Improving Reliability for Steel Mill Furnace Rolls
JEREMY J. RYDBERG – VP BUSINESS DEVELOPMENT ATLAS MACHINE AND SUPPLY | 7000 Global Dr. Louisville KY 40258 IMPROVING FURNACE ROLL RELIABILITY ESTAD 2017
IMPROVING FURNACE ROLL RELIABILITY AUTHOR: JEREMY RYDBERG Abstract Atlas was approached by a major steel manufacturer and asked to develop methods to improve overall reliability of steel mill furnace rolls. To accomplish this it was determined we would need to develop methods to inspect new, used and failed furnace rolls to determine why they fail and how to avoid it. Destructive examination tests were performed on used and failed furnace rolls to determine causes of failure. Phased array ultrasonic testing and fluorescent dye penetrant testing were developed to identify these flaws through non-destructive examination (NDE) methods that could be applied to rolls that may be put back into service. Through the destructive and non-destructive examination methods it was determined that the leading causes of premature roll failure stem from poor weld filler metal choices, poor welding craftsmanship and low quality castings. To improve furnace roll reliability Atlas has qualified materials, suppliers, developed qualified weld procedures, and developed NDE methods to qualify new and used and rolls for service. It is expected that implementation of these methods will lead to improved furnace roll reliability.
IMPROVING FURNACE ROLL RELIABILITY AUTHOR: JEREMY RYDBERG Premature furnace roll failure cannot be reliably prevented with current manufacturing and inspection methods. Atlas Machine and Supply conducted a study of the furnace roll supply chain to evaluate current practices in order to increase reliability and reduce the risk of in-service failures. Furnace rolls, also known as hearth rolls, are used extensively in the steel making industry. They are a critical component for any continuous aluminizing or galvanizing line and responsible for safely handling the sheet though the heat treating furnace. A plants ability to produce quality steel is directly related to the performance of these rolls. Poor roll designs, manufacturing flaws and the increased physical stress put on rolls by Advanced High Strength Steels all contribute to the premature failure of furnace rolls and ultimately unplanned plant outages. It is reported that a single roll failure can cost a plant as much as $750,000 in lost production, ruined product, and maintenance labor. Current methods of evaluating new and used rolls to predict and prevent failure in service are inadequate and rarely performed. In May of 2015, Atlas Machine and Supply began working on more effective NonDestructive Examination (NDE) processes to discover what causes premature roll failure and develop solutions for these causes. Traditional NDE practices for furnace rolls frequently went unused by suppliers and end users but included verifying roll material chemistry, a visual inspection, and fluorescent liquid dye penetrant (FLP) testing. Our initial focus was to begin utilizing phased array ultrasonic testing (PAUT) in addition to the traditional methods in order to see flaws below the surface of the roll. Traditionally PAUT does not perform well in materials with long grain structure materials like cast high temperature stainless steel, but recent developments in hardware and PAUT technique for the nuclear industry had shown promise and it was believed we could apply these advancements to the steel industries’ furnace rolls. In order to begin testing advanced PAUT methods a calibration block was constructed from a failed furnace roll provided by a major steel producer. Since all failure examples that had been seen up to this point were in the weld joint between the static cast end bell and the centrifugally cast body, the calibration block material was taken from this joint in the roll. Known defects where machined into the block per ASME guidelines and included two drilled holes, one in the body side weld fusion line and one at the end bell weld fusion line, and an ID notch. With the calibration block Atlas began working with a 3rd party NDE service provider and the PAUT equipment OEM to develop an effective procedure for detecting the known flaws in the block. The methods developed by this team has proven effective with some minor limitations. Since the materials are cast and are of varying chemistry and grain structure the dimensional accuracy of the PAUT is reduced. The reduction in accuracy is not substantial enough to prevent a welder or machinist from finding and excavating a
Figure 1 - Phased Array UT Coupon and Scanner
defect but would prevent them from accurately reporting the wall thickness of the roll. The other limitation is the PAUT is not effective at seeing the first .100” of examined surface and can only see defects .100” or deeper. Because of this limitation, surface inspections with Fluorescent Dye Penetrant are still required. Once the PAUT inspection method was developed, a major American Steel manufacturer began supplying Atlas with new and used rolls from a variety of manufacturers for inspection. There were several purposes for these inspections including learning how effective the new PAUT process was at finding indications, determining if a technician could use the PAUT report to excavate the flaw, and if the roll was qualified to return to service. Of the first three rolls inspected, two had minor indications showing from the FLP and the third was clean. The PAUT also found the third roll to be free of defects but found significant sub-surface defects in the other two rolls, one defect was severe enough that we believe the roll body was 60% detached from the end bell. These defects were later determined to be cracks forming from the inside out that propagated from lack of fusion and incomplete joint penetration at the weld root. Attempts to excavate the cracks from the outside ultimately separated the rolls in two. Continued testing of new and used furnace rolls proved that the combination of FLP and PAUT was an effective method for determining if a roll was qualified for service and could be used to determine a repair procedure for the roll. Following the inspection of several furnace rolls, Atlas began auditing material suppliers and discussing issues with end users to identify what factors reduced furnace roll reliability. Several factors came into play but the most significant were poor welding craftsmanship, poor casting quality and a poor choice of filler metals. By design, furnace roll materials have high yield strengths at elevated temperatures but this also means they are not very ductile at room temperature. Materials with low ductility are inherently difficult to weld and prone to solidification cracks and heat affected zone cold cracking. To remedy the welding difficulty many roll manufactures choose filler materials that are over-alloyed with higher nickel content than the base metals. This improves weldability but ultimately the inconsistency in chemistry in the weld joint reduces roll reliability. The variances in chemistry in the weld joint create differences in the coefficient of thermal expansion between the weld area and base materials. As rolls are thermally cycled in service these variances create very high levels of stress in the weld joint causing minor defects to grow eventually leading to premature roll failure. To improve roll reliability, rolls should be manufactured with as few defects as possible with filler metals of matching chemistry. In May of 2015, Atlas Machine and Supply began developing weld procedures and material suppliers for furnace rolls that would allow for producing a roll with filler metal that had similar chemistry and CTE to the base material and that could be welded nearly defect free. Atlas audited
Figure 2 - Failed HN Furnace Roll
several foundries and procured sample weld coupons for weld testing. During the audits a wide range of quality control procedures and methods where observed. Pricing also varied greatly, the same pair of end bells was quoted from $10,800-$38,700 between each of the audited foundries. The most common practices to verify casting quality was visual inspection and LP. The foundries that appeared to produce the most reliable castings also used FEA pattern making and pouring software and radiography to insure parts are produced properly. The differences in the
Figure 3 – Coupon from foundry without FEA design or Radiography
Figure 4 – Coupon from foundry with FEA design or Radiography
sample provided correlated with cost and improved processes. The cleanest sample and best weld joint came from a foundry that used FEA software and radiography but was also towards the higher end of the price scale. The cast weld coupons took several weeks to produce, during the down time Atlas developed approximate weld procedures by welding on readily available tubes of 304ss. Initial weld tests were performed using existing weld procedure for similar alloys in MIG and Sub Arc but with filler metal similar to HN and MO-RE1. The adopted weld procedures, as expected, had issues that needed to be solved. These include solidification cracks, cold cracking in the heat effected zone, and stop cracks at each weld termination.
Figure 5 - Example of Solidification Cracking (Vertical) and HAZ Cracking (Horizontal)
Figure 6 - Example of Stop Crack
As weld development continued, the key to eliminating issues proved to be to keep the heat input as a low as possible. Experiments were performed with low amperage spray transfer GMAW welding and short arc GMAW welding. Atlas was not able to produce a weld with consistent fusion utilizing short arc processes and decided on a spray transfer GMAW process for further testing. The spray transfer GMAW process was used to test weld the coupons provided by the foundries. The same weld procedure was used on each sample in order to demonstrate how the base metal effected weld quality. Figure 9 shows the weld in material from a lower cost foundry. This weld sample had cracks in the dilution zone between the sample casting and weld metal. The casting from a higher cost foundry (Figure 8) with more thorough foundry control methods did not have any surface indications, metallography examination was performed for further evaluation and three indications were found in the weld. They included a root crack and two HAZ cracks.
Figure 9 - Weld Sample with less foundry controls
Figure 11 - Root Crack
Figure 8 - Weld sample with more complete foundry controls
Figure 10 - HAZ Crack 1
The weld joint design was successfully changed to an open root design in order to eliminate the root crack. Testing continued with adjustments to a spray transfer GMAW welding process to remove the HAZ cracks. During testing it was determined that the process window for successfully GMAW welding was very narrow and that when the HAZ cracks were eliminated other defects where created. The GMAW process was abandoned in favor of GTAW welding. Figure 7 - HAZ Crack 2
GTAW welding development is ongoing. This process was selected because of lower heat inputs and the ability to separate arc control and wire feed. Weld coupons have been produced in HN and MO-RE1 without defects detectable by PAUT, LP or Metallography. Developing methods for improving furnace roll reliability has proven challenging, but successful. The implementation of PAUT can reliably detect weld joint defects and Figure 12 - Weld Coupon being TIG welded can be used to determine if a roll should be put in to service. Improvements in raw material supply can help improve material weldability and defect free welding with matching chemistry and can be accomplished on many furnace roll materials. Implementing these practices can provide visibility to the risk of premature roll failure and offer methods to reduce those risks.
IMPROVING FURNACE ROLL RELIABILITY AUTHOR: JEREMY RYDBERG Abstract Atlas was approached by a major steel manufacturer and asked to develop methods to improve overall reliability of steel mill furnace rolls. To accomplish this it was determined we would need to develop methods to inspect new, used and failed furnace rolls to determine why they fail and how to avoid it. Destructive examination tests were performed on used and failed furnace rolls to determine causes of failure. Phased array ultrasonic testing and fluorescent dye penetrant testing were developed to identify these flaws through non-destructive examination (NDE) methods that could be applied to rolls that may be put back into service. Through the destructive and non-destructive examination methods it was determined that the leading causes of premature roll failure stem from poor weld filler metal choices, poor welding craftsmanship and low quality castings. To improve furnace roll reliability Atlas has qualified materials, suppliers, developed qualified weld procedures, and developed NDE methods to qualify new and used and rolls for service. It is expected that implementation of these methods will lead to improved furnace roll reliability.
IMPROVING FURNACE ROLL RELIABILITY AUTHOR: JEREMY RYDBERG Premature furnace roll failure cannot be reliably prevented with current manufacturing and inspection methods. Atlas Machine and Supply conducted a study of the furnace roll supply chain to evaluate current practices in order to increase reliability and reduce the risk of in-service failures. Furnace rolls, also known as hearth rolls, are used extensively in the steel making industry. They are a critical component for any continuous aluminizing or galvanizing line and responsible for safely handling the sheet though the heat treating furnace. A plants ability to produce quality steel is directly related to the performance of these rolls. Poor roll designs, manufacturing flaws and the increased physical stress put on rolls by Advanced High Strength Steels all contribute to the premature failure of furnace rolls and ultimately unplanned plant outages. It is reported that a single roll failure can cost a plant as much as $750,000 in lost production, ruined product, and maintenance labor. Current methods of evaluating new and used rolls to predict and prevent failure in service are inadequate and rarely performed. In May of 2015, Atlas Machine and Supply began working on more effective NonDestructive Examination (NDE) processes to discover what causes premature roll failure and develop solutions for these causes. Traditional NDE practices for furnace rolls frequently went unused by suppliers and end users but included verifying roll material chemistry, a visual inspection, and fluorescent liquid dye penetrant (FLP) testing. Our initial focus was to begin utilizing phased array ultrasonic testing (PAUT) in addition to the traditional methods in order to see flaws below the surface of the roll. Traditionally PAUT does not perform well in materials with long grain structure materials like cast high temperature stainless steel, but recent developments in hardware and PAUT technique for the nuclear industry had shown promise and it was believed we could apply these advancements to the steel industries’ furnace rolls. In order to begin testing advanced PAUT methods a calibration block was constructed from a failed furnace roll provided by a major steel producer. Since all failure examples that had been seen up to this point were in the weld joint between the static cast end bell and the centrifugally cast body, the calibration block material was taken from this joint in the roll. Known defects where machined into the block per ASME guidelines and included two drilled holes, one in the body side weld fusion line and one at the end bell weld fusion line, and an ID notch. With the calibration block Atlas began working with a 3rd party NDE service provider and the PAUT equipment OEM to develop an effective procedure for detecting the known flaws in the block. The methods developed by this team has proven effective with some minor limitations. Since the materials are cast and are of varying chemistry and grain structure the dimensional accuracy of the PAUT is reduced. The reduction in accuracy is not substantial enough to prevent a welder or machinist from finding and excavating a
Figure 1 - Phased Array UT Coupon and Scanner
defect but would prevent them from accurately reporting the wall thickness of the roll. The other limitation is the PAUT is not effective at seeing the first .100” of examined surface and can only see defects .100” or deeper. Because of this limitation, surface inspections with Fluorescent Dye Penetrant are still required. Once the PAUT inspection method was developed, a major American Steel manufacturer began supplying Atlas with new and used rolls from a variety of manufacturers for inspection. There were several purposes for these inspections including learning how effective the new PAUT process was at finding indications, determining if a technician could use the PAUT report to excavate the flaw, and if the roll was qualified to return to service. Of the first three rolls inspected, two had minor indications showing from the FLP and the third was clean. The PAUT also found the third roll to be free of defects but found significant sub-surface defects in the other two rolls, one defect was severe enough that we believe the roll body was 60% detached from the end bell. These defects were later determined to be cracks forming from the inside out that propagated from lack of fusion and incomplete joint penetration at the weld root. Attempts to excavate the cracks from the outside ultimately separated the rolls in two. Continued testing of new and used furnace rolls proved that the combination of FLP and PAUT was an effective method for determining if a roll was qualified for service and could be used to determine a repair procedure for the roll. Following the inspection of several furnace rolls, Atlas began auditing material suppliers and discussing issues with end users to identify what factors reduced furnace roll reliability. Several factors came into play but the most significant were poor welding craftsmanship, poor casting quality and a poor choice of filler metals. By design, furnace roll materials have high yield strengths at elevated temperatures but this also means they are not very ductile at room temperature. Materials with low ductility are inherently difficult to weld and prone to solidification cracks and heat affected zone cold cracking. To remedy the welding difficulty many roll manufactures choose filler materials that are over-alloyed with higher nickel content than the base metals. This improves weldability but ultimately the inconsistency in chemistry in the weld joint reduces roll reliability. The variances in chemistry in the weld joint create differences in the coefficient of thermal expansion between the weld area and base materials. As rolls are thermally cycled in service these variances create very high levels of stress in the weld joint causing minor defects to grow eventually leading to premature roll failure. To improve roll reliability, rolls should be manufactured with as few defects as possible with filler metals of matching chemistry. In May of 2015, Atlas Machine and Supply began developing weld procedures and material suppliers for furnace rolls that would allow for producing a roll with filler metal that had similar chemistry and CTE to the base material and that could be welded nearly defect free. Atlas audited
Figure 2 - Failed HN Furnace Roll
several foundries and procured sample weld coupons for weld testing. During the audits a wide range of quality control procedures and methods where observed. Pricing also varied greatly, the same pair of end bells was quoted from $10,800-$38,700 between each of the audited foundries. The most common practices to verify casting quality was visual inspection and LP. The foundries that appeared to produce the most reliable castings also used FEA pattern making and pouring software and radiography to insure parts are produced properly. The differences in the
Figure 3 – Coupon from foundry without FEA design or Radiography
Figure 4 – Coupon from foundry with FEA design or Radiography
sample provided correlated with cost and improved processes. The cleanest sample and best weld joint came from a foundry that used FEA software and radiography but was also towards the higher end of the price scale. The cast weld coupons took several weeks to produce, during the down time Atlas developed approximate weld procedures by welding on readily available tubes of 304ss. Initial weld tests were performed using existing weld procedure for similar alloys in MIG and Sub Arc but with filler metal similar to HN and MO-RE1. The adopted weld procedures, as expected, had issues that needed to be solved. These include solidification cracks, cold cracking in the heat effected zone, and stop cracks at each weld termination.
Figure 5 - Example of Solidification Cracking (Vertical) and HAZ Cracking (Horizontal)
Figure 6 - Example of Stop Crack
As weld development continued, the key to eliminating issues proved to be to keep the heat input as a low as possible. Experiments were performed with low amperage spray transfer GMAW welding and short arc GMAW welding. Atlas was not able to produce a weld with consistent fusion utilizing short arc processes and decided on a spray transfer GMAW process for further testing. The spray transfer GMAW process was used to test weld the coupons provided by the foundries. The same weld procedure was used on each sample in order to demonstrate how the base metal effected weld quality. Figure 9 shows the weld in material from a lower cost foundry. This weld sample had cracks in the dilution zone between the sample casting and weld metal. The casting from a higher cost foundry (Figure 8) with more thorough foundry control methods did not have any surface indications, metallography examination was performed for further evaluation and three indications were found in the weld. They included a root crack and two HAZ cracks.
Figure 9 - Weld Sample with less foundry controls
Figure 11 - Root Crack
Figure 8 - Weld sample with more complete foundry controls
Figure 10 - HAZ Crack 1
The weld joint design was successfully changed to an open root design in order to eliminate the root crack. Testing continued with adjustments to a spray transfer GMAW welding process to remove the HAZ cracks. During testing it was determined that the process window for successfully GMAW welding was very narrow and that when the HAZ cracks were eliminated other defects where created. The GMAW process was abandoned in favor of GTAW welding. Figure 7 - HAZ Crack 2
GTAW welding development is ongoing. This process was selected because of lower heat inputs and the ability to separate arc control and wire feed. Weld coupons have been produced in HN and MO-RE1 without defects detectable by PAUT, LP or Metallography. Developing methods for improving furnace roll reliability has proven challenging, but successful. The implementation of PAUT can reliably detect weld joint defects and Figure 12 - Weld Coupon being TIG welded can be used to determine if a roll should be put in to service. Improvements in raw material supply can help improve material weldability and defect free welding with matching chemistry and can be accomplished on many furnace roll materials. Implementing these practices can provide visibility to the risk of premature roll failure and offer methods to reduce those risks.
