Testing, cost and choosing thermit

A thermit weld is finished in minutes, but proving it sound takes longer. Most of a thermit weld is out of sight inside its collar, and the defects that matter, lack of fusion at a face, a shrinkage cavity in the gap, slag from an early tap, are all internal. Before the crusher returns to service, the engineer needs evidence that the weld is sound, and before the weld is made at all, the engineer needs to be sure it is the right repair, compared with an arc weld repair or a new frame.

This chapter looks at what testing the early welders did and what can be done now, builds a cost comparison of the three repair options, and sets out the decision and the record.

How the early welds were tested

Hart reported a test made at a railway shop: a section of a cast steel locomotive frame, 4 by 5½ inches, was thermit welded with 75 pounds of thermit, 12 pounds of punchings and 1½ pounds of manganese in a firebrick mould. After it had cooled, the collar was planed back on two sides below the original surface to show where the metals had joined, the riser was cut off, and the weld was found to be solid, without a single blowhole, even in the riser. The welded section, with collar left on two sides, was then broken in a wheel press as a beam. According to the report Hart quoted, the fracture showed “perfect amalgamation of the metals”.

That test contains the elements of modern testing: cut the weld to see the fusion line, look for porosity, and load it to failure to see whether it breaks in the weld or the parent and what the fracture face shows. The difference is that a destructive test can be done on a trial weld, not on the frame that must go back into service.

Testing the finished crusher-frame weld

On the actual weld, only non-destructive tests are possible. This is an illustration of the engineer’s inspection plan, agreed with the contractor and a qualified inspector:

Figure 1Inspection of the crusher-frame thermit weld

1. 1Visual and dimensionalcollar complete, no cracks or run-outs; riser pipe confined to the riser
2. 2Frame measurementsreference marks compared with readings taken before cutting (Chapter 7)
3. 3Magnetic particle inspectioncollar edges, dressed surfaces and frame corners for surface cracks
4. 4Ultrasonic testingthrough the weld for lack of fusion, shrinkage and slag
5. 5Hardnesson dressed surfaces of weld, heat-affected zone and parent
6. 6Recordsportion batch, preheat temperatures, timing, cooling, results

Ultrasonic testing is the key test, because it can see lack of fusion at the faces and shrinkage cavities in the gap, the two defects the earlier chapters worked hardest to prevent. The collar makes it harder: its rough cast surface must be dressed in places to give the probe a contact surface, and its geometry complicates the sound paths. The inspector plans the scan from the member’s faces beyond the collar and from dressed areas of the collar itself, and calibrates on a reference block of similar steel.

A trial weld adds evidence the actual weld cannot give. The engineer required the contractor to make a qualification weld on a block of cast steel of similar grade and section, using the same portion, mould design and procedure, and to cut it up for a macro-etched section, a hardness survey and tensile and bend specimens, before welding the frame. The cost of that block test, about A$6,500 in the illustration below, is part of the thermit option’s cost.

What the riser and the collar can tell you

Some evidence comes free with the weld. When the riser is cut off, its cut face shows where the shrinkage pipe ended: a pipe confined to the riser, well above the collar, means the riser did its job (Chapter 5); a pipe running down towards the collar is a warning to look closely at the gap with ultrasonic testing. A sample drilled from the riser can be analysed to check that the weld metal’s composition is in the expected range, although the riser’s metal, the last to freeze, is not identical to the metal in the gap. The collar’s outer surface shows whether the mould filled completely and evenly, and whether there were leaks at the seal.

Who sets the acceptance criteria

Inspection is only useful against stated acceptance criteria: what size and type of indication is acceptable, and what must be repaired. For rail welds, the criteria come from the infrastructure manager’s standards and the requirements adopted through AS 1085.20. For a one-off repair like the crusher frame, there is no standard written for that weld, so the criteria are set by the engineer with the qualified inspector and, here, the manufacturer’s metallurgist, taking account of where the member is loaded and how a defect would grow.

Ultrasonic testing of cast weld metal also has limits the criteria must respect. Coarse cast grains scatter sound, which reduces sensitivity in the weld metal compared with wrought steel, and the collar’s shape leaves zones the probe cannot reach well. The inspector’s report states the techniques used, the sensitivity achieved and any areas not fully covered, so that the engineer knows what the inspection has and has not shown.

Is the block test worth A$6,500?

The qualification block test is the largest single extra cost of the thermit option. One way to judge it is to compare its cost with the cost of the failure it helps prevent.

This is an illustration. If the repair failed within a year, the quarry would face a second repair of about A$38,000, five days down at A$15,000 a day, A$75,000, and perhaps A$40,000 of secondary damage to the crusher, about A$153,000 in all. Suppose the engineer judges the chance of such a failure at about 10% without the block test, because the portion and procedure have not been proved on this steel, and about 2% with it. The expected cost of failure falls from 0.10 × 153,000 ≈ A$15,300 to 0.02 × 153,000 ≈ A$3,060, a reduction of about A$12,240, nearly twice the cost of the test.

The probabilities are judgements, not measurements, and a different engineer might choose different ones. But the calculation makes the judgement explicit, so that it can be discussed and recorded rather than left as a feeling, and it shows that the test would be worthwhile across a wide range of reasonable assumptions.

Three ways to repair the crusher

This is an illustration. The engineer compared three options. All costs exclude GST. The crusher’s lost production while it is down is valued at A$15,000 a working day, the quarry’s estimate of lost margin.

Thermit weld. The contractor’s lump sum for preparing the joint, making the mould, preheating, welding and dressing is A$28,000, with A$6,500 for the qualification block test and A$3,500 for inspection. The crusher is down for about three working days: preparation and jacking, the weld, and slow cooling and inspection.

Arc weld repair. A multi-pass arc weld across the 200 × 150 mm section, with a double-bevel preparation, would need the joint preheated and held hot for the whole of the welding, many runs with interpass cleaning, and careful control of distortion and cracking in heavy, restrained cast steel. The welding contractor quotes A$18,000, inspection A$3,500, and about five working days down.

Replacement frame. The manufacturer can supply a new frame for A$180,000 in six weeks, about 30 working days, and changing it would take four days. Inspection of the new installation is about A$1,500.

Figure 2Cost of the three crusher repair options

	Thermit weld	Arc weld repair	Replace frame now
Contract or part	A$28,000	A$18,000	A$180,000
Qualification block test	A$6,500	included in procedure	none
Inspection	A$3,500	A$3,500	A$1,500
Days down	3	5	34 (30 waiting + 4 change)
Lost production	A$45,000	A$75,000	A$510,000
Total	A$83,000	A$96,500	A$691,500

The thermit weld is cheapest, mainly because it keeps the crusher down for the shortest time: the weld is made in one pour, without the days of multi-pass welding a heavy arc repair needs. Replacing the frame now is out of proportion, because of the six-week wait. But the costs are not the whole decision.

Risk, not just cost

Each option carries risks that the cost table does not show.

• The thermit weld depends on the frame’s material being suitable for the portion and procedure (Chapter 8), on fusion across the whole 200 × 150 mm faces (Chapters 2 and 6), on shrinkage being compensated in a closed frame (Chapter 7), and on the hazards being controlled in a damp crusher pit (Chapter 3). The qualification block test and ultrasonic testing address the first two; jacking and frame inspection the third; the job plan the fourth.
• The arc weld repair carries the risks of a long, heavy, multi-pass weld in restrained cast steel: hydrogen cracking if preheat or electrode control lapses, distortion and cracking from the many shrinkage cycles, and lack of fusion between runs. Weld Repairs That Hold covers those risks and their controls.
• The replacement is the lowest technical risk but the highest cost, and leaves the quarry without a crusher for weeks.

The cost table also assumes each option works first time. A second outage after a failed repair would change the ranking, which is why the block test, valued earlier in this chapter, matters.

The engineer also asked the crusher’s manufacturer for advice, because the frame is their design and they know its material and loads. Their metallurgist reviewed the contractor’s procedure and the frame’s material grade and advised that a thermit weld was acceptable for this member, provided the qualification block test met stated requirements, the weld passed ultrasonic testing, and the frame was inspected for cracks at the next two planned shutdowns.

When thermit is the wrong choice

The crusher frame suited thermit: a heavy, compact section, a single joint, a closed frame where one controlled contraction was easier to manage than many, and a high cost of downtime. Many repairs do not suit it. Viall pointed out in 1921 that neither the plastic nor the fusion method was practical for very small sections or long seams in thin sections, which oxy-acetylene or electric welding did better, and that remains true. Thermit is a poor choice for thin plate, long seams, parts that cannot carry a mould and collar, materials no portion has been qualified for, and locations where its hazards cannot be controlled. A good engineer considers it, as one option among several, for the jobs it suits.

The decision and the record

This is an illustration. The engineer recommended the thermit repair, subject to the manufacturer’s conditions, and also recommended ordering a replacement frame as a long-lead spare to be fitted at a planned shutdown, so that the repaired frame becomes a rebuildable spare rather than a permanent fixture. The quarry’s management approved both.

The repair record the engineer assembled, so that the next person to look at the crusher could understand what had been done and why, contained:

1. the crack inspection report and photographs, and the frame’s material grade from the manufacturer;
2. the options comparison above and the manufacturer’s advice and conditions;
3. the contractor’s procedure, portion details and qualification block test results;
4. the job plan with its hazard controls and permits;
5. the preheat temperatures, timing, jacking measurements and cooling record from the weld itself;
6. the inspection results: visual, magnetic particle, ultrasonic and hardness;
7. the frame measurements before and after, and the inspection schedule for the next two shutdowns.

A failure: the repair nobody could explain

This is an illustration. When the engineer first looked at the crusher, the maintenance files contained a photograph of an old weld on the same frame’s opposite side, with a collar shape that suggested a thermit repair decades earlier. There was no record of who made it, what was used, or how it had been inspected.

That weld had served for many years, but the engineer had no way of judging its condition or remaining life except by inspecting it, and no way of knowing whether its shrinkage had left the frame stressed in a way that contributed to the new crack. Ultrasonic testing of the old weld, added to the inspection plan, found it sound. The episode is the reason the repair record above is as complete as it is: the next engineer should not have to guess.

Exercise: a higher value of lost production

During a period of high demand, the quarry values lost production at A$25,000 a working day. Recalculate the total cost of each option with the same days down, and the difference between the thermit and arc options.

The thermit weld costs 28,000 + 6,500 + 3,500 + 3 × 25,000 = A$113,000. The arc weld repair costs 18,000 + 3,500 + 5 × 25,000 = A$146,500. Replacing the frame now costs 180,000 + 1,500 + 34 × 25,000 = A$1,031,500.

The gap between the thermit and arc options widens from A$13,500 to A$33,500, because the thermit weld saves two days of downtime, now worth A$50,000. The higher the value of production, the more the decision is driven by time out of service, and the more the qualification block test and inspection, which protect against a failed repair and a second outage, are worth their cost.