Occupational exposure to benzene at the ExxonMobil Refinery in Baytown, TX (1978–2006)

Abstract

Although occupational benzene exposure of refinery workers has been studied for decades, no extensive analysis of historical industrial hygiene data has been performed focusing on airborne concentrations at specific refineries and tasks. This study characterizes benzene exposures at the ExxonMobil Baytown, TX, refinery from 1978 to 2006 to understand the variability in workers' exposures over time and during different job tasks. Exposures were grouped by operational status, job title, and tasks. More than 9000 industrial hygiene air samples were evaluated; approximately 4000 non-task (>3 h) and 1000 task-related (<3 h) personal samples were considered. Each sample was assigned to one of 27 job titles, 29 work areas, and 16 task bins (when applicable). Process technicians were sampled most frequently, resulting in the following mean benzene concentrations by area: hydrofiner (n=245, mean=1.3 p.p.m.), oil movements (n=286, mean=0.23 p.p.m.), reformer (n=575, mean=0.10 p.p.m.), tank farm (n=9, mean=0.65 p.p.m.), waste treatment (n=446, mean=0.13 p.p.m.), and other areas (n=460, mean=0.062 p.p.m.). The most frequently sampled task was sample collection (n=218, mean=0.40 p.p.m.). Job title and area did not significantly impact task-related exposures. Airborne concentrations were significantly lower after 1990 than before 1990. Results of this task-focused study may be useful when analyzing benzene exposures at other refineries.

Introduction

Benzene is a natural, minor constituent of crude oil, where it is present in amounts varying from 0.1% to 3% (Verma and des Tombe, 1999). During the distillation of crude oil, the lighter weight fractions of the crude are extracted; because benzene has a low molecular weight and boiling point, it is removed with the lower boiling fractions. Benzene can also be produced in refining processes, such as reforming. Once produced, pure benzene is isolated in the refinery and transported through an enclosed and continuous process to the chemical plant, primarily for use as an intermediate feedstock in the production of other chemicals. Although the refining process is generally “closed and continuous,” occupational exposure to benzene in the oil refining industry has been evaluated for more than 60 years (OSHA, 2003). Existing literature reporting benzene exposures during refining operations indicate that full-shift exposures have been typically less than 1 part per million (p.p.m.) since the 1980s (Weaver et al., 1983; Buchet et al., 1984; Runion and Scott, 1985; CONCAWE, 1987, 1994; Nordlinder and Ramnas, 1987; Rappaport et al., 1987; HEI, 1988; Glass et al., 1994, 2000, 2001; Armstrong et al., 1996; Verma et al., 2001). Although these studies provide useful information regarding potential occupational exposures to benzene in the petroleum industry, they do not provide detailed job- or task-specific exposure data or information specific to workers at individual facilities. As a result, the industry-wide data sets are difficult to use if one wishes to conduct more detailed dose reconstruction studies or to estimate the exposures of individual workers on the basis of their job descriptions. This study expands upon the currently available literature by analyzing industrial hygiene data collected from one of the world's largest refineries on the basis of the jobs and tasks being performed.

In the United States, ExxonMobil currently owns and operates seven petroleum refineries in Baton Rouge, LA; Joliet, IL; Baytown, TX; Beaumont, TX; Chalmette, LA; Torrance, CA; and Billings, MT. As part of an ongoing historical benzene exposure assessment program, analyses have been conducted to understand both non-task (approximately full-shift time weighted averages) and task level exposures for refinery employees for these facilities (Panko et al., 2009). This paper presents the quantitative estimates of potential exposure by job category and task for the largest domestic ExxonMobil refinery, located in Baytown, TX.

Background

The Baytown refinery was built by Humble Oil in 1919. Humble Oil was eventually purchased by Standard Oil of New Jersey, later named Exxon, and thus upon the merger of Exxon and Mobil in 1999, operation of the Baytown refinery continued under the ExxonMobil Corporation (Henderson and Benjamin, 1996; Segal, 1999). The Baytown refinery is one of the largest in the United States capable of processing over 500,000 barrels of oil per day (b.p.d.). Nearly 95% of the crude oil processed at the Baytown refinery arrives through ships and barges. Both fuels and lube products are produced at this refinery. The Baytown refinery produced asphalt for product sales until 2002. Fuel products are primarily exported through pipeline, whereas specialty products, such as lubes, leave through barge. There are no rail or truck loading locations within the refinery.

Discussion of general refining processes have been detailed elsewhere; however, all refineries are unique (OSHA, 2003; Meyers, 2004). Information specific to the process configuration at the Baytown facility is therefore provided in Figure 1. In addition, a description of each area in the refinery and its associated functions are provided in Table 1.

Figure 1
figure1

Process flow diagram for the ExxonMobil refinery in Baytown, TX. This diagram identifies the process streams associated with the major units of the refinery, including benzene content by weight.

Table 1 Description of areas within the ExxonMobil refinery in Baytown, TX that are represented in the database.

Historical Engineering and Process Changes that Reduced Occupational Exposure to Benzene

The greatest potential for benzene exposure during refinery operations results from direct contact with benzene-containing process or waste streams. Accordingly, certain jobs or tasks, such as those involving the opening of lines or equipment containing benzene, provide the highest likelihood of worker exposure to benzene. Occupational health practitioners at the Baytown refinery have a history of limiting potential worker exposure to benzene; they have been evaluating worker exposure to benzene since the 1930s, and have well-established industrial hygiene guidelines/procedures for minimizing potential exposure to benzene. The use of personal protective equipment such as respirators, protective clothing and gloves, and the containment of process streams in closed systems has been a focus of the industrial hygiene and operations groups since the refinery was built. In addition, numerous operational changes and engineering controls have been implemented to reduce the potential for worker exposure to benzene and to limit environmental emissions of benzene.

The Baytown refinery was one of the first to institute the use of ventilated and enclosed sampling boxes. These boxes were installed in 1988 at process stream sampling points, and these allow workers to sample process streams in a closed system, thereby reducing the potential for benzene exposure during sampling. A benzene stripper was installed in 1996 to remove benzene from any water stream that contained greater than 0.5% benzene. Such water streams are likely to be those that contact process streams, particularly those associated with fuels production.

As pure benzene is produced at and exported from the Baytown refinery and associated chemical plant complex, much of the focus of engineering control projects at the refinery was placed on lowering benzene exposures at the loading docks, particularly at the dock where benzene was loaded onto barges.

Industrial Hygiene Program for Benzene

In addition to implementing process modifications designed to reduce benzene exposure in the workplace, ExxonMobil established benzene programs and practices to ensure that employee exposures were less than the generally accepted exposure limits and to comply with the Occupational Safety and Health Administration (OSHA) standards. As OSHA changed the benzene exposure limits, similar adjustments were made to the Baytown refinery's benzene program to maintain compliance. These programs included the identification of benzene-containing streams and associated tasks that potentially resulted in worker exposure to benzene, employee monitoring and hazard communication, medical surveillance of potentially exposed workers, and the implementation of benzene exposure controls including handling practices for the performance of tasks where there may be a potential for exposure to benzene, including preparing equipment for maintenance work, opening and blinding equipment, enclosed space entry, and clean-up of spills or leaks. These guidelines include gas testing before performing the tasks and establishing a temporary benzene “regulated area” if test results are greater than 1 p.p.m. benzene.

Overview of Industrial Hygiene Monitoring Strategy

Industrial hygiene programs at the ExxonMobil refineries across the United States are guided by an exposure assessment strategy (EAS) that incorporates both qualitative and quantitative aspects for assessing worker exposure. The EAS is a multistep process that starts with an evaluation of the chemical constituents of the process streams to which workers may be exposed, and then assesses the level of potential (independent of the use of respirators) on the basis of the tasks that may bring the workers into contact with streams or other potentially hazardous materials. Quantitative measurements are then targeted to assess tasks and activities with higher potential for exposure, or to reduce the uncertainty in exposure assessment results. As part of the benzene program, much of the monitoring has historically been focused on activities where a worker is working on equipment containing streams with higher percentages of benzene (i.e., reformate) and where the activities involve open handling of streams (e.g., opening/draining equipment versus operating closed processing equipment), Similarly, the industrial hygiene monitoring has historically targeted workers who are potentially exposed to process streams more frequently than those who are not. This sampling strategy has resulted in the monitoring of selected jobs and tasks that potentially have higher levels of benzene exposure than would result from a random sampling approach. Furthermore, once an area, job, or task is well characterized and determined to be of low exposure potential, or after controls have been implemented, collecting additional data for that area, job, or task becomes infrequent. As a result of this focused or targeted monitoring strategy (e.g., in contrast to an industrial hygiene program that involves collecting random samples, or that requires quarterly or annual sampling of every employee), the benzene air sampling data collected under this program is likely to over-represent workers and work activities with higher benzene exposures.

Personal Protective Equipment

Current and historic ExxonMobil respiratory protection standards were reviewed to identify those tasks and activities that have historically required respiratory protection. In the past, a few areas of the Baytown refinery required respirators (parts of the environmental control house, docks, and the oil movements control center) and were designated as permanent benzene-regulated areas in accordance with 29 CFR 1910.28. These areas have changed over time as engineering controls have reduced the exposure potential. On the basis of gas testing results, in instances when benzene concentrations are greater than 1 p.p.m. (conditions most likely to occur during periodic tasks involving work with open process equipment), appropriate respirators are required to provide protection to the workers. In addition to these instances, certain specific tasks in the oil movements area require respiratory protection regardless of gas testing results; such tasks include tank gauging, tank water draws, and mechanical work in tanks (For tanks containing benzene, steam cracked naphtha, hydroformer feed, and/or reformate.). Temporary areas requiring benzene respiratory protective equipment have historically been used for selected tasks and activities by standard operating and maintenance procedures including opening any process equipment that has contained more than 0.1% benzene and could not be completely drained and flushed before opening or in areas where gas testing detects benzene concentrations in air above the exposure limit. In these cases, respirators are required to be worn and the type of respirator is dependent on the measured air concentration.

The potential for dermal exposure to benzene may exist for anyone working with an open benzene-containing process stream. A review of all available current and historical refinery industrial hygiene documentation was performed to determine whether specific benzene-related jobs or tasks actually offered a significant opportunity for dermal exposure to benzene, and whether PPE was recommended for these tasks. On the basis of this review, it appears that refinery workers were always required to wear protective clothing and gloves to prevent dermal exposure. The industrial hygiene records support this assessment as records after about 1970 indicate that when a potential dermal exposure to benzene existed, workers wore gloves. In addition, the electronic industrial hygiene records indicate that more than 99% of the tasks and jobs that have been monitored for benzene exposure did not involve dermal contact. Therefore, absorption of benzene due to dermal contact was judged to be de minimis, and no attempts to quantify exposure were performed.

Methods

Data Collection

Historical industrial hygiene data used in this analysis were gathered from multiple sources. A total of 9650 (total records in database) post-1978 samples were gathered from three electronic ExxonMobil databases: Personal Computer Industrial Hygiene System (PCIHS, 1978–1998), Medgate (1999–2003), and EAS (2005-present). Information contained in these databases pertinent to our analysis included the following: sample date, sample duration, sample type (personal/area), analytical result, sample media, job title, department, area and task associated with the sample. The accuracy of the information contained in all three databases was verified through an independent and random review of the corresponding paper records associated with 25% of the air sample results. An error was considered critical if it involved the sample result (concentration, laboratory result, units, qualifier). All other errors, including typographical errors, were considered non-critical, given that they would not lead to miscalculation of the airborne concentration. Overall, based on the fact that there were fewer than 0.8% critical errors in the database, it was concluded that the database accurately reflected the original documentation. For any errors (typically transcription errors) found during the verification process, appropriate changes were made to the database.

In addition to the electronic databases, information regarding the refinery processes, job and task descriptions, benzene process changes, industrial hygiene surveys, and exposure assessment initiatives relevant to the study were gathered from numerous sources. Industrial hygiene surveys conducted over time throughout the refinery were reviewed for both additional exposure data and details regarding jobs and tasks. Furthermore, both historical and current industrial hygiene programs specific to benzene and respiratory protection were reviewed to gain an understanding of personal protection practices throughout the refinery. Finally, refinery process engineers, industrial hygienists, and operations personnel were consulted regarding information not found in written records.

Data Analysis

Air samples collected during routine and turnaround operations were organized by area (Table 1), job title (Table 2), and task description (Table 3). Air samples were also classified by sample duration (i.e., <180 min and ≥180 min) and type of sample (i.e., personal and area). Samples less than 180 min were typically collected to capture periods in which specific task activities were underway and were considered task samples representative of peak or task-specific exposures. Samples with durations of 180 min or greater were considered non-task samples, as they most likely characterized more than one task performed by a worker as part of routine job duties and were not targeted in monitoring programs as task-specific. Samples were also classified as non-task samples when the sample collection time could not be determined.

Table 2 Description of job titles within the ExxonMobil refinery in Baytown, TX that are represented in this database. The areas associated with each job title (as represented in the database) are presented as well.
Table 3 Description of task bins. The task bins were created to group similar tasks with similar exposure potential together, and were reviewed by industrial hygienists from ExxonMobil. Both the job titles and areas associated with each task (as represented in the database) are presented.

Each personal air sample was assigned to one of 27 job titles and one of 29 work areas. In addition, the numerous tasks conducted by workers at the refinery were consolidated into 16 task bins on the basis of the nature of the task. All samples were classified by a refinery operational status of routine or turnaround. The protocol for assigning job title, work area, and task assignments was reviewed by ExxonMobil industrial hygienists to ensure that samples were properly characterized. Traditionally, worker exposure groups have been defined by ExxonMobil as job title-area-task combinations. These worker exposure groups were evaluated to determine whether any of them could be combined to increase statistical power for a quantitative assessment of exposures using an analysis of variance (ANOVA) model. The ANOVA was used to determine whether the average benzene concentrations differed significantly by operational status, by work area for a given job title (non-task samples), and by job title or work area for a given task bin (task samples).

Censored data (i.e., those samples less than the limit of detection (LOD)) were included in the statistical analysis using the regression on order statistics (ROS) method (Helsel, 2005; U.S. EPA, 2007), which is equivalent to the robust log probit regression method presented in the IH literature (Hewett, 2007; Ignacio and Bullock, 2006). The general approach of the ROS method includes fitting a linear regression model of the detected values of the data set to the quantiles of the assumed distribution (values from the y axis of a probability plot) and replacing the values for samples less than the LOD with the values extrapolated from the linear regression. As this data set has multiple limits of detection associated with it, the robust ROS method developed by Helsel and Cohn (1988) was used. This method was used instead of the typical substitution method — substituting the LOD by the LOD divided by 2 or the —because it produces fairly robust estimates of the mean and SD even with modest departures from the lognormal distribution and even if 50–70% of the data are below the LOD (Huybrechts et al., 2002; Lubin et al., 2004; Baccarelli et al., 2005; Ignacio and Bullock, 2006).

The benzene concentration data for the task and non-task data sets were tested for distribution fit using the Kolmogorov–Smirnov goodness-of-fit test for a normal, lognormal, and gamma distribution, and none of these distributions (P<0.05) were found at a 95% confidence level. However, the data were found to be approximately lognormal on the basis of probability plots. Therefore, the robust ROS method for a lognormal distribution was used for these two data sets.

As these data sets were approximately lognormal, the natural log-transformed sample results including non-detect values estimated by the ROS model were used in the ANOVA analyses. The Tukey multiple comparison test was used to identify any significant pairwise differences between area for a given job title (non-task samples) and between job title or area for a given task bin (task samples) at a 95% confidence level (Neter et al., 1990). Job categories and task bins characterizing potential differences by area (non-task and task samples) or job title and area (task samples only) were established on the basis of the results of the ANOVA and post hoc pairwise analysis.

Standard industrial hygiene descriptive summary statistics including geometric SD, geometric mean and arithmetic mean were calculated for the resulting job categories and task bins. However, the primary purpose of this study was to present comprehensive job category and task bin benzene concentration data for use in historical exposure reconstruction. Therefore, the results and discussion emphasize the arithmetic mean, which is considered the best metric for worker dose (Ignacio and Bullock, 2006).

To determine whether there was a trend over time by any job category, pairwise comparisons were carried out to identify statistical differences between samples collected from 1978 to 1989 and those collected from 1990 to 2006. The task data set does not contain refinery personal samples from 2004 to 2006. The year 1990 was used as the cutoff point, as regulations intended to reduce employee exposure to benzene and to reduce benzene emissions to the ambient environment from petroleum refineries were mostly implemented by 1990 (OSHA, 1987; U.S. EPA, 1989).

Results

In total, 9650 samples were initially assembled to perform this analysis: 8845 from the ExxonMobil data bases and 805 from archived hard copy files. Before analysis, duplicate entries, samples lacking quantification for benzene, field blanks, instantaneous or grab samples, samples taken during emergency response or fire training, and any samples rejected by the ExxonMobil industrial hygienists for being invalid or not representative of personal exposure were eliminated from the data set (Figure 2). Of the remaining 9310 data points, 7402 were taken in the refinery and 1908 were collected at the docks or loading areas. Docks and loading area samples are reported separately. Of the 7402 refinery samples, 1988 were area samples, 5239 were personal samples, 18 were source-specific samples, and 157 lacked sufficient data to classify sample type. Source-specific or unclassifiable samples may not be characteristic of typical exposures of refinery employees, and therefore not considered further in this analysis.

Figure 2
figure2

Exposure Monitoring Data Categorization (1978–2006): Breakdown of the data included in the database. For this analysis, only those personal samples associated with the refinery were considered.

The refinery data set used for the analysis contained 5239 personal samples, of which 4160 were considered non-task samples (≥180 min) and 1079 were characterized as task samples (<180 min) (Figure 2). Of the non-task samples, 3897 were taken during routine operations and 263 during turnarounds. Of the task samples, 1027 were taken during routine operations, 52 during turnaround operations.

Air sampling for benzene at Baytown was conducted by ExxonMobil industrial hygienists according to the standard operating procedures involving the use of either charcoal sorbent tubes or passive organic vapor badges. Samples were analyzed by a laboratory accredited by the American Industrial Hygiene Association according to NIOSH or other methods consistent with the internal standard operating procedures (National Institute of Occupational Safety and Health, 2003). In general, over the past 20 years, data indicate that the majority of benzene concentrations were below the LOD. A detailed description of detection frequencies and the limits of detection for each data set can be found in the Table 4.

Table 4 Detection frequency and average limit of detection for non-task and task samples in the refinery, docks, and loading rack data sets.

Non-Task Concentration Estimates in the Refinery Data Set

The results of the non-task benzene concentrations by job category are presented in Table 5. Not surprisingly, the ANOVA and pairwise comparisons indicated that air concentrations for five job titles (Process Technician, Machinist, Pipefitter/Welder, Instrument Technician, and Contractor-Catalyst) were influenced by the area of the plant in which the employee was working. As such, separate job categories were created for each job title/area combination that had benzene concentrations that were significantly different than that particular job title with all areas combined.

Table 5 Summary statistics for the non-task data set for (a) ExxonMobil employees, (b) contractors, including job title, operational status, and area.

Concentrations of airborne benzene for non-task samples were analyzed for each job category during routine and turnaround operations. Of those job categories that had a sufficient sample size (n≥10), the highest mean benzene air concentration (3.4 p.p.m.) was associated with the pipefitter/welder working during routine operations in the tank farm (n=11). Of the 11 samples associated with this job category, three had results >1 p.p.m.; these samples were collected in 1984 during blinding and gas-freeing of the light cat naphtha tank.

The airborne benzene concentration distributions for those job categories with sufficient data (n≥10) are presented in Figure 3. The most frequently sampled job categories included those with the greatest potential for contact with benzene-containing process streams, such as process technician, machinist, pipefitter-welder, and laboratory technician.

Figure 3
figure3

Distribution of non-task benzene air concentrations by job category. This figure demonstrates the overall distribution of air concentrations of benzene for those job categories with greater than 10 samples.

Influence of Operating Status

The ANOVA comparing benzene concentrations during routine and turnaround operations indicated that there was a statistical difference (P<0.05) between operational statuses when all job categories were considered. Within job categories, differences between operational status were not meaningful or were attributable to specific targeted activities, such as catalyst skimming at the hydrofiner or hydraulic excavation at the waste treatment plant. For most job categories, potential differences in exposure during turnaround could not be discerned because too few samples were available for the job category or job categories in specific areas. One exception was the Process Technician category (n>50, for routine and turnaround groups), where the mean benzene concentration for all areas combined is lower during turnaround operations (n=102, mean=0.067 p.p.m.) than routine operations (n=2021, mean=0.27 p.p.m.). Data collected in areas such as the hydrofiner, oil movements, and tank farms during routine operations drives the statistical difference in benzene concentration between the operation statuses (Table 5). This result was consistent with operational practices that require draining/purging of liquid streams before the commencement of turnaround operations. Therefore, during much of the turnaround there is little material remaining in the equipment to generate significant exposure to benzene.

Benzene Concentrations over Time

The distributions of personal airborne benzene concentrations by job category with sufficient sample size (n>10) for the two time periods (1978–1989 and 1990–2006) are shown in Figure 4. Statistical differences (P<0.05) in mean benzene concentration by time period were observed for process technician (hydrofiner (1.5 and 0.010 p.p.m.), waste treatment (0.14 and 0.012 p.p.m.), and all other areas (0.070 and 0.010 p.p.m.)), machinist (waste treatment (0.37 and 0.019 p.p.m.) and all other areas (0.10 and 0.012 p.p.m.)), laboratory technician (0.12 and 0.13 p.p.m.), and mobile equipment operator (0.23 and <0.11 p.p.m.) during routine operations. For all of these job categories, mean benzene air concentrations were statistically significantly lower for the time period from 1990 to 2006 than for 1978–1989 with the exception of the laboratory technician (Figure 4). In this case, the simple arithmetic mean for the laboratory technician increased over time. Although the simple arithmetic mean is an appropriate and unbiased predictor of the true mean, for subsets of data with large sample sizes and large geometric SDs, such as this one, it is less applicable to the ANOVA comparison than other estimates (Ignacio and Bullock, 2006). Reanalysis of the arithmetic mean for this group on the basis of the lognormal minimum variance unbiased estimator revealed a decrease over time for the laboratory technician from 0.13 to 0.02 p.p.m.

Figure 4
figure4

Distribution of non-task benzene air concentrations over time by job category. This figure demonstrates the overall distribution of air concentrations of benzene for each window of time (1978–1989 and 1990–2006). Distributions are presented by job category, and are limited to those job categories with greater than 10 samples in each time window. *Denotes statistical significance by time period (P<0.05). Calculation of the MVUE of arithmetic mean revealed a decrease over time for the laboratory technician from 0.13 p.p.m to 0.02 p.p.m. ††All results for samples collected between 1990–2006 (n=12) were below detection limit; mean detection limit plotted.

Task Exposure Concentration in the Refinery Data Set

The average benzene air concentrations and related summary statistics for task samples are summarized in Table 6. The ANOVA and pairwise comparisons indicated that the area in which these tasks were performed did not influence the benzene concentrations associated with the tasks. In addition, the job title of the individual performing the task appears to have had no influence on the benzene concentrations associated with the tasks.

Table 6 Summary statistics for the task data set, including job title and area.

Of the task bins with sufficient sample size (n≥10), the laboratory sample analysis bin had the highest mean benzene air concentration (1.9 p.p.m.). To perform this task, the technician must work directly with small quantities of process stream samples, and work is usually performed under controlled conditions. However, this result is driven by nine sample results above 10 p.p.m. that were collected in 1988 during the filtration of reagent grade benzene for saybolt color test while wearing powered air-purifying respirators.

The distributions of benzene air concentrations for those task bins with sufficient data (n≥10) are presented in Figure 5. The tasks with a potential for contact with a process stream were the most frequently sampled, and include the following: sample collection (n=218), sample analysis (n=147), liquid transfer (n=137), blinding and breaking (n=97), gauging (n=65), equipment cleaning and repair (n=58), equipment preparation (n=47), and waste treatment plant maintenance (n=46).

Figure 5
figure5

Distribution of task benzene air concentrations by task bin. This figure demonstrates the overall distribution of air concentrations of benzene for those task bins with greater than 10 samples.

Influence of Operating Status

The ANOVA that compared benzene concentrations during routine and turnaround operations indicated that there was a statistical difference (P<0.05) between operational statuses when all task bins were considered. However, for most task bins, potential differences in exposure during turnaround could not be discerned because too few samples were available for the task bin. Nonetheless, task level samples were analyzed separately on the basis of operations status (Figure 5).

Benzene Concentrations over Time

The distributions of personal airborne benzene concentrations by task bins with statistical differences (P<0.05) in mean benzene concentration for the two time periods (1978–1989 and 1990–2003) are shown in Figure 6. For those task bins with sufficient sample size (n≥10), a statistical difference in mean benzene concentration by time period was observed for sample collection (0.87 and 0.079 p.p.m.) and sample analysis (4.6 and 0.10 p.p.m.). Although the repair leak task bin only had a sample size of eight, it too was statistically significant by time period (36 and 0.037 p.p.m.). For these task bins, benzene air concentrations were statistically significantly lower for the time period from 1990 to 2003 than for 1978–1989.

Figure 6
figure6

Distribution of benzene air concentrations over time by task. This figure demonstrates the overall distribution of air concentrations of benzene for each window of time (1978–1989 and 1990–2003). Distributions are presented by task bin, and are limited to those bins with greater than 10 samples in each time window. *Denotes statistical significance by time period (P<0.05).

Discussion

This study presents the results of 25 years of personal air monitoring for benzene at the Baytown, TX ExxonMobil refinery. Air concentrations of benzene collected while performing specific jobs and tasks were found to be consistent with the results reported in other studies of benzene exposure in the petroleum industry (Weaver et al., 1983; Buchet et al., 1984; Runion and Scott, 1985; CONCAWE, 1987, 1994; Nordlinder and Ramnas, 1987; Rappaport et al., 1987; HEI, 1988; Verma et al., 2001). When comparing the results of this study to the industry-wide data sets, the average exposures at the Baytown refinery decrease within the range of those reported in the literature, despite the fact that the Baytown data set was targeted to over-represent benzene handling activities (Verma et al., 2001). The average benzene air concentration at the Baytown refinery for all non-task samples is 0.23 p.p.m. (during routine operations), compared with the arithmetic means reported in the literature, which range from 0.05 to 1.62 p.p.m., with an average of 0.22 p.p.m. (Runion and Scott, 1985; Verma et al., 2001).

Fifty-nine percent of the non-task samples showed benzene air concentrations below the LOD. Given the large number of samples in the data set and the targeted nature of the sampling program (sampling is focused on jobs where direct contact with benzene-containing streams is most likely), the low detection frequency is indicative of generally overall low concentrations of benzene in the air. Specifically, nearly 98% of the non-task sample results were less than 1 p.p.m., a finding consistent with an industry-wide study indicating that 95% of benzene results from air samples taken at several US refineries were below 1 p.p.m. (Runion and Scott, 1985). Furthermore, 65 of the samples classified as non-task in these data set had unknown sampling times, and were assumed to represent samples collected for over 180 min. This is a conservative assumption, because any error in the analysis associated with this assumption would overestimate long-term exposure.

Analysis by refinery area indicates that with the exception of five job categories, the area in which an individual worked did not influence the average benzene air concentration. These findings are similar to data from other refinery studies, and demonstrate that the nature of the work being performed is a major determinant of overall exposure to benzene (Verma et al., 2001). However, because there are often very few samples associated with a job category and area, the absence of any statistical difference may be a function of a small sample size. For the Baytown refinery job categories that displayed area dependence during routine operations, the sample sizes were large (n=30–2021), and the results were consistent with expectations; for example, in areas where the benzene content of the process stream was low, occupational exposure was low, and vice versa.

The analysis of air concentration trends over time indicated that for most job categories, exposures were slightly higher in the 1978–1989 time period than in 1990–2003, and were statistically significantly higher for three of the job categories (Figure 4). These results are consistent with expectations, given the general time period in which various process changes were implemented. However, decreases in airborne concentrations of benzene in the specific areas where the changes occurred could not be discerned from the relatively small data sets available.

The benzene air concentrations associated with the 16 task bins were largely non-detectable. Only two of ten task bins (n≥10) had a detection frequency greater than 50%. Overall, the results from the analysis of the task-level data indicated that tasks that may require interaction with process streams, such as sample collection, sample analysis, liquid transfer, repairing leaks or blinding, and breaking in tank farms generally have higher air concentrations than those where exposure is minimal (Table 6). These results reinforce the historic focus of the benzene exposure control program described earlier. Of the 1079 task level samples, only 39 (3.6%) had results greater than 5 p.p.m. In comparison, in a study of European refineries, 14% of the benzene air concentrations associated with the tasks conducted by refinery operators exceeded 5 p.p.m. (CONCAWE, 1994).

Each task bin was created to characterize subsets of all tasks performed at the refinery, with the intention of including in each task bin activities sharing a similar exposure potential. As each job category is defined by the types of activities that an individual may perform, each task bin has only a few job titles associated with it. Nevertheless, there was no difference between job titles within each task bin, indicating that the tasks were the primary determinant of exposure. The analysis of the task bin by area further supports this concept, as it also indicated no statistical difference by area.

Task data can also be used to gain a better understanding of exposure by job title or area. In this data set, the ANOVA and pairwise comparisons for non-task concentration estimates indicated that operational status and area of the plant was a determinant of exposure. By examining these results, it can be seen that in some instances, while the concentrations are low, the variability within a given job title, area, and operational status combination is appreciable. In these instances, it is not possible to ascertain which activities might be associated with the underlying cause of variability without evaluating the task data. For example, the air concentrations associated with the process technician in the waste treatment area under routine conditions was characterized by a geometric mean of 0.045 p.p.m., geometric SD of 5.6, and an arithmetic mean of 0.13 p.p.m. By reviewing the tasks performed by process technicians and associated job titles and areas, it can be seen that waste treatment plant maintenance task, with a geometric mean of 0.25 p.p.m., geometric SD of 7.2 p.p.m., and arithmetic mean of 0.85 p.p.m., could be an appreciable contributor to variability. As process technicians only rarely conduct this task, this potentially important determinant of exposure would not have been obvious by review of the non-task samples alone.

Targeted Monitoring Strategy and Use of the Data

According to the ExxonMobil monitoring strategy and standard industrial hygiene practice, air sampling for benzene is focused on jobs where there is a potential for exposure based on the nature of the work or the work area. This sampling approach is evident in the number of samples taken for each job, area, and task. Those areas with process streams that contained the highest percentage of benzene were the most frequently sampled; similarly, the jobs and tasks with the highest likelihood of exposure to process streams containing benzene were most frequently sampled (see Tables 1, 2, 3).

On the basis of the targeted sampling approach used by the industrial hygienists at this refinery, the non-task averages calculated for each job category may not represent long-term (over the span of months or years) averages for all job titles. For those job categories where the individuals have regular contact with process streams (e.g., process technicians, pipefitter-welders, and so on) and are therefore more likely to be targeted for sampling, these daily averages may be more representative of long-term averages. However, for those job categories where individuals do not have regular contact with process streams (e.g., instrument technicians, mobile equipment operators, and so on), the benzene air concentrations are more likely to represent the unique activities being performed on the sampling day, rather than those activities being performed on an average day, and therefore the daily averages, as predicted by the data presented here, would not be representative of long-term averages.

As of the targeted nature of the monitoring strategy, care must be taken when using the data presented in this analysis to estimate employee exposures to benzene at this refinery. As for some job categories, the non-task daily averages may not accurately represent long-term averages, it is necessary to reconstruct these exposures on a case-by-case basis. This reconstruction of daily exposures may require using task-based data, time-activity patterns, and exposure frequencies. Furthermore, by reconstructing an individual's exposure on a case-by-case basis, considerations can be made, if appropriate, for respirator use.

Strengths and Limitations

This study provides analysis of a robust data set of benzene air samples, and considers several factors: operational status of the refinery; area of the refinery, job title, and task performed. Of the studies available in the existing literature, none have attempted to consider all of these factors in determining potential benzene exposure; rather, most of the literature has focused only on one variable, such as refinery area or job. Furthermore, within the review of the data, an exhaustive independent review of the data was conducted to ensure data quality and accuracy with respect to original documentation. Of the existing published literature, no other studies have ensured this level of data quality review on such a large body of data.

The analysis by task bin is unique to this study; other refinery studies have focused primarily on long-term exposure estimates. Although using task-based exposure levels in epidemiology has its limitations (Cherrie, 1996: Verma et al 2001), which summarized data from several studies evaluating benzene exposure to employees in petroleum industries, explicitly recommend moving toward task-based exposure assessments as opposed to long-term time weighted average estimates for the petroleum industry (Verma et al., 2001). This study addresses the recommendation by Verma et al. (2001) by providing a detailed task-level analysis of benzene air concentrations at the Baytown refinery. In addition, because analysis of the task level data suggests that exposure to benzene is driven by the task or by the work being performed (as opposed to other factors such as area, job, or operational status); the results from this study support that observation.

Although this study improves upon the currently available literature, it is not without limitations. Although this facility has been actively involved in assessing worker exposures to benzene since the 1950s, only limited documentation of the results before 1978 were available. As a result, quantification of workers' exposures before 1978 will require some extrapolation from the current data set (Proctor et al., 2004). Similarly, industrial hygiene benzene data exist for 1978–2006; however, although personal samples were collected from 2004 to 2006, task-related samples were unable to be identified as such in the data base, necessitating extrapolation from 2004 to the present. In addition, samples were not collected every year for each job category. Interpolation will therefore be necessary to estimate exposures for specific job categories during years when data were not collected. Finally, respiratory protection factors or adjustments have not been applied to the measured air concentrations, although, on the basis of the available documentation, respiratory protection is part of the standard work practice for some tasks, and should be taken into account when using the data for exposure reconstruction.

Conclusions

The data presented here indicate that the benzene air concentrations at the ExxonMobil refinery in Baytown, TX, are generally low for both non-task and task-based sampling events. When considering these data for exposure reconstruction, it is especially important to bear in mind the nature of the samples analyzed in this data set, which were primarily collected using a targeted strategy. As such, these data are likely to represent the upper end of occupational exposure distributions at the refinery, rather than air concentrations representative of typical exposures. This study demonstrates the necessity of analyzing refinery benzene exposures on the task level, and is useful for understanding which activities drive refinery worker exposures to benzene.

References

  1. Armstrong T.W., Pearlman E.D., Schnatter A.R., Bowes S.M., Murray N., and Nicolich M.J. Retrospective benzene and total hydrocarbon exposure assessment for a petroleum marketing and distribution worker epidemiology study. Am Ind Hyg Assoc J 1996: 57: 333–343.

    CAS  Article  Google Scholar 

  2. Baccarelli A., Pfeiffer R., Consonni D., Pesatori A.C., Bonzini M., Patterson D.G., Bertazzi P.A., and Landi M.T. Handling of dioxin measurement data in the presence of non-detectable values: overview of available methods and their application in Seveso chloracne study. Chemosphere 2005: 60: 898–906.

    CAS  Article  Google Scholar 

  3. Buchet J.P., Van Eyken J., and Lauwerys R. Evalation de l′exposition au benzene des travailleurs preposes aux qais de chargement de l'essence dans une raffinerie. Cahiers de Medecine de Travail Cahiers Voor Arberdsgeneeskunde 1984: 21: 25–27.

    CAS  Google Scholar 

  4. Cherrie J.W. Are task-based exposure levels a valuable index of exposure for epidemiology? Ann Occup Hyg 1996: 40 (6): 715–722.

    CAS  Article  Google Scholar 

  5. CONCAWE. A Survey of Exposures to Gasoline Vapour. Oil Companies European Organization for Environment, Health and Safety (CONCAWE), The Haguc, The Netherlands, 1987.

  6. CONCAWE. Review of European oil Industry Benzene Exposure Data (1986–1992). Oil Companies European Organization for Environment, Health and Safety (CONCAWE), Brussels, 1994.

  7. Glass D.C., Adams G.G., Manuell R.W., and Bisby J.A. Retrospective exposure assessment for benzene in the Australian petroleum industry. Ann Occup Hyg 2000: 44 (4): 301–320.

    CAS  Article  Google Scholar 

  8. Glass D.C., Gray C.N., Adams G.G., Manuell R.W., and Bisby J.A. Validation of exposure estimation for benzene in the Australian petroleum industry. Toxicol Ind Health 2001: 17 (4): 113–127.

    CAS  Article  Google Scholar 

  9. Glass D.C., Spurgeon A., Calvert I.A., Clark J.L., and Harrington J.M. Retrospective assessment of solvent exposure in paint manufacturing. Occup Environ Med 1994: 51 (9): 617–625.

    CAS  Article  Google Scholar 

  10. HEI. Gasoline Vapor Exposure and Human Cancer: Evaluation of Existing Scientific Information and Recommendations for Future Research. Cambridge, MA: Health Effects Institute (HEI), 1998.

  11. Helsel D. More than obvious: better methods for interpreting nondetect data. Environ Sci Technol 2005: 39: 419A–423A.

    CAS  Article  Google Scholar 

  12. Helsel D.R., and Cohn T. Estimation of descriptive statistics for multiply censored water quality data. Water Resour Res 1988: 24: 1997–2004.

    CAS  Article  Google Scholar 

  13. Henderson W., and Benjamin S. Standard Oil: The First 125 Years. Motorbooks International, New York, 1996. pp 12.

    Google Scholar 

  14. Hewett P. User Guide for IH Data Analyst. Available from: http://www.oesh.com/x%20Software/IHDA.php. 2007.

  15. Huybrechts T., Thas O., Dewulf J., and Van Langenhove H. How to estimate moments and quantiles of environmental data sets with non-detected observations? A case study on volatile organic compounds in marine water samples. J Chrom 2002: 975: 123–133.

    CAS  Article  Google Scholar 

  16. Ignacio J.S., and Bullock W.H., (eds.) A Strategy for Assessing and Managing Occupational Exposures. AIHA Press, Fairfax, VA, 2006. pp 415–421.

    Google Scholar 

  17. Lubin J.H., Colt J.S., Camann D., Davis S., Cerhan J.R., Severson R.K., Bernstein L., and Hartge P. Epidemiologic evaluation of measurement data in the presence of detection limits. Environ Health Perspect 2004: 112 (17): 1691–1696.

    CAS  Article  Google Scholar 

  18. Meyers R.A., (ed.). Handbook of Petrolem Refining Processes. McGraw-Hill, New York, 2004.

    Google Scholar 

  19. National Institute of Occupational Safety and Health. NIOSH manual of analytical methods: aromatic Hydrocarbons, Method 1501. 4th edn. 2003.

  20. Neter J., Wasserman W., and Kutner M.H. Applied Linear Statistical Models, 3rd edn. Irwin, Burr Ridge, IL, 1990.

    Google Scholar 

  21. Nordlinder R., and Ramnas O. Exposure to benzene at different work places in Sweden. Ann Occup Hyg 1987: 31 (3): 345–355.

    CAS  PubMed  Google Scholar 

  22. OSHA 29 CFR 1910.1028: Benzene 1987.

  23. OSHA. OSHA technical manual: petroleum refining processes. Vol Section IV, Chapter 2. Occupational Safety and Health Administration (OSHA) 2003.

  24. Panko J.M., Gaffney S.H., Burns A.M., Unice K.M., Kreider M.L., Booher L.E., Gelatt R.H., Marshall J.R., and Paustenbach D.J. Occupational exposure to benzene at the ExxonMobil refinery at Baton Rouge, Louisiana (1977–2005). J Occup Environ Hyg 2009: 6 (9): 517–529.

    CAS  Article  Google Scholar 

  25. Proctor D.M., Panko J.P., Liebig E.W., and Paustenbach D.J. Estimating historical occupational exposure to airborne hexavalent chromium in a chromate production plant: 1940–1972. J Occup Environ Hyg 2004: 1 (11): 752–767.

    CAS  Article  Google Scholar 

  26. Rappaport S.M., Selvin S., and Waters M.A. Exposures to hydrocarbon components of gasoline in the petroleum industry. Appl Ind Hyg 1987: 2 (4): 148–154.

    CAS  Article  Google Scholar 

  27. Runion H.E., and Scott L.M. Benzene exposure in the United States 1978–1983: an overview. Am J Ind Med 1985: 7: 385–393.

    CAS  Article  Google Scholar 

  28. Segal D. Exxon-Mobil Merger Wins FTC Approval. Washington Post. Washington, D.C, 1999.

    Google Scholar 

  29. US EPA. 40 CFR 61 Subpart FF: National Emission Standards for Benzene in Waste Operations, 1989.

  30. US EPA. Statistical Software ProUCL 4.0 for Environmental Applications for Data Sets with and without Nondetect Observations. Technical Support Center for Monitoring and Site Characterization, Las Vegas, NV, 2007. Available from: http://www.epa.gov/nerlesd1/tsc/tsc.htm.

  31. Verma D.K., and des Tombe K. Measurement of benzene in the workplace and its evolution process, Part I: overview, history, and past methods. Am Ind Hyg Assoc J 1999: 60 (1): 38–47.

    CAS  Article  Google Scholar 

  32. Verma D.K., Johnson D.M., Shaw M.L., and des Tombe K. Benzene and total hydrocarbons exposures in the downstream petroleum industries. Am Ind Hyg Assoc J 2001: 62 (2): 176–194.

    CAS  Google Scholar 

  33. Weaver N.K., Gibson R.L., and Smith C.W. Occupational exposure to benzene in the petroleum and petrochemical industries. In: Mehlman M.A. (Ed.). Advances in Modern Environmental Toxicology, Vol IV. Princeton Scientific Pub, 1983.

    Google Scholar 

Download references

Acknowledgements

We thank Paul Scott of ChemRisk for his assistance with the statistical analysis and Matt Le and Kathleen Navarro for their assistance in assembling and organizing the electronic data.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Shannon H Gaffney.

Ethics declarations

Competing interests

This work was funded by ExxonMobil, a firm that has been involved in the study of benzene for several decades. At least two of the authors have served or are likely to serve as expert witnesses for ExxonMobil on matters relating to industrial hygiene, exposure assessment, risk assessment, or toxicological issues related to benzene.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Gaffney, S., Panko, J., Unice, K. et al. Occupational exposure to benzene at the ExxonMobil Refinery in Baytown, TX (1978–2006). J Expo Sci Environ Epidemiol 21, 169–185 (2011). https://doi.org/10.1038/jes.2009.53

Download citation

Keywords

  • benzene
  • refineries
  • exposure assessment
  • industrial hygiene

Further reading

Search