John A. Talbott, P.E., and Keith Cronrath, P.E.
Talbott Associates Consulting Firm
67th General Meeting in Anaheim, California
Summary: The following article is a part of National Board Classic Series and it was published in the National Board BULLETIN. (5 printed pages)
In the summer of 1994, when the Port of Ridgefield, Washington, found that they needed to replace the boiler that provided process steam to their tenant, they were willing to accept a used but reconditioned boiler. A thermal systems company found a 600 hp firetube boiler initially rated for 150 psi internal working pressure, which was manufactured in 1966*. The length of time the boiler had been out of service and the condition of its storage during this time were both unknown factors.
The reconditioned boiler was offered to the Port of Ridgefield, and the offer was accepted. All new instrumentation and safety controls were installed on the boiler and, with the exception of the furnace tube, it was completely retubed. The boiler was inspected and tested to 150 psi September 13, 1994. The tubesheets were dye-tested for cracks, and none were identified. The boiler was then rejacketed and was installed at the Port of Ridgefield. The installation was nearing completion in December of 1994 when additional difficulty with the existing boiler prompted the Port to ask the thermal systems company if the reconditioned boiler could be put in service a little earlier to preclude the expense of repairing the old boiler. The reconditioned boiler was fully operational December 13, 1994, after completion of the adjustment of emission controls. The Port added a chemical injection unit December 15, to prevent corrosion within the boiler. The operator injected an oxygen scavenger. At 11:45 p.m. December 19, the Port security officer observed that the pressure in the boiler was indicated to be 125 psi instead of the 110 psi setting. The security officer called the maintenance superintendent of the Port who was told to shut down the boiler and then restart it in 30 minutes. The boiler water level was reported to be at the proper level before the night maintenance man restarted the boiler. At 1:00 a.m. December 20, the next shift security officer called the maintenance superintendent at home and told him that the boiler was not making a sound, and its internal pressure was indicated to be 0 psi. The maintenance superintendent said not to restart the boiler and he would check it in the morning.
At about 7:45 a.m. December 20, the maintenance superintendent examined the boiler that had been shut down by its own safety controls. He flipped all of the resets without determining which of the safety controls had been tripped. These controls did not have external telltales to indicate which control had been tripped. The only way to determine that would have been to open the control panel to examine physically the markers that are inside. The water level was normal so he restarted the boiler at 8:30 a.m. It was operated under close supervision for about an hour and appeared to be operating satisfactorily. From 10:00 a.m. to 11:30 a.m., two employees periodically inspected the boiler and found nothing unusual. At 11:30 a.m. and 1:00 p.m., the maintenance supervisor found the boiler operating normally and the water level was within the correct range.
At about 1:30 p.m., two employees of the Port tenant were in the room that housed the boiler. One of the men heard the boiler pop, and he yelled to his coworker and then ran out of the building. The boiler exploded, and its rear door was blown about 300 feet, striking a railroad tank car. The boiler was propelled about 40 feet in the direction of its burner end, rupturing the steam and natural gas lines, and striking and breaking a large double wooden column supporting the roof trusses, striking and killing the other worker.
The 8-foot-diameter rear door with its refractory insulation weighed about 3 tons. It had been held to the shell of the boiler by 16 bolts that were 7/8" in diameter. Examination of the bolts revealed that they all broke in tension overload. The flue passages had to have been pressurized.
Examination of the fractured joint between the furnace tube and the tubesheet revealed that the 5/8" thick tubesheet had been beveled and welded to the outer diameter of the 7/16" thick furnace tube. A backing bar had been used and left in place, but it was separated from the weld in many areas. The fracture was through the metal of the furnace tube, beginning near the toe of the weld on the water side and then progressed in steps that had little or no deformation. Examination of this stepped fracture indicated that it was a result of laminations in the steel plate from which the furnace tube had been formed.
Laminations as used herein are planes within the steel plate across which there is no metallic bond. They are typically a result of nonmetallic inclusions and gas pockets formed in the ingot when it has been cast and as it solidifies. These most often occur in the top of the ingot in the "pipe end" formed as the ingot solidifies. They may include oxide coating of the bubbles, slag inclusions, refractory inclusions from erosion of the furnace, its spout, the ladle and its spout or orifice, and the passing through of slag on the surface of the molten steel in the ladle.
In rimming steel practice, this pipe end, or cavity, is visible. In killed steel practice, as when aluminum is dropped in the top of the molten ingot to deoxidize the steel and initiate crystallization of the steel by providing sites for the formation of the dendrites, the pipe end is often concealed at least in part. In the customary ingot practice, the pipe end is cropped (i.e., it is cut off and returned for remelting, leaving sound steel only in the ingot). If the crop is not deep enough, the end of the ingot that goes to the rolling mill will still have in it a significant amount of gas bubbles and nonmetallic inclusions. As the ingot is rolled into a bloom and the bloom is rolled into plate, these bubbles and nonmetallic inclusions will be broadened approximately in proportion to the increase in width from the ingot to the finished plate width. At the same time, the gas pockets and nonmetallic inclusions will be extended longitudinally in the plate. As an example, a 1" gas pocket in an ingot having a 2' x 3' cross-section which is rolled into 7/16" plate, 10' wide will have that 1" gas pocket widen to about 3-3/8" and length to almost 55".
Notable in this fracture is the extent of the lamination with up to four laminations occurring in the same thickness, and though not connected continuously in planes, overlapping substantially so that the weld applied to the outer diameter was only attached to the outermost lamination in the steel. Since there was no shear across the plane of the lamination, the only part of the plate that was effective in transferring stress was that between the outside diameter and the nearest lamination. In spots, this reduced the strength of the connection to about one-fifth of what it would have been without the laminations.
There were indications of corrosion products in the zone on the water side of the tubesheet where it joined the furnace tube. There were also areas in which corrosion penetrated into the furnace tube. Although this corrosion was very minor and would have presented no problem if the plate was free of laminations, with the laminations present, the corrosion caused significant further reduction in the strength of the connection.
The slight sagging of the new firetubes and the collapse of the furnace tube were thought by the manufacturer to have been a result of overheating as a result of loss of water. There was also the allegation that pearlite colonies, grain growth, and intergranular voids found in metallurgical samples cut from the boiler tubes supported the hypothesis of overheating as having been the causal factor in the boiler failure. To address this issue, Michael York, P.E., was engaged to test examplar boiler tube sections when exposed to 1,350 degrees F. for various lengths of time to determine if growth of the pearlite colonies or the grain size or intergranular voids could be associated with overheating as could have been possible in the temperature regime of the boiler during the period of use.
The result of this testing was that the pearlite colonies which were found in the furnace tubes spanned the normal range of variations in microstructure of the tube as manufactured. The grain growth was observed exclusively at the inside diameter of the firetube specimens in the subject boiler and found to be a result of the manufacturing process and was consistent with heat from the normal boiler operation when the hot flue gases flow through the firetubes. The metallurgy of the tubes did not indicate they had experienced a high heat condition. The intergranular voids alleged to have occurred were not substantiated in any of the tests.
Under the direction of the chief boiler inspector in Washington, all of the instruments and controls were tested for performance after the accident. All performed properly. The low water cutoffs functioned properly, and the relief valves relieved properly. A major complication in the study was the fact that the Port maintenance supervisor had reset all the controls when he arrived on the morning of the accident without recording which of the controls had been tripped.
Specimens cut from the furnace tube and firetubes were tested chemically and for mechanical properties. All of these tests showed the steel to be in the specified range for ASME SA-285-A for the furnace tube and ASME SA-178-A and SA-273 for the firetubes.
Along the fracture, the tubesheet was found to be flared somewhat.
The laminations reduced the strength of the weld joint between the furnace tube and the tubesheet to less than 20% of the thickness of the plate comprising the furnace tube. The further reduction in strength caused by very moderate corrosion allowed the initial rupture to occur in the firetube at the inner end of the weld to the tubesheet. This rupture caused rapid loss of water from the boiler, allowing steam to fill the flue chamber, blowing off the rear door, and at the same time, propelling the boiler violently in the direction of its burner.
The laminated plate was determined to be defective for use as a furnace tube; this defect was the primary cause of the boiler failure. These defects were latent at the time that the boiler was manufactured. Indeed, they were latent at the time that the steel was manufactured. Because the end of the steel had received a light weld coating, the laminations were not discoverable by dye-checking, as there was no crack at the surface. It is further unlikely that magnetic particle inspection would have detected these flaws. The defects were not discoverable by short-term testing of the boiler with water under 150 psi internal pressure. No defect was found in the design or workmanship of the manufacturer.
Though it is not required under the ASME Code, ultrasonic testing would probably have identified the laminations so that the defective condition of the boiler could have been known and its use prevented until the furnace tube was replaced. Pressure testing of the reconditioned boiler at 150% of its rated pressure may also have caused the boiler to fail the test and prevent its use until it was properly repaired.
There is some question whether the laminations exceeded those allowed in the material specifications. In the opinion of the writers, they were excessive, but there were well-qualified people who disagreed. That issue was not adjudicated since the case was settled without trial.
- The allowed laminations in boiler furnace tube plates should probably be reduced from that which the present standard allows. The language must be clearer to prevent disputes in interpretation.
- Testing at the rated pressure of repaired and reconditioned boilers is probably inappropriate; the test pressure should probably be on the order of 1.25 to 1.5 times the rated pressure of the boiler.
- Where the end of the furnace tube plate has been coated with weld material, neither magnetic particle inspection nor dye-penetrant testing are adequate to identify flaws. Ultrasonic testing is the only means at hand that could have identified those flaws.
Fortunately, the major problem of laminations is being eliminated by the change in steel mill practice. Ingot practice is giving way to pressure casting and continuous casting, neither of which has the potential for gas pockets and nonmetallic inclusions anything like that which is inherent in ingot practice.
(*Note: The initial posting of this article contained an error, stating that the subject boiler had been manufactured in 1996. The text should have read 1966, as it does now.)
Editor's note: Some ASME Boiler and Pressure Vessel Code requirements may have changed because of advances in material technology and/or actual experience. The reader is cautioned to refer to the latest edition and addenda of the ASME Boiler and Pressure Vessel Code for current requirements.