EH-19 Welding Blankets: The Unexpected Hazards ENVIRONMENT, SAFETY & HEALTH BULLETIN Assistant Secretary for U.S. Department of Energy Environment, Safety, & Health Washington, D.C. 20585 DOE/EH-0039 Issue No. 19 March 1987 ------------------------------------------------------------------------------ Welding Blankets: The Unexpected Hazards Some blankets lose too much overall weight to be considered heat resistant and some release copious amounts of toxic fumes. ------------------------------------------------------------------------------ Four workers were overcome by fumes from the welding blanket they were using during a routine flame-cutting operation. The blanket had been placed around an insulation pipe to protect it from the welder's torch as he worked nearby but when the flames struck the blanket as he worked, it released a cloud of fumes. Suddenly and without warning, he and the three other workmen in the room were disabled. Although three of them recovered quickly and resumed work the same day, one was hospitalized. The accident prompted Lawrence Livermore National Laboratory to test more than 50 different non-asbestos welding blankets to learn how well they resist heat and what amounts of toxic products they generate at temperatures reached during routine cutting and welding operations. The first test, on a remnant of the same blanket that caused the accident, revealed that copious amounts of hydrochloric acid (HCl), cyanide (HCN), nitrogen oxide (NOx), and carbon monoxide (CO) were formed at a temperature as low as 500 degrees C. By the time the blanket reached 700 degrees C, which is several thousand degrees below the temperature of a cutting torch flame, 40% of its weight had been lost. The speed with which the workers were overcome suggests that the toxic agents were quick-acting and disabling, such as cyanide or carbon monoxide. Blankets made of silica cloth are extremely resistant to high temperatures, produce minimal amounts of toxic products, and, therefore, probably are the best substitute for the previously used asbestos blankets. At Least 30 Different Types of Blankets Are Sold The researchers, Anne Lipska-Quinn and Stephen J. Priante, found that 31 different types of blankets were being sold in their part of northwestern California alone. Although universally sold as "welding blankets," most do not have accompanying information regarding their limitations, even though they should be expected to suffer from intermittent periods of exposure to cutting torch flames. Most are made of fiberglass or silica materials ranging from felt-like composites and coated laminates to tightly woven fabrics. The fiberglass materials are often coated with silicon rubber, hypalon, neoprene, or acrylics; the silica-based materials are often covered with proprietary coatings. Various manufacturers reported that both silica and fiberglass cloths are washed in acid (which acts as an organic binder) after they are woven. Although most of the acid is then removed with water, residual hydrochloric acid often remains. Some manufacturers heat the blankets to a high temperature to remove the excess binder prior to distribution. If the excess binder is not removed, both HCN and NOx will be formed when the fabric is subsequently subjected to high temperatures. Some Lose Too Much Overall Weight to be Considered Heat Resistant First, Lipska-Quinn and Priante tested their 31 samples to determine at what temperatures degradation began, the rates of weight loss, and the total weight losses. They used a thermal gravimetric analyzer and conducted the tests from ambient temperature to 900 degrees C at a heating rate of 20 degrees C/min. They found that initial degradation of samples varied between temperatures of 40 degrees and 689 degrees C. Total weight losses also varied considerably among the samples, however the silica fabrics showed little weight change after an initial weight reduction (probably due to water desorption) at 100 degrees C. Some Release Substantial Amounts of Toxic Fumes After making their thermogravimetric tests, Lipska-Quinn and Priante chose a number of the best performers to analyze for toxic gasses. In the list above, they eliminated the first eight samples because those samples degraded at low temperatures, they lost too much overall weight to be considered heat resistant, and they were coated with either neoprene or hypalon which evolve large amounts of HCl upon degradation. Sample #21 was eliminated because it, too, was coated with neoprene. They chose to test Termonol cloth because of its relative resistance to heat; and Siltemp, Silica fabric #2-3700, and Thermoglass cloth because of their low overall weight losses. Each sample was heated to 500 degrees C and its smoke drawn with a hand pump into a Draeger tube. Two grams of each sample were tested for each individual gas - CO, CO2, HCN, HCl, and NOx - using a gas chromatograph and a mass spectrometer. Concentrations could be determined by the degree and extent of the color changes in the sensing material. Thermogravimetric analysis of 31 non-asbestos welding blankets ---------------------------------------------------------------------------- Temperature leading to Total weight initial loss weight loss Sample identification (degrees C) (%) ---------------------------------------------------------------------------- 1. Duck heavy-weight hard finish 311 94.30 CFM approved 2. Weld Tex vinyl-coated fabric 276 97.60 3. Fro-Prene nylon fabric coated with 266 84.00 Hypalon 4. Armor-Pleated heavy-duty Hypalon fabric 285 92.07 5. Fro-Lon #1 fiberglass cloth with 277 30.18 Hypalon coating 6. Fro-Lon #2 fiberglass cloth with 220 19.06 Neoprene coating 7. Fro-Sil fiberglass cloth with rubber coating 220 7.96 8. Jaxcolite glass cloth 289 40.36 9. Thermonol cloth 569 71.26 10. Weld flex 205 41.07 11. Weld-O-Glass #2300 (fiberglass) 289 44.60 12. Fro-Flex special fiber blend 291 43.20 13. Fromelt, silica fabric 328 19.64 (9.81)(a) 14. Siltemp. silica fabric 182 15.33 (9.92)(a) 15. Silica fabric #2-3700 182 18.29 (9.89)(a) 16. Thermoglass cloth 40 5.25 17. LLNL Stores stock welding blanket, 307 3.46 Sample #1 18. LLNL Stores stock welding blanket, 218 17.95 Sample #2 19. LLNL Stores stock welding blanket, 207 10.38 Sample #3 20. Weld-O-Glass #2-1900 (same as Silica 182 18.29 #2-3700) (9.89)(a) 21. Weld-O-Glass #3000R 452 13.22 22. Weld-O-Glass #2000G 274 21.75 23. Nor-Fab material 689 61.05 24. Thermo-Sil Style #1550 299 22.61 25. West Mac Co. #HR100-96 324 22.03 26. Weld-O-Glass #900FC 346 2.85 27. Weld-O-Glass #1000FC 380 2.54 28. Weld-O-Glass #2650 381 0.82 29. Weld-O-Glass #1850 503 0.63 30. Zetex 1100 245 3.42 31. Zetex 1200 313 2.39 ---------------------------------------------------------------------------- (a)Percent moisture loss. Silica Cloth Found to be Best Although they caution that care must be taken in predicting the specific performance of any given blanket based on their small-scale laboratory tests, Lipska-Quinn and Priante have concluded that unless welding blankets are composed of silica cloth, they will lose a considerable amount of weight when exposed to temperatures as low as 300 degrees C and will form substantial amounts of toxic products. Blankets made of silica cloth are extremely resistant to high temperatures, produce minimal amounts of toxic products, and, therefore, are probably the best substitutes for the previously used asbestos blankets. Suggested Actions Based on the findings of the researchers, LLNL changed its purchasing policy and now buys only those welding blankets that demonstrate acceptable levels of out-gasses when exposed to high heat. Testing is relatively quick and easy using measuring devices that are designed for individual gases, such as Draeger tubes. For best results, the indicator tube should be placed directly in the smoke stream. Note: There was a ventilation fan in the area where the men were working but they did not use it. (Since the accident, LLNL management has strengthened procedures to assure that workers use ventilating fans while they are welding and flame-cutting inside poorly ventilated areas.) In addition, the workers had no cooling water or fire extinguishers within easy reach. Such equipment might have been useful to cool the work and quench the production of decomposition products. Finally, they mistakenly assumed the blanket could effectively endure direct exposure to the cutting-flame. No welding blanket, even if made of asbestos, can prevent heat transfer and eventual degradation if it is subjected to the flame directly. ------------------------------------------------------------------------------ Toxic-product formation from welding blanket samples heated in air at 500 degrees C ------------------------------------------------------------------------------ Toxic products and amounts ----------------------------------------- CO CO HCN HCl NOx Sample identification (%) (%) (ppm) (ppm) (ppm) ------------------------------------------------------------------------------ 1. Thermonol 0.5 5 150 50 N.A.(a) 2. Siltemp 0 0 2 20 1 3. Silica fabric #2-3700 0.25 0 9 20 6 4. Thermoglass 0 0 10 0 3 5. LLNL Stores stock blanket sample 0.5 5.0 30 50 10 6. Weld-O-Glass #900FC 0.5 2.5 75 10 7 7. Weld-O-Glass #1000FC 0.5 2.0 75 100 0 8. Weld-O-Glass #2650 0.1 0 15 10 7 9. Weld-O-Glass #1850 0.1 0 30 10 15 10. Zetex-1100 0.5 trace 50 0 trace 11. Zetex-1200 0.4 trace 25 10(b) 0.5 ------------------------------------------------------------------------------ Threshold Limit Value 50 ppm 5000 ppm 3 ppm 5 ppm 5 ppm ------------------------------------------------------------------------------ (a)Too much interference by other compounds; cannot interpret results (b)Color change is due more to Cl2 than to HCl. The above materials were not coated with a special organic finish, thus the HCN and NOx were degradation products from an organic binder used during the fiber-manufacturing process. For comparison, the bottom row of the table lists the allowed threshold limit value (TLV) of these agents as defined by OSHA. The blanket which caused the accident was sample #5. ------------------------------------------------------------------------------ Information for this Bulletin was provided by Norman Alvarez, Hazards Control Department, at DOE's Lawrence Livermore National Laboratory. ------------------------------------------------------------------------------ Bulletin is published so that DOE program managers and contractors can share information about potential health and safety problems relevant to DOE operations. For additional copies, contact Nona Shepard, Editor, Office of Operational Safety, Assistant Secretary for Environment, Safety & Health, U.S. Department of Energy, Washington, DC 20545. Telephone: FTS 233-2958; Commercial 301-353-2958.