- Email: [email protected]
Residual strain distribution in bent composite boiler tubes
Materials Science and Engineering A 437 (2006) 83–87
Residual strain distribution in bent composite boiler tubes Fei Tang a,∗ , Camden R. Hubbard a , Gorti Sarma b , James Keiser a a
Oak Ridge National Laboratory, Metals & Ceramics Division, Oak Ridge, TN 37831, United States b Oak Ridge National Laboratory, Computing & Computational Sciences Division, Oak Ridge, TN 37831, United States Received in revised form 15 April 2006; accepted 15 April 2006
Abstract Kraft recovery boilers are typically constructed of carbon steel boiler tubes clad with a corrosion resistant layer, and these composite tubes are bent and welded together to form air port panels which enable the combustion air to enter the boiler. In this paper, the through-thickness residual strain in the carbon steel layer of non-heat-treated and heat-treated composite bent tubes were measured by neutron diffraction techniques and modeled by finite element modeling. The results can be used to optimize material selection and manufacturing processes to prevent stress corrosion and corrosion fatigue cracking in the boiler tubes. © 2006 Published by Elsevier B.V. Keywords: Residual stresses; Neutron diffraction; Heat treatment
1. Introduction Co-extruded 304L stainless steel/SA210 carbon steel tubes were first used to make recovery boiler walls in Nordic countries in the early 1970s, and by the end of the decade, application of the composite tubes had been extended to service in many boilers in North America as floor tubes and wall tubes due to improved resistance to environments that caused severe thinning of carbon steel tubes [1]. But later experience gained with composite tubes in kraft recovery boiler tubes led to the realization that these tubes could be subject to different corrosion problems and failure mechanisms than the carbon steel tubes they replaced. These include accelerated preferential corrosion of the stainless steel outer layer in recesses around port openings, and cracking of the stainless layer in tubes that formed spout openings in some boilers. As the widespread nature of the cracking problem in composite tubes became apparent in North American and Nordic countries’ boilers [2], a study funded by the U.S. Department of Energy was undertaken to try to identify the cracking mechanism and to recommend solutions.
∗
Corresponding author. Tel.: +1 865 241 1898; fax: +1 865 574 3940. E-mail address: [email protected] (F. Tang).
0921-5093/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.msea.2006.04.050
Residual stresses on the outer surface of alloy 825/carbon steel composite bent tubes with or without heat-treatment were measured by X-ray diffraction technique in a previous study [3]. The results showed the surface residual stresses are compressive on the as-received non-heat-treated bent tubes, and change to tensile after heat-treated at either 615 or 920 ◦ C. Previous observations of cracked floor tubes also revealed that some cracks initiated in the outer layer can continue to propagate into the inner carbon steel (CS) layer. Most cracks were circumferential and hence their growth would be aided by axial tensile stresses [4]. For preventing cracks growing across the CS/825 interface, compressive residual stresses in the CS layer can be very beneficial. In this paper, room temperature axial residual strain in the carbon steel (CS) layer of alloy 825/carbon steel composite bent tubes were measured by neutron diffraction. Three types of composite bent tubes were investigated: a non-heat-treated tube; a tube heat-treated at 615 ◦ C; and a tube heat-treated at 920 ◦ C. These residual strain measurements are intended for validation of stress modeling to predict stress evolution in the CS layer of bent tubes and welded air ports during normal operation at high temperatures or with some localized thermal excursions. The study also assesses if heat-treatment procedures can be optimized or alternate materials selection can be used to avoid large tensile axial stress during operation or upon return to room temperature to prevent crack propagation into CS tube.
84
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
2. Materials and experimental procedure 2.1. Materials Three bent tube samples are all of the Kaevener design [5], where the bend is less extreme than with air ports from other sources. The extruded bent tube consists of a clad out-layer (825 steel) 2 mm thick and a carbon steel layer 5 mm thick (Fig. 1). The locations of neutron strain measurements in the CS layer are shown in Table 1 and Figs. 2 and 3. The distance between adjacent measurement locations is about 50 mm. For the neutron residual strain measurement, 9 locations on Kaevener tube, 13 locations on GPKV2 tube, and 11 locations on GPKV5 tube on the bent side with relatively high residual stresses were selected for study. Because previous observation shows crack growth was mainly aided by axial tensile stresses, only axial residual strain were measured on the Kaevener and GPKV2 bent tubes, while both axial and radial residual strain were measured on GPKV5 bent tube.
Fig. 3. Photos of GPKV5 bent tube showing the neutral and bent sides. Circles on the tube indicate the locations (approximately 50 mm apart) where surface stress and internal carbon steel strain were measured.
Fig. 4. Neutron strain measurement locations in the carbon steel layer as a function of depth. The measurement volume is shown schematically by the shaded diamond. With translation of the tube, multiple depths within the CS were sampled.
2.2. Strain measurement by neutron diffraction Fig. 1. Cross-section of a bent tube. Clad 825 steel layer thickness is 2 mm and the carbon steel core is 5 mm thick.
Table 1 Bent tube samples and measurement locations Bent tube sample name
Heat-treatment
Locations measured by neutron in carbon steel layer
Kaevener GPKV2 GPKV5
As-received, non-heat-treated 920 ◦ C heat-treated 615 ◦ C heat-treated
#1 to #9 #1 to #13 #1 to #6, #8 to #12
The strain as a function of depth in the carbon steel layer (as shown in Fig. 4) was measured at the neutron residual stress facility (NRSF2) at the High Flux Isotope Reactor (HFIR) of Oak Ridge National Laboratory [6]. The bent tubes were mounted horizontally and residual strains were measured using the computer controlled goniometer with a single position sensitive detector (PSD). The diffraction peak of Fe (2 1 1) and neutron wavelength of 0.173 nm was used. Nominal gauge volume of 1 mm × 1 mm ×5 mm for the neutron measurement is defined by slits close to the specimen. A small steel bar (2 mm × 2 mm × 20 mm) was cut from the bent tube and used as the stress-free d0 standard for strain calculation. As shown schematically in Fig. 4, translation stages move the specimen relative to the gauge volume which is fixed in space. Effectively the gauge volume of the neutron beam moved through the carbon steel layer from the CS/825 interface to the inner (ID) free surface with a step size of 1 mm. 3. Results 3.1. Strain measurement results
Fig. 2. As-received Kaevener bent tube and measurement locations #1 to #9 along bent side. There is about 50 mm between each measurement location.
The measurement results of axial residual strain at nine different locations along the length and through the thickness of the CS in the Kaevener bent tube are shown in Fig. 5. For a given
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
Fig. 5. Axial residual strain in as-received Kaevener bent tube carbon steel layer as a function of depth below 825/CS interface from locations #1 to #9. This tube was not heat-treated.
location, the axial strain distribution does not have a significant gradient from the CS/825 interface to CS inner free surface. However, the axial strains gradually changed from compressive to tensile as the location changed from #1 to #9 (along the length of the tube). Fig. 6 shows the axial residual strain distributions as a function of depth and location along the tube in the 920 ◦ C heattreated GPKV2 tube CS layer, which is quite different from that in the tube without heat-treatment (Fig. 5). The strain distributions for all 13 locations along the length of the tube have the same trend. The largest compressive axial strain is nearest the CS/825 interface; the axial strain decreases sharply as the distance from the interface increases. After that sharp strain reduction, the axial strain becomes nearly constant throughout the remainder of the CS layer, showing only a small further decrease close to the inner CS free surface. Figs. 7 and 8 present the axial and radial residual strain distribution in the CS layer of the 615 ◦ C heat-treated GPKV5 bent tube. Generally, the residual strains are still compressive as seen
Fig. 6. Axial residual strain in GPKV2 bent tube carbon steel layer as a function of depth from locations #1 to #13 (heat-treated at 920 ◦ C).
85
Fig. 7. Axial residual strain in GPKV5 bent tube carbon steel layer as a function of depth from locations #1 to #6 and from locations #8 to #12 (heat-treated at 615 ◦ C).
in the GPKV2 bent tube. However, unlike both the Kaevener and GPKV2 bent tubes, at most of locations there is no region of constant axial strain. Near the CS/825 interface, there are relatively large axial compressive strains, and there is a continuous decreasing trend from the CS/825 interface to the inner free surface. The gradients of radial strain profiles are larger than that of axial strain. 3.2. Modeling of heat treatment In an effort to study the effect of heat treatment on the stresses in the composite tube, finite element modeling of the temperature and stress distributions in the alloy 825/carbon steel composite tube was conducted using commercial software ABAQUS. A three-dimensional model was developed for a quarter section of the tube assuming axi-symmetry, and temperature values were prescribed on the outer surface corresponding to normal boiler operation and localized heating. Three cases were considered corresponding to no heat treatment, and heating treatments at
Fig. 8. Radial residual strain in GPKV5 bent tube carbon steel layer as a function of depth at location #6 and from locations #8 to #12 (heat-treated at 615 ◦ C).
86
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
Fig. 9. Axial stress variation at the mid-thickness of the carbon steel layer for a composite tube subjected to 615 and 920 ◦ C heat treatment followed by a normal operating cycle.
615 and 920 ◦ C, before exposure to operating conditions. Initial stresses in the tube were prescribed based on measurements through the tube thickness and at the outer surface using neutron and X-ray diffraction [7]. Fig. 9 shows the axial stress variation at the mid-thickness of the carbon steel layer for a tube subjected to 615 and 920 ◦ C heat treatment, before going through a normal operating cycle. The axial stress in the carbon steel is initially compressive, and heating causes stress to become less compressive or more tensile while cooling causes the stress to become more compressive. This is because of the lower thermal expansion of the carbon steel compared to the outer alloy 825 clad layer. Heat treatment at 615 ◦ C does not alter the stress, whereas heat treatment at 920 ◦ C causes plastic deformation and the axial stress becomes more compressive. During subsequent operation, axial stress again becomes less compressive/more tensile on heating and returns to more compressive value on cooling. The prediction of compressive stress in the carbon steel layer even after heat treatment is consistent with neutron diffraction strain measurements. While results are shown only for carbon steel layer at the mid-thickness, stress variation has also been examined at the surface of the alloy 825 clad layer. Modeling has shown that heat treatment at 615 ◦ C does not alter the initial stress, while heat treatment at 920 ◦ C causes highly tensile stresses at the surface, which is not desirable from the standpoint of stress corrosion cracking of the tubes [4]. Therefore heat treatment of the tubes is not recommended as a means to alter the stresses in the composite tubes used to make primary air ports for recovery boilers.
concave regions of the bent tube were selected for comparison. As shown in Fig. 10, the residual axial strain in the CS layer changes from highly compressive in the convex region to less compressive in the straight region and finally to slightly tensile in the concave region. In short, the residual strain distribution in the as-received bent tube is significantly affected by the mechanical bending applied and the level of bending curvature. The residual strains observed arise primarily due to plastic deformation and the subsequent spring back after the bending. During high temperature heat treatment at either at 615 or 920 ◦ C, most of the bending induced residual strain were relieved in the CS since the residual stresses at the heat treat temperature are higher than the yield strength of carbon steel. But, due to the different coefficient of thermal expansion (CTE) between alloy 825 layer and the CS layer, residual stresses were still present after the bent tube is cooled from the heat treatment temperature. Since the initial residual strain which resulted from mechanical bending was relieved after heat treatment at 615 or 920 ◦ C, one finds a more uniform strain/stress distribution. Therefore, as shown in Figs. 6–8, the axial and radial residual strain through CS layer at all measured locations are consistently compressive,
4. Discussion First, we will discuss the residual axial strain results from the Kaevener bent tube, which is as extruded and has no post-heattreatment. In this tube, one expects to see that residual strain would mostly result from the mechanical extrusion and especially the bending. The bending should make the residual strain distribution different at different locations along the length of the tube. From the results shown in Fig. 5, the residual axial strain changes from one location to another location, although the residual strain level is relatively flat through the thickness of the CS layer at each location. In Fig. 10, the convex, straight, and
Fig. 10. Comparison of axial residual strain of Kaevener bent tube at locations with different bending conditions.
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
87
(b) Close to the CS/825 interface, the residual strains are highly compressive. (c) The axial residual strains gradually decrease through the CS layer thickness. (d) Axial strain values in GPKV2 and GPKV5 are similar, except the strain value variation in GPKV2 is smaller due to higher temperature heat-treatment.
Fig. 11. Comparison of axial strain as a function of depth in the GPKV2 and GPKV5 bent tubes carbon steel layer.
irrespective of whether the location is in the convex, straight or concave region. Fig. 11 compares axial strain between GPKV2 and GPKV5 at corresponding locations of #1 to #9 on the bent tube, which covers half the length of the tube on the bent side. The GPKV5 shows greater strain variation. The smaller variations are reasonable since the GPKV2 bent tube experienced a much higher temperature (920 ◦ C) heat treatment than GPKV5 (615 ◦ C). The higher the heat treatment temperature, the more similar or homogeneous will be the residual stress/strain distribution due to greater plastic deformation and relief of the residual stresses due to bending. 5. Summary Neutron through-thickness strain measurements in the carbon steel layer of bent boiler tubes revealed the following: (1) For the as-received, non-heat-treated Kaevener bent tube: (a) Axial strains are quite uniform throughout the CS thickness at most of the measured locations. (b) Strain levels are mainly affected by mechanical bending condition or curvature. (c) In the convex and straight regions, the residual strains are compressive throughout the CS thickness. (d) In the concave regions, the residual strains in the CS layer are slightly tensile. (2) For the heat-treated GPKV2 (920 ◦ C HT) and GPKV5 (615 ◦ C HT) bent tubes: (a) Axial and radial strains are compressive through the CS thickness at all measured locations.
Based on the axial and radial strain data reported in this paper, heat treatment of as-received bent tubes may seem to be desirable for producing compressive residual stress in the CS layer to resist crack growth across the CS/825 alloy interface. However, X-ray measurements of the stress state in the clad layer become increasing tensile and further the stresses in composite tube will change during high temperature operation. The room temperature residual strain measured can provide the initial stress state for further stress/strain modeling of boiler tube and air ports at normal operation temperature or with some localized thermal excursion. Acknowledgments Research is sponsored by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Industrial Technologies Program, Industrial Materials for the Future (IMF) and the Forest Products Industries of the Future (IOF). Research is performed at the Oak Ridge National Laboratory managed by UT-Battelle, LLC under contract DE-AC05-00OR22725. References [1] W.B.A. Sharp, TAPPI J. 64 (1981) 113–115. [2] A. Klarin, TAPPI J. 76 (1993) 183. [3] J.R. Keiser, X.-L. Wang, R.W. Swindeman, G.B. Sarma, C.M. Hoffmann, P.J. Maziasz, D.L. Singbeil, R. Prescott, L.A. Frederick, P.M. Singh, J. Mahmood, Status report on studies of recovery boiler floor tube cracking, in: Proceedings from the 1999 TAPPI Engineering/Process and Product Quality Conference, Atlanta, GA, 1999. [4] J.R. Keiser, G.B. Sarma, X.-L. Wang, C.R. Hubbard, R.W. Swindeman, D.L. Singbeil, P.M. Singh, Why do kraft recovery boiler composite floor tubes crack? in: Proceedings from the 2000 TAPPI Engineering Conference, Atlanta, GA, 2000. [5] R. Shenassa, K. Haaga, J. Tuiremo, Primary air port tube integrity—a critical review of primary air port design and the effect of boiler design parameters, in: Proceedings from the TAPPI 2002 Fall Technical Conference, San Diego, CA, 2002. [6] C.R. Hubbard, A.D. Stoica, M.C. Wright, H. Choo, S. Craig, W. Bailey, F. Tang, K. An, Design and performance of the second generation neutron residual stress mapping facility (NRSF2) at ORNL, Mater. Sci. Eng. A, submitted for publication. [7] X.-L. Wang, C.R. Hubbard, S. Spooner, B. Taljat, J.R. Keiser, Residual stresses due to processing of composite tubes, in: Proceedings from the Fifth International Conference on Residual Stress, Linkoping, Sweden.
Residual strain distribution in bent composite boiler tubes Fei Tang a,∗ , Camden R. Hubbard a , Gorti Sarma b , James Keiser a a
Oak Ridge National Laboratory, Metals & Ceramics Division, Oak Ridge, TN 37831, United States b Oak Ridge National Laboratory, Computing & Computational Sciences Division, Oak Ridge, TN 37831, United States Received in revised form 15 April 2006; accepted 15 April 2006
Abstract Kraft recovery boilers are typically constructed of carbon steel boiler tubes clad with a corrosion resistant layer, and these composite tubes are bent and welded together to form air port panels which enable the combustion air to enter the boiler. In this paper, the through-thickness residual strain in the carbon steel layer of non-heat-treated and heat-treated composite bent tubes were measured by neutron diffraction techniques and modeled by finite element modeling. The results can be used to optimize material selection and manufacturing processes to prevent stress corrosion and corrosion fatigue cracking in the boiler tubes. © 2006 Published by Elsevier B.V. Keywords: Residual stresses; Neutron diffraction; Heat treatment
1. Introduction Co-extruded 304L stainless steel/SA210 carbon steel tubes were first used to make recovery boiler walls in Nordic countries in the early 1970s, and by the end of the decade, application of the composite tubes had been extended to service in many boilers in North America as floor tubes and wall tubes due to improved resistance to environments that caused severe thinning of carbon steel tubes [1]. But later experience gained with composite tubes in kraft recovery boiler tubes led to the realization that these tubes could be subject to different corrosion problems and failure mechanisms than the carbon steel tubes they replaced. These include accelerated preferential corrosion of the stainless steel outer layer in recesses around port openings, and cracking of the stainless layer in tubes that formed spout openings in some boilers. As the widespread nature of the cracking problem in composite tubes became apparent in North American and Nordic countries’ boilers [2], a study funded by the U.S. Department of Energy was undertaken to try to identify the cracking mechanism and to recommend solutions.
∗
Corresponding author. Tel.: +1 865 241 1898; fax: +1 865 574 3940. E-mail address: [email protected] (F. Tang).
0921-5093/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.msea.2006.04.050
Residual stresses on the outer surface of alloy 825/carbon steel composite bent tubes with or without heat-treatment were measured by X-ray diffraction technique in a previous study [3]. The results showed the surface residual stresses are compressive on the as-received non-heat-treated bent tubes, and change to tensile after heat-treated at either 615 or 920 ◦ C. Previous observations of cracked floor tubes also revealed that some cracks initiated in the outer layer can continue to propagate into the inner carbon steel (CS) layer. Most cracks were circumferential and hence their growth would be aided by axial tensile stresses [4]. For preventing cracks growing across the CS/825 interface, compressive residual stresses in the CS layer can be very beneficial. In this paper, room temperature axial residual strain in the carbon steel (CS) layer of alloy 825/carbon steel composite bent tubes were measured by neutron diffraction. Three types of composite bent tubes were investigated: a non-heat-treated tube; a tube heat-treated at 615 ◦ C; and a tube heat-treated at 920 ◦ C. These residual strain measurements are intended for validation of stress modeling to predict stress evolution in the CS layer of bent tubes and welded air ports during normal operation at high temperatures or with some localized thermal excursions. The study also assesses if heat-treatment procedures can be optimized or alternate materials selection can be used to avoid large tensile axial stress during operation or upon return to room temperature to prevent crack propagation into CS tube.
84
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
2. Materials and experimental procedure 2.1. Materials Three bent tube samples are all of the Kaevener design [5], where the bend is less extreme than with air ports from other sources. The extruded bent tube consists of a clad out-layer (825 steel) 2 mm thick and a carbon steel layer 5 mm thick (Fig. 1). The locations of neutron strain measurements in the CS layer are shown in Table 1 and Figs. 2 and 3. The distance between adjacent measurement locations is about 50 mm. For the neutron residual strain measurement, 9 locations on Kaevener tube, 13 locations on GPKV2 tube, and 11 locations on GPKV5 tube on the bent side with relatively high residual stresses were selected for study. Because previous observation shows crack growth was mainly aided by axial tensile stresses, only axial residual strain were measured on the Kaevener and GPKV2 bent tubes, while both axial and radial residual strain were measured on GPKV5 bent tube.
Fig. 3. Photos of GPKV5 bent tube showing the neutral and bent sides. Circles on the tube indicate the locations (approximately 50 mm apart) where surface stress and internal carbon steel strain were measured.
Fig. 4. Neutron strain measurement locations in the carbon steel layer as a function of depth. The measurement volume is shown schematically by the shaded diamond. With translation of the tube, multiple depths within the CS were sampled.
2.2. Strain measurement by neutron diffraction Fig. 1. Cross-section of a bent tube. Clad 825 steel layer thickness is 2 mm and the carbon steel core is 5 mm thick.
Table 1 Bent tube samples and measurement locations Bent tube sample name
Heat-treatment
Locations measured by neutron in carbon steel layer
Kaevener GPKV2 GPKV5
As-received, non-heat-treated 920 ◦ C heat-treated 615 ◦ C heat-treated
#1 to #9 #1 to #13 #1 to #6, #8 to #12
The strain as a function of depth in the carbon steel layer (as shown in Fig. 4) was measured at the neutron residual stress facility (NRSF2) at the High Flux Isotope Reactor (HFIR) of Oak Ridge National Laboratory [6]. The bent tubes were mounted horizontally and residual strains were measured using the computer controlled goniometer with a single position sensitive detector (PSD). The diffraction peak of Fe (2 1 1) and neutron wavelength of 0.173 nm was used. Nominal gauge volume of 1 mm × 1 mm ×5 mm for the neutron measurement is defined by slits close to the specimen. A small steel bar (2 mm × 2 mm × 20 mm) was cut from the bent tube and used as the stress-free d0 standard for strain calculation. As shown schematically in Fig. 4, translation stages move the specimen relative to the gauge volume which is fixed in space. Effectively the gauge volume of the neutron beam moved through the carbon steel layer from the CS/825 interface to the inner (ID) free surface with a step size of 1 mm. 3. Results 3.1. Strain measurement results
Fig. 2. As-received Kaevener bent tube and measurement locations #1 to #9 along bent side. There is about 50 mm between each measurement location.
The measurement results of axial residual strain at nine different locations along the length and through the thickness of the CS in the Kaevener bent tube are shown in Fig. 5. For a given
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
Fig. 5. Axial residual strain in as-received Kaevener bent tube carbon steel layer as a function of depth below 825/CS interface from locations #1 to #9. This tube was not heat-treated.
location, the axial strain distribution does not have a significant gradient from the CS/825 interface to CS inner free surface. However, the axial strains gradually changed from compressive to tensile as the location changed from #1 to #9 (along the length of the tube). Fig. 6 shows the axial residual strain distributions as a function of depth and location along the tube in the 920 ◦ C heattreated GPKV2 tube CS layer, which is quite different from that in the tube without heat-treatment (Fig. 5). The strain distributions for all 13 locations along the length of the tube have the same trend. The largest compressive axial strain is nearest the CS/825 interface; the axial strain decreases sharply as the distance from the interface increases. After that sharp strain reduction, the axial strain becomes nearly constant throughout the remainder of the CS layer, showing only a small further decrease close to the inner CS free surface. Figs. 7 and 8 present the axial and radial residual strain distribution in the CS layer of the 615 ◦ C heat-treated GPKV5 bent tube. Generally, the residual strains are still compressive as seen
Fig. 6. Axial residual strain in GPKV2 bent tube carbon steel layer as a function of depth from locations #1 to #13 (heat-treated at 920 ◦ C).
85
Fig. 7. Axial residual strain in GPKV5 bent tube carbon steel layer as a function of depth from locations #1 to #6 and from locations #8 to #12 (heat-treated at 615 ◦ C).
in the GPKV2 bent tube. However, unlike both the Kaevener and GPKV2 bent tubes, at most of locations there is no region of constant axial strain. Near the CS/825 interface, there are relatively large axial compressive strains, and there is a continuous decreasing trend from the CS/825 interface to the inner free surface. The gradients of radial strain profiles are larger than that of axial strain. 3.2. Modeling of heat treatment In an effort to study the effect of heat treatment on the stresses in the composite tube, finite element modeling of the temperature and stress distributions in the alloy 825/carbon steel composite tube was conducted using commercial software ABAQUS. A three-dimensional model was developed for a quarter section of the tube assuming axi-symmetry, and temperature values were prescribed on the outer surface corresponding to normal boiler operation and localized heating. Three cases were considered corresponding to no heat treatment, and heating treatments at
Fig. 8. Radial residual strain in GPKV5 bent tube carbon steel layer as a function of depth at location #6 and from locations #8 to #12 (heat-treated at 615 ◦ C).
86
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
Fig. 9. Axial stress variation at the mid-thickness of the carbon steel layer for a composite tube subjected to 615 and 920 ◦ C heat treatment followed by a normal operating cycle.
615 and 920 ◦ C, before exposure to operating conditions. Initial stresses in the tube were prescribed based on measurements through the tube thickness and at the outer surface using neutron and X-ray diffraction [7]. Fig. 9 shows the axial stress variation at the mid-thickness of the carbon steel layer for a tube subjected to 615 and 920 ◦ C heat treatment, before going through a normal operating cycle. The axial stress in the carbon steel is initially compressive, and heating causes stress to become less compressive or more tensile while cooling causes the stress to become more compressive. This is because of the lower thermal expansion of the carbon steel compared to the outer alloy 825 clad layer. Heat treatment at 615 ◦ C does not alter the stress, whereas heat treatment at 920 ◦ C causes plastic deformation and the axial stress becomes more compressive. During subsequent operation, axial stress again becomes less compressive/more tensile on heating and returns to more compressive value on cooling. The prediction of compressive stress in the carbon steel layer even after heat treatment is consistent with neutron diffraction strain measurements. While results are shown only for carbon steel layer at the mid-thickness, stress variation has also been examined at the surface of the alloy 825 clad layer. Modeling has shown that heat treatment at 615 ◦ C does not alter the initial stress, while heat treatment at 920 ◦ C causes highly tensile stresses at the surface, which is not desirable from the standpoint of stress corrosion cracking of the tubes [4]. Therefore heat treatment of the tubes is not recommended as a means to alter the stresses in the composite tubes used to make primary air ports for recovery boilers.
concave regions of the bent tube were selected for comparison. As shown in Fig. 10, the residual axial strain in the CS layer changes from highly compressive in the convex region to less compressive in the straight region and finally to slightly tensile in the concave region. In short, the residual strain distribution in the as-received bent tube is significantly affected by the mechanical bending applied and the level of bending curvature. The residual strains observed arise primarily due to plastic deformation and the subsequent spring back after the bending. During high temperature heat treatment at either at 615 or 920 ◦ C, most of the bending induced residual strain were relieved in the CS since the residual stresses at the heat treat temperature are higher than the yield strength of carbon steel. But, due to the different coefficient of thermal expansion (CTE) between alloy 825 layer and the CS layer, residual stresses were still present after the bent tube is cooled from the heat treatment temperature. Since the initial residual strain which resulted from mechanical bending was relieved after heat treatment at 615 or 920 ◦ C, one finds a more uniform strain/stress distribution. Therefore, as shown in Figs. 6–8, the axial and radial residual strain through CS layer at all measured locations are consistently compressive,
4. Discussion First, we will discuss the residual axial strain results from the Kaevener bent tube, which is as extruded and has no post-heattreatment. In this tube, one expects to see that residual strain would mostly result from the mechanical extrusion and especially the bending. The bending should make the residual strain distribution different at different locations along the length of the tube. From the results shown in Fig. 5, the residual axial strain changes from one location to another location, although the residual strain level is relatively flat through the thickness of the CS layer at each location. In Fig. 10, the convex, straight, and
Fig. 10. Comparison of axial residual strain of Kaevener bent tube at locations with different bending conditions.
F. Tang et al. / Materials Science and Engineering A 437 (2006) 83–87
87
(b) Close to the CS/825 interface, the residual strains are highly compressive. (c) The axial residual strains gradually decrease through the CS layer thickness. (d) Axial strain values in GPKV2 and GPKV5 are similar, except the strain value variation in GPKV2 is smaller due to higher temperature heat-treatment.
Fig. 11. Comparison of axial strain as a function of depth in the GPKV2 and GPKV5 bent tubes carbon steel layer.
irrespective of whether the location is in the convex, straight or concave region. Fig. 11 compares axial strain between GPKV2 and GPKV5 at corresponding locations of #1 to #9 on the bent tube, which covers half the length of the tube on the bent side. The GPKV5 shows greater strain variation. The smaller variations are reasonable since the GPKV2 bent tube experienced a much higher temperature (920 ◦ C) heat treatment than GPKV5 (615 ◦ C). The higher the heat treatment temperature, the more similar or homogeneous will be the residual stress/strain distribution due to greater plastic deformation and relief of the residual stresses due to bending. 5. Summary Neutron through-thickness strain measurements in the carbon steel layer of bent boiler tubes revealed the following: (1) For the as-received, non-heat-treated Kaevener bent tube: (a) Axial strains are quite uniform throughout the CS thickness at most of the measured locations. (b) Strain levels are mainly affected by mechanical bending condition or curvature. (c) In the convex and straight regions, the residual strains are compressive throughout the CS thickness. (d) In the concave regions, the residual strains in the CS layer are slightly tensile. (2) For the heat-treated GPKV2 (920 ◦ C HT) and GPKV5 (615 ◦ C HT) bent tubes: (a) Axial and radial strains are compressive through the CS thickness at all measured locations.
Based on the axial and radial strain data reported in this paper, heat treatment of as-received bent tubes may seem to be desirable for producing compressive residual stress in the CS layer to resist crack growth across the CS/825 alloy interface. However, X-ray measurements of the stress state in the clad layer become increasing tensile and further the stresses in composite tube will change during high temperature operation. The room temperature residual strain measured can provide the initial stress state for further stress/strain modeling of boiler tube and air ports at normal operation temperature or with some localized thermal excursion. Acknowledgments Research is sponsored by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Industrial Technologies Program, Industrial Materials for the Future (IMF) and the Forest Products Industries of the Future (IOF). Research is performed at the Oak Ridge National Laboratory managed by UT-Battelle, LLC under contract DE-AC05-00OR22725. References [1] W.B.A. Sharp, TAPPI J. 64 (1981) 113–115. [2] A. Klarin, TAPPI J. 76 (1993) 183. [3] J.R. Keiser, X.-L. Wang, R.W. Swindeman, G.B. Sarma, C.M. Hoffmann, P.J. Maziasz, D.L. Singbeil, R. Prescott, L.A. Frederick, P.M. Singh, J. Mahmood, Status report on studies of recovery boiler floor tube cracking, in: Proceedings from the 1999 TAPPI Engineering/Process and Product Quality Conference, Atlanta, GA, 1999. [4] J.R. Keiser, G.B. Sarma, X.-L. Wang, C.R. Hubbard, R.W. Swindeman, D.L. Singbeil, P.M. Singh, Why do kraft recovery boiler composite floor tubes crack? in: Proceedings from the 2000 TAPPI Engineering Conference, Atlanta, GA, 2000. [5] R. Shenassa, K. Haaga, J. Tuiremo, Primary air port tube integrity—a critical review of primary air port design and the effect of boiler design parameters, in: Proceedings from the TAPPI 2002 Fall Technical Conference, San Diego, CA, 2002. [6] C.R. Hubbard, A.D. Stoica, M.C. Wright, H. Choo, S. Craig, W. Bailey, F. Tang, K. An, Design and performance of the second generation neutron residual stress mapping facility (NRSF2) at ORNL, Mater. Sci. Eng. A, submitted for publication. [7] X.-L. Wang, C.R. Hubbard, S. Spooner, B. Taljat, J.R. Keiser, Residual stresses due to processing of composite tubes, in: Proceedings from the Fifth International Conference on Residual Stress, Linkoping, Sweden.