One such application is the high speed welding[1] of LPG (liquid petroleum gas) cylinders in India.  Operating under the classification F65A0-EL8,  of AWS/SFA 5.17[2]., it is obvious that whereas the electrode specification is highlighted (EL8[3]), as is the mechanical properties of the expected weld metal, no details are forthcoming about the other consumable, that is welding flux, which is a critical component for achieving a sound weld.

The plate[4] thickness being barely 3mm, the problem of burn-through & weld defects involving oxide inclusions & porosity was a serious time & cost multiplier.

As process parameters required 2 runs, (circumferential & bung ) the challenge was, therefore, to design a uniform melting & fast-self peeling  flux able to handle welding abuse (unclean surfaces, no pre-heat, no flux pre-heat etc) which is very endemic within the unregulated welding sectors of India.

Utilizing a VOC (Voice of Customer) methodology, an operational data collection drive was set-up to monitor cylinder welding. Defects were then stratified & plotted as Pareto charts. Fish bone (Ishikawa) diagrams were then set up to arrive at the root causes of the defects.

This was of-course, not a 2 step process, but a repetitive process of continuous improvement identified as PDCA (plan, Do, Check, Act) & incorporating Design of Experiments (DOE[5]) protocols.

It was determined that uniformity of the welding flux components, flux fluidity, flux reducing capability & flux surface tension after welding were the Critical to Quality components.

• If a uniform mixture was to be ensured, the mineral components would have to be melted. Physical mixing would never result in a 100% mixed material.  This validated the customers’ contention that agglomerated/bonded fluxes had already been tried & were rejected due to persistent operational inadequacies.

The solution was to offer a Fused SAW flux.

• High welding speeds & high currents resulted in molten flux run-off. Flux fluidity therefore required reduction through adjustment of input proportions.
• Lack of joint cleanliness required a superior de-oxidation capability to address both the oxide inclusions & porosity due to ambient moisture. Components that increased Si & Mn deposition in the weld metal were then added or increased.
• At roughly 45seconds/revolution, Slag not detaching immediately after the first run would cause defects in the form of inclusions in the next run.   Components that influenced the surface tension at the metal-slag interface were then added or increased.

 

 A fused Mno-SiO2-CaO system based SAW flux, duly tweaked, as enumerated above was offered & accepted.

Whereas a customized variation of this fused welding flux (Fluxomelt BRD-1[6]) became the standard flux used for this process in India (& Bangladesh); from the Lean[7] point of view though, there were still random weld defects that could not be accounted for even after welding parameters were all fine tuned.

As users of SAW flux are aware, fluxes are sold in mesh size fractions such as 8X48, 12X100 or 20X150. These fractions indicate particle size distribution within the upper & lower control limits of the mesh sizes indicated. Thus 8X48 implies 100% of the material passed a sieve of mesh size 8 & 0% passed the sieve of mesh size 48. Mesh sizes are measured in various standards worldwide such as Tyler, ASTM (US), BSS (UK) etc.

A general rule of thumb is of decreasing the flux particle size as the welding speed (as is the current) increases.

For this application, unfortunately, finer sizes created porosity based weld defects, probably due to the flux not being heated prior to use. As the chemical/mineral components of the flux had already been fine tuned, attention was therefore focused on the mesh size (particle size) of the flux granules.

Mesh (sieve) size fractions were then generated, with materials ranging from that passing 8 mesh to that passing 48 mesh.

Welding runs were then made with each individual fraction & a pareto-bar chart generated stratifying each defect.

The optimum mesh size fraction, as per the data generated was that between 18 mesh & 24 mesh.

Control Limits of 18 mesh (LCL) & 24 mesh (UCL) were then set at the flux manufacturing facility & welding defects reduced to a minimum.

This narrow fraction to be sold at the same competitive price, of-course required a Lean- Six Sigma[8] based data collection & analysis action plan, to optimize the logistics of this production reducing processing waste, which in this case also included any deviation from the ideal customer requirement. This was of course our own production concern & not a welding issue, & was handled along similar lines as discussed above.

 Conclusion:

Regardless of Specifications, any product must operate effectively & efficiently[9] to satisfy the customers’ expectations.

The mineral/chemical composition & particle (granular) size of a Submerged Arc welding flux are 2 very critical to quality (CTQ) parameters for ensuring defect free welds. Welding of CS (carbon Steel), SS (Stainless steel) or ultra low C steels require fluxes of different chemical components to facilitate the optimal weld microstructure. Whereas similar welding results are possible from 2 or more fluxes having totally different components; the user should explore which set of flux components are most optimal for their application, based on the chemistry & microstructure of the parent metal & that desired of the weld metal.

Similarly the particle/granular size affects the rate of melting & wetting the parent metal (within that limited time window of arcing)  & is unique to the user & needs to be optimized.  

 

It is therefore important for SAW shops to take maximum advantage of this versatile & highly efficient process by implementing a Six sigma based data collection & analysis drive to arrive at both the optimal base composition  of the flux required, &, the optimal granule size fraction, required for their specific process parameters. The goal being, to ensure a robust, defect & breakdown free production process, insulated from human error. Not only are these processes simple & economical[10]  to implement, but in this economical situation, squeezing out processing waste would go a long way in addressing operational profitability & international competitiveness.

This case study relates to fine tuning of upper & lower control limits of a welding consumable, namely Submerged arc welding flux (fused variety) which has  applications for use in fabrication of Oil, Gas & water pipelines, gas cylinders, ship building  & surfacing of worn out surfaces of mining & material handling equipment. Of all automated welding processes, this (SAW) is the most versatile.

Background: Submerged arc welding flux is a granular ceramic mixture manufactured by mixing mineral & chemical components in proprietary proportions & fusing the mixture in an electric furnace. The molten product is then quenched in water, dried in a rotary kiln, crushed & sieved to yield a granular product. The final product is sold in sieve mesh[1] fractions, spread over a number of sizes. This granular composition physically added to cover the welding arc zone (hence the name Sub-merged arc) shields the weld arc zone from the atmosphere, refines the weld metal & peels off on solidification after floating to the top.

Problem Statement:

We were approached by our flux using customers with an unusual problem. Increasing welding productivity through higher welding speeds & in effect higher operating currents & voltages resulted in a dramatic increase in weld defects. This was with a product (consumable)  that they had used for over 10 years with very satisfactory results.

A data collection team was immediately deputed to their companies & based on its analysis; the root cause was identified as the mesh (particle) size fraction. Whereas we had been supplying them a product designated as Mesh 8X48[2]; the new optimal size fraction was mesh 18 X 30. This fraction had upper & lower control limits almost 50% less than what was being currently supplied. This was easily said than done as this meant eliminating 20% of our currently sell-able product. India being a very cost sensitive market, price increase was not an option. Thus began our simultaneous Six Sigma & Lean initiative to reduce product variation to the new customer requirement & simultaneously drastically reducing our operational costs.

The Goal was thus to reduce product variation to customer acceptable limits of UCL mesh 18 & LCL mesh 30, within 2 months.

Data was collected and a Pareto bar chart of causes plotted, stratifying manufacturing process points against defects, in this case the weight fraction not conforming to the new control limits.

An Action Plan was set up focusing on optimizing the operation of the roller crusher & the sieve decks. A simultaneous 5S & TPM program were initiated for both machinery stations.

 This included regular scheduled vibration & alignment checks &, calibration of measurement tools.

Data Collection was carried out varying the operational parameters of the crusher; namely: the revolution (rpm); the gap between the crushers & the feed rate. The data was Pareto charted & Ishikawa (fish bone) diagrams were then set up to address the process limitations.

A DOE (design of experiments) varying the three parameters (roller gap, roller revolution speed & material feed rate) were carried out & the ideal set of parameters were arrived at after trials.

Line balancing, movement analysis & Takt time studies were also undertaken to ensure a uniform production rate between work stations & breakdown free production.

 

One example of process improvements based on the data driven campaign was the placement of Photo cells connected to warning buzzers to indicate non conforming gap between the crusher rollers. This failsafe (Poka Yoke) system ensured a properly calibrated machine at all times. The rollers themselves were weld-overlayed/surfaced with a special alloy & duly machined to ensure smooth vibration free operation. 

 

Literature had suggested that strategic application of Six Sigma protocols would result in a more than a 30% improvement in the bottom line & our experience in implementing our data driven Six Sigma & Lean campaign exceeded all of our expectations.