- Email: [email protected]
An investigation into the likely impact of oxy-coal retrofit on fire-side corrosion behavior in utility boilers
International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Contents lists available at ScienceDirect
International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc
An investigation into the likely impact of oxy-coal retrofit on fire-side corrosion behavior in utility boilers Andrew Fry a,∗ , Brad Adams a,1 , Kevin Davis a,1 , Dave Swensen a,1 , Shawn Munson a,1 , William Cox b,2 a b
Reaction Engineering International, 77 West 200 South, Suite 210, Salt Lake City, UT 84101, United States Corrosion Management Ltd., 21 Sedlescombe Park, Rugby CV22 6HL, United Kingdom
a r t i c l e
i n f o
Article history: Received 7 January 2011 Received in revised form 5 May 2011 Accepted 6 May 2011 Available online 8 June 2011 Keywords: Oxy-coal combustion Fire-side corrosion Utility boilers
a b s t r a c t Real-time electrochemical measurements of corrosion rate were performed to evaluate the respective corrosion rates of one boiler waterwall material (SA210) and three boiler superheater materials (T22, P91 and 347H) while firing Utah Western bituminous, Illinois high-sulfur bituminous and Powder River Basin (PRB) sub-bituminous coals in a 1.5 MW pulverized coal-fired furnace. The raw average measured corrosion rates were very low, between 0.0003 and 0.016 mm/year (0.012 and 0.63 mils/year) for most materials under air- and oxy-fired conditions. For some high-sulfur conditions measured corrosion rates were as high as 0.72 mm/year (28 mils/year). Waterwall corrosion rates decreased consistently when converting from air- to oxy-firing while superheater corrosion rates generally increased, although they were less than twice the air-fired rate under most conditions. Corrosion rates for the lower alloyed materials (SA210 and T22) increased significantly during transients from reducing to oxidizing conditions. Measured increases in the corrosion rate of 347H material under high sulfur and low temperature conditions, and associated decrease in corrosion rate at higher temperatures on this alloy, were consistent with the formation of trisulphates in the superheater deposits. The increase of corrosion rate with increased metal temperatures was demonstrated, as was the consistently repeatable nature of the observed results. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Fire-side tube corrosion is a topic of great concern for US utility boilers. Some furnaces have experienced local tube metal loss rates on the order of 2.5 mm/year (100 mils/year). EPRI estimated that fire-side corrosion costs the electric power industry in the USA up to $590 million per year and is responsible for approximately half of unscheduled outages in steam generation units (Syrett and Gorman, 2003). During the last twenty years, the introduction of low NOx firing systems (low NOx burners and overfire air) and efforts to improve thermal efficiency through the use of higher pressure/temperature steam conditions have resulted in growing concern about fire-side corrosion and have led to increasing pressure to increase the high-temperature performance of superheater tube materials. At the same time, concern about greenhouse gas emissions has led to the evaluation of oxy-fuel combustion, as the smaller volume of exhaust combustion gases simplifies the collection of CO2 . Oxy-fuel combustion can change the heat balance
∗ Corresponding author. Tel.: +1 801 364 6925x21; fax: +1 801 364 6977. E-mail address: [email protected] (A. Fry). 1 Tel.: +1 801 364 6925x21. 2 Tel.: +44 0 1788816231. 1750-5836/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijggc.2011.05.021
within the boiler, especially if applied on existing boilers. Additionally, the combustion of coal with oxygen and flue gas recycle (FGR) may affect corrosion behavior in a number of different ways. The mechanism of corrosion often is specific to waterwall, superheater and economizer regions and is affected by factors such as material composition, fuel properties, boiler design and operating conditions. Waterwall corrosion mechanisms that have been identified include oxidation, chlorine-related reactions, gas-phase sulfur interactions, and the deposition and interaction of reduced sulfur forms (Shim et al., 2008; Valentine et al., 2007). In the superheater/reheater area oxidation, active oxidation, high temperature chloride, molten sulfate, carburization, and sulfidation mechanisms have all been indicated as possible mechanisms for tube attack (Reid, 1971; Badin, 1984). A simplistic consideration of oxy-coal combustion indicates that a number of factors may affect corrosion tendencies. Thermodynamics indicate that flame temperature is a function of oxygen concentration in the reactant gases. In addition, a difference in peak combustion temperatures will lead to changes in heat flux and temperature gradients. Each of these factors may influence corrosion rates as a result of one or more of the mechanisms indicated above. The removal of the nitrogen diluent and the consequent increase in concentration of minor species (especially sulfur and chlorine species) during flue gas recycle can provide another
S180
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185 Table 1 Coal properties (values in mass percent unless specified). Utah
Fig. 1. University of Utah’s 1.5 MW pilot-scale pulverized coal furnace (L1500).
variant in the corrosion environment. In addition to the thermal and concentration issues noted, deposit chemistry/stability is known to be strongly affected by alternating oxidizing and reducing conditions (Bakker and Kung, 2000). The presence of extremes in oxygen concentration during oxy-firing could further magnify these effects. Experiments were performed using University of Utah’s 1.5 MW furnace to investigate the possible impacts of oxy-coal combustion retrofit of utility boilers on the fire-side corrosion of heat transfer surfaces. The goals of these tests were two-fold: 1. To demonstrate differences, if any, in corrosion behavior of materials typically utilized in US electric power generation boilers between air-coal and oxy-coal combustion conditions; 2. To develop a data set suitable for validation of mechanisms to predict the corrosion rate of heat transfer surfaces under airand oxy-fired coal combustion conditions. During this testing, Utah bituminous, Illinois bituminous and Powder River Basin (PRB) sub-bituminous coals were fired with air and with a blend of oxygen and flue gas recycle. The corrosion rate of heat transfer surfaces was determined using electrochemical corrosion monitoring, generating real-time data for materials relevant for US utility boilers. 2. Materials and methods The 1.5 MW pilot-scale combustor (L1500) at the University of Utah is a PC-fired furnace that was designed to simulate combustion in low emission, pulverized coal-fired boilers. This unit has been used for many investigations of technologies for NOx and particulate control, including: staging, reburning, SNCR and burner development. The reaction zone of this furnace has a onemeter, square cross section and is approximately 14 m in length. The length is divided into 10 sections, each with various sampling and injection ports. The furnace is refractory lined, with cooling panels in the first four sections to maintain realistic boiler temperature profiles. The multiple ports located in each of the reactor sections allow numerous alternative configurations of sampling, overfire air and reagent injection. The pilot-scale combustor configuration is presented in Fig. 1 with some of its features and sample locations detailed. As detailed in Fig. 1, the L1500 combustion test facility was retrofitted with a stainless steel flue gas recycle pipe, fan and control system to allow exhaust gas to be returned to the burner. These modifications, along with an O2 and CO2 supply and control system, enabled oxy-combustion experimentation to be undertaken on the unit. Three different coals were fired for these experiments, a western bituminous coal from the Utah Skyline mine, a bituminous coal from the Illinois Shay #1 mine, and a sub-bituminous coal from the PRB North Antelope deposit. The purpose of using three coals was to provide a range of sulfur and chlorine concentrations as well
Coal analysis C H N S O Ash Moisture Cl, g/g Volatile matter Fixed carbon HHV, Btu/lb Mineral matter analysis Al Ca Fe Mg Mn P K Si Na S Ti
PRB
Illinois
70.60 5.05 1.42 0.53 10.39 8.38 3.18 290 38.60 49.39 12,606
53.72 3.57 0.78 0.23 13.07 4.94 23.69 <10 33.36 38.01 9078
64.67 5.59 1.12 3.98 16.65 7.99 9.65 790 36.78 45.58 11,598
14.52 6.11 5.09 1.39 0.02 0.59 0.57 60.89 1.41 2.33 0.88
14.78 22.19 5.20 5.17 0.01 1.07 0.35 30.46 1.94 8.83 1.30
17.66 1.87 14.57 0.98 0.02 0.11 2.26 49.28 1.51 2.22 0.85
as mineral matter composition. The properties of these coals are presented in Table 1. In order to obtain a real-time indication of corrosion, a measurement system based on electrochemical noise (EN) sensing was utilized. The principle of operation of this corrosion sensing technique is that spontaneous fluctuations in the measured electrical potential and current signals are generated during electrochemical corrosion activity. The transients are monitored and converted to digital signals and supplied to a computerized data acquisition unit. An estimate of the rate of corrosion can be obtained using the corrosion current density Icorr , which is obtained by replacing the polarization resistance (Rp ) in the standard Stern-Geary equation with the noise resistance (Rn ) value and converting the corrosion current value obtained to an equivalent metal loss rate by application of Faraday’s Law. (Bakker et al., 1992) Icorr =
B Rn
where Icorr – Corrosion current density; B – Stern-Geary coefficient; Rn – Noise resistance. Corrosion rate is computed as a product of the corrosion current density and the material constant. The material constant is a term encompassing the atomic mass of the sensor electrode material, Faraday’s constant, the number of electrons produced in the anodic reaction (2 electrons in this case), and the density of the material. This sensing technique was performed on corrosion elements fabricated from materials representative of heat transfer surfaces in high temperature combustion applications. This technique has been used previously to measure corrosion rates in coal-fired furnaces and validation has been performed against measurements of loss of material measurements (Davis et al., 2004; Linjewile et al., 2003). For this program, one waterwall material and three superheater tube materials were chosen that are typical of those used in US utility boilers. These materials were used for fabrication of the sensor elements in four separate sensor probes. The specifications of the materials chosen are presented in Table 2. The waterwall and superheater sensors differ in physical construction. The sensor elements of the waterwall probe were constructed from the SA210 material and were located at the tip
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
S181
Table 2 Specification of the materials used for corrosion sensor elements. Material
SA 210
T22
P91
347H
HT surface C, % Si, % Mn, % Ni, % Cr, % Mo, % S, % P, % Cu, % Al, % Nb/Cb, % V, % N, % Ta, %
Waterwall 0.07 0.23 0.42 – – – 0.002 0.009 – – – – – –
Superheater 0.11 0.2 0.44 – 2.21 0.95 0.003 0.01 – – – – – –
Superheater 0.10 0.32 0.47 0.15 8.52 0.93 0.002 0.018 0.11 0.01 0.07 0.22 0.044 –
Superheater 0.048 0.40 1.32 9.73 17.45 – 0.008 0.026 – 0.005 0.63 0.078 – 0.02
of the probe on a face orthogonal to the probe axis. When installed in a furnace the probe was positioned so that the elements of the probe were in the same plane as the wall of the furnace. This configuration positions the corroding surface of the sensor elements as if they are a waterwall tube. The elements of the superheater probe were rings fabricated from actual boiler tube, i.e. they were fabricated from commercial T22, P91 and 347H tubes. The corrosion surface of these elements retained the tube manufacturers’ factory finish. These elements were mounted at the end of a long, cooled stainless steel lance. When inserted into the furnace, the orientation of the sensor elements is that of a tube in cross flow, just like a superheater tube in a utility boiler. The corrosion sensor probe and furnace configuration for these tests is presented in Fig. 2. The sensor elements of the waterwall probe typically were maintained at 655 K and the probe was installed in section 2 of the L1500 with gas temperatures about 1533 K. The sensor elements on the superheater probes typically were controlled at 761 K and were installed in a location where the gas temperature was approximately 1255 K. To target this gas temperature the superheater probes were installed in section 10 of the L1500. For all fire-side corrosion tests, the firing rate was 1.03 MW. For the majority of these tests, the burner was staged with a stoichiometric ratio of 0.9. The overall stoichiometric ratio was sufficient to produce 3.0% O2 (dry) in the flue gas, dependent on air or oxy-fired conditions. The overfire air was introduced in section 6 of the furnace. These operating conditions are referred to as the baseline conditions for this study.
Fig. 3. Fate of sulfur for each coal and firing condition.
The average corrosion rates measured by this method assume that the metal loss occurs uniformly over the surface of the corrosion sensor element. In reality, the loss of material is more likely to occur in discrete locations on the sensor element associated with deposit and flow characteristics. For the superheater probes in short-term tests, this discrete area probably was limited to about one-third of the overall surface on the leading and trailing sides of the tube in cross flow. Peak corrosion rates therefore are likely to be approximately 3 times higher than the overall average rates reported here. However, this was a pilot-scale facility that was operated only for an eight to ten hour period each day and the corrosion probes were removed each night. Furthermore, the objective of the trials was to obtain comparative data on corrosion behavior in the various service environments, rather than definitive materials corrosion rate estimates. Thus, the relative corrosion rate values were more relevant for comparison purposes than were the absolute rate values. 3. Results 3.1. Sulfur balance The sulfur concentrations in both the baghouse ash and the entrained ash samples were measured along with all of the major mineral species using a Bruker S4 X-ray fluorescence spectrometer. The sulfur contents in the water condensate samples were determined using EPA method 200.7. These data, along with the flue gas SO2 measurements, were used to perform a sulfur material balance for each of the conditions of interest. Flue gas SO2 concentrations were determined by averaging the measured values corresponding to each test condition across the entire test campaign. These data, along with furnace operational data collected during the testing, were used to close a sulfur material balance. The results of the sulfur material balance are presented in Fig. 3. In Fig. 3 the PRB/Oxy condition shows a significant amount of sulfur in the entrained ash. This is likely due to the high calcium content of the PRB mineral matter as described in Table 1. 3.2. Overall corrosion results
Fig. 2. Corrosion sensor probe configuration.
Over a period of six weeks of furnace operation, extensive measurements of the corrosion rates for each of the materials were performed at baseline conditions using the EN technique. The EN
S182
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 5. Superheater probe corrosion rates for oxidizing and reducing conditions while oxy- and air-firing Illinois coal. Fig. 4. Summary of average corrosion rates for the baseline conditions.
data were evaluated during windows of time where the furnace and probes were operating stably at the desired conditions. All of the data from these windows of operation were reduced to an average corrosion rate for every combination of material, coal and firing condition. The average corrosion rates measured for the baseline conditions are presented in Fig. 4. Fig. 4 shows comparative rates under air- and oxy-fuel firing for each of the materials under test. Relatively low corrosion rates were evident for all of the materials for all coals and firing conditions, though the sensors were extremely responsive to instantaneous changes in the corrosivity of the furnace environment. However, in five conditions – four with T22 material and one with 347H material, the corrosion rates were considerably higher. These conditions included (for example) the T22 material with Skyline and Illinois coals for both air and oxy-fired conditions and the 347H material under Illinois coal with oxy-firing conditions. These five service condition and tube material combinations gave indicated rates on the order of 0.21–0.72 mm/year (8.3–28.3 mils/year). Corrosion rates for the remaining material and coal combinations ranged from 0.0034 to 0.016 mm/year (0.13–0.62 mils/year). Additional experiments were performed to determine the effects of transitions between oxidizing and reducing conditions. For these experiments, the superheater probes were moved to section five of the furnace, just inside of the OFA ports. The burner stoichiometric ratio was varied between 0.9 and either 1.16 for air-fired, and 1.09 and oxy-fired conditions. The upper limit of the variation was designed to produce 3% O2 (dry) in the furnace exit flue gas. As the furnace was varied between staged and unstaged operation, air or O2 and FGR was moved between the burner secondary and the air ports. Therefore a good indicator of transitions between staged and unstaged conditions was the flow rate of the outer secondary air register. This flow rate, along with superheater probe corrosion rates, is plotted in Fig. 5 for the staging experiments described above.
Fig. 6. Temperature dependence of the superheater corrosion probes while oxyfiring Illinois coal.
Further experiments to determine the dependence of corrosion rate on temperature were performed. During these experiments the set-point temperatures of the superheater probes were varied between the baseline condition and the maximum recommended service temperature for each material while firing Illinois coal. Some of these data are presented in Fig. 6. In Fig. 6, the operating temperatures and corrosion rates of each of the superheater probes are presented. The grayed areas on each plot indicate periods over which the rate data were averaged and used to obtain the overall average for that condition, which results are tabulated in Table 3. Further experiments were performed to determine the temperature dependency of the corrosion rates. For these experiments, the controlled temperature set-points of the corrosion sensor elements were varied from their baseline condition up to the maximum operating temperature of the material. The maximum temperature set points were 728 K, 839 K, 866 K and 978 K for the SA210, T22,
Table 3 Conditions and measured corrosion rates during temperature variation experiments with the superheater probes. Time period
Firing condition
Temperature condition
T22 corrosion rate (mm/year)
P91 corrosion rate (mm/year)
347H corrosion rate (mm/year)
A B C D E
Illinois/Air Illinois/Oxy Illinois/Oxy Illinois/Oxy Illinois/Oxy
Baseline Baseline Maximum Baseline Maximum
0.661 1.079 1.462 0.552 1.174
0.004 0.011 0.059 0.008 0.057
0.010 0.001 0.050 0.001 0.042
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 7. Inverse temperature dependence of 347H superheat probe corrosion rate on temperature while oxy-firing Illinois coal.
P91 and 347H materials, respectively (the differing temperatures being related to the manufacturers’ recommended service temperature for each of the materials). The first temperature variation experiment returned surprising results for the 347H material while oxy-firing Illinois coal. Immediately when the probe was heated, the corrosion rate fell dramatically. This behavior is depicted in Fig. 7. At the conclusion of the pilot-scale testing, the elements were removed from the corrosion probes for metallographic analysis. Samples of the as received materials from which the elements were fabricated were also analyzed. Morphological analyses were performed using scanning electron microscopy (SEM) with energydispersive X-ray area analysis. Backscattered electron images and the averaged SEM morphological compositions of the preand post-test 347H material are presented in Fig. 8 and Fig. 9 respectively. 4. Discussion The sulfur material balance in Fig. 3 shows the variability of the overall amount of sulfur present in the furnace due to the different coals and also indicates whether the sulfur was present in the gasphase or was particulate bound. However, Fig. 3 does not provide information about the variability of the concentration of SO2 in the
S183
Fig. 9. Comparison of 347H composition pre- and post-test from the morphological analysis. Table 4 Average measured gas-phase SO2 . Condition
Average SO2 concentration (ppmv, dry)
PRB – Air PRB – Oxy Utah – Air Utah – Oxy Illinois – Air Illinois – Oxy
129 289 446 1754 3219 17642
gas-phase due to the inherent differences between air- and oxyfiring. When oxy-firing, the SO2 formed from sulfur in the coal is diluted by only oxygen and moisture, instead of nitrogen, oxygen and moisture as under the air-fired condition. If various mechanisms for sulfur removal from the gas phase are neglected, this leads to an increase of approximately 5 times in gas-phase pollutant concentrations. To illustrate this point, the average of the measured SO2 concentrations for all coals and firing conditions are presented in Table 4. The change in the furnace environment characterized by the change in SO2 concentrations detailed in Table 4 undoubtedly play a role in the associated measured corrosion rates. For the Illinois-oxy condition, the measured SO2 concentration may not be
Fig. 8. Backscattered electron images of the 347H material, pre- and post-test.
S184
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 10. Average increase in corrosion rate when converting from air- to oxy-firing.
acceptable for utility boiler operation and changes to the gas handling equipment may have to be considered to remove the sulfur. Fig. 4 details a significant trend between the air and oxyfired conditions. For the waterwall probe (SA210), the corrosion rate decreased when the combustion conditions were changed from air- to oxy-firing for all conditions tested. By contrast, for all of the superheater probes (T22, P91 and 347H) the corrosion rate increased when combustion conditions in the furnace were changed from air- to oxy-firing for all but one condition tested. The only superheater material and coal that exhibited behavior that was contrary to the general trend of increasing corrosion rates under oxy-firing was the T22 material with Illinois coal. To examine these trends in more detail it is useful to look at the plot of the increase in corrosion rate when converting from air- to oxy-firing for all materials and conditions. This plot is presented in Fig. 10. The decrease in waterwall corrosion rate when converting from air to oxy-fired conditions, although involving minor corrosion rates, could be a significant circumstance. An expected decrease in corrosion rate to these heat transfer surfaces when retrofitting for oxy-combustion could be a facilitating result for the technology. The difference in behavior between the waterwall and superheater probes may be explained by the difference between oxidizing and reducing conditions. For these data, the waterwall probe was installed in a region with stoichiometry of 0.9, whereas the superheater probes experienced superstoichiometric conditions. The corrosion rate behavior of the superheater probes was not concerning for most of the materials and coals tested. The increase in corrosion rates was small, compounding an already small corrosion rate. The results observed here that may be concerning and warrant further investigation are the P91 material with Skyline coal and the 347H material with Illinois coal. The dramatic increase in corrosion rate for the 374H probe may be explained by considering the temperature response detailed in Fig. 7. These data showed that the corrosion rate was unexpectedly high (1.4 mm/year) for the 347H material, which typically would be considered to be quite corrosion-resistant at high temperature, at its ‘baseline condition’ temperature while oxy-firing Illinois coal. However, the monitoring data also show that at the elevated temperature condition of 978 K, the corrosion rate again was low. These data were consistent with a reported mechanism whereby trisulphate species attack the high nickel alloy at low temperatures, but are decomposed and volatilized at high temperatures (Viswanathan and Bakker, 2000). The morphological analysis presented in Fig. 9 showed that for the 347H material, the nickel
content of the scale decreased while the sulfur content increased. These observations are consistent with a mechanism for sulfur attack on nickel. This is considered to be the first occasion, however, that such behavior has been confirmed by real-time corrosion rate measurements. Once the behavior was identified, the average corrosion rates for the Illinois/oxy firing condition with the 347H material were scrutinized again. It was determined that the data producing the overall average had been bimodal. Significantly lower corrosion rates were measured for these conditions following high temperature operation, or after cleaning of the corrosion elements to remove residual corrosion/ash deposits from earlier lower-temperature exposure. For the initial hour and a half of the staging tests summarized in Fig. 5, the corrosion rates appeared to be coming to steady state after the probes had been reinstalled in the furnace. This was probably because the deposit characteristics were different for this probe location and the surface condition on the probes required a longerthan-normal period to equilibrate to the new flue gas environment. During the first half of the day, air-fired conditions were tested. At this time, two variations from reducing to oxidizing operation were tested. The first variation occurred while the probes were still equilibrating. The second variation showed a strong impact on corrosion rate for the T22 material precisely as the conditions were transitioned from reducing to oxidizing. For the second half of the day, oxy-fired conditions were maintained. Conditions were varied from oxidizing to reducing on two occasions. Under oxy-fired conditions, the T22 corrosion rate spiked during the transient from oxidizing to reducing conditions, which was the opposite behavior to that observed under the air-fired conditions. It appeared that the higher alloy materials were unaffected by transients between oxidizing and reducing conditions. Similar behavior was observed for the waterwall sensor made from the SA 210 material. These results indicate that the corrosion rate of low-alloyed materials may be very high in regions of variable stoichiometric ratio, such as the near burner region and around OFA ports. The results presented above describe consistent transient behavior obtained in short-duration exposures that coupon tests with synthesized gas and deposits cannot resolve. Data such as these can be used in advanced combustion modeling to develop and tune corrosion mechanisms for predicting accurately the impacts of fluctuations between oxidizing and reducing conditions. The corrosion data presented in Fig. 6 and Table 3 indicate that in the absence of trisulphates, the temperature dependence of the 347H corrosion rate was proportional, as expected. These data also illustrate the sensitivity and repeatability of electrochemical noisebased corrosion rate measurements.
5. Conclusions Electrochemical sensing technology was used to determine the fire-side corrosion rates of one waterwall (SA210) and three superheater (T22, P91 and 347H) materials while air- and oxy-firing Utah western bituminous, Illinois high-sulfur bituminous and PRB sub-bituminous coals. The raw average measured corrosion rates were very low, between 0.0003 and 0.016 mm/year (0.012 and 0.63 mils/year) for most conditions. The corrosion rates were higher for the T22 material while air- and oxy-firing Utah and Illinois coals and for the 347H material when oxy-firing Illinois coal; and were in the range of 0.21–0.72 mm/year (8.3–28.3 mils/year). These corrosion rates assume that the loss of material had occurred evenly over the entire sensor element. Actual peak corrosion rates would be expected to be on the order of 3 times higher. Waterwall (SA210) corrosion rates consistently decreased when converting from air to oxy-firing for all coals. An expected decrease in the rate of corrosion attack on these heat transfer surfaces when
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
retrofitting for oxy-combustion could be a facilitating result for the technology. Superheater corrosion rates increased when converting from air- to oxy-firing for all conditions with the exception of the T22 material when firing Illinois coal. During these tests, the flue gas environment was sub-stoichiometric for the waterwall probes and super-stoichiometric for the superheater probes. The increase in superheater corrosion rate when comparing the oxy-fired condition with the air-fired condition was generally very small (double or less) with the exception of the P91 material with Utah coal (5.2 times higher) and 347H material with Illinois coal (52.2 times higher). Corrosion rates for the lower alloyed materials (SA210 and T22) were shown to increase significantly during transients from reducing to oxidizing conditions when air-firing, and from oxidizing to reducing conditions when oxy-firing. Such transients are likely to contribute greatly to practical in-plant corrosion rates in the nearburner and near-OFA port regions. These effects cannot be resolved using coupon tests. Measured increases in the corrosion rate of 347H material under high sulfur and low temperature conditions were consistent with the presence of trisulphate deposits on the superheater probes. It was demonstrated conclusively, and for the first time by real-time corrosion monitoring, that these species are promptly decomposed by operating at higher material temperatures, reducing the subsequent corrosion rates to much lower levels, confirming earlier theoretical predictions based in laboratory coupon exposures. The increase in corrosion rates with increased probe material temperatures was demonstrated as was with the consistently repeatable nature of the results. Acknowledgements This material is based upon work supported by the Department of Energy under Award Number DE-NT0005288. Project Manager is
S185
Timothy Fout. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. References Badin, E.J., 1984. Coal Combustion Chemistry – Correlation Aspects. Elsevier, New York. Bakker, W.T., Kung, S.C., 2000. Waterwall corrosion in coal-fired boilers – a new culprit: FeS. In: Proceedings of the NACE International Symposium , Orlando, FL, March 26–31. Bakker, W.T., et al., 1992. High-temperature fireside corrosion monitoring in the superheater section of a pulverized-coal-fired boiler. In: EPRI , Palo Alto, CA, TR-101799. Davis, K.A., et al., 2004. A multi-point corrosion monitoring system applied in a 1,300 MW coal-fired boiler. Anti-Corros. Methods Mater. 51 (5). Linjewile, T., et al., 2003. Prediction and real-time monitoring techniques for corrosion characterization in furnaces. Mater. High Temp. 20, 175. Reid, W.T., 1971. External Corrosion and Deposits: Boilers and Gas Turbines. Elsevier, New York. Syrett, B.C., Gorman, J.A., 2003. Cost of corrosion in the electric power industry – an update. Mater. Perform. 42 (2), 32–38. Shim, H.-S., et al., 2008. Development of fire-side waterwall corrosion correlations using pilot-scale test furnace. Fuel 15–16, 3353–3361. Valentine, J.R., et al., 2007. CFD evaluation of waterwall wastage in coal-fired utility boilers. Energy Fuels 21, 242. Viswanathan, R., Bakker, W.T., 2000. Materials for boilers in ultra supercritical power plants. In: Proceedings of the International Joint Power Generation Conference , Miami Beach, FL, July 23–26.
Contents lists available at ScienceDirect
International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc
An investigation into the likely impact of oxy-coal retrofit on fire-side corrosion behavior in utility boilers Andrew Fry a,∗ , Brad Adams a,1 , Kevin Davis a,1 , Dave Swensen a,1 , Shawn Munson a,1 , William Cox b,2 a b
Reaction Engineering International, 77 West 200 South, Suite 210, Salt Lake City, UT 84101, United States Corrosion Management Ltd., 21 Sedlescombe Park, Rugby CV22 6HL, United Kingdom
a r t i c l e
i n f o
Article history: Received 7 January 2011 Received in revised form 5 May 2011 Accepted 6 May 2011 Available online 8 June 2011 Keywords: Oxy-coal combustion Fire-side corrosion Utility boilers
a b s t r a c t Real-time electrochemical measurements of corrosion rate were performed to evaluate the respective corrosion rates of one boiler waterwall material (SA210) and three boiler superheater materials (T22, P91 and 347H) while firing Utah Western bituminous, Illinois high-sulfur bituminous and Powder River Basin (PRB) sub-bituminous coals in a 1.5 MW pulverized coal-fired furnace. The raw average measured corrosion rates were very low, between 0.0003 and 0.016 mm/year (0.012 and 0.63 mils/year) for most materials under air- and oxy-fired conditions. For some high-sulfur conditions measured corrosion rates were as high as 0.72 mm/year (28 mils/year). Waterwall corrosion rates decreased consistently when converting from air- to oxy-firing while superheater corrosion rates generally increased, although they were less than twice the air-fired rate under most conditions. Corrosion rates for the lower alloyed materials (SA210 and T22) increased significantly during transients from reducing to oxidizing conditions. Measured increases in the corrosion rate of 347H material under high sulfur and low temperature conditions, and associated decrease in corrosion rate at higher temperatures on this alloy, were consistent with the formation of trisulphates in the superheater deposits. The increase of corrosion rate with increased metal temperatures was demonstrated, as was the consistently repeatable nature of the observed results. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Fire-side tube corrosion is a topic of great concern for US utility boilers. Some furnaces have experienced local tube metal loss rates on the order of 2.5 mm/year (100 mils/year). EPRI estimated that fire-side corrosion costs the electric power industry in the USA up to $590 million per year and is responsible for approximately half of unscheduled outages in steam generation units (Syrett and Gorman, 2003). During the last twenty years, the introduction of low NOx firing systems (low NOx burners and overfire air) and efforts to improve thermal efficiency through the use of higher pressure/temperature steam conditions have resulted in growing concern about fire-side corrosion and have led to increasing pressure to increase the high-temperature performance of superheater tube materials. At the same time, concern about greenhouse gas emissions has led to the evaluation of oxy-fuel combustion, as the smaller volume of exhaust combustion gases simplifies the collection of CO2 . Oxy-fuel combustion can change the heat balance
∗ Corresponding author. Tel.: +1 801 364 6925x21; fax: +1 801 364 6977. E-mail address: [email protected] (A. Fry). 1 Tel.: +1 801 364 6925x21. 2 Tel.: +44 0 1788816231. 1750-5836/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijggc.2011.05.021
within the boiler, especially if applied on existing boilers. Additionally, the combustion of coal with oxygen and flue gas recycle (FGR) may affect corrosion behavior in a number of different ways. The mechanism of corrosion often is specific to waterwall, superheater and economizer regions and is affected by factors such as material composition, fuel properties, boiler design and operating conditions. Waterwall corrosion mechanisms that have been identified include oxidation, chlorine-related reactions, gas-phase sulfur interactions, and the deposition and interaction of reduced sulfur forms (Shim et al., 2008; Valentine et al., 2007). In the superheater/reheater area oxidation, active oxidation, high temperature chloride, molten sulfate, carburization, and sulfidation mechanisms have all been indicated as possible mechanisms for tube attack (Reid, 1971; Badin, 1984). A simplistic consideration of oxy-coal combustion indicates that a number of factors may affect corrosion tendencies. Thermodynamics indicate that flame temperature is a function of oxygen concentration in the reactant gases. In addition, a difference in peak combustion temperatures will lead to changes in heat flux and temperature gradients. Each of these factors may influence corrosion rates as a result of one or more of the mechanisms indicated above. The removal of the nitrogen diluent and the consequent increase in concentration of minor species (especially sulfur and chlorine species) during flue gas recycle can provide another
S180
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185 Table 1 Coal properties (values in mass percent unless specified). Utah
Fig. 1. University of Utah’s 1.5 MW pilot-scale pulverized coal furnace (L1500).
variant in the corrosion environment. In addition to the thermal and concentration issues noted, deposit chemistry/stability is known to be strongly affected by alternating oxidizing and reducing conditions (Bakker and Kung, 2000). The presence of extremes in oxygen concentration during oxy-firing could further magnify these effects. Experiments were performed using University of Utah’s 1.5 MW furnace to investigate the possible impacts of oxy-coal combustion retrofit of utility boilers on the fire-side corrosion of heat transfer surfaces. The goals of these tests were two-fold: 1. To demonstrate differences, if any, in corrosion behavior of materials typically utilized in US electric power generation boilers between air-coal and oxy-coal combustion conditions; 2. To develop a data set suitable for validation of mechanisms to predict the corrosion rate of heat transfer surfaces under airand oxy-fired coal combustion conditions. During this testing, Utah bituminous, Illinois bituminous and Powder River Basin (PRB) sub-bituminous coals were fired with air and with a blend of oxygen and flue gas recycle. The corrosion rate of heat transfer surfaces was determined using electrochemical corrosion monitoring, generating real-time data for materials relevant for US utility boilers. 2. Materials and methods The 1.5 MW pilot-scale combustor (L1500) at the University of Utah is a PC-fired furnace that was designed to simulate combustion in low emission, pulverized coal-fired boilers. This unit has been used for many investigations of technologies for NOx and particulate control, including: staging, reburning, SNCR and burner development. The reaction zone of this furnace has a onemeter, square cross section and is approximately 14 m in length. The length is divided into 10 sections, each with various sampling and injection ports. The furnace is refractory lined, with cooling panels in the first four sections to maintain realistic boiler temperature profiles. The multiple ports located in each of the reactor sections allow numerous alternative configurations of sampling, overfire air and reagent injection. The pilot-scale combustor configuration is presented in Fig. 1 with some of its features and sample locations detailed. As detailed in Fig. 1, the L1500 combustion test facility was retrofitted with a stainless steel flue gas recycle pipe, fan and control system to allow exhaust gas to be returned to the burner. These modifications, along with an O2 and CO2 supply and control system, enabled oxy-combustion experimentation to be undertaken on the unit. Three different coals were fired for these experiments, a western bituminous coal from the Utah Skyline mine, a bituminous coal from the Illinois Shay #1 mine, and a sub-bituminous coal from the PRB North Antelope deposit. The purpose of using three coals was to provide a range of sulfur and chlorine concentrations as well
Coal analysis C H N S O Ash Moisture Cl, g/g Volatile matter Fixed carbon HHV, Btu/lb Mineral matter analysis Al Ca Fe Mg Mn P K Si Na S Ti
PRB
Illinois
70.60 5.05 1.42 0.53 10.39 8.38 3.18 290 38.60 49.39 12,606
53.72 3.57 0.78 0.23 13.07 4.94 23.69 <10 33.36 38.01 9078
64.67 5.59 1.12 3.98 16.65 7.99 9.65 790 36.78 45.58 11,598
14.52 6.11 5.09 1.39 0.02 0.59 0.57 60.89 1.41 2.33 0.88
14.78 22.19 5.20 5.17 0.01 1.07 0.35 30.46 1.94 8.83 1.30
17.66 1.87 14.57 0.98 0.02 0.11 2.26 49.28 1.51 2.22 0.85
as mineral matter composition. The properties of these coals are presented in Table 1. In order to obtain a real-time indication of corrosion, a measurement system based on electrochemical noise (EN) sensing was utilized. The principle of operation of this corrosion sensing technique is that spontaneous fluctuations in the measured electrical potential and current signals are generated during electrochemical corrosion activity. The transients are monitored and converted to digital signals and supplied to a computerized data acquisition unit. An estimate of the rate of corrosion can be obtained using the corrosion current density Icorr , which is obtained by replacing the polarization resistance (Rp ) in the standard Stern-Geary equation with the noise resistance (Rn ) value and converting the corrosion current value obtained to an equivalent metal loss rate by application of Faraday’s Law. (Bakker et al., 1992) Icorr =
B Rn
where Icorr – Corrosion current density; B – Stern-Geary coefficient; Rn – Noise resistance. Corrosion rate is computed as a product of the corrosion current density and the material constant. The material constant is a term encompassing the atomic mass of the sensor electrode material, Faraday’s constant, the number of electrons produced in the anodic reaction (2 electrons in this case), and the density of the material. This sensing technique was performed on corrosion elements fabricated from materials representative of heat transfer surfaces in high temperature combustion applications. This technique has been used previously to measure corrosion rates in coal-fired furnaces and validation has been performed against measurements of loss of material measurements (Davis et al., 2004; Linjewile et al., 2003). For this program, one waterwall material and three superheater tube materials were chosen that are typical of those used in US utility boilers. These materials were used for fabrication of the sensor elements in four separate sensor probes. The specifications of the materials chosen are presented in Table 2. The waterwall and superheater sensors differ in physical construction. The sensor elements of the waterwall probe were constructed from the SA210 material and were located at the tip
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
S181
Table 2 Specification of the materials used for corrosion sensor elements. Material
SA 210
T22
P91
347H
HT surface C, % Si, % Mn, % Ni, % Cr, % Mo, % S, % P, % Cu, % Al, % Nb/Cb, % V, % N, % Ta, %
Waterwall 0.07 0.23 0.42 – – – 0.002 0.009 – – – – – –
Superheater 0.11 0.2 0.44 – 2.21 0.95 0.003 0.01 – – – – – –
Superheater 0.10 0.32 0.47 0.15 8.52 0.93 0.002 0.018 0.11 0.01 0.07 0.22 0.044 –
Superheater 0.048 0.40 1.32 9.73 17.45 – 0.008 0.026 – 0.005 0.63 0.078 – 0.02
of the probe on a face orthogonal to the probe axis. When installed in a furnace the probe was positioned so that the elements of the probe were in the same plane as the wall of the furnace. This configuration positions the corroding surface of the sensor elements as if they are a waterwall tube. The elements of the superheater probe were rings fabricated from actual boiler tube, i.e. they were fabricated from commercial T22, P91 and 347H tubes. The corrosion surface of these elements retained the tube manufacturers’ factory finish. These elements were mounted at the end of a long, cooled stainless steel lance. When inserted into the furnace, the orientation of the sensor elements is that of a tube in cross flow, just like a superheater tube in a utility boiler. The corrosion sensor probe and furnace configuration for these tests is presented in Fig. 2. The sensor elements of the waterwall probe typically were maintained at 655 K and the probe was installed in section 2 of the L1500 with gas temperatures about 1533 K. The sensor elements on the superheater probes typically were controlled at 761 K and were installed in a location where the gas temperature was approximately 1255 K. To target this gas temperature the superheater probes were installed in section 10 of the L1500. For all fire-side corrosion tests, the firing rate was 1.03 MW. For the majority of these tests, the burner was staged with a stoichiometric ratio of 0.9. The overall stoichiometric ratio was sufficient to produce 3.0% O2 (dry) in the flue gas, dependent on air or oxy-fired conditions. The overfire air was introduced in section 6 of the furnace. These operating conditions are referred to as the baseline conditions for this study.
Fig. 3. Fate of sulfur for each coal and firing condition.
The average corrosion rates measured by this method assume that the metal loss occurs uniformly over the surface of the corrosion sensor element. In reality, the loss of material is more likely to occur in discrete locations on the sensor element associated with deposit and flow characteristics. For the superheater probes in short-term tests, this discrete area probably was limited to about one-third of the overall surface on the leading and trailing sides of the tube in cross flow. Peak corrosion rates therefore are likely to be approximately 3 times higher than the overall average rates reported here. However, this was a pilot-scale facility that was operated only for an eight to ten hour period each day and the corrosion probes were removed each night. Furthermore, the objective of the trials was to obtain comparative data on corrosion behavior in the various service environments, rather than definitive materials corrosion rate estimates. Thus, the relative corrosion rate values were more relevant for comparison purposes than were the absolute rate values. 3. Results 3.1. Sulfur balance The sulfur concentrations in both the baghouse ash and the entrained ash samples were measured along with all of the major mineral species using a Bruker S4 X-ray fluorescence spectrometer. The sulfur contents in the water condensate samples were determined using EPA method 200.7. These data, along with the flue gas SO2 measurements, were used to perform a sulfur material balance for each of the conditions of interest. Flue gas SO2 concentrations were determined by averaging the measured values corresponding to each test condition across the entire test campaign. These data, along with furnace operational data collected during the testing, were used to close a sulfur material balance. The results of the sulfur material balance are presented in Fig. 3. In Fig. 3 the PRB/Oxy condition shows a significant amount of sulfur in the entrained ash. This is likely due to the high calcium content of the PRB mineral matter as described in Table 1. 3.2. Overall corrosion results
Fig. 2. Corrosion sensor probe configuration.
Over a period of six weeks of furnace operation, extensive measurements of the corrosion rates for each of the materials were performed at baseline conditions using the EN technique. The EN
S182
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 5. Superheater probe corrosion rates for oxidizing and reducing conditions while oxy- and air-firing Illinois coal. Fig. 4. Summary of average corrosion rates for the baseline conditions.
data were evaluated during windows of time where the furnace and probes were operating stably at the desired conditions. All of the data from these windows of operation were reduced to an average corrosion rate for every combination of material, coal and firing condition. The average corrosion rates measured for the baseline conditions are presented in Fig. 4. Fig. 4 shows comparative rates under air- and oxy-fuel firing for each of the materials under test. Relatively low corrosion rates were evident for all of the materials for all coals and firing conditions, though the sensors were extremely responsive to instantaneous changes in the corrosivity of the furnace environment. However, in five conditions – four with T22 material and one with 347H material, the corrosion rates were considerably higher. These conditions included (for example) the T22 material with Skyline and Illinois coals for both air and oxy-fired conditions and the 347H material under Illinois coal with oxy-firing conditions. These five service condition and tube material combinations gave indicated rates on the order of 0.21–0.72 mm/year (8.3–28.3 mils/year). Corrosion rates for the remaining material and coal combinations ranged from 0.0034 to 0.016 mm/year (0.13–0.62 mils/year). Additional experiments were performed to determine the effects of transitions between oxidizing and reducing conditions. For these experiments, the superheater probes were moved to section five of the furnace, just inside of the OFA ports. The burner stoichiometric ratio was varied between 0.9 and either 1.16 for air-fired, and 1.09 and oxy-fired conditions. The upper limit of the variation was designed to produce 3% O2 (dry) in the furnace exit flue gas. As the furnace was varied between staged and unstaged operation, air or O2 and FGR was moved between the burner secondary and the air ports. Therefore a good indicator of transitions between staged and unstaged conditions was the flow rate of the outer secondary air register. This flow rate, along with superheater probe corrosion rates, is plotted in Fig. 5 for the staging experiments described above.
Fig. 6. Temperature dependence of the superheater corrosion probes while oxyfiring Illinois coal.
Further experiments to determine the dependence of corrosion rate on temperature were performed. During these experiments the set-point temperatures of the superheater probes were varied between the baseline condition and the maximum recommended service temperature for each material while firing Illinois coal. Some of these data are presented in Fig. 6. In Fig. 6, the operating temperatures and corrosion rates of each of the superheater probes are presented. The grayed areas on each plot indicate periods over which the rate data were averaged and used to obtain the overall average for that condition, which results are tabulated in Table 3. Further experiments were performed to determine the temperature dependency of the corrosion rates. For these experiments, the controlled temperature set-points of the corrosion sensor elements were varied from their baseline condition up to the maximum operating temperature of the material. The maximum temperature set points were 728 K, 839 K, 866 K and 978 K for the SA210, T22,
Table 3 Conditions and measured corrosion rates during temperature variation experiments with the superheater probes. Time period
Firing condition
Temperature condition
T22 corrosion rate (mm/year)
P91 corrosion rate (mm/year)
347H corrosion rate (mm/year)
A B C D E
Illinois/Air Illinois/Oxy Illinois/Oxy Illinois/Oxy Illinois/Oxy
Baseline Baseline Maximum Baseline Maximum
0.661 1.079 1.462 0.552 1.174
0.004 0.011 0.059 0.008 0.057
0.010 0.001 0.050 0.001 0.042
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 7. Inverse temperature dependence of 347H superheat probe corrosion rate on temperature while oxy-firing Illinois coal.
P91 and 347H materials, respectively (the differing temperatures being related to the manufacturers’ recommended service temperature for each of the materials). The first temperature variation experiment returned surprising results for the 347H material while oxy-firing Illinois coal. Immediately when the probe was heated, the corrosion rate fell dramatically. This behavior is depicted in Fig. 7. At the conclusion of the pilot-scale testing, the elements were removed from the corrosion probes for metallographic analysis. Samples of the as received materials from which the elements were fabricated were also analyzed. Morphological analyses were performed using scanning electron microscopy (SEM) with energydispersive X-ray area analysis. Backscattered electron images and the averaged SEM morphological compositions of the preand post-test 347H material are presented in Fig. 8 and Fig. 9 respectively. 4. Discussion The sulfur material balance in Fig. 3 shows the variability of the overall amount of sulfur present in the furnace due to the different coals and also indicates whether the sulfur was present in the gasphase or was particulate bound. However, Fig. 3 does not provide information about the variability of the concentration of SO2 in the
S183
Fig. 9. Comparison of 347H composition pre- and post-test from the morphological analysis. Table 4 Average measured gas-phase SO2 . Condition
Average SO2 concentration (ppmv, dry)
PRB – Air PRB – Oxy Utah – Air Utah – Oxy Illinois – Air Illinois – Oxy
129 289 446 1754 3219 17642
gas-phase due to the inherent differences between air- and oxyfiring. When oxy-firing, the SO2 formed from sulfur in the coal is diluted by only oxygen and moisture, instead of nitrogen, oxygen and moisture as under the air-fired condition. If various mechanisms for sulfur removal from the gas phase are neglected, this leads to an increase of approximately 5 times in gas-phase pollutant concentrations. To illustrate this point, the average of the measured SO2 concentrations for all coals and firing conditions are presented in Table 4. The change in the furnace environment characterized by the change in SO2 concentrations detailed in Table 4 undoubtedly play a role in the associated measured corrosion rates. For the Illinois-oxy condition, the measured SO2 concentration may not be
Fig. 8. Backscattered electron images of the 347H material, pre- and post-test.
S184
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
Fig. 10. Average increase in corrosion rate when converting from air- to oxy-firing.
acceptable for utility boiler operation and changes to the gas handling equipment may have to be considered to remove the sulfur. Fig. 4 details a significant trend between the air and oxyfired conditions. For the waterwall probe (SA210), the corrosion rate decreased when the combustion conditions were changed from air- to oxy-firing for all conditions tested. By contrast, for all of the superheater probes (T22, P91 and 347H) the corrosion rate increased when combustion conditions in the furnace were changed from air- to oxy-firing for all but one condition tested. The only superheater material and coal that exhibited behavior that was contrary to the general trend of increasing corrosion rates under oxy-firing was the T22 material with Illinois coal. To examine these trends in more detail it is useful to look at the plot of the increase in corrosion rate when converting from air- to oxy-firing for all materials and conditions. This plot is presented in Fig. 10. The decrease in waterwall corrosion rate when converting from air to oxy-fired conditions, although involving minor corrosion rates, could be a significant circumstance. An expected decrease in corrosion rate to these heat transfer surfaces when retrofitting for oxy-combustion could be a facilitating result for the technology. The difference in behavior between the waterwall and superheater probes may be explained by the difference between oxidizing and reducing conditions. For these data, the waterwall probe was installed in a region with stoichiometry of 0.9, whereas the superheater probes experienced superstoichiometric conditions. The corrosion rate behavior of the superheater probes was not concerning for most of the materials and coals tested. The increase in corrosion rates was small, compounding an already small corrosion rate. The results observed here that may be concerning and warrant further investigation are the P91 material with Skyline coal and the 347H material with Illinois coal. The dramatic increase in corrosion rate for the 374H probe may be explained by considering the temperature response detailed in Fig. 7. These data showed that the corrosion rate was unexpectedly high (1.4 mm/year) for the 347H material, which typically would be considered to be quite corrosion-resistant at high temperature, at its ‘baseline condition’ temperature while oxy-firing Illinois coal. However, the monitoring data also show that at the elevated temperature condition of 978 K, the corrosion rate again was low. These data were consistent with a reported mechanism whereby trisulphate species attack the high nickel alloy at low temperatures, but are decomposed and volatilized at high temperatures (Viswanathan and Bakker, 2000). The morphological analysis presented in Fig. 9 showed that for the 347H material, the nickel
content of the scale decreased while the sulfur content increased. These observations are consistent with a mechanism for sulfur attack on nickel. This is considered to be the first occasion, however, that such behavior has been confirmed by real-time corrosion rate measurements. Once the behavior was identified, the average corrosion rates for the Illinois/oxy firing condition with the 347H material were scrutinized again. It was determined that the data producing the overall average had been bimodal. Significantly lower corrosion rates were measured for these conditions following high temperature operation, or after cleaning of the corrosion elements to remove residual corrosion/ash deposits from earlier lower-temperature exposure. For the initial hour and a half of the staging tests summarized in Fig. 5, the corrosion rates appeared to be coming to steady state after the probes had been reinstalled in the furnace. This was probably because the deposit characteristics were different for this probe location and the surface condition on the probes required a longerthan-normal period to equilibrate to the new flue gas environment. During the first half of the day, air-fired conditions were tested. At this time, two variations from reducing to oxidizing operation were tested. The first variation occurred while the probes were still equilibrating. The second variation showed a strong impact on corrosion rate for the T22 material precisely as the conditions were transitioned from reducing to oxidizing. For the second half of the day, oxy-fired conditions were maintained. Conditions were varied from oxidizing to reducing on two occasions. Under oxy-fired conditions, the T22 corrosion rate spiked during the transient from oxidizing to reducing conditions, which was the opposite behavior to that observed under the air-fired conditions. It appeared that the higher alloy materials were unaffected by transients between oxidizing and reducing conditions. Similar behavior was observed for the waterwall sensor made from the SA 210 material. These results indicate that the corrosion rate of low-alloyed materials may be very high in regions of variable stoichiometric ratio, such as the near burner region and around OFA ports. The results presented above describe consistent transient behavior obtained in short-duration exposures that coupon tests with synthesized gas and deposits cannot resolve. Data such as these can be used in advanced combustion modeling to develop and tune corrosion mechanisms for predicting accurately the impacts of fluctuations between oxidizing and reducing conditions. The corrosion data presented in Fig. 6 and Table 3 indicate that in the absence of trisulphates, the temperature dependence of the 347H corrosion rate was proportional, as expected. These data also illustrate the sensitivity and repeatability of electrochemical noisebased corrosion rate measurements.
5. Conclusions Electrochemical sensing technology was used to determine the fire-side corrosion rates of one waterwall (SA210) and three superheater (T22, P91 and 347H) materials while air- and oxy-firing Utah western bituminous, Illinois high-sulfur bituminous and PRB sub-bituminous coals. The raw average measured corrosion rates were very low, between 0.0003 and 0.016 mm/year (0.012 and 0.63 mils/year) for most conditions. The corrosion rates were higher for the T22 material while air- and oxy-firing Utah and Illinois coals and for the 347H material when oxy-firing Illinois coal; and were in the range of 0.21–0.72 mm/year (8.3–28.3 mils/year). These corrosion rates assume that the loss of material had occurred evenly over the entire sensor element. Actual peak corrosion rates would be expected to be on the order of 3 times higher. Waterwall (SA210) corrosion rates consistently decreased when converting from air to oxy-firing for all coals. An expected decrease in the rate of corrosion attack on these heat transfer surfaces when
A. Fry et al. / International Journal of Greenhouse Gas Control 5S (2011) S179–S185
retrofitting for oxy-combustion could be a facilitating result for the technology. Superheater corrosion rates increased when converting from air- to oxy-firing for all conditions with the exception of the T22 material when firing Illinois coal. During these tests, the flue gas environment was sub-stoichiometric for the waterwall probes and super-stoichiometric for the superheater probes. The increase in superheater corrosion rate when comparing the oxy-fired condition with the air-fired condition was generally very small (double or less) with the exception of the P91 material with Utah coal (5.2 times higher) and 347H material with Illinois coal (52.2 times higher). Corrosion rates for the lower alloyed materials (SA210 and T22) were shown to increase significantly during transients from reducing to oxidizing conditions when air-firing, and from oxidizing to reducing conditions when oxy-firing. Such transients are likely to contribute greatly to practical in-plant corrosion rates in the nearburner and near-OFA port regions. These effects cannot be resolved using coupon tests. Measured increases in the corrosion rate of 347H material under high sulfur and low temperature conditions were consistent with the presence of trisulphate deposits on the superheater probes. It was demonstrated conclusively, and for the first time by real-time corrosion monitoring, that these species are promptly decomposed by operating at higher material temperatures, reducing the subsequent corrosion rates to much lower levels, confirming earlier theoretical predictions based in laboratory coupon exposures. The increase in corrosion rates with increased probe material temperatures was demonstrated as was with the consistently repeatable nature of the results. Acknowledgements This material is based upon work supported by the Department of Energy under Award Number DE-NT0005288. Project Manager is
S185
Timothy Fout. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. References Badin, E.J., 1984. Coal Combustion Chemistry – Correlation Aspects. Elsevier, New York. Bakker, W.T., Kung, S.C., 2000. Waterwall corrosion in coal-fired boilers – a new culprit: FeS. In: Proceedings of the NACE International Symposium , Orlando, FL, March 26–31. Bakker, W.T., et al., 1992. High-temperature fireside corrosion monitoring in the superheater section of a pulverized-coal-fired boiler. In: EPRI , Palo Alto, CA, TR-101799. Davis, K.A., et al., 2004. A multi-point corrosion monitoring system applied in a 1,300 MW coal-fired boiler. Anti-Corros. Methods Mater. 51 (5). Linjewile, T., et al., 2003. Prediction and real-time monitoring techniques for corrosion characterization in furnaces. Mater. High Temp. 20, 175. Reid, W.T., 1971. External Corrosion and Deposits: Boilers and Gas Turbines. Elsevier, New York. Syrett, B.C., Gorman, J.A., 2003. Cost of corrosion in the electric power industry – an update. Mater. Perform. 42 (2), 32–38. Shim, H.-S., et al., 2008. Development of fire-side waterwall corrosion correlations using pilot-scale test furnace. Fuel 15–16, 3353–3361. Valentine, J.R., et al., 2007. CFD evaluation of waterwall wastage in coal-fired utility boilers. Energy Fuels 21, 242. Viswanathan, R., Bakker, W.T., 2000. Materials for boilers in ultra supercritical power plants. In: Proceedings of the International Joint Power Generation Conference , Miami Beach, FL, July 23–26.