European Commission logo
English English
CORDIS - EU research results

Development of oxy-hydrogen flame for welding, cutting and brazing

Final Report Summary - SAFEFLAME (Development of oxy-hydrogen flame for welding, cutting and brazing)

Executive Summary:
Brazing is a process of joining two materials together at a temperature higher than 450°C without melting the parent metal. Filler metal is melted into the joint which when solidified causes all the components to be joined together. In traditional flame brazing, an oxyacetylene or propane source is used to generate the flame. Due to the safety concerns associated with storage and transport of oxyacetylene, there has been a trend away from the use of acetylene. A new method of generating a flame is required on the market, which is safe and low cost.

In the SafeFlame approach, water is electrolysed into hydrogen and oxygen, and then recombined at the torch tip and burned. Gases are kept completely separate until they are recombined, and at no time are the gases stored. This provides a safer solution, which is easy to use.

The project has focused on the development of the technology and the systems, and a demonstrator has been successfully produced. The SafeFlame unit generates oxygen and hydrogen gases from a water electrolysis process, and is designed to deliver up to 9 litres per minute of hydrogen gas, and 4.5 litres per minute of oxygen gas to a maximum pressure of 0.5barg (7psig) via a dedicated flame torch nozzle. When the torch is ignited, a high temperature flame is produced which can be used for a variety of laboratory and workshop applications including heating, brazing and soldering.

SafeFlame is designed for use by trained personnel familiar with the use of similar equipment and with knowledge of safety requirements for the use of hydrogen and other industrial gases. When handling, installing or operating this equipment, personnel must employ safe working practices and observe all related local regulations, health and safety procedures and legal requirements for safety.

Project Context and Objectives:
The SafeFlame project addresses the needs of large communities of SMEs in the huge global market of brazing. The consortium represents approximately 1000 SMEs using brazing, part of a sector that employs 125,000 people in Europe with a turnover of 20bn Euros . Currently, brazing and many other flame processes use highly flammable bottled gases such as acetylene and propane. The use of high pressure cylinders containing flammable gases increases the risk of major accidents and the cost of insuring operators and vehicles transporting the gases. Such processes cover a wide range of industries including refrigeration, air conditioning, ship building, rail stock, nuclear, jewellery, dentistry and polishing.
Each year in the EU there are hundreds of serious incidents involving bottled gases such as acetylene and propane. These gases are routinely used in well over one hundred thousand individual applications using fuel and oxygen flames, for example in manual hand torches for fabrication by welding, cutting and brazing of items such as refrigeration and air-conditioning plants.
Those cylinders known to have failed catastrophically generally split axially and fragment into several pieces; these have been known to travel distances in excess of 200 metres, from the explosive propulsion of the burning gas which they contained.
Kept at numerous sites across the EU, acetylene is a versatile gas used by workers such as welders and mechanics, but it can be lethal if ignited or trapped inside a building that has caught fire, as the high pressure cylinders are highly explosive. The gas can still pose a risk after a fire is extinguished, as the acetylene inside the cylinders can begin to chemically decompose if the blaze was hot enough, effectively creating a time-bomb.
The project has developed a ‘new safer flame from water’ technology, which uses a portable electrolyser, operating on a standard single phase electricity supply. The electrolyser dissociates water with high efficiency to supply both hydrogen and oxygen at low pressure, in two separate streams to a flame torch, so they can be individually controlled and combusted at the face of a surface mixed burner. Using state-of-the-art electrolysing technology designed for the hydrogen economy, water molecules have been separated into hydrogen and oxygen, which have been recombined and combusted in proportions optimised for each application. By this means, the flame can be surface mixed and the chemistry of the flame has been controlled between reducing (hydrogen rich) and oxidising (oxygen rich) to suit the needs of the application.

The SafeFlame electrolyser uses deionised water which is separated by the application of electricity into streams of hydrogen and oxygen gases. The electrolyser has been developed by ITM Power. The gases are delivered from the electrolyser as separate streams, they are not stored, but instead are generated separately and supplied in well metered and controlled proportions to suit burning applications. Several designs of torches have been modelled, simulated, constructed and tested. The design of the torch has proved critical to the correct operation of the heat application for a given process. Several torches have been constructed in order to cover a range of applications.
Existing hand torches employed for combusting acetylene and oxygen utilise pre-mixed burners and it is this pre-mixing of the gases that is the main safety concern. Oxy-fuel-gas mixtures are inherently dangerous and operators can easily achieve “light back” at weak mixture ratios resulting in small explosions and potential damage or injury. Surface mixed burner designs have been used for many years in the glass manipulation industry to enable hydrogen and oxygen flames to be used safely with no fear of light back. The combination of technologies has resulted in systems which are safe and easy to use for novice and expert users alike.
During the project, the SafeFlame system has been developed which has the following characteristics:

• User friendly flame system.
• Portable.
• Safer than gas cylinders.
• Low running costs.
• Adjustable flame chemistry.
• High flame/adjustable flame temperature.
• Good heat transfer characteristics.

SafeFlame technologies have addressed the needs of large communities of SME-AGs and SME’s. There are many applications in industry today where many industrial (manufacturing, repair, engineering, servicing etc) companies are forced to use gases to produce flames in particular for brazing copper, aluminium and other metals for refrigeration, air conditioning and other construction applications. SafeFlame has demonstrated a relevant technology which can replace gas cylinders for many of the applications which use pipe work up to 32mm diameter.

The aims of the work packages were to consider and address the needs of the SME-AGs and their body of companies which comprise a large work force across Europe engaged in the brazing industry. To that end the SME-AGs undertook some surveys of market requirements, and helped to formulate the specification of the devices to be produced.

The RTD ITM Power undertook a series of tasks and these led to various electrolysers being produced, which were incorporated into test stack systems and then into prototypes. The tasks commenced with modelling of gas flows. Throughout the development process, membrane technologies were either being trialed or experimental systems were under development. A new patent associated with findings in membrane technologies was applied for.

ITM Power conducted studies to investigate different electrolytes and the concentration of the electrolytes with respect to performance of the various electrolysers studied. This work led to the identification of optimal process conditions for the various electrolytes required. It had been noted there was always going to be a trade-off between performance and the effects of increased concentrations. This work was carried forward to the manufacture of tests stacks. Duty cycle analysis was conducted on many test stack systems running concurrently. Therefore, in conjunction with concentration studies and membrane technologies, many systems could be assessed at the same time. This achieved optimisation of various parameters, resulting in stacks which could be used in the SafeFlame system.

The stacks arising from the optimisation studies which had passed longevity testing were then able to be recommended for incorporation in the balance of plant (BoP) design and construction. The BoP design considered both the aesthetic appearance of the external structure and all the internal equipment to make a fully functioning device. The BoP was an extensive system which incorporated the liquid and air handling systems, electrical and electronic systems, the electrolyser and gas handling capabilities. This resulted in the testing of a completed system.

Several systems were produced during the course of the project with their respective advantages and disadvantages. One variation required potassium hydroxide as an electrolyte while an alternative only required water. The water based system used acid which was captured and retained in the membrane, therefore it did not leach away. The water based system was preferred by project partners because it seemed safer to use as there was no free potassium hydroxide in the system.

Work Package 3 considered the design and construction of the combustor. Several designs were produced and many others had been modelled. There had been an iterative development cycle of the flame torch, which considered the handle aesthetics and the nozzle ports. In different embodiments of similar designs, trials were undertaken to experiment with the arrangements of one, two, and three port burners. Single port burners were preferred because of the simplicity of manufacture. However three port burners were preferred for heating larger diameter pipes.

Modelling of the combustion process had been implemented to assess differences between oxyhydrogen and other gas types. Modelling of nozzle geometry and the influences of port sizes and spatial alignment was considered. This gave a significant body of knowledge to validate the positioning and understanding of how to best construct the nozzle structure. Additionally, the designs could be optimised to meet processing requirements for the speed of flame arising from the nozzle, this has influence on the Reynolds numbers produced and therefore the velocity of the flame which may impact the workpiece.

Modelling had also been undertaken to assess the flame chemistry or gas species which were arising in the flames. The gas species and the velocity of impact could give rise to a known mechanism of heat transfer. The gas species in the flame would also be available to do work such as surface modification. Because of the high temperature of the flames involved, some of the flame attributes are acting as a plasma, therefore the possibility of processing or surface treating other materials also became viable. VTT’s modelling work led to the optimisation of nozzle conditions for the torch nozzles. ITM Power also produced a range of aesthetically pleasing, easy to use handles for the torches, which received high praise across all the EU countries in which they had been tested.

Process trials have been undertaken by many of the project partners to validate and experience the outputs of the project. These included working with dry gases from gas cylinders metered through gas panels and burned via the project’s torches, and conducting similar experiments through the SafeFlame prototypes which were producing wet gases. It was interesting for partners to understand the difference in processing conditions which may be available under the different scenarios. During the course of experiments, gas stability diagrams were constructed to present the safe working envelope of a range of conditions for oxyhydrogen, oxypropane and oxyacetylene. Pictures of flames at many different gas flow ratios and flow rates were established and these were linked to the stability diagrams. Various tests of brazed samples were conducted, these included polished micro sections of brazed joints, for the purpose of studying flow conditions of the filler materials. Analysis of joint strengths as a function of process gases had also been conducted; these had shown very compatible results between all the gases, however where a fast heating rate had been achieved, less annealing took place, therefore a higher joint strength was achieved.

Investigations into joint geometry were also conducted, this led to interesting findings in that the traditional joining practices widely adopted across the EU produces joints which are overly large. This means that excess braze filler is being used in almost every joint undertaken. Sometimes these joints use precious metal fillers.

During the project there was a process of integration, assembly and commissioning; this led to three prototype systems, these were tested and evaluated by partners. Various partners were able to consider the health and safety implications of the systems, ITM Power had produced HAZOP and operational reports for the project. These were transitioned into reports and manuals which could later be used by project partners. A preliminary life cycle assessment was also undertaken to show the total material breakdown by type of material and the relative equivalent carbon dioxide content of that material.

Application demonstrators were made to assess the ease of construction by using either oxyacetylene or oxyhydrogen systems. The demonstrators consisted of brazing a radiator coil, radiator pipework, and a copper coil system. It was commented that constructing systems using oxyhydrogen is far easier than using oxyacetylene. It was noted that while oxyacetylene is quicker to use, the user has to be skilled and quick to construct the joint without destroying the pipework. Novice and expert users both preferred using oxyhydrogen systems because of the level of confidence and the clarity of vision it gave the user.

Extensive training materials have arisen through the course of the project, these are to help train operators, train the trainers and construct presentations for training and dissemination. Various presentations have previously been reported and these materials have been arising from the project partners.

Many dissemination events have been held and news and media broadcasts have had wide coverage across Europe for the SafeFlame project, these have included articles by Euro news, coverage with a UK regional MEP, and an article featuring the EU Commissioner for science. A website has been constructed initially by ITM Power, then being passed to TWI and now residing with the SME-AGs.

Project Results:
The electrolyser stack is the part of the system where the electrolysis process takes place and hydrogen and oxygen gases are produced from water. All other parts of any electrolyser system have the sole purpose of ensuring that the optimum conditions are provided to the stack so that electrolysis can take place.

The electrolyser produces hydrogen and oxygen gases by splitting apart water molecules using the power provided by the Power System. Each cell in the stack therefore needs to be supplied with water for this reaction to take place. It is important that the flow of water is even across all cells in the stack to ensure even cell performance and prevent premature cell failure. It is also important that the flow of water through the cells is able to facilitate the transport of the produced gases out of the cells. The water flow is also important in regulating the temperature of the stack. Heaters and cooling systems are used to control the temperature of the water in the electrolyser system and heat is transferred into or out of the cells as required to ensure the stack remains at the optimum operating temperature.

It is important that the hydrogen and oxygen gases are removed from the stack and water system. The gases are separated from the water pumped from the electrolyser stack. The gas is pressurised, and dried to the desired level and then presented to the user in a manner that is convenient and safe.

ITM Power had an existing stack design, the LAM stack, which consists of a number of cells with a 130cm2 active area. The LAM stacks are built up with varying numbers of cells depending on the required gas output. The LAM stack design seen in Figure 1 has been tested extensively at pressures of 15bar. Because of this requirement for high pressure gas, the LAM stack has thick stainless steel end plates which add cost, weight and size to the electrolyser. The LAM stack is based around acidic cation-exchange (CE) membranes which require expensive platinum-based catalysts. Due to the nature of the SafeFlame application, the SafeFlame electrolyser stack only needs to produce gas at low pressures (<0.5bar) and given the fact that low weight and cost are both highly beneficial, the LAM stack can be seen as over engineered for SafeFlame.

Figure 1 LAM Stack

A new stack design was conceive. It was able to keep the well proven internal electrode components from the LAM stack, but was better suited to the needs of the SafeFlame application. This new stack design (the Petal stack) was able to use smaller, lower cost, lighter weight end plates as high pressures were not being produced inside the stack. The Petal stack design significantly reduces the overall size of the stack by smart positioning of the tie rods needed to clamp the cells together. This size reduction was a great benefit for SafeFlame. The petal stack design can be seen in Figure 2. The difference in size between a LAM stack and an equivalent petal stack is shown in Figure 3.

Figure 2 Petal Stack

Figure 3 Petal and LAM Stacks
The stacks are constructed from cells, numerous cells are use in a stack. Each cell is constructed from several material layers. Sealing of the layers has taken careful consideration, and new methods to allow an even pressure to be applied over the faces of the layers have been developed. Fluid flow over the face of the surfaces has to be considered. The electrodes need to be able to distribute the water supply evenly over the entire active area of each cell and be able to let the generated gases move away from the membrane and out of the cell. This movement of water and gas can be achieved by using either a flow field behind the electrode or by using an electrode material that is porous and allows fluids to pass through.

Flow fields require expensive machining to produce and therefore are not suited to a cost-sensitive product such as SafeFlame. Sintered metal structures have an open porous structure that allows liquids and gasses to pass through easily. They would also allow large contact areas with other components for improved electrical conduction between components (membrane, foil etc.). Sintered materials can be made with varying pore sizes and can be produced from compressing metal powders together to form a porous solid, or by compressing many layers of mesh together to give a single part with a much more open structure than is achievable by sintering powder.

The membrane is the cell component that has the greatest effect on the performance of the electrolyser cell. The membrane has the functions of keeping the gases generated on each side of the cell separate and providing electrical insulation between the two electrodes. However, the membrane also needs to be able to allow the movement of ions from one side of the cell to the other in order for the electrolysis reaction to take place. These functions all have an important impact on the operation of the cell.

If the gases were allowed to mix the result would be a mixture of hydrogen and oxygen in the system which is an explosive combination, this should be avoided for safety reasons. If the gases were able to pass to the other side of the cell they would quickly cause corrosion of the electrodes and catalysts which would decrease the performance of the stack and dramatically reduce the lifetime of the product. It is therefore vital that the membrane is able to prevent the gases passing from one side of the cell to the other.

The efficiency of the cells is directly related to the separation between the electrodes. The membrane acts as an electrical insulation layer between the electrodes which are both pushed up against the membrane. The cell efficiency can therefore be improved by decreasing the thickness of the membrane. However, as the membrane becomes thinner it also becomes weaker and more likely to fail. It is important to make the membrane as thin as possible while making sure that it has the necessary strength to be able to survive in the stack without suffering splits or other damage.

The cell efficiency is also directly related to the ionic conductivity of the membrane. The higher the ionic conductivity, the easier it is for ions to pass from one side of the cell to the other allowing the electrolysis reaction to take place at a higher rate. Materials with high ionic conductivity are traditionally very mechanically weak, it is therefore important that a membrane is developed which has the necessary mechanical strength but also has high ionic conductivity.

Modelling of fluid flow through the stack
In order for the stack to work effectively it is important that water is supplied to all cells in the stack. The stack design was modelled using CFD software. This analysis was performed to give graphical representations of the flow of electrolyte through the manifolds, cross-drilled holes and cell cavities within the stack.

If the flow into a cell was too low, less heat would be removed from the cell, which would cause that particular cell to run at a higher temperature. An increase in temperature would increase the efficiency of the cell and would lead to that cell producing more gas and working harder than the other cells in the stack. This could cause the cell to have a shorter lifetime and cause the stack to fail earlier than expected. CAD models of the six cell 200cm² Petal stack were generated and imported into the CFD software.

From the CAD models produced a potential pump candidate had to be found. The flow in the six cell stack was simulated at three different input flow rates of 1.25lpm 2.5lpm and 5lpm; a sample is given in Figure 4. This indicated whether the flow rate given by the pump would be able to provide consistent flow into all cells within the stack, and whether there would be a benefit to finding a pump with a higher or lower flow rate. The simulation of flow across a cell is given in Figure 5.

Figure 4 Cell flow rate simulation.

Figure 5 Flow of liquid across cell surface.

Cell current density
Test evidence suggested that a current density of 1A/cm2 was high enough to produce good quantities of gas, but low enough that the potential damage could be minimised. This current density is higher than used in many commercially available alkaline electrolyser systems which typically operate in the region of 0.3A/cm2 enabling smaller, cheaper stacks to be used.

The increased current density is possibly due to the advantages of the ITM alkaline ionic membrane. Other alkaline electrolyser systems generally do not use an ionic membrane and because of this they are generally less efficient. The ITM ionic membrane allows the gases produced on each side of the cell to be kept separate; non-separated alkaline electrolysers are not able to provide separated gas streams. Many other alkaline electrolysers use a separator in each cell to keep the gases from mixing. These separators are generally non-ionic and result in a larger separation between the electrodes, there by adding a large resistance to each cell. This resistance causes the cells to suffer a large drop in efficiency. Ionic membranes such as the ITM AE membrane allow the gases to be kept separate and have a much reduced resistance due to the ionic properties of the material. The membranes are also able to be made very thin which allows the separation to be reduced between the electrodes. These factors allow cells with ITM ionic membranes to work at much higher efficiencies and current densities than other alkaline electrolyser systems.

Power supply
Detailed analysis was carried out calculating the cost of manufacture for stacks with different numbers of cells and cell sizes and combining the stack costs with the relevant Power Supply Units (PSUs). This analysis took into account the costs associated with using membranes of different types, different catalyst options, the cost of cells as a function of size and the effect of running stacks at different current density levels. It was found that the lowest cost stack and PSU combination for a system capable of producing 9lpm of hydrogen was a stack with six cells of 200cm² active area combined with a 12V 200A power supply.

Stack testing
The components for a six cell 200cm² petal stack were manufactured. For initial testing purposes the stack was built using acidic CE membranes so that the performance of the stack could be directly compared to the extensive test data that ITM has collected for the CE LAM stack. The stack was connected up to a test rig so that preliminary testing could be carried out as shown in Figure 6.

Figure 6 200cm² six cell Petal stack connected to a test rig

The 200cm² Petal stack was then connected to a SafeFlame prototype system shown in Figure 7. The stack was run using the SafeFlame control system to the maximum current of 200A, producing the required 9lpm of hydrogen gas and 4.5lpm of oxygen gas. The gases were piped to a burner and ignited.

Figure 7 Early SafeFlame prototype system with Petal stack attached.

AE membrane development
Traditional CE electrolysers contain platinum based catalysts which adds cost to the stack and which may not be suitable for use in the SafeFlame stack. AE electrolysers are able to use lower cost catalyst materials such as stainless steel and nickel which dramatically reduce the cost.

Anion Exchange (AE) membranes are composed of a polymeric backbone along which are attached pendant groups containing fixed cationic species. Mobile anionic counter-ions are associated with the fixed cations to ensure an electrically neutral system. The choice of polymeric backbone is based on amongst other things chemical stability, ease of synthesis, mechanical properties and cost. The ionic species for AE membranes are almost exclusively based on quaternary ammonium salts (quaternary phosphonium and tertiary sulfonium salts are also theoretically possible but have yet to achieve commercialisation).

Traditionally, membranes used for electrochemical applications have been based on hydrocarbon copolymers of styrene (or vinylpyridine) and divinylbenzene into which the fixed ionic charges can be introduced after polymerisation. Divinylbenzene acts as a chemical cross-linker, joining the polystyrene chains together to form a three dimensional polymer network. The styrene groups allow the incorporation of chloromethylene groups by a Friedel-Crafts alkylation reaction. This chloromethyl species allows the formation of a quaternary ammonium functionality which acts as the fixed cationic group. However this multi-step process is technically challenging, time consuming and uses highly toxic intermediates.

Simplification of the traditional method of producing poly(styrene-co-divinylbenzene) was undertaken using vinylbenzyltrimethylammonium chloride; a commercially available styrene based monomer which possesses an ionic quaternary amine group. Co-polymerisation with divinylbenzene results in a highly efficient single step process which eliminates the need for post-polymerisation chemical reactions and the need for highly toxic intermediates. The copolymer membrane is naturally hydrophilic due to the highly polar ionic species within vinylbenzyltrimethylammonium chloride. The membrane expansion and water uptake capability is governed by the cross linker content, which varies between 10 and 20%. If the cross-linker concentration is too low then the membrane resembles a gel, the mechanical strength is too low and the membrane easily breaks when handled. High cross-linker levels however result in a brittle polymer prone to fracturing when handled or subjected to differing compressive forces within the electrolyser stack. In order to further control the hydrophilic nature and mechanical properties of the membrane, styrene is added to the formulation.

Brittleness can in many polymers be reduced by the addition of plasticizers such as dioctylphthalate. These hydrophobic plasticizers tend to be incompatible with the hydrophilic solvents required to dissolve the vinylbenzyltrimethylammonium chloride ionomer. These plasticizers also leach out of the membrane over time. Polymer mechanical properties can be improved by grafting the growing polymer onto polybutadiene rubber; again this hydrophobic rubber is incompatible with the monomer formulation.

Dissolution of polysulfone or polyvinylalcohol into modified monomer formulations results in a heterogeneous mixture of the two polymers in the final membrane giving no improvement in mechanical properties. Composite reinforced membranes formed by curing the ionic membrane around a supporting mesh or nanoporous film proved technically challenging as the ionic membrane expands upon hydration. This results in severely distorted membranes or the membrane pulling itself away from the substrate completely. It was found that the detrimental expansion could be controlled by hydrating directly into concentrated alkaline solutions.

Membranes were made with nylon or polypropylene reinforcing meshes. The meshes used varied in thickness from 100μm to 300μm with an open area of ca. 50%. Similar thickness membranes show a significant loss in electrolyser performance for the mesh reinforced membrane due to the dilution of ionic content of the membrane by the mesh. These efficiency losses can be limited by making thinner membranes, however the production of large, thin and uniform membranes again proved technically challenging.

Mesh reinforced membranes showed significant degradation after 1000 hours of electrolyser operation with micro-cracking of the membrane corresponding to the apex of the mesh threads. These cracks allow gas to pass through the membrane resulting in cell and electrode corrosion as evidenced by the brown rust on the membrane surface.

A new cross-linker synthesized from 2,4,6-tris(dimethylaminomethyl)phenol and vinylbenzyl chloride provides a molecule capable of creating three cross-linking bonds. Substitution of divinylbenzene for this new ionic cross-linker produces membranes with reduced expansion and water uptake. Usually this reduced expansion would result in a reduction in electrolyser efficiency; however in the new ITM12 membrane efficiency is maintained due to the increased ionic content of the membrane. This reduced polymer expansion results in improved gas separation and the greater polymer density gives the membrane greater mechanical strength. The ionic nature of the cross-linking bonds imparts better strain properties (i.e. less brittle) than the covalently linked divinylbenzene membrane. The ITM12 AE membrane has been developed for use in the SafeFlame product and shows a significant improvement over previous ITM AE membranes. The membrane gives good electrolyser performance and has much improved mechanical properties, making it well suited for use in the SafeFlame Petal stack.

Surface mix combustor system design
The purpose of the combustor system is to burn gases, in this case oxygen and hydrogen, and to burn them safely allowing an appropriate transfer of heat to a work piece. SafeFlame uses an electrolyser to produce the gases instead of using bottle gas. The combustor is a handheld burner allowing an operator to easily move a flame around a working geometry, the handle is usually connected to two hoses which provide the appropriate supplies of gas.

The combustor has to perform a number of functions which are listed below:

• Receive two or more separate streams of gas.
• Either keep the streams separate or combine varying ratios of gases.
• Regulate the flow of gases.
• Isolate the user from possible chilling arising from the gas flow.
• Isolate the user from heat produced at the nozzle.
• Be lightweight, ergonomic and easy to use.
• Provide a gas at a flow rate suitable for the intended application.
• Be robust for manual day to day use and misuse.

The burner development which has taken place in SafeFlame has simplified and improved the safety of burning hydrogen for the end user. The mechanisms employed keep the hydrogen and oxygen gases separate, both at the point of production and while being transported to the burning point. The point at which gases can mix, ignite and burn is outside the burner itself. Because of this key fact (separate gas streams) safety has been improved because any risk of explosion has been eliminated.

It is well-known that oxy/acetylene, which is the hottest burning gas mixture available and has been traditionally used for many applications for which it could be considered to be unnecessarily hot and therefore could be inappropriate for many operations. Often the acetylene flame must be used strongly reducing ie cooler and kept moving in order to reduce the chance of melting the parent metal. Using oxy/acetylene equipment is a skilled operation with many considerations and risks; a good level of knowledge/training should be in place before attempting work.

SafeFlame burns hydrogen. Although hydrogen burns in pure oxygen with a high temperature, designs steps have been taken to keep the flame temperature as usable as possible, and suitable for brazing copper and aluminium.

A key aim of SafeFlame has been to increase brazing appeal to a wider market, therefore its operation needs to be simple, clean and effective. The temperature of the SafeFlame flame has been tailored to be the most effective temperature for brazing applications.

Flames produced by surface mixed burners are considerably richer than flames produced by pre-mixing, even though they have the same oxygen ratio. Surface mixed burners are capable of burner port loadings 15% higher than pre-mixed burners.

The initial trials had given good pointers as to effective and less effective ways to burn gases. Computer simulations also helped to further define the conditions where gas flows should focus. Computational Fluid Dynamics (CFD) was initially used to simulate concentric burner ports. A section view of a concentric burner port is shown in Figure 8. CFD was used to investigate different burner geometries and establish likely burner orifice ratios.

The inner tube internal diameter (ID) was 1.194mm the outer tube ID was 2.55mm.

Figure 8 Section view of a concentric burner port.

Figure 9 Simulation (single hole burner).

Torch Design 1
An example of a single port burner is given in Figure 10. Two concentric holes can be seen, hydrogen flows from the central hole and oxygen through the annulus. The flame from the burner can be seen in Figure 11, it had a long thin aspect. Heating of copper tube or pipe was very effective, providing the pipe was in the narrower range 6 to 12mm diameter but it could be used to braze pipe up to 22mm diameter. The flame provided an appropriate heat source, but it was suggested that if it was wider (shorter and fatter) then heating larger diameter pipes would be made easier. A short wide flame would result in less movement for an operator, which in turn reduced possible heat losses as the flame hits free space rather than the work piece; it also increased the flame’s ability to wrap around larger pipes. These important factors gave the operator the opportunity to hold the burner reasonably still.

Figure 10 Single port burner (mk1).

Figure 11 Oxyhydrogen flame from the single port burner (mk1).

The burner shown in Figure 11 was initially used with bottle gas and produced a laminar flame when the total gas flow was below a total of 10.5 litres per minute. The production of a brazed joint is shown in Figure 12, the flame is clear blue as it exits the burner and fluxes and braze metals turn it green/blue. Two (mk1) burners working in parallel were seen to speed up brazing of larger pipe sizes, however there was still mixed success above 22mm diameter.

Figure 12 Brazed joint with single port burner (mk1).

During later simulations of Design 1, it was found that changing the dimensions from:
• Core diameter 1.194 mm.
• Annulus inner diameter 1.65 mm.
• Annulus outer diameter 2.55 mm.

• Core diameter 1.372 mm.
• Annulus inner diameter 1.829 mm.
• Annulus outer diameter 3.0 mm.
gave an improvement to the flow when used at a higher flow rates of 9.0 litres per minute of hydrogen whereas the previous values were very good with flow rates of 4.5 6.0 and 7.0 litres per minute of hydrogen.

Torch Design 2
In Design 2 the number of ports was increased to three and spread linearly across the burner. A CAD drawing of the burner is shown in Figure 13 and a close-up of the burner ports is shown in Figure 14. The burner ports had become slightly dislodged over time, which is why they are no longer perfectly concentric. The small diameter inner burner tubes have bent because they are unsupported and of narrow wall thickness, square tubes were proposed earlier-on to overcome this issue. The corners of square tube self-locate in the holes but allow gas to pass around the gaps. However it was measured that not enough gas flowed around the square tubes for this application.

Figure 13 Torch design 2 CAD.

Figure 14 Torch Design 2 front face

The broad flame was capable of brazing pipes of up to 28.5mm and reduced the time to braze smaller pipes when compared to the single burner. As a further addition to the burner handle, a switch was added to control gas production at the electrolyser which increased or decreased flame intensity via the handle. Interactive demonstrations with welding and brazing engineers demonstrated that the flame control functionality was greatly appreciated by all who used the burner.

Thermal imaging was undertaken of the Design 2 oxy/hydrogen flame; the results of these trials are shown in Figure 15 and Figure 16. The hydrogen flame had a low emissivity and the images show a low temperature reading because of the emissivity. Figure 15 shows three separate flames exiting the burners and the peak temperature recorded is 335°C. Figure 16 shows a view across the three burner ports, a peak temperature of 288°C is recorded. There is also a significant swirl effect generated which had not previously been recognised.

Figure 15 Infrared image (Design 2) burning gases from three ports.

Figure 16 Infrared image (Design2) side view of three ports.

Design 3
Design 3 shown in Figure 17 used a tri port burner with a 120° port arrangement. The flame produced was less effective at heating than from a linear port arrangement, however the flame had a longer throw. The design was not taken further, but did result in the swan neck design shown in Figure 18 which was carried forward to the final designs.

Figure 17 Tri port burner CAD.

Figure 18 Tri port burner used in demonstration.

Design 4
The three port linear burner provided a wide flame which was suitable for heating a wide range of pipes. The largest size of refrigeration grade copper pipe which it could effectively heat at 9 litres of hydrogen per minute was found to be 28mm although 32mm could also be achieved with practice. Refrigeration grade material has a thicker wall than the more common water pipes used for homes and commercial buildings. The three port linear burner was also effective at heating small pipes or tubes.

Figure 19 shows the orientation and size of the burner ports as a CAD model. Figure 20 shows the actual burner head.

Figure 19 Handle design 4, 3 port linear.

Figure 20 3 port linear.

Design 5
Brazing aluminium is technically challenging because the melting point of aluminium is usually close to that of the filler, therefore it is easy to melt the aluminium, resulting in a large hole (worst case) or deformation of the parent metal. A second design of single port burner (mk 2) shown in Figure 21 and Figure 22 was undertaken to produce a laminar flame at 9 litres hydrogen per minute. The flame from this burner was very soft at 9 litres and quietly wrapped around the aluminium tubes. The gentle flame proved to be good at heating aluminium, offering advanced warning of the impending melting point by the yellowing of the flame, offering an increased visual guide to the operator. The operator also benefited from an ability to use the flame at a greater standoff distance as the laminar flame has an increased length of operation.

Figure 21 Single port burner (mk 2).

The larger port version of the single port burner was specifically designed for a hydrogen flow rate of 9 litres per minute the following dimensions were used: core diameter 1.372mm annulus inner diameter 1.829mm annulus outer diameter 3.0mm. The diameter of the burner nozzle at the burner port was 10mm.

Figure 22 Single port burner (mk 2).

Figure 23 shows a cutting operation with a mk1 single port burner using 9 litres hydrogen and 4.5 litres oxygen per minute on 3mm thick steel. The angle of the cut is important as this helps push the melt pool along and through the metal.

Figure 23 Cutting operation with the single port burner (mk1).

Gas panels
Gas panels as shown in Figure 24 have been used to verify results from other gas processes, whether outputs from SafeFlame or direct cylinder usage. The gas panels have been used to construct stability diagrams for oxyhydrogen, oxyacetylene and oxypropane, in conjunction with various burners.

Stability diagrams are used to show the effective ‘sweet spot’ (gas flow), of a fuel gas and oxygen, at a range of flow rates, for a given size of burner port and ancillary equipment. Examples are given below to explain how the diagrams were created and their meaning. Diagrams have been produced for oxy/acetylene, oxy/propane and oxy/hydrogen burners. Stability diagrams are important because they demonstrate the correct operational parameters or more specifically the operational envelope of a piece of equipment for a known type of fuel gas.

Figure 24 Gas panels.

Stability diagrams
Each fuel has its own unique values, which determine its usable flame range for a given orifice size. It will be possible to deduce from the data provided that hydrogen using a surface mix burner has a very wide operational range, in fact, so wide that within the context of equipment delivery parameters there were no points of failure. This is in comparison to propane for example, where a pre-mixed torch gives the flame a usable performance over a limited range of gas flows. A new operator using a hydrogen system would be able to use any flame they were able to obtain, however an oxy/propane flame requires careful setup. The operator may not be able to establish correct parameters.

The stability diagrams for oxy/acetylene use three conditions of flame known as: yellow flame, yellow tipping and blue flame. The yellow flame can be seen in Figure 25 the flame is bright yellow, it is intense to look at and full of unburnt carbon which is why it is yellow, as there is not enough oxygen in the gas mix to achieve complete combustion. Figure 26 shows the condition where the oxygen has been increased. The yellowness of the flame is now contained to just the inner cone, ideally it should be just at the tip of the inner cone, hence the term yellow tipping. The flame in this condition is the first stage of a usable flame, it provides a reducing atmosphere and has a lower temperature than the maximum 3600°C. In the third scenario, the oxygen has been increased again and the inner cone is smaller and hotter with no yellow component. The flame in this condition is said to be neutral as shown in Figure 27. Increasing the oxygen content still further (not shown) produces a lilac/purple colour to the inner cone, this is said to be oxidising, it did not form part of the study.

Figure 25 Oxy acetylene yellow flame.

Figure 26 Oxy acetylene blue flame with inner cone transitioning from yellow to blue.

Figure 27 Oxy acetylene blue flame (no yellow component).

The chart in Figure 28 shows the start of a stability diagram. The x axis shows the amount of acetylene gas consumed per minute in litres, while the y axis shows the amount of oxygen provided by the gas delivery system (but does not take into consideration any oxygen consumed from free space). The reading started at 0.25 litres per minute because this was the minimum recordable delivery of the system. The chart contains two traces; the lower line represents the minimum useable flame, ie at yellow tipping: The flame would not be used with a lower level of acetylene for the given oxygen supply. The upper trace gives the points at which the gas flow through the nozzle forces the flame to ‘lift-off’ which means it is extinguished. Above the upper trace a flame cannot exist for that nozzle and gas combination, therefore the flame does not exist to the right hand side of the chart either. The upper and lower limits of the charts are shown in Figure 29 which produce a boundary; within the boundary the flame can exist, outside the boundary the flame cannot exist. However this statement is not strictly true because the left hand side (lower gas limit) is determined by equipment measurement capabilities. A flame used at the low gas flow rates would be of limited use being so small, reducing the gas flow levels further, and at some point the low gas flow would cause the flame to burn back in to the torch, causing a pop as the flame extinguished itself in a flashback.

Figure 28 Oxy/acetylene, Tip 1 - 0.6mm.

Figure 29 Oxy/acetylene, Tip 1 - 0.6mm with upper and lower limits.

Figure 30 builds more information into the graph by showing a likely upper limit of operation of a ratio of 2.5:1 oxygen to acetylene. The rate of flow shown in the curve is not likely to be oxidising but higher rates of oxygen would start to produce an oxidising flame. The region bounded by the ratios 2.5:1 and 1.2:1 are the most usable and keeping the acetylene to below 0.75l/minute increases the usability of the flame. The lift-off trace shows a decrease after 0.75l/minute; the flame became increasingly unstable above this point.

Figure 30 Oxy/acetylene, Tip 1 - 0.6mm with the addition of upper gas ratio.

Figure 31 displays a graph for the oxy/acetylene burner using Tip 2 which has a port diameter size of 0.7mm or 0.1mm larger than the previous example. The maximum flow rate was increased to just over 1.6l/minute from just under 0.9l/minute previously. The slightly increased port size had a large effect on the flame stability at lower flow rates. Whilst the area under the graph increased, the useful components of the flame are bounded by the ratio 2.5:1 and 1.2:1 lines. Because the port size increased the flame was more usable over a greater range of flow rates.

Figure 31 Oxy/acetylene, Tip 2 - port size 0.7mm diameter.

Figure 32 gives the experimental findings for the Tip 3 burner. The maximum flow rate has effectively been doubled from Tip 2, however the 2.5:1 ratio rapidly approaches lift off at 2.5l/minute acetylene. This could effectively be taken to define the maximum flow that Tip 3 would use, however in practice it is likely to be lower. Tip 3 gave a good range of stability and made a useful contribution to the burner torch tool kit.

Figure 32 Oxy/acetylene, Tip 3 - port size 0.7mm diameter.

Figure 33 shows the results for oxy/acetylene Tip 5. The port diameter was 1.2mm the largest of those used for the oxy/acetylene trials. Tip 5 could use a wide range of flow rates; it was necessary to change the flow meters during the experiment because the flow rates exceeded the equipment capabilities for one set of meters. Therefore there are two traces for ‘lift off’ which, while overlapping, do not have the same values, however they are within experimental error. The maximum flow of acetylene rose to 5l/min and the lift off minimum limit rose to 3l/minute of acetylene.

Figure 33 Oxy/acetylene, Tip 5 - port size 1.2mm diameter.

A similar set of experiments was undertaken for oxypropane. Again, different sizes of burner ports were used. One example is given in Figure 34, which shows the graph having a reasonable region were the gases could be effectively used. Increasing the gas flow in conjunction with Tip 6 collapsed the usable region, meaning the gas flow had to be critically controlled.

By comparison, the stability diagram for oxyhydrogen is given in Figure 35, almost the entire area of the graph represents acceptable gas usage.

Figure 34 Oxy/propane, Tip 5 – 1.9mm port diameter.

Figure 35 Oxy/hydrogen stability diagram.

During trials it was possible to establish that the amount of hydrogen could be reduced to a level approaching zero without any deleterious effects. The flame did not extinguish or flashback. The lower trace along the x axis represented burning pure hydrogen. This is not problematic for the surface mix burner, but produces a lower temperature flame. The flame is slightly more difficult to see and because there is no oxygen flow, the flame is soft and rises quickly as the hydrogen disipates in the air.

The stoichiometric line of Figure 35 shows the region of highest processing temperature, the area below this is said to be ‘gas rich’ or reducing and the flame temperature will reduce progressively with falling oxygen input. Increasing the oxygen above the stoichiometric ratio, the flame temperature starts to fall. As more oxygen is used, the flame length reduces, as the fuel is consumed faster. Because the flame velocity is increasing there may be additional benefit to the heating process, however once the flame has become small, the chilling effect of the oxygen reduces the flame’s capacity to keep a work piece hot. This was particularly relevant to cutting operations where extra oxygen could be employed to clear the melt pool and associated slag. The area below the stoichiometric trace or the ‘gas rich’ region was found to be of particular interest. Reducing the oxygen content produced a different range of gas species, which would impart their energy of formation to the work piece, resulting in an effective heating process in conjunction with a protective atmosphere.

Hydrogen embrittlement
Hydrogen embrittlement of materials can result in a material which will fail prematurely. It was necessary to investigate whether the hydrogen contained within an impingent flame heating process using both pure hydrogen and oxy/hydrogen, surrounded by an air atmosphere, could result in hydrogen absorption.

The table below gives the operating parameters for Hydrogen testing undertaken. The equipment used consisted of: oxy/hydrogen gas metering panel, the gas being delivered via the larger single port burner. The gas provided from gas bottles was dry and had the flow rates given in Table 1 below. The samples were all of the same low carbon steel, the grade being specifically used for hydrogen analysis; each sample was 25x15x15mm. A Bruker G4 Phoenix gas analyser was used to define the amount of hydrogen captured in the samples.The processing temperature was 400°C.

Table 1 Hydrogen absorption

The hydrogen levels found constitute no risk to the operator or to the joint in operation. Should a time lag of greater than one minute exist before a service condition is met there is no risk from effects from hydrogen embrittlement.

Mechanical testing
Mechanical testing in the form of destructive tensile tests was undertaken using an Instron tensile testing machine. The samples which had been previously brazed together were pulled apart, the main mode of failure was via failure in the heat-affected-zone of the parent material. This indicates the joint strength to be higher than the heated material near the joint.

Figure 36 shows the tensile strength of 5/8” copper tube brazed with acetylene, hydrogen and propane samples. All the samples failed in the parent metal, none of the brazed joints failed. There are three areas shown corresponding to acetylene, hydrogen and propane (left to right), the results for acetylene and propane are very similar.

It had been proposed that the hydrogen samples failed by a different mechanism and therefore the results are generally slightly lower. The hydrogen sample which failed correctly (parent metal failure denoted by a circle and arrowed label) produced a result equal to the acetylene and propane. The other samples are likely to have failed because of longer heating times, and corresponding annealing of the parent metal.

Figure 36 Maximum tensile strength of 5/8” copper tube brazed with acetylene, hydrogen and propane.

In further trials on larger diameter tubes, which are shown in Figure 37, the results were consistent, and oxyhydrogen performed equally well.

Figure 37 Maximum tensile strength of 7/8” copper tube brazed with acetylene, hydrogen and propane.

Joint geometry testing
Mechanical testing has been carried out to investigate how joint strength is affected by reducing the overlap of the expanded joint. The effective amount of joint expansion was made using a hand activated expanding tool, the length of tube placed over the expanding tool was reduced for each subsequent set of three tube joints time. The tubes used 3/4” copper pipe (refrigeration grade).

The sizes of the overlapped part of the joint produced were:
• 14mm.
• 10mm.
• 7mm.
• 3.5mm.

Figure 38 Joint strength results plot of the geometry test.

Figure 38 shows the results an experiment to evaluate decreasing the overlapped length. Ignoring outlying points the result are broadly similar, which gave confidence that the overlap distance could be significantly reduced. Reducing the overlap results in a significant reduction in braze alloy consumption and therefore represents a major saving to the operator.

Influence of moisture on flame
Many simulations have been undertaken for the project, one of which considers the effect of moisture in the flame. The influence of moisture in the inflowing hydrogen and oxygen streams was studied, through simulation of the single port and three port linear burner, with hydrogen flow rate of 9 l/m and oxygen flow rate of 4.5 l/m. The inflows were measured at ITM (SafeFlame Mk3) to have a humidity of 100% at temperature 25°C. For the linear burner, two alternatives were considered: a case with 9 l/m H2 (4.5 l/m O2) plus the humidity (H2+H2O flow rate 9.29 l/m, O2+H2O flow rate 4.65 l/m) and a case where 9 l/m included H2 and water vapour and, similarly, 4.5 l/m included O2 and water vapour. For the three port burner only the latter case was considered based on the results from the single port burner simulations. At 100 % relative humidity at 25°C the mass fraction of H2O in hydrogen flow is 0.22 and that in the oxygen flow is 0.018.

Figure 39 to Figure 50 compares the simulation results for the dry gases and the two moist gas cases for the single port burner. The simulations indicate that moisture has only a small influence on the predicted flame length and shape.

Figure 39 Dry gases H2 9 l/m.

Figure 40 Moist gases H2+H2O 9 l/m.

Figure 41 Moist gases H2+H2O 9.29 l/m.

Figure 39, Figure 40 and Figure 41 show the computed isosurface temperature of 2163°C for the single port burner and compare of the dry and moist gas conditions.

Figure 42 Dry gases H2 9 l/m.

Figure 43 Moist gases H2+H2O 9 l/m.

Figure 44 Moist gases H2+H2O 9.29 l/m.

Figure 42, Figure 43 and Figure 44 show computed isosurface of OH mole fraction 0.017 for the single port burner, giving a visual comparison of the dry and moist gas conditions.

Figure 45 Dry gases, hydrogen 9 l/m.

Figure 46 Moist gases hydrogen + water 9 l/m.

Figure 47 Moist gases hydrogen + water 9.29 l/m.

Figure 45, Figure 46 and Figure 47 show the computed temperature field (°C) for the single port burner. Cross sections are shown through the centre of the flame and the comparison of dry and moist gas cases.

Figure 48 Dry gases H2 9 l/m.

Figure 49 Moist gases H2+H2O 9 l/m.

Figure 50 Moist gases H2+H2O 9.29 l/m.

Figure 48, Figure 49 and Figure 50 show the computed mole fraction of hydrogen for the single port burner as a cross section view through the centre of the flame. Visual comparison of dry and moist gas cases can be seen. The scale is logarithmic and the flame length is given between 0 mm to 600 mm.

Potential Impact:
Socio-economic impact and the wider societal implications and dissemination activities
• Benefits to users:
o Ease of operation.
o High visibility of work piece, less obscuring by flame.
o Wider adoption of flame processing technology.
• Benefits to the public:
o Reduced risk of property damage arising from industry.
o Less transport
o Educational - schools and colleges high temperature flames.
o Increased potential work force for flame processing, plumbing, air-con etc.
• Industry benefits:
o Reduced transportation of gas cylinders.
o Could be used on sites where cylinders are banned.
o Greater access to a wider range of processing conditions.
o Easier to transport relative to larger cylinders.
o Easier to process aluminium over other gases, use of aluminium is increasing.
o Can be run directly from renewable energies.
o Lower running cost when compared to hire and delivery of cylinders.
o SMEs do not need to hold stock of cylinders, or make unnecessary travel to replace empties.
o Removes the need to store cylinders in vehicles which leads to several explosions a year.

List of Websites:
Project website details
The SafeFlame website went live on 21 December 2011 using
and is redirected to the .eu address.

For General and System queries contact Steve Kowalski at EABS, Tel: +44 1625 431 160