Anatomy of a Welded Repair

We, welders, are a curious lot. Our typical response to the question, “Can you weld this?” is “I can weld anything except the crack of dawn or a broken heart!” Self-confidence is never in short supply. And for the most part, we get the job done.

This is a story of a welded repair I was involved with nearly a decade ago. Any repair that survives a decade of use is a success.

Background

A customer called the office to tell me that an essential piece of equipment had suffered a catastrophic failure. Being somewhat cautious, I asked, “What is the material of construction?” The answer, “It’s steel.” From my desk, I can hear the wheels in the caller’s head whirling around, and he’s thinking, “What else would it be if I’m asking him if he can weld it?”

I like to believe that with age comes wisdom, so I respond with, “It can probably be welded, but how we weld it will depend on what the alloy is.” 
“That’s no problem, we have one of those positive material identification (PMI) guns,” the customer responded. “The PMI gun isn’t going to tell me what I need to know. We need a chemical analysis from a metallurgical laboratory. We need to send a sample out for a chemical analysis,” I said.

The demeanor changed, and the panic is evident in his voice. “How long is that going to take?” he asked. “A couple of days is typical,” I told him. The client said, “But we need it right away; production is at a standstill.”

I’m a problem solver by nature, but I resisted the urge to jump in with both feet because I needed to do a little research to find out what caused the failure and which alloy needed to be repaired. If the repair is unsuccessful and the component fails when it’s returned to service, the customer isn’t going to shrug and say, “At least he tried.” The reality is I will never hear from that customer again.

Chemical Analysis

My typical customers know from experience that one of my first questions will be, “What is the material we’ll be welding?” In many cases, they have already sent a sample to a laboratory for analysis.

When a chemical analysis is needed, people think in terms of the various metallic alloys added to iron. They often don’t understand how certain alloying or trace elements can adversely affect steel’s weldability. The lighter elements, such as carbon, sulfur, and phosphorus, are “bad actors” when it comes to welding. The laboratory and the customer tend to gravitate toward using PMI guns because they are quick and relatively inexpensive. A PMI gun is fine for separating 304 austenitic stainless steel from 316 or 317 stainless steel. Still, the test method is not sensitive enough to quantify the amount of lighter elements present in the parts, like carbon or sulfur, so it will not differentiate low carbon, medium carbon, or high carbon steel — or a free machining steel containing sulfur that has poor weldability, or a steel alloy with low carbon and low sulfur that is easily welded. No, I need a chemical analysis that identifies all the elements listed by the carbon equivalent formula we choose to use. There are a few carbon-equivalent formulas in use, so I tell the laboratory what elements must be reported. I make it clear that the results from a PMI gun are not acceptable.

Chemical analysis is used to determine the minimum preheat temperature needed to mitigate cracking, and it helps me select a filler metal with a chemical composition similar to the steel being welded. Each carbon equivalent formula provides useful guidelines, but these guidelines are specific to that formula. Rule number one is never mix the guidelines from one carbon equivalent formula with the guidelines of a different formula.

I also consider the method used to manufacture the broken component. The part may be cast, forged, hot rolled, or cold rolled. Each manufacturing method influences mechanical properties such as tensile strength, yield strength, ductility, and toughness. Casting can be notorious for having a high sulfur content. We can accommodate the higher sulfur content by using a filler metal with a high manganese content. Fortunately, in this case, the component was a forging, so issues with high sulfur are rarely a problem.

Heat Treatment

The state of heat treatment also plays a role in determining a component’s properties. It is important to know whether the original part was supplied in the annealed, normalized, tempered, or quenched and tempered condition, as it may require a heat treatment before welding, or a postweld heat treatment to develop the properties required after welding. The drawings in this case stated that the forging was normalized. Welding heats the metal to the melting point; how the weld and adjacent areas cool, and any subsequent postweld heat treatment, must be considered when developing the repair procedure.

Residual Stress

Consider the residual stress associated with the welded repair. The magnitude of the residual stress is comparable to the base metal’s yield strength. Multiple welds intersecting at different angles can produce a resultant stress that exceeds the tensile strength of the base metal, increasing the potential for cracking. Intersecting welds can be treated as vectors with both direction and magnitude to predict the resultant (stress) magnitude and direction. My usual analysis considers the residual stress in the longitudinal direction of each intersecting weld. The higher the strength of the base metal, the greater the potential for cracking at weld intersections. Cracking potential can be reduced by undermatching the filler metal. Typically, filler metals with lower strength have better ductility so that the weld can undergo some plastic deformation (yield) and the residual stress will “relax” without cracking. However, thick components with thick welds can develop triaxial stresses that can initiate a crack at an abnormality (e.g., a small slag inclusion or incomplete fusion) in the weld.

In-process stress relief can be used to reduce residual stresses to a level less likely to trigger a crack. It’s not that “thinner” components do not develop triaxial stresses; it’s just that thicker sections are more rigid and so they tend to be more troublesome and more prone to cracking.

These are the primary factors I consider when developing a repair: chemistry, manufacturing method, heat treatment, and residual stresses. With due consideration, we are ready to develop a welding procedure specification to facilitate the repair.

Component Analysis

Next, we needed to examine the component that needs repair. This repair involved a component weighing around two tons. It was a machined forging with integral gear teeth that engaged another gear that opened and closed the jaws of a very large pair of tongs that gripped large chunks of steel weighing up to several tons. The tongs pick up and move large, heavy pieces of metal to a furnace that heats them to high temperatures, then move the hot metal to a forge where it is pounded into shape. The cross section of the part where the fracture was located was about 14 in. wide by 18 in. thick. Figure 1 shows the cross section through the fracture. The failure may have initiated in a couple of locations, but the fracture appeared brittle, with no visible deformation.

In preparation for welding, the fracture surfaces were machined to form a double U-groove. About 1 in. of the fracture was not machined to leave sufficient root face to facilitate fit-up and alignment of the two broken pieces. To maintain alignment and critical dimensions, a fixture was fabricated to hold the two pieces — Fig. 2.

Figure 3 shows the part being placed in the fixture; notice the relative size of the workers and the component.

Preheat

Now that we have addressed the mechanics of the repair, how was the preheat temperature determined? We performed the chemical analysis, so I knew exactly what the chemistry was. Now it is time to put that information to use.

The chemistry was determined to be as follows:

Carbon – 0.43%, Chrome – 0.99%, Manganese – 0.2%, Molybdenum – 0.2%, and Nickel – 0.19%. The balance was Fe.

The chemistry is similar to that of American Iron and Steel Institute (AISI) 4140 steel. When those values reported by the laboratory are plugged into the carbon equivalent formula typically included in AWS, American Society for Testing Materials (ASTM), and American Society of Mechanical Engineers (ASME) standards (see Fig. 4), the calculated value was 0.7. That is “off the scale” or, simply stated, the carbon equivalent formula isn’t reliable for the alloy that had to be welded.

A different approach was needed to determine the preheat temperature required to weld this particular alloy. The solution I’ve used for many years has been to use an isothermal transformation diagram (available from the American Society for Materials) for the alloy system I need to weld.

Figure 5 is the isothermal transformation diagram for AISI 4140 published by the American Society for Metals (ASM), which includes several important features I would like to point out. The line labeled A_f is the temperature at which the steel is fully austenitized; i.e., it is the temperature necessary to ensure that all the steel is in the face-centered cubic condition, and, if held for sufficient time, all the alloying constituents will go into solution. The iron is the solvent; the alloying elements are the solute. The process of transforming from body-centered cubic to face-centered cubic begins when the temperature reaches A_s. The transformation to austenite is complete once the steel is heated to the austenite finish temperature, A_f.

Three horizontal lines at the bottom of the diagram are labeled M_s, M₅₀, and M₉₀. Upon cooling from the austenitizing temperature, M_s is the temperature where the retained austenite starts to decay into martensite, M₅₀ is the temperature at which 50% of the retained austenite has decayed into martensite, and M₉₀ is the temperature where 90% of the retained austenite has decayed into martensite. The material is considered to be fully hardened.

When austenite decays into martensite, the crystalline structure becomes body-centered tetragonal, instead of being body-centered cubic, as is the case for ferrite. Martensite is shaped like a shoebox; that is, the cubic shape is elongated like a shoebox to accommodate the extra carbon. This happens while the metal is in the solid state, so it has to deform under high stress. As the atoms’ crystalline arrangement changes, it becomes susceptible to cracking. The trick is to ensure the parts being welded are hot enough to prevent retained austenite from decaying into martensite.

There’s more to be discovered if the isothermal transformation diagram for 4140 is studied. Let’s assume a preheat of 800°F is used. If one follows a horizontal line from 800°F on the left axis and traces a straight horizontal line toward the right axis, we see the austenite begins to decay into ferrite and cementite after a lapse time of about three seconds. After a lapse of 100 minutes, all the austenite has decayed into ferrite and cementite; i.e., there is little to no retained austenite left to decay into martensite upon cooling below 650ºF, thus a hard, brittle microstructure and crack-inducing strain are avoided.

Figure 6, extracted from the Welding Procedure Specification (WPS), indicates that the specified preheat temperature is 700°F. This is slightly above the Ms temperature of 650°F. The high preheat/interpass temperature ensures Martensite does not form while welding is underway. Both the fixture and the welded-repaired pieces were heated in an oven and wrapped in blankets to maintain the preheat temperature while the parts were positioned on the welding positioner and readied for welding.

The Repair

Electric resistance heating blankets were used to prevent the welds and base metal from cooling below 700°F during welding. The parts were initially welded from one side until a sufficient weld deposited to hold the pieces securely in position. Then the second side was back gouged to sound metal. Once the back gouge was completed, continuous-current dry-particle magnetic particle testing was performed to verify that the back gouge depth was sufficient to remove any compromised base metal and the entire crack. The second side was then welded until a total of three inches of weld was deposited.

The assembly, including the fixture, was transported to the heat-treating facility for an in-process stress-relief operation. The WPS indicates the temperature for in-process stress relief was 1150° to 1175°F for roughly three hours, based on the thickness of the deposited weld. The weldment was wrapped in insulating blankets during transport to and from the heat-treat facility.

Then it was back into the welding positioner, and a check was made to verify the interpass temperature was at least 700°F before welding resumed. The fixture holding the parts was turned side to side while welding to ensure the residual forces on both sides of the butt joint were balanced and maintained proper alignment. Another in-process stress relief was performed once another 
3 in. of weld had been deposited.

Next, it was back into the welding positioner for more welding. The cycles of welding followed by in-process stress relief were repeated each time 3 in. of weld were deposited. Once the weld was completed, a normalizing operation was performed to ensure a uniform microstructure was attained, the residual stresses were reduced to a minimum to mitigate, and any diffusible hydrogen was minimized to mitigate the probability of delayed cold cracking, and ensure any martensite that may have inadvertently formed was tempered.

The selection of filler metal was uncomplicated because the choices were very limited. A filler metal that produced a low-hydrogen deposit and a close match in properties between the base metal and the filler metal after thermal stress relief was available from one supplier. The flux-cored electrode chosen met AWS A5.29, E101T2-GM-H4. The tradename of the flux-cored electrode was Dual Shield II 4130 SR, with shielding gas composed of 75% argon and 25% carbon dioxide.

The WPS presented the welding parameters as a graph showing the permitted ranges for wire feed speed, arc voltage, contact tip-to-work distance (CTWD), and amperage. Because the flux cored arc welding process uses a constant-voltage power supply, the arc voltage remains essentially constant once set by the welder. Likewise, once set, the wire feed speed is a constant unless the welder changes it. The CTWD is a constant as long as the welder visually monitors the distance from the end of the contact tip to the surface of the weld pool. This isn’t usually a problem when the welder understands that changing the CTWD profoundly affects the amperage.

Reputable electrode manufacturers provide recommended ranges for welding parameters on their websites. The parameters are usually presented in a tabular format; 
I convert the parameters into a graph that is easier for the welder to use. I’ve yet to have a welder complain about the presentation. The graphic included by the WPS is for this specific flux cored electrode is depicted in Fig. 7.

Conclusion

The repair was successful, and it took about 10 days to complete and to reinstall the component. The machine has been in operation since 2017 without further problems. Proper planning and coordination enabled us to repair a key component that was no longer in production. The successful repair salvaged a 
multimillion-dollar machine that was integral to the plant’s continued operation. Key to the success of the repair was the cooperation of the welding contractor and their insistence that their welders faithfully followed the WPS.

ALBERT J. MOORE JR. (amoore999@comcast.net) is owner of NAVSEA Solutions, Burlington, Conn. He is an AWS Senior Certified Welding Inspector and an ASNT NDT Level III. He is also a NOCTI certified welding instructor.

This article is based on a presentation by the author at a joint meeting of ASNT and AWS in Groton, Conn.