Prioritizing Factors Associated with Thermal Stresses Imposed on Workers in Steel and Iron Casting Industries Using the Monte Carlo Simulation and Sensitivity Analysis

Abstract

Objectives: The aims of this study were to develop approaches for monitoring and prioritizing factors associated with thermal stresses imposed on workers in iron and steel casting industries, and to eventually purpose effective control strategies. Methods: The whole study was completed in the furnace areas of two steel casting and two iron casting plants, where the air temperature (T_a), radiant temperature (T_r), air velocity (V_a) and partial water vapor pressure (P_a) were measured continuously during two consecutive work cycles. Simultaneously, the metabolic rates (M) of all workers in the furnace area were also measured. Results: Using the WBGT as an index for screening purposes, our results suggest that all furnace area workers in both types of casting plants might experience severe heat stress. The predicted heat strain (PHS) model proposed by ISO 7933 was further adopted for detailed analysis from the physiological aspect. Through use of the Monte Carlo simulation and sensitivity analysis, both M and T_r were found to be the two most important factors associated with workers' thermal hazard. Therefore, two effective control strategies were suggested, including reducing workloads of workers and reducing radiant heat transmitting from furnaces to workplace environments. Conclusions: The approach developed in the present study would be beneficial to many other industries for initiating strategies to avert the thermal hazard imposed on workers.

(J Occup Health 2014; 56: 505–510)

Introduction

The four environmental factors of air temperature, radiant temperature, humidity and air movement¹⁾ and two personal factors of metabolic heat and worker clothing are primary factors affecting the extent of thermal stress imposed on workers. In 2000, the American Conference for Governmental Industrial Hygienists proposed a new decision-making process for managing thermal hazards²⁾. It first uses the wet bulb globe temperature (WBGT) as a heat stress index. When workers' actual work time exceeds that proposed by the ACGIH³⁾, a physiological-based analysis is required to further assess the workers' thermal hazard in more detail. For the physiological-based analysis, the International Standards Organization (ISO) published its first ISO 7933 standard in 1989⁴⁾ and revised it in 2004⁵⁾. This standard incorporates environmental factors, workload, and clothing ensemble to interpret the thermal stress imposed on workers based on a predicted heat strain (PHS) model. If excessive heat strain is found by using the PHS model, then direct physiological monitoring should be conducted for each individual worker.

It is true that the above decision-making protocol provides a useful tool for assessing worker's heat stress. But to the best of our knowledge, the above approach is still not widely used in the field, and its application for thermal hazard control still remains limited. Indeed, the ACGIH does provide guidelines for both “general control” and “job-specific control” with thermal hazard control at different stages during the decision-making process. However, it should be noted that both guidelines simply provide general concepts for thermal hazard control. Therefore, developing approaches for prioritizing factors associated with heat stress imposed on workers would be helpful for initiating more detailed strategies for thermal hazard control.

Steel and iron casting are important processes in many machinery manufacturing industries. In Taiwan, there are 73,847 workers currently employed by 1,279 iron and steel casting companies. These companies are involved in at least three thermal-related operations: (1) melting raw material, (2) transferring hot metal into molds and (3) mold cooling and shakeout. In these operations, furnace area workers associated with raw material melting are subjected to the most severe thermal stress^6–8).

In the current study, the decision-making process proposed by the ACGIH3) was adopted to assess thermal stresses imposed on furnace area workers in two iron casting and two steel casting plants. Although conducting direct physiological monitoring would provide a better understanding of workers' health effects, it does not allow us to identify the contribution of a specific factor to the thermal stress imposed on workers. Hence, direct physiological monitoring was not conducted in this study. Instead, we first used the WBGT index for screening purposes. If a worker was identified to be overexposed to heat stress through the screening process, a systematic approach, combining both the Monte Carlo simulation and sensitivity analysis, was adopted and applied to the ISO 7933 PHS model to prioritize the contributions of various factors to the heat stress imposed on the worker. Finally, the resultant prioritized factors were used to initiate effective thermal hazard control strategies for workers in both types of casting industry.

Materials and Methods

Sampling strategy

The furnace areas of two steel casting plants (denoted as S1 and S2, respectively) and two iron casting plants (denoted as I1 and I2, respectively) were selected in this study. Electric arc furnace (EAFs) were adopted for melting metal with capacities of 1.0 and 1.2 tons and operating temperatures of ∼1,650 and 1,800°C in S1 and S2, respectively. For the two selected iron casting plants, less automated cupola furnaces were used for melting iron with capacities of 1.0 and 1.0 tons and furnace operating temperatures of ∼1,700 and 1,750°C in I1 and I2, respectively. Based on our field observations, furnace workers stayed in the furnace area for ∼8 hours per day to continuously conduct their work cycles. During one work cycle, workers undertook a 20-min task for tapping melted metal or slag and then had a 10-min standby period in a standing posture (i.e., 67% work and 33% rest). All furnace workers had more than two years work experience in the furnace area and had acclimatized to the heat.

For environmental monitoring, a thermal environment monitor (measurable range 5 to 100°C, precision 0.01°C) (QT-15, Quest Technologies, Oconomowoc, WI, USA) was used to measure the air temperature (Ta), globe temperature (Tg) and corresponding WBGT. The relative humidity (RH) was determined by using an Assman and Sling psychrometer (measurable range 20 to 60°C, precision ± 3%) (TES-1361, TES, Taipei, Taiwan). The air velocity (Va) was measured by using an anemometer (measurable range 0 to 45.0 m/s, precision ± 3%) (AVM-07, TES, Taipei, Taiwan). This equipment allowed us to continuously collect data every 30 s. Field sampling was conducted during two consecutive work cycles. The mean radiant temperature (T_r) and the partial water vapor pressure (P_a) were calculated using the equations provided in ISO 7726⁹⁾. To properly characterize worker exposures, five monitoring points (uniformly distributed around the furnace) were selected from each casting plant for monitoring.

For personal sampling, 4–6 furnace workers were chosen from each selected plant to determine personal factors associated with the imposed heat stresses. Personal characteristics including age, height, and weight were determined (Table 1). The heart rate of each subject was measured using a real-time heart rate monitor (precision ± 1 beats per min) (Accurex Plus, Polar, Kempele, Finland) every 15 seconds during field sampling. The metabolic rate (M) of each worker was determined according to ISO 8996¹⁰⁾ based on the measured heart rate. All selected workers wore similar clothing ensembles, including a cotton brief, T-shirt, fitted cotton overalls, socks and heavy work shoes. No personal protective equipment was adopted in the present study. Based on the above information, the clothing insulation (I_cl) was calculated as 0.7 clo (0.11 m²K/W²) for all selected workers according to ISO 9920¹¹⁾. Each participant was briefed on the purpose, design, and experimental procedures of the study, and informed consent was obtained from each participant. The research protocol were reviewed and approved by the Research Ethics Committee of China Medical University and Hospital (No. CMUH102-REC3-006).

Table 1. Workers' personal background information (mean ± SD) for all selected casting plants

Industry	Steel casting		Iron casting
	S1 (n=6)	S2 (n=5)	I1 (n=6)	I2 (n=4)
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
	Age (year)	33.0 ± 2.5	33.8 ± 3.2	33.3 ± 4.3	34.0 ± 3.2
	Height (cm)	169.0 ± 5.9	175.0 ± 2.4	165.7 ± 4.5	173.0 ± 3.2
Body weight (kg)	69.5 ± 6.3	70.8 ± 2.8	63.3 ± 3.7	72.5 ± 3.0

Data analysis

The mean and standard deviation (SD) were calculated for measured environmental factors (i.e., T_a, T_g, RH, V_a, T_r, P_a and WBGT) and personal factors (i.e., M) for each selected plant. The resultant mean WBGT and M values were used to determine the allowable work/rest regimen according recommendations of the ACGIH³⁾. If a worker's actual work time exceeded the allowable time, the ISO 7933 PHS model⁵⁾ was used to further estimate the allowable exposure time (AETs) for each furnace area worker according to two criteria: the maximum heat storage (AET₁) and the maximum water loss (AET₂). In principle, AET₂ can be determined according to the maximum water loss allowable for either the 50th percentile or 95th population percentile. In this study, the 95th population percentile was determined for practical reasons. The lower value of the resultant AET₁ and AET₂ (i.e., (AET₁, AET₂)_min) was selected as the AET for any given furnace area worker. Both the mean AET and its corresponding 95% confidence interval were calculated for all furnace workers of each selected casting plant.

Monte Carlo simulation and sensitivity analysis

In principle, the extent of thermal stresses imposed on workers was affected by the four environmental factors of T_a, V_a, T_r and P_a and two personal factors of M and I_cl. Considering all the selected workers wore a similar clothing ensemble (i.e., similar I_cl), the factor I_cl was not considered for further sensitivity analysis. In the present study, Monte Carlo simulation and sensitivity analysis were adopted for prioritizing the sensitivities of the above five factors in determining AET based on the PHS model. The technique is widely used by professionals in many different fields, including finance, energy, manufacturing, engineering, transportation and the environment. The use of Monte Carlo simulation and sensitivity analysis requires for the application of a range of values—a probability distribution—to the selected any factor that has inherent uncertainty. Using thousands or more sets of random values or more form the probability functions of all selected factors, the distribution of possible outcome values can then be obtained. Finally, the effect of any given factor on the outcome results can be estimated through deterministic analysis. In this study, we used Monte Carlo simulation to develop 10,000 simulations for furnace area workers at each selected casting plant according to the measured T_a, V_a, T_r, P_a and M values obtained from field samplings. Sensitivity analysis was used to prioritize the sensitivities of the above five factors in determining AET based on the PHS model. The above approach was successfully used in our previous research¹²⁾ (Tsai et al. 2003) to identify the significant environmental factors affecting the efficacy of a thermal exposure chamber designed for assessing the thermal hazards for workers.

Results and Discussion

Characteristics of thermal stress-related personal and environmental factors

Table 2 summarizes thermal stress-related personal and environmental factors obtained from all the selected casting plants. The mean metabolic rates (M) for workers of the two selected steel casting plants (=147.0 and 141.0 W/m² for S1 and S2, respectively) were lower than those for the workers of the two selected iron casting plants (=158.9 and 159.1 W/m² for I1 and I2, respectively). It is known that workers of both types of casting plants conducted very similar work tasks (i.e., tapping melted metal and slag from a furnace). Therefore, it was expected that workers of both types of casting plants would share a very similar M. Because a much less automated cupola furnace was used in the iron casting plant than the electric arc furnace used in the steel casting plant, it was not so surprising to see that the workload required for workers of the former was lower than that of the latter. Nevertheless, we found that the metabolic rates for workers in both types of casting plant were rated in the same category (i.e., moderate workload).

Table 2. Environmental and personal factors (mean ± SD) obtained from the selected casting plants

Factors	Steel casting		Iron casting
	S 1	S 2	I 1	I 2
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
	Personal factor
	M (W/m2)	147.0 ± 17.6	141.0 ± 12.6	159.1 ± 22.1	158.9 ± 28.2
Environmental factors
Ta (°C)	36.6 ± 0.92	38.7 ± 4.76	41.5 ± 2.18	37.1 ± 3.66
Tg (°C)	38.9 ± 1.27	45.5 ± 6.45	42.9 ± 1.48	42.9 ± 3.61
RH (%)	49.7 ± 1.52	48.9 ± 2.26	38.2 ± 4.86	41.5 ± 7.91
Va (m/s)	0.09 ± 0.02	1.03 ± 0.56	0.30 ± 0.09	0.20 ± 0.12
Tr (°C)	41.1 ± 1.52	58.2 ± 9.15	43.6 ± 2.49	47.2 ± 3.70
Pa (kPa)	2.74 ± 0.12	2.75 ± 0.12	2.41 ± 0.27	2.71 ± 0.27
WBGT (°C)	31.8 ± 0.65	34.1 ± 3.32	33.9 ± 0.83	32.8 ± 0.83

Among the various environmental factors, only those with independent effects on heat stress are discussed in this study (i.e., T_a, V_a, T_r and P_a). We found that T_r was consistently higher than Ta, indicating that radiated heat was quite severe in both types of casting plants. However, S2 was found to have a much higher T_r (=58.2°C) than S1 (=41.1°C), even though both plants used the same type of furnace (i.e., electric arc furnace). The above results might have been obtained because the operating temperature for S2 (=1,800°C) was higher than that of S1 (=1,650°C) and the work area of S2 was much smaller than that of S1. Regarding the two selected iron plants, the T_r obtained from I2 was higher than that obtained from I1. This might have been due to the intrinsic differences in the amount of reflective material dispersed in both plants. The highest V_a was obtained at S2 (=1.03 ± 0.56 m/s), and this might have been because an industrial fan was installed in the workplace. The values for the rest of the plants were consistently lower than 0.5 m/s, which were in accordance with those found in most indoor workplaces^{13, 14)}. The mean P_a values for all the selected plants were quite low (=2.41–2.75 kPa), indicating that all the selected casting plants could be regarded as dry-heat environments. The lowest P_a values, found at I1, could have been due to it having the highest T_a among all the selected plants.

Assessing heat stresses imposed on casting industry workers

In the current study, the WBGT index was adopted as a heat stress screening index. As shown in Table 2, the mean WBGT for S1, S2, I1 and I2 were 31.8, 34.1, 33.9 and 32.8°C, respectively. Since all selected workers were rated to have a moderate workload and were acclimatized to heat, a work/rest time regimen of 25% work/75% rest was suggested according to the criteria proposed by the ACGIH in 2012³⁾. Comparison of this with the actual work/rest time regimen of all the selected workers (i.e., 67% work/33% rest) indicated that all the casting workers might suffer from severe heat stresses. Therefore, the PHS model was adopted for further detailed analysis according to the decision-making process suggested by the ACGIH in 2012³⁾.

Table 3 shows the means and their corresponding 95% confidence intervals (95%CIs) for both AET1 (i.e., based on the maximum heat storage in the human body) and AET2 (i.e., based on the maximum water loss) for workers of each selected casting plant. For workers of the two selected steel casting plants (i.e., S1 and S2), the AET₂ values (=273.2 and 261.2 minutes, respectively) were consistently lower than the AET₁ values (=353.2 and 480 minutes, respectively), indicating the predominance of the maximum water loss criteria (i.e., AET₂) in determining AETs. On the other hand, the AETs for workers of both of the selected iron casting plants (i.e., I1 and I2) were determined based on the maximum heat storage criteria (i.e., AET₁=57.8 and 144.0 minutes, respectively), rather than AET₂ (=203.2 and 244.3 minutes, respectively). Yet it is true that the environmental conditions for both types of casting plants were quite similar (i.e., dry-heat environment). Furthermore, it is known that the workers in the iron casting plants had higher M values (or workload) than the workers in the steel casting plants (see Table 2). At this stage, whether the higher M (or workload) had a more significant effect on heat storage than water loss warrants further investigations (because of the complexity in physiological responses to various heat-stress-related factors).

Table 3. Estimated means and their corresponding 95% confidence interval (95% CI) for both AET₁ and AET₂ for furnace workers of each selected casting plant

Industry		Steel casting		Iron casting
		S 1	S 2	I 1	I 2
		AET1	Mean 95% CI	353.2 222.9–480.0	480.0 —*	57.8 28.1–87.6	144.0 36.7–251.3
AET2	Mean 95% CI	273.2 248.5–297.8	261.2 247.0–275.4	203.2 121.4–285.0	244.3 228.8–259.7

*  A 95% CI range could not be estimated since the mean AET₁=480 min.

We found that the workers of S1, S2, and I1 and I2 allowed for 56.9% (=273.2/480), 54.4% (=261.2/480), 12.0% (=57.8/480) and 30.0% (=144/480) work time according to the resultant AETs obtained from the PHS model. The above results further confirm that workers of all the selected plants might suffer from severe heat stresses under their current designated work/rest time regimens (i.e., work time =67%). The work/rest time regimens obtained from the PHS model were somewhat different from that obtained from the WBGT criterion (i.e., 25% work/75% rest). Our results further confirm that the WBGT index can be only regarded as a screening tool for identifying subjects with excessive heat stresses.

The 95% CI of the resultant AETs for workers of each of the selected casting plants exhibited quite a wide range (Table 3). Great variations can also be observed in workloads (or M) for workers of the same casting plant (see Table 2). At this stage, whether the dramatic variation in AETs was caused by the inherent variation in M warrants further investigations.

Sensitivity analysis of various factors associated with heat stresses imposed on workers

In principle, the main purpose of occupational hygiene is not simply for analysis of exposure for its own sake but is also for hazard prevention. Therefore, it is important to use the collected information to prioritize the contributions of various factors to heat stresses imposed on workers. Through the Monte Carlo simulation and sensitivity analysis, we were able to prioritize the sensitivities of the above five parameters in estimating AETs. As shown in Fig. 1, we found that a decrease in the magnitude of all selected parameters, with the exception of Va, would lead to an increase in the magnitude of the AETs. The above results were theoretically plausible since (1) a decrease in Ta and Tr would reduce the heat being transferred from the environment to the human body, (2) a decrease in Pa would increase the evaporation of water during the sweating process and lead to a decrease in heat storage in the human body and (3) a decrease in M would lead to a decrease in heat storage in the human body because less internal heat would be generated. However, the effect of Va on AETs might be worth further discussion. In principle, an increase in V_a would result in an increase in the evaporation rate (i.e., a decrease in heat storage) due to a low P_a. Conversely, an increase in Va might also result in an increase in the convection heat transferred from the environment into the human body (i.e., an increase in heat storage) when the Ta is higher than the body temperature. In the current study, we found that an increase in V_a would result in a decrease in AETs, suggesting that the net effect of V_a could be more significant on evaporation than on convection.

Fig. 1

Sensitivity (dimensionless) analysis for assessing the effects of the four environmental factors (including T_a, P_a, V_a, and T_r) and one personal factor (M) on determining AETs.

Figure 1 shows that M was the most sensitive factor in determining AETs for the four selected casting plants. The above results further confirm our expectation (i.e., the variation in AETs was caused mainly by the inherent variation in M). Figure 1 also shows that Tr was the second sensitive factor for S1, S2 and I2. However, it ranked third for I1. Again, due to the complexity (or competition) of the various factors affecting the heat transfer process between the ambient environment and human body, we might not be able to explain why the role of T_r in I1 was different from that in the other three casting plants. The above results clearly indicate that reduction of workload and control of T_r transmission from the furnace to the workplace environment are the two most important measures to avert the thermal hazard for furnace area workers.

Conclusions

For casting industries, the use of a less automated cupola furnace will result in workers' workloads being higher than those of workers using an electric arc furnace. The mean P_a values measured for all the selected casting plants were quite low; therefore, their furnace areas could be regarded as a dry-heat work environments. For steel casting plant workers, the predominant criteria for determination of their AETs were predominated by the maximum water loss criteria. On the other hand, the AETs for iron casting plant workers were determined mainly based on the maximum heat storage criteria. The work/rest time regimens obtained from the PHS model were somewhat different from that obtained from the WBGT criterion. It is concluded that the WBGT index can be only regarded as a screening tool for identifying subjects with excessive heat stresses. Through Monte Carlo simulation and sensitivity analysis, we found that a decrease in T_a, T_r, P_a and M would lead to an increase in AETs. On the other hand, an increase in Va would result in a decrease in AETs, indicating that the net effect of V_a could be more significant on evaporation rather than on heat convection. In conclusion, reduction of workload and control of T_r transmission from the furnace to the workplace environment are the two most important measures to avert the thermal hazard for furnace area workers.

Acknowledgment:    The authors wish to thank the Institute of Occupational Safety and Health (IOSH) of the Council of Labor Affairs in Taiwan for funding this research project. The contribution of C.-P. Chen (co-author) is as important as that of the corresponding author in this research work.

References

• 1)   Parsons  K. Human thermal environments. In: The effects of hot, moderate and cold temperatures on human health, comfort and performance. 2nd ed., London: Taylor & Francis; 2003.
• 2)  American Conference of Governmental Industrial Hygienists (ACGIH) Worldwide, Threshold Limit Values for chemical substances and physical agents and biological exposure indices. Cincinnati (OH): ACGIH Worldwide; 2000.
• 3)  American Conference of Governmental Industrial Hygienists (ACGIH) Worldwide, 2012. Threshold Limit Values for chemical substances and physical agents and biological exposure indices with 7^th Edition Documentation. Cincinnati (OH): ACGIH Worldwide; 2012.
• 4)  ISO 7933 “Hot Environments Analytical Determination and Interpretation of Thermal Stress Using Calculation of Required Sweat Rate”. Geneva: International Standards Organization; 1989.
• 5)  ISO 7933 “Ergonomics of the Thermal Environment -- Analytical Determination and Interpretation of Heat Stress Using Calculation of the Predicted Heat Strain”. Geneva: International Standards Organization; 2004.
• 6)   Chen  M-L,  Chen  C-J,  Yeh  W-Y,  Huang  J-W,  Mao  I-F. Heat stress evaluation and worker fatigue in a steel plant. Am Ind Hyg Ass J 2003; 64: 352-9.
• 7)   Jay  O,  Kenny  GP. Heat exposure in the Canadian workplace. Am J Ind Med 2010; 53: 842-53.
• 8)   Chung  SH,  Lee  LH. The Establishment of the Guidance Technology for Occupational Safety and Health in the Iron and Steel Industry(I). Taipei (Taiwan): Institute of Occupational Safety and Health; 2011 (in Chinese).
• 9)  ISO 7726: “Ergonomics of the Thermal Environments -- Instruments and Methods for Measuring Physical Quantities”. Geneva: International Standards Organization; 1998.
• 10)  ISO 8996: “Ergonomics of the Thermal Environments -- Determination of Metabolic Rate”. Geneva: International Standards Organization; 2004.
• 11)  ISO 9920: “Ergonomics of the Thermal Environment -- Estimation of the Thermal Insulation and Water Vapout Resistance of a Clothing Ensemble”. Geneva: International Standards Organization; 2007.
• 12)   Tsai  PJ,  Lo  CL,  Sun  YM,  Juang  YJ,  Liu  HH,  Chen  WY,  Yeh  WY. Evaluating the efficacy of a thermal exposure chamber designed for assessing worker's thermal hazard. J Occup Health 2003; 45: 153-9.
• 13)   Berry  RD,  Froud  S. An investigation of wind conditions in workplaces to assess their effect on the quantity of dust inhaled, Health and Safety Executive Research Laboratory and Services Division, No: IR/L/DS/89/3: 1989.
• 14)   Sleeth  DK,  Vincent  JH. Proposed modification to the inhalable aerosol convention applicable to realistic workplace wind speeds. Ann Occup Hyg 2011; 55: 476-84.