Human error identification and risk assessment in loading and unloading of petroleum products by road trucks using the SHERPA and fuzzy inference system method

Human error constitutes one of the primary causes of accidents, particularly in the context of loading and unloading operations involving road trucks, especially those carrying petroleum products. The process of identifying and evaluating human errors within these operations involves several key steps. Initially, all sub-tasks associated with loading and unloading are meticulously identified and analyzed utilizing Hierarchical Task Analysis (HTA), achieved through direct observation, document examination, and interviews. Subsequently, potential human error modes within each task are delineated using the Systematic Human Error Reduction and Prediction Approach (SHERPA). Finally, essential data for determining the criticality, probability, and severity of each error are gathered through expert elicitation and the application of Fuzzy Inference Systems (FIS). Through the analysis of SHERPA worksheets, a total of 37 errors during loading operations and 14 errors during unloading operations of petroleum products were identified. Among these errors, the predominant category during loading operations was action errors, comprising 31 instances, while communication errors were the least frequent, occurring only twice. Similarly, action errors were most prevalent during unloading operations, constituting 13 instances. These errors were further categorized and ranked based on their risk levels, resulting in 27 levels for loading operations and 12 levels for unloading operations. The consistent occurrence of action errors underscores the need for implementing control measures to mitigate their frequency and severity. Such strategies may include periodic training sessions to reinforce proper work procedures and the development of monitoring checklists, among other interventions.


Introduction
In the petroleum industry, ensuring safety is crucial due to the potential severe impacts on human life and the environment resulting from accidents.Establishing comprehensive safety measures is essential to ensure the safe and efficient operation of a refinery.Consequently, a critical objective is to minimize or eliminate risks in various areas of the refinery, including loading terminals [1].Loading terminals play an important role in receiving petroleum products from both domestic and foreign oil refineries, and subsequently distributing them by tank trucks to meet industrial and individual needs.The distribution process involves loading the products into tanks of truck trucks, transporting them, and unloading them at the destination tanks [2,3].
The transportation of petrochemicals by road truck is a common practice globally.However, it is important to note that this activity carries inherent risks and can be a significant factor contributing to major accidents [4,5].Not paying attention to the loading threshold, overfilling the tank, truck spill, rupture of the filling and unloading pipe, not paying attention to the equipment used when pumping fuel to the truck/unloading tank, followed by fire and explosion during loading and unloading of the truck have occurred, which has increased awareness of safety issues in the transportation of products by road trucks [1,6].Statistics show that at least 2 major fires occur in oil refineries and loading terminals in the worldwide every year, and these incidents mostly occur in petroleum product loading platforms [7].According to the International Truck Owner Pollution Federation (ITOPF) report, it was determined that there is the highest probability of explosion and fire during loading and unloading operations [8].The analysis of incidents over two years of release of petroleum products in seven American states by the Agency for Toxic Substances and Disease Registry (ATSDR) showed that out of a total of 1369 incidents related to petroleum products, 512 injuries and 36 deaths were recorded [9].Advances in petroleum industry programs may have significant social and economic benefits.However, Risks in the transportation of petroleum products have potentially devastating effects on the natural environment, vulnerable ecosystems, and problems associated with recovery and clean-up operations [10][11][12][13].Before 1956, there were limited laws to protect the control of contamination from petroleum products.These included the Refusal Act of 1899, the oil Pollution Act of 1924, and the Truck Act.However, over time, the importance of protecting natural and human resources to prevent pollution of petroleum products became more apparent [14].
Since some of the main steps of loading and unloading operations are still carried out by operators, special attention should be given to potential human errors during such operations.Human errors play a significant role in the majority of accidents that occur in the petrochemical industry [15][16][17].Statistics recorded by the National Toxic Substances Incident Program (NTSIP) in the United States, showed that more than 40 % of petroleum accidents are caused by human practices, and human error is the most common cause of these accidents [9,18].Ignoring the human element in the workplace not only result in loss of human performance, but also increases in the number of injuries and harm, which leads to significant financial losses [19,20].Accordingly, anticipating and preventing potential human errors in critical operations conducted in the petrochemical industry is of paramount importance.
The Systematic Human Error Reduction and Prediction Approach (SHERPA), is one of the most practical methods for error classification in identifying valid errors related to a sequence of human activities.SHERPA belongs to a family of human error identification tools that have a psychological approach [21][22][23].In their studies, Kirwan (1998) and Stanton (2006) mentioned the relative advantage of the SHERPA technique as ranking highest overall in a comparative study among human error detection techniques, including HEART, THERP, SLIM, and CREAM techniques, based on different performance criteria.Some advantages of this method include its ease of implementation and short execution time, determining the level of risk, identifying the consequences of errors, and ultimately providing control measures [24,25].This systematic approach for predicting human errors was introduced by Embrey in 1986 [26] and is one of the most effective methods in various safety conditions, including the chemical industry [27], petrochemical [28][29][30], surgery [31,32], aviation [33,34], etc.
However, despite the high capability of the SHERPA technique in identifying and predicting human errors, it has not been able to quantitatively assess the risks associated with these errors and requires integration with other methods.The utilization of uncertain knowledge and subjective judgment creates a significant challenge in the application of this qualitative technique [35].Uncertain knowledge arises due to a lack of awareness or insufficient awareness of the work process, the modifications in industrial methods and processes, and unreliable information.Using decision-based fuzzy models provides a solution for such a challenge.Fuzzy models based on the decision-making system can be mentioned as fuzzy AHP techniques, fuzzy TOPSIS, and FIS MATLAB tools.Studies have shown that the FIS method is superior to other methods based on decision-making systems in terms of validation and reduction of computing time.FIS is robust when dealing with uncertain, imprecise, and qualitative information, especially in situations where there is ambiguity surrounding a subject [36][37][38].The probabilities and severities employed in the risk assessment process possess a certain level of uncertainty.By incorporating a fuzzy logic system, this methodology ensures logical consistency in the application of the conventional qualitative risk matrix approach by the risk assessor, achieved through the implementation of a fuzzy rule base.Based on the mentioned content, the objective of the current research is to identify and assess human errors in the loading and unloading operations of petroleum products by road trucks using the SHERPA and FST techniques.

Methods
This descriptive cross-sectional study was conducted at the loading terminal and fuel station of a petroleum products distribution company in Hamadan, Iran.The methodology employed in the study consists of three main phases (Fig. 1).
Phase 1: Identification of human errors in loading and unloading operations using the SHERPA technique.Phase 2: Estimation of the risks associated with the identified errors using FIS.Phase 3: Prioritization of the identified errors based on their risk numbers.

Human error identification
The process of human error identification involves identifying all the tasks and sub-tasks associated within the loading and unloading operations of petroleum products.This phase includes regular and frequent observations of the operations, reviewing working conditions, and analyzing past data to gain a comprehensive understanding of the operation.The hierarchical task analysis  (HTA) technique is then used to divide the tasks and sub-tasks into work components.Upon drawing the HTA, the SHERPA technique is employed to identify the types of human errors in each subtask.Table 1 demonstrates the categorization of human errors into five categories: action errors, checking errors, retrieval errors, communication errors, and selection errors [39,40].Using the SHERPA technique, possible errors are listed for each stage of the HTA.After determining the error modes, outcome analysis, error recovery analysis, error probability analysis, and error severity analysis are performed based on the SHERPA worksheet [41].Additionally, a FIS is utilized to determine the probability and consequences of each error, and suggestions for reducing each error are provided.Finally, the error modes are quantified in terms of associated risk.

Expert elicitation
To calculate the risk level of the identified errors in the loading and unloading of petroleum products, information regarding the probability of occurrence and the severity of the consequences of these errors is required.However, due to the limited available knowledge and the absence of documentation and records on the probability and occurrence of human errors, as well as the uncertainty inherent in human verbal judgments, additional methods are utilized.
Fuzzy logic and expert elicitation techniques are employed to address these challenges [42,43].A carefully selected group of experts with diverse experiences, relevant expertise in the field of human error and operations of loading and unloading petroleum products, varying ages, and educational backgrounds play a crucial role in providing their judgments.These experts evaluate the probability of occurrence and severity of the identified errors based on their extensive knowledge in the field.Based on Table 2, each expert is assigned a specific score for each characteristic, considering five specific characteristics obtained from the experts.The job position is assigned a score ranging from 1 to 5, the years of work experience receive a score ranging from 1 to 5, the educational qualifications are given a score ranging from 1 to 5, and the age of the expert is considered for a score ranging from 1 to 4. Each expert accumulates scores for these five characteristics, resulting in a sum of five values for each expert.The weight coefficient for each expert is determined using the following formula:.The weight coefficient(WC Expert )for each expert is calculated by dividing the sum of scores (S Expert,j ) for each feature by the total sum of scores for all features across all experts.This normalization ensures that the weight coefficients reflect the relative importance of each expert's scores within the overall context.The numerator captures the individual expert's scores for all features, while the denominator represents the total sum of scores for all experts and all features.This formula provides an effective method for deriving the weight coefficients, enabling robust analysis and decision-making processes based on the aggregated scores of the features.By utilizing this approach, the expertise and qualities of each expert are effectively integrated into the risk assessment process, leading to a more accurate evaluation of the risk level associated with the identified errors in the loading and unloading of petroleum products.
The risk assessment matrix employs five fuzzy language variables with linguistic terms, including "very low (VL)," "low (L)," "medium (M)," "high (H)," and "very high (VH)," to classify the probability and severity of the consequences associated with identified errors (Table 3).The severity of consequences and frequency are categorized and scaled based on the specific activity or processes being evaluated, taking into account the nature of the risks involved.
For simpler risk assessments, a 3 × 3 cells matrix can be utilized, while larger structures like process plants may require a 5 × 5 or even a 7 × 4 matrix.In this particular study, a 5 × 5 cells risk matrix is recommended, indicating the presence of five distinct levels for both probability and severity of consequences.
The relationship between frequency, severity, and risk categories is determined by risk-based engineering rules.These rules establish the correlation between different levels of frequency and severity, enabling the assignment of appropriate risk categories.
In conclusion, the fuzzy ratings in the risk assessment matrix are determined by categorizing and scaling the severity of consequences and frequency, utilizing a 5 × 5 cells risk matrix, and applying risk-based engineering rules to establish the relationship between frequency, severity, and the corresponding risk categories [44,45].

Fuzzy inference system
FIS are based on a combination of fuzzy if-then rules, which assign fuzzy inputs to fuzzy outputs [46].FIS is a method rooted in traditional logical reasoning using 0 and 1, and it was initially proposed by Zadeh in 1965.In 1975, Mamdani and Assilian applied fuzzy logic and fuzzy reasoning to control a steam engine, demonstrating its practical use in real-life applications.Since then, fuzzy theory has found wide application in industrial processes, including the petroleum industry [47].There are various fuzzy combination methods available for fuzzy inference in the Mamdani fuzzy model.In this study, the max-min Mamdani combination method was employed [46,48].Subsequently, the center of area (COA) calculation method was used for the defuzzification process, which converts fuzzy sets into crisp values [49,50].
In this study, FIS was utilized to analyze the criticality of the identified errors by determining the risk number associated with each error.This involved combining the probability of occurrence of the error with the severity of the damage caused by the error.

Application of the methodology: a case study
Road trucks are vehicles used worldwide for transporting petroleum products in the petrochemical industry.For the purposes of this study, the loading operations of petroleum products by road trucks at a loading terminal, which serves as a distributor of various petroleum products such as gasoline, gas oil, and kerosene, and manages fuel stations, have been considered.The loading terminal consists of 10 gasoline loading platforms, 10 gas oil loading platforms, and 4 kerosene loading platforms.Each loading platform associated with a specific product is identified by a distinctive color, with red indicating gasoline loading platforms, yellow indicating gas oil loading platforms, and blue indicating kerosene loading platforms.Fig. 2 illustrates a visual representation of a loading platform.The unloading operations are also considered at a fuel station located in Hamadan city.This fuel station is equipped with four 60,000-L tanks for petroleum products, consisting of two tanks for gasoline and two tanks for gas oil.The operational staff at this fuel station consists of six individuals.The unloading of petroleum products is carried out by road trucks, facilitated by the station's personnel (Fig. 3).

Human error identification
Based on observations of work operations, worksite analysis, expert opinions, and literature review, a HTA was conducted for the loading and unloading operations of petroleum products.Table 4 provides details of the HTA for the loading operation, which consists of six sub-sections.Sub-sections 1 to 6 respectively describe the following tasks: checking the documents of the truck driver, preparing the truck for loading, loading of petroleum products, conducting quality control, sealing oil trucks, and checking driver information and issuing forms.The sub-section "preparing the truck for loading" is further divided into three tasks, while "sealing" is divided in to seven tasks.Additionally, the task of washing the truck is divided into 12 sub-tasks, and the task of sinking the truck is divided into seven sub-tasks.Placement of the truck in the loading line -Place the truck in the designated route for loading, as indicated by markings and guide signs.
-Place the truck parallel or perpendicular to the loading line and maintain a suitable distance from other trucks.

2.2
Check the truck bill of lading and record information -Verify the truck's bill of lading to ensure that the truck aligns with the specified cargo or product mentioned in the bill.-Record the relevant information from the bill of lading in the logbook

2.3
The oil tank is washed 2.3.1 Place the truck in the washing area Drive the truck to the designated washing area and place it at the specified location within the washing zone.

2.3.2
Turn off and pull the handbrake -Switch off the engine of the truck.
-Engage the handbrake to secure the truck 2.3.3 Connecting the earth connection cable to the truck Attach the earth connection cable to the appropriate grounding point of the truck.

2.3.4
Loading volume of 400-300 L of kerosene Fill the tank with 400-300 L of kerosene, taking into account the capacity and requirements of the tank.

2.3.5
Move the truck for 5 min in the specified direction Start the engine and drive the truck in the specified direction within the washing area for a duration of 5 min, allowing the kerosene to mix with the tank during the truck's movement, effectively cleaning the tank.

2.3.6
Placement of the truck in the unloading platform After the 5-min movement for cleaning the truck's tanks, drive the truck to the designated unloading platform for the disposal of the kerosene in the tanks.

2.3.7
Turn off and pull the handbrake -Switch off the engine of the truck.
-Engage the handbrake to secure the truck 2.3.8 Connecting the earth connection cable to the truck Attach the earth connection cable to the appropriate grounding point of the truck.2.3.9 Connect the drain pipe to the tank Connect the drain pipe sequentially to each tank on the truck containing kerosene for cleaning purposes, allowing the effective discharge of kerosene from the tanks.

2.3.10
Open the drain valve Carefully open the drain valve of each tank that the drain pipe is connected to, allowing the contents of the tank to be freely released through the drain pipe.
(continued on next page) M.M. Aliabadi et al.
Similarly, the unloading operation of petroleum products is divided into two sub-sections: checking the driver's license and completing the product issuance and unloading form.These details are presented in Table 5.
It states that each stage of the HTA analysis was conducted by a team of experts to preliminarily identify errors based on tasks, code errors, and error types separately for the loading and unloading operations.The SHERPA technique was utilized for this purpose, and Close the drain valve After completing the discharge and disconnecting the drain pipe, close the drain valve on the truck's tank.

2.4
Removing the products remaining in the tanks of oil trucks 2.4.1 Placement of the truck in place The driver should place the truck in the designated place to remove the remaining products from the truck.

2.4.2
Turn off and pull the handbrake -Switch off the engine of the truck.
-Engage the handbrake to secure the truck 2.4.3 Connecting the earth connection cable to the truck Attach the earth connection cable to the appropriate grounding point of the truck.

2.4.4
Connect the drain pipe to the truck tank Connect the drain pipe to each tank of the truck in sequence to drain the tanks effectively.2.4.5 Open the drain valve Carefully open the drain valve of each tank that the drain pipe is connected to, allowing the contents of the tank to be freely released through the drain pipe.

2.4.6
Separate the drain pipe from the tank Disconnect the drain pipe from the tank's outlet once the draining process is complete.

2.4.7
Close the drain valve After completing the discharge and disconnecting the drain pipe, close the drain valve on the truck's tank.

2.5
Operator awareness of the truck entering the loading platform The truck's entry to the loading platform is communicated to the operator of the platform.

3
Loading of petroleum products 3.1 Placement of the truck on the loading platform The truck is placed in the specified loading platform.

3.2
Parking the truck by the driver on the loading platform -The driver aligns the truck with the designated parking area on the loading platform.
-The driver ensures the truck is properly positioned and stationary on the loading platform.

3.3
Turn off the truck and pull the hand brake and place the wedge next to the tires -Switch off the engine of the truck.
-Engage the handbrake to secure the truck.
-The driver places wedges under the tires to prevent sudden movement of the truck.

Presence of firefighter next to the loading platform
The firefighter is placed next to the loading platform to prevent accidents.

3.5
Check the bill of lading -Verify the truck's bill of lading to ensure that the truck aligns with the specified cargo or product and matches the type of fuel required by the loading platform.-Record the relevant information from the bill of lading in the logbook.

3.6
Check the truck drain valves by the loading operator All truck drain valves are inspected by the loading operator to ensure that no residue is present in the tank.

3.7
Connecting the earth connection cable to the truck Attach the earth connection cable to the appropriate grounding point of the truck.

3.8
Open the truck tank lid Open the truck's inlet tank lid for loading the petroleum cargo.

3.9
Inserting the loading arm into the truck tank The loading arm is placed at the first inlet tank of the truck for loading the petroleum cargo.

3.10
Locking the loading arm The loading arm lever is locked to prevent sudden detachment of the arm from the truck tank.

3.11
Setting the Volumetric device the Volumetric device is adjusted according to the volume written in the bill of lading.

3.12
Pump the product into the tank The start button of the volumetric device is pressed to pump the petroleum product into the tank.

3.13
Place the opening of the loading arm in the aluminum bucket After loading the product, the loading arm lever is detached from the tank and placed inside an aluminum bucket attached to the loading arm.

Measurement of product level in tank by Brass Rod
The operator manually measures the level of the loaded petroleum product in the tank using a brass rod.

3.15
Close the loaded tank lid The operator closes the truck tank lid, which contains the loaded petroleum product.

3.16
Separate the earth connection cable from the truck After loading all the truck's tanks, disconnect the earth connection cable from the truck.

3.17
Loading data recording Record all the information of the loaded content in the truck's tanks.

Quality Control
The information of the bill of lading and the loading performed by the quality control personnel is systematically recorded.5 Sealing oil trucks 5.1 Placement of the truck at the place of sealing The truck is placed in the sealing area by the driver and parked.

5.2
Turn off and pull the handbrake -Switch off the engine of the truck.
-Engage the handbrake to secure the truck 5. 3 Check documents and record information The bill of lading information is checked by the sealing operator.

5.4
Receive numbered seals by the driver Numbered seals are provided to the driver by the sealing personnel for securing the truck.

Product level measurement in the inlet tank by Brass Rod
To ensure the product level in the tank, the truck tanks are measured and checked by a brass rod.

Installation of seals on loading and unloading tanks
The sealing nuts are placed by the sealing operator on all the valves of the loading and unloading tanks of the truck.

5.7
Exit the truck from the sealing area After installing all the sealing nuts, the truck is taken out of the sealing place by the driver.

Check driver information and issuance form
The truck and driver information is reviewed by the personnel in the exit section, and permission to exit is granted to the driver for the truck's departure from the loading terminal.

M.M. Aliabadi et al.
Tables 6-8 provide the relevant information.According to these tables, a total of 37 human errors were identified for the six subsections in the loading operation, while a total of 14 human errors were identified for the two sub-sections in the unloading operation.These findings are summarized in Table 9.Oil truck park on the unloading platform The truck is placed in the specified unloading platform.

2.2
Turn off the truck and pull the hand brake and place the wedge next to the tires -Switch off the engine of the truck.
-Engage the handbrake to secure the truck.
-The driver places wedges under the tires to prevent sudden movement of the truck 2. 3 Connecting the earth connection cable to the truck Attach the earth connection cable to the appropriate grounding point of the truck.

2.4
Connect the drain hose to the truck Connect the drain hose to each tank of the truck in sequence to drain the tanks effectively.

2.5
Open the drain valve Carefully open the drain valve of each tank that the drain hose is connected to, allowing the contents of the tank to be freely released through the drain hose.

2.6
Disconnect the drain hose from the tank Disconnect the drain hose from the tank's outlet once the draining process is complete.

2.7
Insert the drain hose into the aluminum bucket After completing the discharge and disconnecting the drain hose, place the drain hose inside an aluminum bucket.

2.8
Separate the earth connection cable from the truck After unloading all the tanks of the truck, disconnect the earth cable from the truck.

Record unloading information
The truck and driver information is reviewed by the personnel in the exit section, and permission to exit is granted to the driver for the truck's departure from the fuel station.

Fuzzy risk assessment
In order to assess the risk associated with the identified errors, the experts were assigned weights based on Table 10.Subsequently, the experts ranked each error in terms of its probability of occurrence and the severity of its consequences using predetermined linguistic variables.Expert opinions on all identified errors can be found in Tables 11 and 12.
Then, if-then fuzzy rules were developed to represent the relationship between input and output variables based on the insights and expertise of the experts.The collected data, including the probability score and severity of the identified errors, were input into the FIS (Fuzzy Inference System) and analyzed by the fuzzy inference engine using the if-then rules.Through the defuzzification process, the fuzzy values obtained were converted into a risk number.The COA method was commonly used for defuzzification to establish the relationship between input and output variables.The structure of fuzzy reasoning for determining the risk number is illustrated in   -.All calculations were performed using the commercial software MATLAB R2018b Finally, the errors were ranked based on the risk number.Tables 11 and 12 provide information on the total risks and the number of calculated risks for the loading and unloading of petroleum products.A low level of risk corresponds to a situation with a low probability of occurrence and low severity, whereas a high level of risk indicates a situation with a very high probability of occurrence and severity.

Table 11
Probability and severity of all loading operation errors based on expert opinions.

Table 12
Probability and severity of all unloading operation errors based on expert opinions.

Results
In this study, a total of 37 human errors were identified in loading operations and 14 human errors in unloading operations through HTA analysis.Specifically, in loading operations, there were 31 errors categorized as action errors, 4 errors categorized as checking errors, and 2 errors categorized as communication errors.Similarly, in unloading operations, there were 13 action errors and 1 checking error.The number and categories of errors are visually presented in Fig. 6.
In the loading operation, most of the action errors occur in the form of performing the operation incompletely (A9) and omitted (A8).In checking errors, the check is usually performed incompletely (C2), and in communication error, information is not exchanged (I) or incorrect information is exchanged (I2).In the unloading operation, most of the action errors occur in the form of performing the operation omitted (A8).In checking errors, the check is usually performed incompletely (C2).Also, according to Fig. 7, in petroleum product loading operations, the number of unrecoverable errors included 21 action errors (56.75 %), 3 checking errors (8.10 %) and 1 communication error (2.70 %).
Observing the errors, it was found that in both operations, most of the errors are not effectively addressed or corrected, as per the definition of error recovery in the SHERPA technique, which entails taking actions in subsequent stages to restore the system to its initial state.Therefore, these errors were considered unrecoverable, as they could not be rectified or mitigated in the subsequent stages of the operations.Furthermore, a significant number of unrecoverable errors in both operations were identified as action errors.Considering this definition, the identification of a significant number of unrecoverable errors in both operations highlights the failure to implement the necessary actions for error recovery.These errors, once they occurred, could not be effectively addressed or corrected in the subsequent stages, resulting in prolonged or irreversible consequences.
To address this issue and minimize the occurrence of unrecoverable errors, it is crucial to implement proactive measures, such as employee training programs, the utilization of checklists, installing audio and visual alarms, developing written work instructions, and paying attention to the work-rest period of employees.By implementing these measures, operators can improve the ability to recover from errors and prevent their escalation into unrecoverable situations.
Failure to implement these preventive measures can lead to severe consequences, such as explosions, petroleum product leakage, and failure to load or unload products according to the required volume.Therefore, it is imperative to prioritize the identification and mitigation of unrecoverable errors to enhance safety and operational efficiency in petroleum terminals.The risk assessment by the FIS showed that in the operation of loading petroleum products (Table 11), 27 ranks were determined for the risk level of identified errors.A number of errors also scored the same risk score.The highest risk score was related to error, checking the drain valves is omitted (error no 16).And the lowest risk score related to errors, removal of the remaining product in the truck is incomplete (error no 7) and exit the loading arm before the loading was complete (error no 18).Also, in the operation of unloading the petroleum product (Table 12), 12 ranks were set for the risk level of identified errors.The highest risk score related to the error, the disconnection of the earth connection cable from the truck is omitted (error no 12).And the lowest risk score for the error was the driver is omitted to turn off the truck (error no 3).

Discussion
Human error is a notable concern in various industries as it significantly impacts the occurrence of accidents.Additionally, a crucial strategy for improving safety performance in the petrochemical industry reducing human error [9,51].Tt is crucial to understand that road truck transfers require significant operator engagement, occur frequently, and pose a considerable risk of serious accidents.As a result, the evaluation and management of human factors play a vital role in ensuring the safety and efficiency of these operations [4].This study presents a human error evaluation method and applies it to petroleum product loading and unloading terminals.Initially, task analysis was performed hierarchically for both operations.Subsequently, the researchers completed the worksheets using the SHERPA technique.
In the petroleum product loading operation, the main errors identified were 31 action errors (83.78 %).checking errors were in the second place with the number of 4 errors (10.81 %) and communication errors were in the third place with the number of 1 error (5.4 %).There were no retrieval errors and selection errors in this operation.Similarly, in the oil product unloading operation, the main errors identified were 13 action errors (92.85 %).checking errors were also in the second place with the number of 1 error (7.14 %).Also, no retrieval errors, selection errors, and communication errors were detected.To reduce the identified errors, solutions such as employee training, preparing a checklist, installing audio and visual alarms, and paying attention to the work-rest period of employees were suggested.Similarly, the study conducted by Sabbaghpoor Azariyan et al., with the aim of identifying, analyzing and management of human errors in the filtration unit of the oil refinery showed that out of 181 identified errors, 154 errors were action errors, 24 errors were checking errors, and 2 errors were communication errors.And 1 error was one of the selection errors.No retrieval errors were detected.Also, to reduce the identified errors, employee training was suggested as a corrective solution [52,53].
SHERPA is considered as a comprehensive and robust technique for detecting and predicting human error.However, it is a qualitative method and has significant uncertainties.In this study, fuzzy set theory was used as a complementary quantitative approach for the assessment of the identified errors by the SHERPA technique.In this way, a three-dimensional risk envelope or surface is created and used to calculate risk values associated with errors.
According to Table 11 in the petroleum product loading operation, the three human errors that allocated the highest risk score are as follows: Checking the drain valves is omitted (14.68),The separation of the earth connection cable from the truck is omitted (14.33), and the sealing of loading and unloading tanks is omitted (13.87).In the petroleum product unloading operation, according to Table 12, 14 human errors were identified and ranked in 12 levels from high to low risk.The findings indicate that the three human errors that allocated the highest risk score are as follows: Disconnecting the earth connection cable from the truck is omitted (15.18),Disconnect the unloading hose before closing the unloading valve (12.90), and the earth connection cable is not connected to the truck at the required time (12.51).
One of the non-recovery errors in both loading and unloading operations was The earth connection cable is not connected to the truck at the required time, which could lead to an explosion and fire.Research showed that static electricity is very important when loading road trucks because of the load generated by the product flow through the pipeline.
The utilization of fuzzy sets is appropriate for managing the imprecision often associated with accident probability and severity data.The total number of rules required to build a fuzzy inference engine is the product of the number of rows and the number of columns for the qualitative risk matrix based on the probability and severity.
The integration of the SHERPA technique as a qualitative method and the fuzzy inference system as a quantitative method has proven highly effective in identifying and quantifying human errors with greater accuracy.While previous studies typically relied on a single probability factor to assess human error, our proposed approach incorporates two factors: the probability of human error and the severity of error occurrences.This integration allows for a more comprehensive and quantitative assessment, resulting in a wellorganized ranking of errors and consideration of various priorities.By employing this approach to identify and prioritize the risks associated with human errors, organizations can allocate resources more efficiently for risk management, thereby improving safety and efficiency in petroleum product terminals.
Nonetheless, it's important to mention that this study didn't classify the identified error risks into acceptable and unacceptable levels, which could be an avenue for future research.Moreover, integrating specific criteria for risk categorization and addressing other aspects of human error management would be beneficial for further studies.
Additionally, one limitation of this study was the formation of an expert team consisting of only five specialists.We had to form an expert team with just five experts as only this number responded to our request to participate.One concern of this study was the potential for an unreasonable aggregation due to the limited number of experts.It's suggested that future studies employ the largescale group method, in which a sufficient number of experts provide their fuzzy rankings in linguistic values.This method fosters effective coordination and collaboration among experts' mindsets, enhancing accuracy in assessing human error [54].

Conclusion
In conclusion, this study focused on the evaluation of human errors in petroleum product loading and unloading terminals using a combination of qualitative and quantitative methods.The SHERPA technique was employed for error identification and ranking, while fuzzy set theory was utilized to assess the associated risks quantitatively.
The findings revealed that the primary types of errors in both loading and unloading operations were action errors and checking errors.Communication errors were also identified in the loading operation.To address these errors, recommendations such as employee training, checklist preparation, installation of audio and visual alarms, and attention to work-rest periods were suggested.
The integration of the SHERPA technique as a qualitative method and the fuzzy inference system as a quantitative method proved to be effective in detecting and quantifying human errors.This approach can assist organizations in allocating resources more efficiently to manage significant risks, thereby enhancing safety and efficiency in petroleum product terminals.
However, it is important to note that this study did not categorize the identified error risks into acceptable and unacceptable levels, which can be a direction for future research.Additionally, incorporating specific criteria for risk categorization and addressing other aspects of human error management would be beneficial for further studies.
Overall, this research contributes to the understanding of human factors in petroleum product terminals and provides insights for improving safety performance.By identifying and addressing human errors, organizations can work towards reducing accidents and enhancing operational effectiveness in the petrochemical industry.

Informed consent statement
The experts present in the study were fully aware of the work process and expressed their consent to participate in the study in writing.

Fig. 4 .
Fig.4.This figure shows the fuzzy sets and its membership function for each variable used in the fuzzy risk assessment matrix.Fig.5depicts the interdependence of the probability and severity variables as a control surface in the FIS based on the defined rules.This figure shows the relationships between input variables (probability and severity) and the output variable (risk number) in the risk matrix.This three-dimensional graphic visually represents the level of risk.-.All calculations were performed using the commercial software MATLAB R2018b Finally, the errors were ranked based on the risk number.Tables11 and 12provide information on the total risks and the number of calculated risks for the loading and unloading of petroleum products.A low level of risk corresponds to a situation with a low probability of occurrence and low severity, whereas a high level of risk indicates a situation with a very high probability of occurrence and severity.

Fig. 5 .
Fig. 5.Control surface of FIS on severity and probability.

Fig. 6 .
Fig. 6.The number and categories of errors in loading and unloading operations.

Fig. 7 .
Fig. 7.The number of errors in terms of recovery in petroleum product loading operations.

Fig. 8 .
Fig. 8.The number of errors in terms of recovery in petroleum product unloading operations.

Table 3
Definition of fuzzy and crisp ratings.

Table 2
Weighting criteria of different experts.

Table 4
HTA petroleum product loading operation.

Check the documents of the truck driver Check
the cargo transport permit, driver's license, truck documents, and record the information 2Preparing the truck for loading 2.1

Table 5
HTA petroleum product unloading operation.

Table 6
Errors caused by HTA in loading operations using SHERPA.

Table 7
Errors caused by HTA in loading operations using SHERPA.

Table 8
Errors caused by HTA in loading operations using SHERPA.

Table 9
Errors caused by HTA in unloading operations using SHERPA.