Correlating Welding Arc Time and Field Derived Generation Rates - PO132 American Industrial Hygiene Association’s 67th Annual Conference and Exposition June 7, 2007, Philadelphia, Pennsylvania F.W. Boelter, CIH, PE C.E. Simmons, CIH G.A. Vos, PhD Boelter Associates, Inc. Park Ridge, IL Problem Statement Would it be feasible to develop a reliable model utilizing anecdotal information related to welding arc time, coupled with laboratory sourced fume generation data to reliably predict concentrations in the breathing zone? To answer this, we performed this investigation, with the goal of answering 2 primary questions: How reliable are subjective estimates of arc time? How does arc time relate to breathing zone concentrations? Potential Variables Welding process Welding consumable Base metal - i.e. stainless steel, mild steel, etc. Presence of coatings or paints on the base metal Whether performed indoors or outdoors Configuration of the space Electrical current and voltage Amount and type of ventilation Welder’s skill and speed Position of the welder Personal protective equipment Fume/Gas quantity, dependent upon several parameters: Consumable diameter Consumable composition (usually alloy matched) Alloy Amperage Voltage Shielding Polarity Welder’s arc time or % time welding Complicating the Measurement of Welding – Many Related Terms with Differing Definitions Arc time Welding time Length of time arc is maintained Non-arc time/handling time Duty cycle Operating factor Operator factor Hours per shift Deposited metal/deposition rate AWS Welding HB– th 9 Ed, V1 Chung 1999 - pg 111 Castner 1998 – pg 228s NIOSH 1979 – pg 14 How reliable are subjective estimates of arc time? Inter-Rater Reliability Reliability of subjective measures could have an impact on any modeling based upon such data Therefore inter-rater reliability of subjective IH welding judgments was assessed 19 video clips of single welding cycles were shown to 9 experienced hygienists (including 3 clips shown in this presentation) Raters were asked to provide judgments for 8 variables per clip: Determination of if welding was being performed (Y/N) Subjective estimate of daily time spent welding in a given job (Hrs.) Estimates of the arc time for a job, as both: Percent of an 8 hour shift (%) Time in minutes (Min) The type of welding being performed (e.g. MIG, TIG, SMAW, etc.) The type of metal being welded (e.g. Steel, Aluminum, etc.) Whether or not the welder was wearing a welding hood (Y/N) And if so, the type of hood being used (e.g. PAPR) Tee Shirt Red Shirt Green Shirt Inter-Rater Reliability Variable Welding Data Type Dichotomous Time Welding (Hrs)* Arc Time (%)* Arc Time (Min)* Welding Type Metal Involved Welding Hood* Continuous Continuous Continuous Nominal Nominal Dichotomous Welding Hood Type Nominal Raters 9 9 6* 6* 5* 9 9 8* 9 9 Cases 19 19 19 19 19 19 19 19 19 19 Statistic KR‐20 Percent Agreement ICC(2,1) Absolute Agreement, Single Measures ICC(2,1) Absolute Agreement, Single Measures ICC(2,1) Absolute Agreement, Single Measures Percent Agreement Percent Agreement KR‐20 Percent Agreement Percent Agreement Value 0.97 93% 0.31 0.56 0.46 73% 73% 0.96 88% 73% 95% CI (71%, 100%) (0.10, 0.56) (0.37, 0.76) (0.24, 0.69) (47%, 100%) (38%,100%) p < 0.001 < 0.001 < 0.001 (66%, 100%) (23%, 100%) * When raters judged any given job as "not welding," other variables were somtimes left blank, resulting in an unbalanced dataset. Thus some statistcs where calculated using all cases but excluding the missing raters “Excellent” reliability for: - Judgments of welding - Whether a hood was worn “Fair to Good” reliability for: “Poor to Fair” reliability for: - Welding type - Metal involved - Hood type - Time spent welding in hours - Arc time (both % of shift and minutes) Reliability measures for all subjective estimates of time related variables were fairly low. This could negatively impact modeling and other related assessments when subjective time estimates are used in lieu of actual quantitative measures. Thus when data is used for modeling and other advanced calculations, actual quantitative measurements of time related variables should be used rather than subjective estimates. How does arc time relate to breathing zone concentrations? Breathing Zone Concentrations Breathing zone concentrations of welding performed by a pipe fitter doing his normal real world welding tasks in the field were evaluated Welding type was SMAW, performed indoors Multiple 15 minute samples were collected during actual arc welding tasks, both inside the welding hood and outside the hood Samples and Percent Arc Time Correlation of Arc Time and Concentrations – In Hood Samples Sample Percent Arc Time Total Particulate Fe (mg/m^3) (mg/m^3) Mn (mg/m^3) 1 2 3 4 5 r p (sig.) r² 0.04 0.12 0.14 0.18 0.11 0.62 0.26 0.39 20% 37% 30% 30% 23% 4 3.37 3.43 4.47 3.4 -0.22 0.72 0.05 0.42 0.59 0.6 0.8 0.48 0.61 0.28 0.37 Correlation of Arc Time and Concentrations – Out of Hood Samples Sample Percent Arc Time Total Particulate Fe (mg/m^3) (mg/m^3) Mn (mg/m^3) 1 2 3 4 5 r p (sig.) r² 20% 37% 30% 30% 23% 0.11 0.11 0.07 0.16 0.08 0.18 0.77 0.03 4.47 3.47 5.2 -0.49 0.67 0.24 0.95 0.51 0.36 0.72 0.37 -0.40 0.50 0.16 Conclusions How reliable are subjective estimates of arc time? Inter-rater reliability measures for subjective estimates of time related variables were low, which could have a negative impact on models using such estimates Thus when arc time data is used for modeling and other advanced calculations, actual quantitative measurements of time related variables would be critical. How does arc time relate to breathing zone concentrations? No significant or strong correlations for arc time and breathing zone concentrations were found Arc time on its own appears to be a poor predictor of exposure Can we develop a model that utilizes anecdotal information related to welding arc time, coupled with laboratory sourced fume generation data to reliably predict concentrations in the breathing zone? Not according to these findings These results emphasize the need for field-derived welding concentrations and emission rates for estimation of exposures Models based primarily upon arc time and laboratory fume generation rates using self reported estimates of welding time would likely yield estimates of exposure grossly in error Recommendations for future study: Determination of what variables do highly correlate with welding fume exposures, so that reliable and reproducible models can be developed Lab based fume generation rates are often developed with protocols and parameters that do not reflect real world welding practices, thus additional research is needed to derive more applicable values so that they may be included in the development of future models Literature KR-20 Streiner, D.L., and G.R. Norman: Health Measurements Scales. A Practical Guide to Their Development and Use, Second Edition. Oxford, U.K.: Oxford University Press, 1995. pp. 102–126. Nunnally, J.C., and I.H. Bernstein: Psychometric Theory, Third Edition. New York: McGraw-Hill, 1994. pp. 83–113. Kuder, G.F., and M.W. Richardson: The theory of the estimation of test reliability. Psychometrika 2:151–160 (1937). ICCs Shrout, P.E., and J.L. Fleiss: Intraclass correlations: Uses in assessing rater reliability. Psychol. Bull. 86:420–428 (1979). McGraw, K.O., and S.P. Wong: Forming inferences about some intraclass correlation coefficients. Psychol. Meth. 1:30–46 (1996). Adjectives that can be used in ICC Assessment: Fleiss, J.L.: The Design and Analysis of Clinical Experiments. New York: John Wiley & Sons, Inc., 1986. Correlating Welding Arc Time and Field Derived Generation Rates Fred W. Boelter, CIH, PE Boelter & Yates, Inc. Park Ridge, Illinois 847/692-4700 fboelter@boelter-yates.com