Solid-state joining and the repair record

Kautny ended his book with a prediction. Of all the methods he had described, the one he considered most important, and apparently destined to revolutionise many industries, was what he called the puddling process: joining heated metal parts below their melting point by hammering, rolling or pressing them together. The physical character of metals often changes greatly when they melt, he argued, so there are real advantages in making a sound joint without melting at all. Metals and alloys then thought unweldable, because they absorbed the gases of the flame when molten and became porous and brittle, might be joined this way.

He was right, though it took most of a century. This chapter explains why solid-state joining avoids so many of the problems of earlier chapters, describes the modern processes that fulfilled his prediction, chooses a process for each pump-set repair and closes with the record that lets anyone else trust the work.

Why not melting helps

Every metal in this book had trouble that came from melting.

• Aluminium’s oxide could not be melted and had to be destroyed; its molten metal absorbed hydrogen that became porosity.
• Copper took up oxygen and hydrogen when molten; brass boiled off its zinc.
• Nickel absorbed gases and was embrittled by sulphur in its molten and hot state.
• Grey cast iron lost silicon and turned white; carbon steels formed martensite in the metal melted and heated around the weld.
• Different metals mixed in the molten pool into compositions neither parent would have chosen, and some formed brittle compounds.

A joint made without melting avoids most of these at once. If the surfaces are clean, or if their oxide films are broken up and pushed out by deformation, the metals bond atom to atom in the solid state. No pool absorbs gas, no element boils off, no filler mixes with the parents, and the heated zone is often narrower. Kautny’s hammer welding of aluminium at about 400 °C, his hammered joints in copper alloys, nickel and silver, and the ancient fire welding of the blacksmith all worked on this principle.

The modern solid-state processes

Several modern processes join metals below their melting point.

• Friction welding. One part is rotated or oscillated against another under pressure until friction heats the interface to a plastic state, then the motion stops and the parts are forged together. The oxide and contamination are expelled in the flash. It is widely used for shafts, bars and tubes, and for joining dissimilar metals that cannot be fusion welded.
• Friction stir welding. A rotating, non-consumable tool is plunged into the joint line and traversed along it, softening the metal by friction and stirring the two sides together without melting. It was invented at TWI in 1991 by Wayne Thomas and his team, as TWI records in its account of thirty years of the process. TWI’s guidance on aluminium alloys notes that it can produce sound welds in heat-treatable alloys that are prone to hot cracking in fusion welding.
• Explosion welding. A controlled explosion drives one plate against another at high speed, bonding them in the solid state. It is used to make clad plate and the transition inserts that allow, for example, aluminium to be welded to steel through a bonded intermediate piece.
• Diffusion bonding. Clean surfaces held together at high temperature and pressure bond by atomic diffusion over time.
• Pressure and resistance methods. Several resistance welding and cold pressure welding processes join metals with little or no melting.

Figure 1Kautny's prediction and the processes that fulfilled it

	What he described	Modern descendant
Hammering heated aluminium at about 400 °C	breaks oxide; bonds without melting	friction stir welding of aluminium
Rolling aluminium sheet together	oxide broken by deformation	roll bonding; clad sheet
Hammering copper alloys, nickel, silver below melting	avoids gas absorption	friction welding; pressure welding
Blacksmith's fire welding	heat and force	forge and friction welding of bars and tubes
Joining metals thought unweldable	no molten mixing	friction, explosion and diffusion bonding of dissimilar metals

Solid-state processes have their own limits. Most need specialised machines, fixturing and joint designs, and many cannot be taken to a part in the field. They suit production and particular joint shapes better than one-off repairs of irregular parts. That is why fusion welding, with all its metal-by-metal care, remains the everyday tool of the maintenance workshop.

Choosing a process for each repair

This is an illustration. The workshop’s final choices for the pump set brought the earlier chapters together.

Figure 2Process and controls chosen for the pump-set repairs

	Process and filler	Key controls
Cast iron casing foot	manual metal arc, nickel butter, nickel-iron fill	whole-casting preheat about 300 °C; short peened runs; slow cool; cleanliness
Medium-carbon steel shaft	not welded; replacement shaft	CE 0.575 and fatigue loading made welding the riskier option
Aluminium cover boss	tungsten inert gas, R-4043A	degrease then brush; dry filler; clamp; thread insert
Copper coil header	tungsten inert gas, deoxidised copper filler	whole-block preheat about 400 °C; insulating support; tube clamped
Stainless spool to mild steel	arc welding, 309-type low-carbon filler	low dilution; stringer runs; paint the carbon steel

Friction stir welding was considered for the aluminium cover and rejected: it is excellent for long seams in aluminium plate but not practical for rebuilding a small boss on an existing part without a dedicated machine and fixture. That is a fair summary of solid-state joining in maintenance: always worth asking about, often not the answer for a single repair, and sometimes the only answer, as for aluminium-to-steel joints.

Keeping the trial pieces

Several chapters relied on trial welds: the scrap casting that revealed the long hot runs and the crayon contamination, the sectioned aluminium boss, the bend-tested copper. Those pieces have value after the job. Labelled with the procedure they tested and kept with the record, they show what a sound repair in that metal looks like in section, they let a later repair be compared with an earlier one, and they are evidence, if a repair is ever questioned, that the procedure was tried before it was used on the customer’s part. A drawer of labelled trial pieces is one of the cheapest quality records a repair workshop can keep.

Inspecting solid-state joints

Solid-state joints fail differently from fusion welds. They have no porosity or solidification cracks, but they can have incomplete bonding where surfaces were not fully cleaned or deformed, which may be invisible on the surface. For production processes such as friction and friction stir welding, quality is controlled largely by monitoring the machine parameters and by destructive tests of sample joints, alongside non-destructive testing suited to the joint. The principle is the same as for every repair in this book: the evidence must match the way the joint can fail.

Procedures and qualification

Every repair in this book has relied on a written plan: the metal, the preheat, the filler, the technique, the cooling and the inspection. In formal welding practice that plan is a welding procedure specification, and its fitness is demonstrated by welding and testing a test piece, recorded in a procedure qualification record. The welder’s ability to follow it is demonstrated separately. Standards set out how this is done; for example, the ISO 15614 series covers specification and qualification of welding procedures for metallic materials, with separate parts for steels and nickel alloys and for aluminium, and the AS/NZS 1554 series covers welding of structural steel in Australia. The applicable standard depends on the work, the industry and the contract. For safety-critical, pressure-containing or structural repairs, qualified procedures, qualified welders and inspection to the governing standard are requirements, not refinements.

Knowledge and skill: Kautny’s other point

James Whiteford’s preface to the English edition made a point that sits beside Kautny’s prediction. In Germany, he noted, welding schools were by then attached to the technical colleges, and Kautny had helped set them up; whether the process succeeded, in Whiteford’s view, depended above all on what the individual welder knew and could do. Their book was written to give welders an understanding of the art, not only instructions.

That remains true, and it is why qualification of welders is separate from qualification of procedures. A procedure written from the properties in this book is only as good as the welder who follows it. In Australia, welder training and certification are offered through vocational education and through the national welding industry body, Weld Australia, and the qualification required for a given job is set by the governing standard and contract. The workshop’s trial welds on scrap castings, aluminium offcuts and copper bar were, among other things, a way of building the specific skill each repair needed.

When to stop and ask

Property-first planning also tells a workshop when a repair is beyond what it should decide alone. The signals are clear: the metal cannot be identified with confidence; the part contains pressure, carries a safety-critical load or falls under a design code or regulator; the crack runs into a region whose failure would be dangerous; no qualified procedure exists for the metal, thickness and process; or the consequence of a hidden defect is severe, as it was for the fatigue-loaded shaft. In any of those cases the right step is to involve a qualified welding engineer, the equipment’s designer or owner, and, where required, the inspection body or regulator, before any heat is applied.

What has not changed since 1915

Reading Kautny’s metal chapters a century on, it is striking how much of his reasoning survives even where his numbers do not.

• Carbon governs iron. His account of carbon dissolved in molten iron and taking different forms on cooling, and of rapid cooling producing a hard, brittle constituent, is the basis of Chapter 4.
• Silicon keeps cast iron grey. His explanation of white, brittle welds in grey iron as a loss of silicon, though wrong about evaporation, points to the right element and the right remedies.
• Oxides must be dealt with. His rule that a metal melting below its oxide needs the oxide destroyed, chemically or mechanically, explains aluminium welding today.
• Phosphorus rescues copper. His phosphorus-bearing copper fillers anticipated deoxidised coppers and fillers.
• Sulphur ruins nickel. His warning about nickel’s affinity for sulphur is now one of the first rules of nickel welding.
• Preheat and sequence control stress. His cast iron expansion calculation and his heated-rim method for a cracked spoke are the reasoning of Chapter 3.
• Joining below melting is powerful. His closing prediction came true.

What has changed is measurement, control and documentation: modern property data, process control, inert gas shielding, qualified procedures and inspection. This book has tried to combine the two: his reasoning, checked against modern numbers, written down so that others can check it.

The repair record

The chapters of this book have each added entries to one record. For the pump set, the workshop’s record answered the questions that anyone, the customer, an inspector or the next repairer, would ask.

Figure 3Repair record for the pump set

• Avoid: Identificationeach metal, how it was confirmed, and certificates
• Avoid: Governing propertythe hazard each repair had to control
• Avoid: Procedureprocess, filler type and batch, preheat, interpass, technique
• Avoid: Thermal recordtemperatures, where measured, heating and cooling times
• Avoid: Cleanlinessdegreasing, markers and crayons excluded, oil burn-out
• Avoid: Inspectiondye penetrant, pressure test, flatness, machining checks
• Avoid: Decisionsshaft replaced rather than repaired, with reasons
• Do: x Handoverrecord supplied to the owner; repaired parts tagged

A failure: the repair nobody knew about

This is an illustration. A year after the pump set returned to service, the factory’s own maintenance contractor found a small crack near the old repair on the cast iron foot, after the pump had been knocked by a forklift. Not knowing what the earlier repair was, the contractor ground it out and welded it with steel electrodes. The foot cracked again beside the new weld within a day.

When the workshop was called back, its record showed exactly what the original repair had been: a nickel-based repair with whole-casting preheat. The contractor’s steel weld on cast iron, without preheat, had repeated the very failure described in Chapter 1. The workshop repaired the foot again with its documented procedure. It also changed its handover: the record now goes to the owner’s maintenance system as well as the job file, and repaired cast iron parts are tagged with a durable label identifying the repair type and the record number. A repair that only its maker can understand is only half finished.

Health and safety across the metals

Each metal in the case brought its own fume: nickel from the cast iron repair fillers, chromium including hexavalent chromium from the stainless spool, copper from the coil, and the hazards of the aluminium process and of hot parts. Safe Work Australia’s model code of practice on welding processes, as adopted in each jurisdiction, sets out the duty to manage these risks, and the consumable and material safety data sheets give the specifics. Hot parts after preheating, compressed gases, ultraviolet radiation and confined spaces inside equipment complete the list. A repair plan that starts from the metal’s properties ends with the controls those properties require.

Exercise: the energy and the time

The two largest preheats in the job were the cast iron casing, about 10,950 kJ to 300 °C, and the copper header, about 267 kJ to 400 °C. Calculate the total in kilowatt-hours and the cost of that energy in an electric furnace at an illustrative A$0.30 per kWh excluding GST. Then say what this suggests about where the cost of such repairs really lies.

The total is about 10,950 + 267 ≈ 11,217 kJ, which is about 11,217 ÷ 3,600 ≈ 3.1 kWh. At A$0.30 per kWh, the energy costs about A$0.93 excluding GST, or about A$1.03 with GST at 10%. The energy is trivial. The cost of these repairs lies in the time: identifying the metals, preparing and cleaning, heating and cooling the casting slowly, trial welds, careful welding and inspection, and recording. The same is true of every metal in this book: the extra care that each metal’s properties demand is mostly time, and the record is what shows that time was spent well.