This article was originally published on February 5, 2015.
Troubleshooting furnace heat-treatment quality issues, such as oxidation and discoloration, can sometimes be challenging.At times, normal, conventional troubleshooting steps and techniques may not always work. This is especially true when precise atmospheric-contaminant analysis is required. This can vary based on the furnace atmosphere, the type of furnace and the metallurgical properties of the materials being heat treated.
This article seeks to inform the reader about the problems that can occur while heat treating in atmosphere furnaces and the analytical equipment necessary to diagnose these problems effectively and efficiently to avoid expensive rebuilds and lost productivity.
Introduction and Background
At one of the world’s leading manufacturers of fine stainless steel cutlery, Dexter-Russell Inc. of Southbridge, Mass., this very challenge was faced during the hardening process of stainless steel cutlery blades. The stainless steel blades were heat treated in a Lindberg® continuous belt furnace at a hot-zone temperature of approximately 1900°F (1038°C) in a H2/N2-based atmosphere. Dexter-Russell (manufacturer) had converted their furnaces to the H2/N2-based atmosphere in the mid-1990s, switching from dissociated ammonia (NH3), primarily for environmental and quality reasons.
The manufacturer attempted to diagnose this discoloration/oxidation issue using tried techniques that had worked in the past: leak testing all of the internal gas piping, muffle retort pressurization, copper strip test and smoke infiltration. Unfortunately, these techniques proved unsuccessful in properly identifying the root cause. The result was lost productivity and increased costs until the problem could be identified and solved.
If we recall from physics and the Ellingham Free Energy oxidation/reduction diagrams, specifically the H2/H2O ratio for chromium is present in varying high percentages in all stainless steel. (Note: This equilibrium ratio needs to be about 500 to 1.)
In other words, it does not take much moisture or oxygen, which is converted to H2O in hydrogen-based atmospheres, to produce oxidation and discoloration of parts. Depending on the temperature at which these oxides form, they can also be quite difficult to subsequently reduce and eradicate.
As a result of lost productivity and frustration with this ongoing discoloration issue, Dexter-Russell approached their hydrogen/nitrogen gas supplier, Air Liquide Industrial U.S. LP and their ALTEC team of specialists in heat-treatment applications. Understanding this issue might require more sophisticated atmosphere analytical equipment and expertise. The aim was to combine their own problem-solving experiences with Air Liquide’s troubleshooting and together be able to effectively diagnose the cause of the problem.
The first step was choosing correct analyzers for the job. Because the furnace was using a hydrogen-based atmosphere for bright annealing, with no hydrocarbon enrichment additives or ammonia gases, dew point and ppm trace oxygen seemed to be the logical choice for atmosphere analysis. Further, since it is important to first eliminate the incoming gases as a possible source of contamination, it would be best to use a low dew point (accurate below -100oF) and a low ppm oxygen sensor (0-5,000 ppm trace oxygen) for these “virgin” gases.
Air Liquide chose a ceramic sensor hygrometer (-148oF to +68oF) and an electrolytic-based trace ppm (parts per million) oxygen analyzer with a non-depleting electrolytic cell for the task. The versatility of these instruments, which also provide quick response times when monitoring in-situ furnace-atmosphere contaminants, made them a good choice.
Step two is to confirm whether the industrial gases (H2/N2) were or were not a contributor to the problem by analyzing the incoming gases before they enter the furnace as a protective atmosphere. Ideally, it is best to check the incoming gases as close to the furnace entry point as possible and work back to the gas supply source should you encounter suspiciously high readings. Since Dexter-Russell had already checked all of their house gas lines for leaks, this second step was basically done to verify that the H2/N2 source gases themselves were at acceptable contaminant levels.
The key to any analysis, whether on incoming gas lines or sampling directly from the furnace, is to allow sufficient time for purging the analyzers and then sufficient time for the readings to stabilize. Impatience with this step can often lead to false high readings. Therefore, we allowed approximately eight hours of sample time on the incoming mixed gases to ensure stability with the final readings.
Once the incoming H2/N2 mixed-gas purity was validated, step three was to take sample readings within the Lindberg muffle (retort) itself. Preferably, this is done starting at the beginning of the hot zone of the furnace, which is where the most damaging oxidation and discoloration can occur. This procedure was done using these same two analyzers, with the addition of a small sampling pump for positive displacement sample extraction at 2 SCFH flow rate. A 99.5-micron particulate filter was used to protect the analyzer cells from particulate contamination. This allowed us to effectively test the atmosphere quality for both dew point and ppm oxygen in both the beginning and middle of the hot zones (1900°F). Based on the specific type of oxidant on the surface of the stainless blades, it appeared that the oxidation was primarily originating from hot-zone oxidation.
Once stable readings with negligible fluctuations were achieved, the results for the incoming mixed gases were -90°F dew point and 0.5 ppm oxygen. Both gases appeared to be well within acceptable limits for industrial-grade nitrogen and hydrogen for both dew-point and oxygen content.
Having eliminated the incoming gas as the potential cause of the oxidation, we then focused our efforts on the Lindberg furnace itself. Again, since Dexter-Russell had already done much of the preliminary testing of the gas lines and furnace-muffle integrity, we did not expect to encounter any mechanical defects, muffle defects or water-jacket leaks on the furnace. The only thing left was to sample the contaminant levels of moisture and oxygen within the furnace to see if there was something else going on that could not be detected by the methods already employed (muffle pressure test/copper oxidation test).
Our initial findings, without changing operating parameters such as temperature, belt speed or atmosphere (H2/N2) flow rates into the furnace, yielded telling results. The dew point in the furnace hot zone was +19°F, while the oxygen levels stabilized at about 3,500 ppm. From prior experience and database records previously established on this furnace, we immediately knew that these contaminant levels were considerably higher than typically expected. We also knew that the stainless knife parts will oxidize at 1900°F under these high-contaminant levels.
With these higher-than-expected contaminant readings, the next logical step was to examine the actual furnace operating parameters. The easiest parameters to evaluate first were the overall furnace gas-flow rates. Based on the Waukee tube and float flow meters, the setpoints used for this test were the same setpoints used prior to the furnace being taken down for maintenance and the hot-zone rebuild. During the rebuild, some of the gas regulators had been upgraded and replaced. This may have resulted in a slightly different delivery pressure to the flow meters, which in turn could alter the actual flow rates based on the calibration pressure established for any given flow meter.
The main purposes of the H2/N2 atmosphere are first and foremost to prevent air (oxygen and moisture) ingress into the furnace muffle (retort). Typically, gas flow rates achieving a few inches water column will suffice. Second, the hydrogen in the atmosphere provides a “getter” reducing gas capability, which should reduce the oxides on the stainless surface and provide a bright finish. The intent is for hydrogen to combine and reduce oxygen levels by forming water vapor. It then becomes a matter of balancing the H2/H2O ratio to achieve “bright” parts.
With this in mind, we next wanted to see the effects of increasing the total mixed-gas flow rate, not the percentages of the gases. After some initial upward flow adjustments, we found that the atmosphere contaminants in the hot zone started to rapidly decrease as we increased the mixed-gas flow. Within an hour after our final flow-meter adjustments, we achieved a dew point of -40°F and an oxygen level of 10 ppm oxygen. Given this dramatic improvement, the 314SS belt started to brighten up almost instantly. We now felt confident that we could achieve similar results on the production blades with these adjusted flow rates. Approximately two hours later, Dexter-Russell was once again producing bright, non-oxidized, annealed stainless steel blades.
Sound engineering and experience will typically go a long way when diagnosing and solving manufacturing issues, whether it is furnace oxidation or some other challenge. In most cases, this process experience and product expertise will allow the customer to intuitively identify the problem and resolve the issue quickly.
However, there are still those cases where an extra set of “experienced eyes” are needed along with the proper analysis equipment. In this case, fast-responding dew-point and oxygen analyzers can sometimes provide the missing piece to the diagnostic puzzle. Fortunately, after carefully eliminating many of the potential variables that can cause or contribute to furnace oxidation, we were able to effectively diagnose the problem and provide the resulting solution.
The author would also like to acknowledge the valuable contribution of Buck Raper, manager of manufacturing and engineering, who, along with Dick Desaulnier, helped bring this evaluation to a successful conclusion.
For more information: Richard F. Speaker, senior business development specialist, Automotive & Fabrication Group; Metal Heat Treating Applications; Air Liquide Industrial U.S. LP Houston, Texas; tel: 800-820-2522; fax: 713-803-7322; e-mail: firstname.lastname@example.org; website: www.us.airliquide.com. Dexter-Russell, Inc. is the largest manufacturer of professional cutlery in the U.S. The company is the proud successor to the two oldest American cutlery manufacturers: The Harrington Cutlery Company and the John Russell Cutlery Company. They maintain a tradition of excellence in both materials and workmanship.
Utilizing Flue-Gas Composition Measurements to Diagnose 10% Combustion-Air Deficiency Aluminum Reverb Melt Furnace with Regenerative Burners
As a follow up to a prior Industrial Heating article (October 2014), one additional melt furnace example is considered. Flue-gas compositions (%CO, %CO2 and %O2) were measured from an aluminum reverb melt furnace operating with two pairs of regenerative burners. This furnace was direct-charged with the burners firing directly into the main hearth toward the scrap pile. All scrap was charged directly into the main hearth. During flue-gas measurements, only clean scrap was charged (no coatings or paint on the scrap).
Flue-gas measurements were made continuously throughout several batch melting cycles. The flue-gas sample probe was located before the dilution air break in the main exhaust stack so that the gas sample was representative of the combustion atmosphere inside the furnace.
Measured flue-gas compositions were 3% CO, 0% O2 and 10% CO2. These values held very consistently throughout the entire melt cycle and were consistent for several heats. During melting, the burners were firing with maximum natural gas firing rate and maximum combustion air flow.
The indicated air/gas ratio was in excess of 10:1 based on flow-meter readings, which would imply oxidizing conditions (excess O2, zero CO). The flue-gas measurements indicated significantly reducing conditions, however, showing that there was not enough combustion air to fully combust the fuel.
Figure A is a V-curve showing calculated flue-gas concentrations for air/gas combustion, assuming CH4 for natural gas. The measured 3% CO and 10% CO2 concentrations correspond to lambda = 0.90 on this V-curve. So, it was concluded that this regenerative burner system was operating at 10% reducing conditions (lambda = 0.90), meaning that there was a 10% deficiency in combustion air supplied.
For this furnace, since only 90% of total combustion-air requirements are being supplied, only 90% of combustion energy is being released inside the furnace. The remaining 10% of fuel input energy burns downstream in the flue exhaust duct after mixing with entrained dilution air.
For this particular furnace, a 10% melt rate increase is desired. O2 enrichment has been proposed to provide the required additional combustion O2 to combust all fuel inside the furnace (bring lambda up to 1.0). An O2-enrichment test is planned, during which the flue-gas measurements will be repeated to accurately establish the amount of O2 required to bring the furnace atmosphere up to neutral conditions and capture 100% of input fuel energy inside the furnace.
Fig. A. Measured and calculated flue-gas compositions from aluminum reverb melter fired with regenerative burners