Suppression of Defects in High Productivity WeldingPatricio F. Mendez and Thomas W. Eagar, MIT Welding and Joining Group

9

10

11

12

13

0% 5% 10% 15% 20% 25%speed increase

wel

ding

exp

ense

s [$

bn]

$800M savings!

10% speed increase

• $12 billion a year is spent on automatic welding• Labor and equipment costs decrease with increasing speed

• Welding is performed late in the manufacturing process a weld defect ruins all previous added value• 300 million tons of metal product are welded each year• Robotic welding is expanding at 10% per year

Increasing the welding speed by 10% could generate almost $1 billion in savings (1% of the payroll in metal products and industrial equipment)

FASTER WELDS SAVE MUCH MONEY WHY CAN’T WE GO FASTER?Good Weld

Bad Weld (humping)Top view Cross sectionbeginning end

Top view

Cross sections

beginning end

increase

d productiv

ity

Good welds

Bad weldsproductivity limit

(undercutting+tunnel porosity)

(undercutting)

EXPERIMENTAL EVIDENCELow Current High Current

speed current DEFECT

Thick weld pool(recirculating flows)

Thin weld pool(longitudinal flows, large depression,

deep penetration)

thickness<100mPREVIOUS STATE OF

THE ART

• Thick weld pool with recirculation• Marangoni*>electromagnetic> gas shear>buoyancy• current electromagnetic may dominate (speculation)• Difficult to model the fully coupled weld, especially with large depression* Marangoni flows are induced by variations in the surface tension of the liquid-gas interface

At high current (>250A) the weld pool surface is very depressed. This depression cannot be explained by the arc pressure alone.

OVERCOMING THE LIMITATIONS• Use the previous insight into the

problem: we already know that the weld pool is very thin• Use orders of magnitude instead of many digits• Use dimensional analysis to reduce the problem• Use the information “hidden” in the governing equations for the weld pool

OUR SOLUTION: THE MAGNITUDE SCALING

METHOD Required• Previous insight into the problem• Ability to write down the governing equations• Solution is “well behaved”• Second order differential equations (or less)• Second derivatives not largeNot required• To simplify the equations• To solve the equationsProvides• Order of magnitude of solution• Relative importance of driving forces• Simple definition of concepts

ANALYSIS OF A THIN WELD POOL

1.00

0.66

0.34

0.08

0.07

0.06

0.03

0.03

7.E

-05

3.E

-04

3.E

-02

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 P2 P1 P7 P10 P5 P4 P9 P8 P6 P3

arc pressureelectromagnetic

hydrostaticcapillary

Marangonibuoyancy

gas shear

“Pec

let”

“Rey

nold

s”

geom

etry

diff

.=/d

iff.

• Driving forces• Effects• Geometry

• Gas shear is the dominant driving force• Marangoni forces are very small• Thickness of the thin liquid layer:

2VD/max)½

• Thickness thin layer thickness mushy zone• The thin layer freezes in 1 ms• The arc acts almost directly over the solid-liquid interface = thickness of liquid layer = liquid metal viscosity

V = welding velocityD = depth of penetration

max = gas shear from arc plasma

MECHANISM OF DEFECT FORMATION

Qfro

nt

flow blockedby prematuresolidification

arc heat losses

humped beadforming

lack of materialbecause the flow

is obstructedsolidifying

tail

Beginning of humping Development

DEFECT SUPPRESSION STRATEGIES• “Forehand” torch angle. This

decreases the gas shear, therefore the weld pool gets thick closer to the arc and will not freeze• Modify gas shear distribution, for example by varying electrode to work distance. This requires a deeper study of the arc (in progress now)• Modify heat input distribution from the arc. For example, if the hot zone reaches more to the rear, humping will be delayed. • Techniques that manipulate Marangoni forces are unlikely to work

Gas shear decreases towards the tail, therefore the weld pool gets thicker. Heat input also decreases towards the end. When heat input decreases faster than the gas shear premature freezing and obstruction of the flow will occur. A total obstruction causes humping, a partial one at the sides causes undercutting, and at the bottom causes tunnel porosity.