl and Ca that show the strongest correlation between concentration and the corrosion rate. It is possible that CaCl2 is formed, which is hygroscopic and could cause corrosion on low-alloyed steel. Other elements, such as K, Zn, and Pb, which normally increases the corrosion, show no such trends in these experiments. In fact, all of them show a slight decrease of the corrosion rate when present in higher concentrations. However, it could be that the effects on the corrosion rate are minor for these elements compared to Cl in these tests.
4.1. Corrosion rate dependence of the elemental concentrations It is well known that some elements and the compounds they form cause corrosion. Commonly mentioned elements are chlorine (Cl), sulfur (S), potassium (K), sodium (Na), lead (Pb), zinc (Zn), and calcium (Ca). All of these elements, except Na, were measured in the deposits using the handheld XRF-instrument, presented in Table 2. By comparing the elemental concentrations to the corrosion rates presented in Fig. 1–Fig. 24 it is possible to see any trends if between the deposit composition and the corrosion. In Fig. 25 the corrosion rates of all samples for the ST45.8-, and the P91-steel are compared to the Cl-concentration of the deposits. For the ST45.8steel the trend is that the corrosion rate increases with the Clconcentration. For the P91-steel the trend is not as obvious but is there to some extent. The same kind of comparison is shown for S in Fig. 26. Here the trend is the opposite for the ST45.8-steel, the
Fig. 23. The corrosion rate for superheater deposits (2) from plant 6 on P91-steel. The corrosion rate reaches 0.04 mm/year after three days.
4.2. Corrosion rate dependence of deposit position The down-time corrosion rate could also be dependent on from which position in the boiler the deposits are taken. There are significant differences between the temperatures, the flue gas composition and surfaces in the different boiler parts. The number of deposits from different positions in the boilers in these experiments was limited. For deposit samples from the furnace (plant 4 using biomass, plant 1, 2a, and 5b using waste, and plant 2b using demolition wood) the corrosion rate varied between 0 to 1.8 mm/year for the ST45.8-steel and from 0.12 to 0.23 mm/year for the P91-steel. The variation was higher within the same fuel type than compared between the three test materials.
Fig. 25. The corrosion rate for the ST45.8-steel increase as the Cl-concentration increase. For the P91-steel the corrosion rate does to some extent increase, as the Cl-concentration increase.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
9
5. Conclusions
Fig. 26. The corrosion rate shows a trend of slightly decreasing when the S-concentration increase for ST45.8-steel. Any correlation between S-concentration and corrosion rate is not obvious for the P91-steel.
Then again, the small number of samples does not give a satisfactory statistical base. For the deposit samples from superheaters (plant 1 and 2a using waste, plant 6 using demolition wood) there is a slight trend that the deposits from the waste boilers corrode more than the ones from the demolition wood boiler. The corrosion rate for the waste boilers is between 0.27–0.55 mm/year and 0.01–0.13 mm/year for ST45.8-, and P91-steel, respectively. The corresponding corrosion rates for the demolition wood boiler are 0.04–0.55 mm/year and 0.01–0.08 mm/year. Deposit samples from the economizer region were only available from waste boilers (plant 1 and 2a). The corrosion rate for the ST45.8-steel was well above the rate for the P91-steel, 0.6–1.61 mm/year compared to 0.06–0.08 mm/year. Only one sample from a loop seal superheater was included in the experimental tests (plant 5a using demolition wood). The measured corrosion rate was 0.12 mm/year. With such a small statistical base no direct conclusions can be drawn. However, for the ST45.8-steel it seems like the deposits from the furnace and economizer are more corrosive than the ones from the superheater region, which in turn is higher than the one from the loop seal. For the P91-steel the corrosion in the furnace seems higher than in the superheaters and economizers. Furthermore, biomass appears to be less corrosive than waste and demolition wood, but the difference is very low. Waste gives rise to somewhat higher corrosion rates than demolition wood for deposits from the superheater region.
From this laboratory study it was concluded that down-time corrosion increases the fastest within 24 h up to four days after the deposit has been exposed to ambient air. However, depending on the type of sample origin (position in the boiler and type of fuel used) the downtime corrosion can have a large variation, from negligible rates up to 1.8 mm/year for the ST45.8-steel. When the corrosion rate measures up to 1.8 mm/year for ST45.8-steel, cleaning of boilers with this or similar materials should be prioritized after boiler shut-down. The study showed that there are, in general, great differences in the corrosion rates between the different steel types: ST45.8-steel corrodes one magnitude faster than the P91-steel, in which in turn corrodes one magnitude faster than 304L-steel, which in turn corrodes one magnitude faster than Alloy 625. In addition, the XRF-study shows that the presence of Cl in the deposits seems to affect the corrosion rate more than other elements present. Finally, the results suggest that the initiation of corrosion is dependent on the moisture content but the corrosion rate is not. Acknowledgements This study was financially supported by Värmeforsk (The Thermal Engineering Research Institute in Sweden) and the Swedish Energy Agency (M12-209). References [1] W.M. Cox, D.M. Farrell, J.L. Dawson, Corrosion monitoring for process control, in: D.S. Holmes (Ed.), Dewpoint Corrosion, Horwood Publishing, Chichester 1985, p. 191. [2] T. Eriksson, In-situ mätningar av rökgassidig korrosion; litteratur och utvecklingsstudie, Material- och Kemiteknik, Värmeforsks orienteringsrapport, 105, 1997 (in Swedish). [3] J. Krueger, Operation experience and future perspectives with degradation at high steam parameters in WTE-plants, Mater. High Temp. 20 (2) (2003) 185–193. [4] N. Bolt, E.J.A. Vogelaar, A.S. de Clerc.q, Fireside corrosion in Dutch waste incinerators, Werkst. Korros. 40 (3) (1989) 142–146. [5] A. Stålenheim, P. Henderson, Korrosion i avfallsförbränningsanläggningar, Materialoch Kemiteknik, Värmeforskrapport, 887, 2004 (in Swedish). [6] A. Hjörnhede., R. Norling., Kalibrering av elektorkemisk korrosionssond i laboratoriemiljö och verifiering med fältförsök, Material- och Kemiteknik, Värmeforskrapport, 1210, 2012 (in Swedish). [7] D. Thierry, A. Thorén, C. Leygraf, Corrosion monitoring techniques applied to cooling water and district heating systems, Paper 463. Corrosion 87, NACE, Houston, 1987. [8] C.E. Jaske, J.A. Beavers., N.G. Thompson., Improving plant reliability through corrosion monitoring, Corrosion Prevention and Control, 49 2002, pp. 3–12. [9] R. Schramm, X-ray fluorescence analysis: practical and easy, Fluxana, 2012, ISBN 978-3-00-04 14 76-3. [10] L. König, X-ray fluorescence — the use of handheld XRF analyzers for a quick measurement of elements in waste incineration ashes and waste fuel, University of Borås, report 18, 2013. [11] S.M. Shackley, X-ray fluorescence in spectrometry (XRF) in geoarchaeology, 2011, ISBN 978-1-4419-6886-9. [12] M. Stern, A.L. Geary, J. Electrochem. Soc. (1957) 56–63 (January issue). [13] K. Hladky, J.L. Dawson, The measurement of localized corrosion using electrochemical noise, Corros. Sci. 21 (1981) 317–322. [14] A. Legat, A. Zevnik, The electrochemical noise of mild and stainless steel in various water solutions, Corros. Sci. 35 (1993). [15] P.R. Roberge, R. Beaudoin, V.S. Sastri, Electrochemical noise measurements for field applications, Corros. Sci. 29 (1989) 1231–1233. [16] F. Mansfeld, H. Xiao, Electrochemical noise-analysis of iron exposed to NaCl solutions of different corrosivity, J. Electrochem. Soc. 141 (1994) 1403–1404. [17] C.A. Loto, R.A. Cottis, Electrochemical noise generation during stress corrosion cracking of the high-strength aluminum AA 7075-T6 alloy, Corrosion 45 (1989) 136–141.
Fig. 27. The corrosion rate for the ST45.8-steel increase as the Ca-concentration increase. Any correlation between Ca-concentration and corrosion rate is not obvious for the P91steel.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Down-time corrosion in boilers Frida Jones ⁎, Daniel Ryde, Anders Hjörnhede SP Technical Research Institute of Sweden, Energy and Bioeconomy, Box 857, SE-501 15 Borås, Sweden
a r t i c l e
i n f o
Article history: Received 22 March 2015 Received in revised form 24 September 2015 Accepted 3 October 2015 Available online xxxx Keywords: Down-time corrosion, Online-corrosion probe, Deposits
a b s t r a c t Down-time corrosion can occur on boiler surfaces, e.g. furnace walls, superheaters, or economizers that are covered with hygroscopic deposits, when the temperature drops under 100 °C. This scenario takes place when a boiler is shut-down for cleaning, maintenance, or other reasons, such as unplanned shut-downs. Initially, the dry deposits will absorb moisture from the surrounding air, potentially creating a corrosive environment. After this, corrosive acids can form in the deposits. In this study modified online-corrosion probes were used in combination with deposits taken from 6 different boilers at various locations (for example, from the furnace, the superheater, and the economizer), where the fuels have been waste, demolition wood or biomass. The deposits were ground and dried in an oven at 160 °C for several hours before exposed to a moist environment (RH 65%) during online measuring of the corrosion rate and the pitting activity. Four types of alloys were tested: low-alloy ST45.8-steel, 9% Cr ferritic P91-steel, austenitic stainless steel 304L, and Ni-based super Alloy 625. The results for ST45.8 show that in biomass boilers a corrosion rate from negligible values up to 0.7 mm/year can be reached within a week, while waste-fired boilers can have rates as high as 1.8 mm/year. Furthermore, for some samples from waste-fired boilers show a high pitting activity already after 24 h. The tests with the P91-steel show values up to 0.16 mm/year, for samples from different locations in the boiler. For 304L and Alloy 625 the downtime corrosion was negligible even after a two-week exposure. The ability to follow the down-time corrosion online has provided data that show that even though thought to be negligible, the risk of down-time corrosion is of significance, especially if the fuel is waste. Also, even if the initial corrosion rate is low, it increases during the first 24 h due to the exposure to moist environment, motivating immediate cleaning of the boilers after shut-down, especially on surfaces of lower steel quality. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Down-time corrosion can occur on boiler surfaces, e.g. furnace walls, superheaters, or economizers that are covered with hygroscopic deposits when the temperature drops under 100 °C. This scenario takes place when a boiler is shut down for cleaning, maintenance, or other reasons, such as unplanned shut downs. Initially, the dry deposits will absorb moisture from the surrounding air, creating a potentially corrosive environment. Furthermore, depending on the chemical composition of the deposits, it is possible that corrosive substances are formed in the deposits when in contact with moisture. Down-time corrosion can also occur if improper procedures are used during boiler shutdown; this is especially true for down-time corrosion on the economizer. To reduce the effects of down-time corrosion it is common to clean the boilers as soon as possible after shut down, either by blasting, brushing, washing or scraping the surfaces. Sometimes the flue gas is diluted during the shut-down procedure, or soot blowing takes place ⁎ Corresponding author. E-mail address: [email protected] (F. Jones).
simultaneously, to minimize the effects. Generally, down-time corrosion is considered negligible compared to the corrosion taking place during the boiler process. However, the field of down-time corrosion still needs investigation and the process need to be further studied as several observations suggest that the corrosion rate could be substantial. If this type of corrosion can be minimized or avoided by simple measures, material investment costs could be decreased and boiler lifetime improved. This study focuses on uniform corrosion by measuring the corrosion rate, and pitting by measuring a parameter denoted pitting corrosion index. These two corrosion mechanisms are thought to be the most important when it comes to down-time corrosion. When the protective oxide layer of steel is broken down and corrosion appears more or less over a whole surface it is considered uniform corrosion. When the corrosion is limited to smaller areas it is local corrosion, also called pitting. The formed pits are often very small but the corrosion rate can be high (several mm/year is not unusual) and results in a small but detrimental puncture of the material. This kind of corrosion is common in acidic environments with access to chloride ions. Although there is a general consent that boilers operating on chloride-rich fuel, such as straw, have problems with down-time
http://dx.doi.org/10.1016/j.fuproc.2015.10.009 0378-3820/© 2015 Elsevier B.V. All rights reserved.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
2
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
corrosion there are only a few available studies. This is mostly due to the difficulties performing measurements on this type of corrosion. Some observations of down-time corrosion include a study of a coal power plant where the corrosion rate was measured with real-time corrosion probes during a three month period. During shut-down, as the temperature decreased, the corrosion went from immeasurable levels to as high as 0.4 mm/year. The corrosion occurred when moisture from the air was absorbed into the deposits forming acids. The corrosion subsided and after twelve weeks it had stopped. The study concluded that the acids were consumed and/or neutralized after this time [1,2]. Another study shows that coatings of austenitic stainless steel can be sensitive to down-time corrosion. If the coating has tension due to residues and is exposed to acid environment it can crack owing to stress corrosion [3]. In a study of a boiler in Haag, The Netherlands a high corrosion rate was measured (7–8 mm/year) on the primary superheater. It was concluded that the corrosion was mainly caused by an adjustment of the primary and secondary air but also by down-time corrosion [4,5]. A previous study has shown that online-corrosion probes are useful when studying down-time corrosion [6]. Since this study a new generation of probes is available, with improved technique and up to fifty times faster measurement cycle. They also have a very high accuracy; at a corrosion rate of 0.5 mm/year the deviation is only 0.1%, although the accuracy is lower at very low corrosion rates. By performing online studies of corrosion rates more qualified information about the corrosion process is gained. The corrosion rate is seldom constant and it is plausible that the most damaging corrosion takes place during 10% of the time causing 90% of the material loss. By measuring online the most unfavorable corrosion conditions can be identified. This cannot be done when assuming that the material loss is divided by the exposure time, which is the conventional way of estimate corrosion rates. This type of online measurements can also be used to characterize different deposits owing to how corrosive they are, or study corrosion rates on different alloys to choose the most suitable for the wanted application. Some examples of techniques [7,8] to measure the corrosion rate online: Electrical Resistance (ER) Zero Resistance Amperemetry (ZRA) or Galvanic Current (GC) Linear Polarization Resistance (LPR)/Electrochemical Impedance Spectroscopy (EIS) Electrochemical Noise (EN) Out of these techniques only Electrochemical Noise can measure pitting, while the others only measure uniform corrosion.
The aim of this study is to characterize the down-time corrosion rate for different alloys with different fuels, and see if it is necessary to take action to minimize the down-time corrosion. Traditionally down-time corrosion is thought to be negligible compared to the corrosion occurring during operation. In addition, boiler shut-down time is limited; from that the material temperature reaches 100 °C during the shut-down to cleaning it is normally not more than a few days, maximum a few weeks. However, the corrosion rate could be high during this time period and cannot be considered completely negligible. Furthermore, different fuels and alloys might show different corrosion rates, hence different material losses (if any) during boiler shut-down. It is also of interest to study how the corrosion rate changes with time. It can be assumed that moisture can be absorbed from the surrounding air when the deposit temperature reaches 100 °C, which will then increase the corrosion rate. By the absorption of moisture it is also possible that acids form in the deposits and depending on the deposit composition different acids with different strength might form at different times. It is therefore of interest to see when the maximum corrosion rate is reached, when it decreases and when it can be considered negligible. If the results would show a significant contribution to the total corrosion at some positions in the boiler, it is easy to prioritize cleaning of these surfaces at boiler shut-down. Since deposits have different composition at different positions in the boiler, such as boiler wall, superheater, and economizer, deposits from all these positions are used in this study. Deposits will also have different compositions owing to the fuel; therefore the study includes several different boilers operating on different fuels.
Table 1 Plant, sample position, fuel and alloy used for the tests.
Table 2 XRF-analysis results for chosen elements in the deposit samples.
2. Materials and methods 2.1. Deposit samples To be able to perform a test that resembles the true conditions as much as possible, the deposit samples were taken from various positions in six different boilers as soon as possible after boiler shut-down. They were directly taken to the laboratory and placed in desiccators. The sample positions varied between the different boilers depending on the availability to certain positions. Table 1 shows from which boiler and position the samples were taken. The sample preparation started with grinding of the deposits, either by hand in a mortar or in an electrical grinder depending on the hardness of the deposit. After this de-ionized water was added to the
Plant Sample position
Fuel
ST45.8 P91 304L Alloy 625
Plant, position, fuel
Element (%)
Cl
1 1 1 2a 2a
Waste Waste Waste Waste Waste
X X X X X
1 1 1 2a 2a
Waste Waste Waste Waste Waste
2.2 4.6 2.9 2.8 4.8 3.5 0.2 2.4 2.8 2.3
Waste Demolition wood Demolition wood Biomass Demolition wood Waste Demolition wood Demolition wood Demolition wood
X X
X
X
X
X X
X
X X
X X
X
X
X
X
2a 2b
Primary superheater Furnace top Economizer Inside center tube Secondary superheater Economizer Furnace
3
Fly ash (superheater)
4 5a
Furnace Loop seal
5b 6
Furnace Superheater (1)
6
Superheater (2)
6
Superheater (3)
X X X X
X
X
X
Primary superheater Furnace top Economizer Inside center tube Secondary superheater 2a Economizer 2b Furnace
Waste Demolition wood 3 Fly ash (superheater) Demolition wood 4 Furnace Biomass 5a Loop seal Demolition wood 5b Furnace Waste 6 Superheater (1) Demolition wood 6 Superheater (2) Demolition wood 6 Superheater (3) Demolition wood
S
K
Pb
1.0 0.3 1.6 0.4 1.1 0.2 0.7 b 0.1 1.0 0.2
Zn
Ca
1.2 22 0.7 20 0.8 21 2.6 3.0 1.0 2.5
2.6 0.4
2.7 8.0
1.7 9.0
1.2 1.0 4.3 1.4 9.2 17
3.4
3.7
2.0
0.9 2.0 16
2.1 0.2
3.5 4.6
8.7 3.1
0.1 2.3 11 9.0 1.7 13
0.1 b0.1 b0.1 10
0.2 8.1
0.1 0.5 21 3.9 6.2 7.4
b 0.1
8.1
3.4 3.7
8.3
1.3 2.4
8.5
0.1
9.5 10
11
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
material until it was firm clay like consistence. This clay was then mounted on the electrodes of the probe, made of one of the virgin materials presented in Section 2.2, and directly put in a 180 °C oven for 2 h. When the samples were dry they were placed in a closed container with a relative humidity of 65%. The corrosion process was monitored from the moment the samples were placed in the container. 2.2. Test materials Four different alloys of varying quality were used for the corrosion tests: ST45.8 (EN 1.0406) ST45.8 is a commonly used steel in boiler walls and economizers in power plants. In this study it also represents the unalloyed and/or low-alloyed steels used. Chemical composition: Fe(0.1–0.35)Si(0.4–1.2)Mn
304L (EN1.4307) The austenitic stainless steel 304L is the most commonly used steel of this quality. It is not normally used in power plants but in this study it represents stainless austenitic steel of the type “18–10”. The corrosion resistance is markedly better than P91. Chemical composition: Fe19Cr10Ni b 2Mn b 0.75Si
P91 (EN 1.4903) The ferritic 9% steel P91 is often used for superheaters in power plants. This steel is far more corrosion resistant than ST45.8. Chemical composition: Fe9Cr0.45Ni0.35Si0.2 V0.1Nb
Alloy 625 (EN 2.4856) Alloy 625 is a nickel-based super alloy that is frequently used in power plants as a protection against corrosion. It is often used for coating of low-alloy. This is the most corrosion resistant steel used in this study. Chemical composition: Ni21.5Cr9Mo3.75Nb1Co0.4Al0.4Ti
3
accuracy of the probes used is very good, and at least 2 μm, in many cases better. The LPR technique is developed based on the Stern–Geary theory [12]. Here the corrosion current is measured at regular intervals over a period of time, and from this it is possible to calculate the total mass of steel lost. The second technique used (EN) consists of recording current and potential fluctuations spontaneously generated by corrosion reactions, so non-intrusiveness is one of the main advantages of this technique [13]. Studies have shown that EN can be used to distinguish between different corrosion types, such as metastable and stable pitting [14], crevice corrosion [15], uniform corrosion [16], and stress corrosion cracking [17]. The virgin alloy samples were exposed to the deposits for at least a few days, some of them for several weeks. During the exposure the corrosion probes measured online and logged data in 30 s intervals. The reason for the extreme measurement time of several weeks was to study long-term effects. However, it is very unlikely that a boiler will be shut-down and not cleaned for this amount of time. 2.5. Experimental schedule Table 1 shows which plant (1–6), sample position, fuel and alloy that was used for each corrosion test. In plants number 2 and 5 samples have been taken from two different boilers, presented as 2a, 2b, 5a, and 5b. In plant 6 three samples were taken from the superheater, presented as superheater (1), (2), and (3), respectively. 3. Results 3.1. Chemical analysis with handheld XRF-instrument
The allyos were in mint condition and had not been used or exposed to higher temperature prior to the experiments. All electrodes were of the same size and shape and polished clean before the samples were mounted on them.
The results of the chemical analysis (of the chosen elements Cl, S, K, Pb, Zn, Ca) of the deposit samples are presented in Table 2. There is no evident correlation between fuel type and the measured concentrations.
2.3. Deposit composition
3.2. Corrosion rate and pitting
Prior to the corrosion test the deposits were chemically analyzed with a handheld XRF-instrument (X-Ray Fluorescence). This is an analysis method measuring radiation from a material excited with x-rays [9]. It gives a fast analysis of the chemical composition of the deposit without consuming or destroying the sample. The measurement technique primarily used in handheld XRF-instruments is Energy Dispersive X-Ray Fluorescence (EDXRF) [10]. A clear limitation with this instrument is that no elements lighter than magnesium (Mg, atom number 12) can be detected. Unfortunately this means that sodium (Na) cannot be measured, and Na is an important element in corrosion, and often found in significant concentration, especially in waste-fired boilers. Furthermore, there are other interference issues that must be accounted for. The XRF is limited when it comes to measure amorphous elements. A non-crystalline structure with mixtures of a variety of chemical elements can affect the behavior of photons in highly complex ways. These effects on light translate directly to complexities in interpreting the fluorescence radiation that is detected in the XRF [11]. The XRFanalysis in this study enabled a quick, inexpensive overview of the sample composition. However, the analysis has the limitation for elements lighter than Mg and possible interference, meaning amorphous elements will not show up in the analysis.
The corrosion rates were measured for all deposits for at least six days. The pitting activity was measured simultaneously but the measurement is not reliable when the corrosion rate is low, hence it is not presented for all samples and steels. Please note the difference in the y-axis of the following figures. In each section Tables 3–8 refers to the samples for the specific boiler, respectively.
2.4. Corrosion tests The down-time corrosion was measured with online probes using Linear Polarization Resistance (LPR) and Electrochemical Noise (EN) calculating the corrosion rate and pitting index internally. For the experiments in this study the maximum pitting activity is set to one “1”. The
3.2.1. Plant 1–3 samples Fig. 1 and Fig. 2 show the corrosion rate and pitting activities for the ST45.8-steel and the P91-steel, respectively, while exposed to the deposit from the superheater region in plant 1. For the ST45.8-steel the Table 3 Samples in plant 1. Plant
Sample position
Fuel
ST45.8
P91
Alloy 625
1 1 1
Primary superheater Furnace top Economizer
Waste Waste Waste
X X X
X X X
X
Table 4 Samples in plant 2. Plant
Sample position
Fuel
ST45.8
P91
304L
2a 2a 2a 2b
Inside center tube Secondary superheater Economizer Furnace
Waste Waste Waste Demolition wood
No corrosion X X No corrosion
X X
X
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
4
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Table 5 Sample in plant 3. Plant
Sample position
Fuel
ST45.8
P91
3
Fly ash (superheater)
Demolition wood
X
X
Table 6 Sample in plant 4. Plant
Sample position
Fuel
ST45.8
P91
304L
4
Furnace
Biomass
X
X
X
Table 7 Samples in plant 5. Plant
Sample position
Fuel
ST45.8
P91
5a 5b
Loop seal Furnace
Demolition wood Waste
X No corrosion
No corrosion
Fig. 2. The corrosion rate for superheater deposits from plant 1 on P91-steel. The corrosion rate exceeded 0.01 mm/year after four days.
Table 8 Samples in plant 6. Plant
Sample position
Fuel
ST45.8
P91
6 6 6
Superheater (1) Superheater (2) Superheater (3)
Demolition wood Demolition wood Demolition wood
X X X
X X No corrosion
corrosion rate increases fastest during the first 48 h and reaches a maximum of 0.25 mm/year while the P91-steel reaches about 0.01 mm/year. Vast pitting occurs three times during the experimental time. For the deposit samples from the furnace (Fig. 3–Fig. 5) the ST45.8-steel shows a corrosion rate of 1 mm/year after 24 h and after approximately 4 days it reaches 1.6–1.8 mm/year before plateauing. The pitting activity is high the first 48 h and then stops. For the P91-steel the maximum corrosion rate is 20 μm/year and the Alloy 625 reaches 0.2 μm/year. Hence, the corrosion rate of Alloy 625 is two magnitudes lower than P91, which in turn is two magnitudes lower than ST45.8. When exposed to deposits from the economizer (Fig. 6 and Fig. 7) the corrosion rate for the ST45.8-steel reaches 1.6 mm/year and pitting activity can be seen the first two days. For the P91-steel the corrosion rate reaches 0.08 mm/year and the pitting is vast during the first few days.
Fig. 1. The corrosion rate and pitting activity for superheater deposits from plant 1 on ST45.8-steel. The maximum corrosion is about 0.25 mm/year.
Fig. 3. The corrosion rate and pitting activity for furnace deposits from plant 1 on ST45.8steel. The corrosion rate reaches about 1.75 mm/year after four days.
3.2.2. Plant 2–4 samples The sample taken from the inside center tube from plant 2 showed no corrosion. The deposit from the superheater was tested with three types of steel and gave a corrosion rate of 0.4 mm/year for the ST45.8steel. The corrosion rate for the P91-steel and the 304L-steel was 0.12 mm/year and 0.035 mm/year, respectively. Pitting activity was initially measured for the ST45.8-steel but not for the two other steels. This is shown in Fig. 8–Fig. 10.
Fig. 4. The corrosion rate for furnace deposits from plant 1 on P91-steel. The corrosion rate almost reaches 0.2 mm/year after three days.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 5. The corrosion rate for furnace deposits from plant 1 on Alloy 625. The corrosion is negligible.
Fig. 6. The corrosion rate and pitting activity for economizer deposits from plant 1 on ST45.8-steel. The corrosion rate is about 1.6 mm/year after three days.
5
Fig. 8. The corrosion rate and pitting activity for superheater deposits from plant 2 on ST45.8-steel. The corrosion rate reached 0.4 mm/year after approximately 30 h and then increased to 0.5 after 75 h.
Fig. 9. The corrosion rate and pitting activity for superheater deposits from plant 2 on P91steel. The corrosion rate reached 0.12 mm/year after five days.
The economizer deposit sample gave a corrosion rate of 0.64 mm/year for the ST45.8-steel and 0.06 mm/year for the P91-steel, see Fig. 11 and Fig. 12. Only limited pitting occurred for the ST45.8-steel and for the P91-steel the pitting is most likely an artifact as the corrosion rate is relatively low. The furnace sample showed no corrosion.
3.2.3. Plant 3–1 sample From plant 3 a fly ash sample was taken from the superheater region. Fig. 13 shows the exposure on ST45.8-steel. A maximum of 0.55 mm/year was reached after approximately 3 days, no pitting activity was registered. For the P91-steel 0.16 mm/year was reached within 24 h and then decreased to approximately 0.8 mm/year after 50 h. During the first 50 h pitting occurred, see Fig. 14.
Fig. 7. The corrosion rate and pitting activity for economizer deposits from plant 1 on P91steel. The corrosion rate reaches about 0.08 mm/year.
Fig. 10. The corrosion rate for superheater deposits from plant 2 on 304L-steel. The corrosion rate reaches 0.035 mm/year after about one day and sustains that level.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
6
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 11. The corrosion rate and pitting activity for economizer deposits from plant 2 on ST45.8-steel. The corrosion rate is steadily increasing during seven days.
Fig. 14. The corrosion rate and pitting activity for fly ash from plant 3 on P91-steel. The maximum corrosion rate 0.16 mm/year is reached after one day.
Fig. 12. The corrosion rate and pitting activity for economizer deposits from plant 2 on P91-steel. The maximum corrosion rate is about 0.06 mm/year The pitting activity is over exaggerated.
Fig. 15. The corrosion rate for furnace deposits from plant 4 on ST45.8-steel. The maximum corrosion rate is 0.7 mm/year and is reached after approximately six days.
3.2.4. Plant 4–1 sample The deposit sample from the only biomass boiler included in the experiments was exposed to three different kinds of steel, ST45.8-, P91-, and 304L-steel, the results shown individually in Fig. 15–Fig. 17, and all together in Fig. 18. For the ST45.8-steel the corrosion rate continued increasing during seven days, and at the experiment end, it had reached 0.7 mm/year. The P91-steel exposure was shorter and was also
increasing slowly, reaching a maximum of 0.12 mm/year before the experiment was stopped. The corrosion rate for the 304L-steel is practically zero. 3.2.5. Plant 5–2 samples For sample 5a the corrosion rate reached 0.12 mm/year on ST45.8steel after approximately two days and then decreased, see Fig. 19. Sample 5b was sintered deposit samples, which was prepared like all other samples, but the corrosion rate was not measurable. 3.2.6. Plant 6–3 samples From plant 6 three different deposit samples from the superheater region was used. For sample (1) the corrosion rate was slowly increasing over approximately five days to 0.37 mm/year on the ST45.8-steel, see Fig. 20. On P91-steel it does not have much effect, only reaching 0.01 mm/year after eight days of exposure shown in Fig. 21. Samples (2) from plant 6 had a lower corrosion rate on the ST45.8steel only reaching 0.15 mm/year after three days and then decrease while the corrosion rate steadily increased for the P91-steel, but only to 0.04 mm/year, see Fig. 22 and Fig. 23. Fig. 24 shows deposit sample (3) from plant 6. The corrosion rate reaches only 0.09 mm/year after approximately three days on ST45.8steel, while the P91-steel did not show any corrosion. 4. Discussion
Fig. 13. The corrosion rate for fly ash from plant 3 on ST45.8-steel. The corrosion rate reaches 0.55 mm/year after three days.
Down-time corrosion starts from no corrosion rate at all and reaches rates of a few hundredths mm/year to a few mm/year. The time it takes
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
7
Fig. 16. The corrosion rate for furnace deposits from plant 4 on P91-steel. The maximum corrosion rate is 0.12 mm/year and is reached after approximately six days.
Fig. 19. The corrosion rate and pitting activity for loop seal deposits from plant 5 on ST45.8-steel. The corrosion rate reaches 0.12 mm/year two days and then decreases.
Fig. 17. The corrosion rate for furnace deposits from plant 4 on 304L-steel. The corrosion is negligible.
Fig. 20. The corrosion rate for superheater deposits (1) from plant 6 on ST45.8-steel. The maximum corrosion rate, 0.037 mm/year is reached after five days and the falls back.
to reach the maximum corrosion rates varies from approximately one day up to a week. In some cases, such as plant 2 (waste boiler, Fig. 11), the corrosion rate is still increasing after several days while others plateaus and some decreases after reaching maximum. The corrosion rate is at its highest the first few days. Some of the measured corrosion rates are considered high. For example, the furnace deposit sample from plant 1 (waste boiler) on ST45.8-steel (Fig. 3) shows corrosion rates between 1.5–1.8
mm/year. In addition the economizer deposit from the same boiler on ST45.8-steel (Fig. 6) shows a corrosion rate of 1.6 mm/year. Also the furnace deposit sample from the biomass boiler (plant 4, Fig. 15) has a corrosion rate of 0.7 mm/year. With corrosion rates of this magnitude it is possible that the down-time corrosion could be worse than the corrosion during operation. In these cases it is of great importance to clean the boiler surfaces without too much delay after boiler shut-down. It is not the material wear that
Fig. 18. A comparison between all three steels, ST45.8, P91, and 304L, in furnace deposits from plant 4.
Fig. 21. The corrosion rate for superheater deposits (1) from plant 6 on P91-steel. The corrosion rate reached 0.01 mm/year.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
8
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
Fig. 22. The corrosion rate and pitting activity for superheater deposits (2) from plant 6 on ST45.8-steel. The corrosion rate reaches 0.15 mm/year after three days.
Fig. 24. The corrosion rate for superheater deposits (3) from plant 6 on ST45.8-steel. The corrosions rate maximum is 0.09 mm/year.
is the main problem, but the possibility for the down-time corrosion to destroy any protective oxide films added to prevent corrosion during operation, especially when pitting occurs simultaneously. Although the corrosion rate is low and in some cases negligible, it is concluded that the corrosion starts during the first 24 h after the boiler has been shut-down. The results suggest that the initiation of corrosion is dependent on the moisture content but the corrosion rate is not. However, it needs to be taken into consideration that all steel test materials used for the experiments were of mint condition and not exposed to high temperature corrosion inside a boiler.
corrosion rate decreases as the S-concentration increases. For the P91-steel there are no clear correlations. The same correlation analyses were made for Ca, see Fig. 27, showing a similar, but weaker, trend as for Cl for both ST45.8-, and P91-steel. Out of all compared elements it is Cl and Ca that show the strongest correlation between concentration and the corrosion rate. It is possible that CaCl2 is formed, which is hygroscopic and could cause corrosion on low-alloyed steel. Other elements, such as K, Zn, and Pb, which normally increases the corrosion, show no such trends in these experiments. In fact, all of them show a slight decrease of the corrosion rate when present in higher concentrations. However, it could be that the effects on the corrosion rate are minor for these elements compared to Cl in these tests.
4.1. Corrosion rate dependence of the elemental concentrations It is well known that some elements and the compounds they form cause corrosion. Commonly mentioned elements are chlorine (Cl), sulfur (S), potassium (K), sodium (Na), lead (Pb), zinc (Zn), and calcium (Ca). All of these elements, except Na, were measured in the deposits using the handheld XRF-instrument, presented in Table 2. By comparing the elemental concentrations to the corrosion rates presented in Fig. 1–Fig. 24 it is possible to see any trends if between the deposit composition and the corrosion. In Fig. 25 the corrosion rates of all samples for the ST45.8-, and the P91-steel are compared to the Cl-concentration of the deposits. For the ST45.8steel the trend is that the corrosion rate increases with the Clconcentration. For the P91-steel the trend is not as obvious but is there to some extent. The same kind of comparison is shown for S in Fig. 26. Here the trend is the opposite for the ST45.8-steel, the
Fig. 23. The corrosion rate for superheater deposits (2) from plant 6 on P91-steel. The corrosion rate reaches 0.04 mm/year after three days.
4.2. Corrosion rate dependence of deposit position The down-time corrosion rate could also be dependent on from which position in the boiler the deposits are taken. There are significant differences between the temperatures, the flue gas composition and surfaces in the different boiler parts. The number of deposits from different positions in the boilers in these experiments was limited. For deposit samples from the furnace (plant 4 using biomass, plant 1, 2a, and 5b using waste, and plant 2b using demolition wood) the corrosion rate varied between 0 to 1.8 mm/year for the ST45.8-steel and from 0.12 to 0.23 mm/year for the P91-steel. The variation was higher within the same fuel type than compared between the three test materials.
Fig. 25. The corrosion rate for the ST45.8-steel increase as the Cl-concentration increase. For the P91-steel the corrosion rate does to some extent increase, as the Cl-concentration increase.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009
F. Jones et al. / Fuel Processing Technology xxx (2015) xxx–xxx
9
5. Conclusions
Fig. 26. The corrosion rate shows a trend of slightly decreasing when the S-concentration increase for ST45.8-steel. Any correlation between S-concentration and corrosion rate is not obvious for the P91-steel.
Then again, the small number of samples does not give a satisfactory statistical base. For the deposit samples from superheaters (plant 1 and 2a using waste, plant 6 using demolition wood) there is a slight trend that the deposits from the waste boilers corrode more than the ones from the demolition wood boiler. The corrosion rate for the waste boilers is between 0.27–0.55 mm/year and 0.01–0.13 mm/year for ST45.8-, and P91-steel, respectively. The corresponding corrosion rates for the demolition wood boiler are 0.04–0.55 mm/year and 0.01–0.08 mm/year. Deposit samples from the economizer region were only available from waste boilers (plant 1 and 2a). The corrosion rate for the ST45.8-steel was well above the rate for the P91-steel, 0.6–1.61 mm/year compared to 0.06–0.08 mm/year. Only one sample from a loop seal superheater was included in the experimental tests (plant 5a using demolition wood). The measured corrosion rate was 0.12 mm/year. With such a small statistical base no direct conclusions can be drawn. However, for the ST45.8-steel it seems like the deposits from the furnace and economizer are more corrosive than the ones from the superheater region, which in turn is higher than the one from the loop seal. For the P91-steel the corrosion in the furnace seems higher than in the superheaters and economizers. Furthermore, biomass appears to be less corrosive than waste and demolition wood, but the difference is very low. Waste gives rise to somewhat higher corrosion rates than demolition wood for deposits from the superheater region.
From this laboratory study it was concluded that down-time corrosion increases the fastest within 24 h up to four days after the deposit has been exposed to ambient air. However, depending on the type of sample origin (position in the boiler and type of fuel used) the downtime corrosion can have a large variation, from negligible rates up to 1.8 mm/year for the ST45.8-steel. When the corrosion rate measures up to 1.8 mm/year for ST45.8-steel, cleaning of boilers with this or similar materials should be prioritized after boiler shut-down. The study showed that there are, in general, great differences in the corrosion rates between the different steel types: ST45.8-steel corrodes one magnitude faster than the P91-steel, in which in turn corrodes one magnitude faster than 304L-steel, which in turn corrodes one magnitude faster than Alloy 625. In addition, the XRF-study shows that the presence of Cl in the deposits seems to affect the corrosion rate more than other elements present. Finally, the results suggest that the initiation of corrosion is dependent on the moisture content but the corrosion rate is not. Acknowledgements This study was financially supported by Värmeforsk (The Thermal Engineering Research Institute in Sweden) and the Swedish Energy Agency (M12-209). References [1] W.M. Cox, D.M. Farrell, J.L. Dawson, Corrosion monitoring for process control, in: D.S. Holmes (Ed.), Dewpoint Corrosion, Horwood Publishing, Chichester 1985, p. 191. [2] T. Eriksson, In-situ mätningar av rökgassidig korrosion; litteratur och utvecklingsstudie, Material- och Kemiteknik, Värmeforsks orienteringsrapport, 105, 1997 (in Swedish). [3] J. Krueger, Operation experience and future perspectives with degradation at high steam parameters in WTE-plants, Mater. High Temp. 20 (2) (2003) 185–193. [4] N. Bolt, E.J.A. Vogelaar, A.S. de Clerc.q, Fireside corrosion in Dutch waste incinerators, Werkst. Korros. 40 (3) (1989) 142–146. [5] A. Stålenheim, P. Henderson, Korrosion i avfallsförbränningsanläggningar, Materialoch Kemiteknik, Värmeforskrapport, 887, 2004 (in Swedish). [6] A. Hjörnhede., R. Norling., Kalibrering av elektorkemisk korrosionssond i laboratoriemiljö och verifiering med fältförsök, Material- och Kemiteknik, Värmeforskrapport, 1210, 2012 (in Swedish). [7] D. Thierry, A. Thorén, C. Leygraf, Corrosion monitoring techniques applied to cooling water and district heating systems, Paper 463. Corrosion 87, NACE, Houston, 1987. [8] C.E. Jaske, J.A. Beavers., N.G. Thompson., Improving plant reliability through corrosion monitoring, Corrosion Prevention and Control, 49 2002, pp. 3–12. [9] R. Schramm, X-ray fluorescence analysis: practical and easy, Fluxana, 2012, ISBN 978-3-00-04 14 76-3. [10] L. König, X-ray fluorescence — the use of handheld XRF analyzers for a quick measurement of elements in waste incineration ashes and waste fuel, University of Borås, report 18, 2013. [11] S.M. Shackley, X-ray fluorescence in spectrometry (XRF) in geoarchaeology, 2011, ISBN 978-1-4419-6886-9. [12] M. Stern, A.L. Geary, J. Electrochem. Soc. (1957) 56–63 (January issue). [13] K. Hladky, J.L. Dawson, The measurement of localized corrosion using electrochemical noise, Corros. Sci. 21 (1981) 317–322. [14] A. Legat, A. Zevnik, The electrochemical noise of mild and stainless steel in various water solutions, Corros. Sci. 35 (1993). [15] P.R. Roberge, R. Beaudoin, V.S. Sastri, Electrochemical noise measurements for field applications, Corros. Sci. 29 (1989) 1231–1233. [16] F. Mansfeld, H. Xiao, Electrochemical noise-analysis of iron exposed to NaCl solutions of different corrosivity, J. Electrochem. Soc. 141 (1994) 1403–1404. [17] C.A. Loto, R.A. Cottis, Electrochemical noise generation during stress corrosion cracking of the high-strength aluminum AA 7075-T6 alloy, Corrosion 45 (1989) 136–141.
Fig. 27. The corrosion rate for the ST45.8-steel increase as the Ca-concentration increase. Any correlation between Ca-concentration and corrosion rate is not obvious for the P91steel.
Please cite this article as: F. Jones, et al., Down-time corrosion in boilers, Fuel Processing Technology (2015), http://dx.doi.org/10.1016/ j.fuproc.2015.10.009