Risk Analysis of Flare Flameout Condition in a Gas Process Facility

Risk Analysis of Flare Flame-out Condition in a Gas Process Facility — Flaring is a common method of disposal of flammable waste gases in the downstream industries. Flare flame out (flame lift-off or blow-outs) often occurs causing toxic vapors to discharge. The toxic gases released may have hazardous effects on the surrounding environment. To study the effect of inhalation exposure of these toxic gases on human health, the four steps of the EPA (Environmental Protection Agency) framework with the field data to quantify the cancer and non-cancer health risks are integrated in this paper. As a part of exposure assessment, gas dispersion modeling using AERMOD and UDM-PHAST is applied in two different conditions of normal flaring and flare flame out during a particular climate condition in Khangiran region. Recommendations to avoid flare flame out conditions are also presented here. Oil & Gas Science and Technology – Rev. IFP Energies nouvelles, Vol. 66 (2011), No. 3, pp. 521-530 Copyright © 2011, IFP Energies nouvelles DOI: 10.2516/ogst/2010027 Risk Analysis of Flare Flame-out Condition in a Gas Process Facility O. Zadakbar1*, R. Abbassi1, F. Khan1, K. Karimpour2, M. Golshani2 and A. Vatani3 1 Faculty of Engineering & Applied Science, Memorial University of Newfoundland, St. John’s, NL, A1B 3X5 Canada 2 Nargan Consulting Engineers, Tehran Iran 3 Institute of Petroleum Engineering, Faculty of Engineering, University of Tehran, Tehran Iran e-mail: o.zadakbar@mun.ca rabbassi@mun.ca fikhan@mun.ca k.karimpour@nargan.com m.golshani@nargan.com avatani@ut.ac.ir * Corresponding author ogst100008_Zadakbar 22/07/11 17:35 Page 521


INTRODUCTION
Refinery flares are used for the safe disposal of flammable waste gases from emergency process upsets as well as for start-up, shut-down and turnaround operations.Flaring could release large quantities of SO 2 and CO into the atmosphere.The magnitude of emissions resulted from flaring process in Iran is not clear, as many refineries do not have flare monitoring and emissions recording procedures in place.Thus, there is a concern about the potential underestimation of reported emissions released during flaring processes [1].
In the downwind plume of sour gas flares, SO 2 and H 2 S exist, in addition to a wide spectrum of sulfur-containing chemicals and the range of unburned hydrocarbons.An annual exposure greater than 4 μg/m 3 H 2 S is associated with spontaneous abortion in humans and animals [2].The odor threshold of H 2 S is approximately 7 μg/m 3 , which is less than the critical concentration.An acceptable daily intake of 1.8 μg/m 3 H 2 S has been presented [3].
Combustion efficiency of a flare is severely affected by wind.Regardless of the regulatory stricture imposed, a conventional flare will never operate above 95% of efficiency except in wind speed of about less than 2 kmph [4].Reduced combustion efficiency must be regarded possible in any operation with flaring.Simultaneous low combustion efficiency causes release of unburned gas.That includes hydrogen sulfide with sour gas flares.
Public health concerns about gas flaring have existed for many years in different regions near natural gas facilities.Some of these concerns are related to potential long term cumulative health effects on humans from exposure to hazardous chemical concentrations released during incomplete combustion of flare gases [5].Environmental Risk Assessment (ERA) evaluates the nature and likelihood of the adverse effects on human health and ecosystems due to the environmental changes [6].Implementing the four steps of EPA (Environmental Protection Agency) framework of risk assessment process with chemical concentrations during flare flame out and normal flaring demonstrated in Figure 1, lead to evaluate the effects of these concentrations emitted due to the flare gases on the human health [7].ERA methodology for evaluating the human health risk of the flare emissions.

Risk characterization
Risk-based evaluation of chemical concentrations during normal flaring as well as flare flame out via a case study in the Khangiran gas process facility (Sarakhs-Iran) is considered in this paper.Recommendations for precluding the flame out conditions are also presented.

Site Specification
The Shahid Hashemi-Nejad (Khangiran) natural gas process facility is one of the most important gas process facilities, located in northeastern Iran in an open inhabitable range land, semi arid and dusty with windblown sand (Fig. 2).The feed gas is supplied from the Mozdouran gas fields.This gas process facility consists of 5 sour gas treating unit, 3 dehydration units, 3 sulfur recovery units, 2 distillation units, 2 stabilizer units and 14 additional units related to other services [8].
The wind direction in the Khangiran vicinity is from northwest to southeast.This condition occurs in the Khangiran gas process facility for 90% of a year.For the remainder of the year, the wind direction changes, and the wind blows from southeast to northwest [10].This means that for about 10% of the year, personnel are in danger of inhaling more toxic gases.During this period, the personnel are exposed to an unacceptable amount of toxic gases which could affect their health.Wind roses were used to give a succinct view of how wind speed and direction are typically distributed and to determine the direction of the prevailing wind in the vicinity of Khangiran.During 10% of the year, the wind direction is 160 degrees [10], which causes the wind to carry toxic gases to the personnel dormitory.This dormitory is located 1050 meters away from the flare stack and it is 200 meters in length.The release time was considered to be 1 h.

Hazard Identification
The toxic gases considered during the release time (1 h) has been studied and calculated for two flaring conditions: -Flare flame out, in which the concern gas is H 2 S; -Normal flaring, in which the concern gas is SO 2 .
H 2 S (Hydrogen Sulfide) is an extremely hazardous and toxic compound.H 2 S is colorless and flammable which can be identified by its characteristic rotten egg odor.Low concentrations of 20-150 ppm H 2 S cause irritation of the eyes, slightly higher concentrations may cause irritation of the upper respiratory tract, and if exposure is prolonged, pulmonary edema may result.The irritant action has been explained on the basis that H 2 S combines with the alkali present in moist surface tissues to form sodium sulfide, a caustic [11].As the concentration approaches 100 ppm, odor becomes impressible due to olfactory fatigue.At these levels, the gas disrupts cellular respiration and may cause profound respiratory depression as well as cardiac dysrhythmias [12].A high concentration of 200 ppm is extremely hazardous and can immediately be life threatening [13].Inhalation of Figure 2 Khangiran gas process facility and surrounding areas [9].500 ppm for 30 minutes produces headache, dizziness, excitement, staggering, and gastroenteric disorders, followed in some cases by bronchitis or bronchial pneumonia.Concentrations above 600 ppm can be fatal within 30 minutes through respiratory paralyses [14].
Furthermore, SO 2 is a colorless gas which smells like burnt matches.It can be oxidized to sulphur trioxide, which in the presence of water vapor is readily transformed to sulphuric acid mist.SO 2 with acute exposure of 5 ppm may cause dryness of nose and throat and a miserable increase in resistance to bronchial air flow.SO 2 increasing up to 6 to 8 ppm causes a decrease in tidal respiration volume.Sneezing, cough & eye irritation occur at 10 ppm.SO 2 concentration of 20 ppm may cause Bronchospasm and 50 ppm causes extreme discomfort, but no injury in less than a 30 minute exposure.Finally, inhalation of 1000 ppm more than 10 minutes causes death [15].The Concentrations of H 2 S and SO 2 in two different conditions considered in this modeling scenario can be seen in Table 1.It is assumed that H 2 S is completely oxidized during normal flaring.

Exposure Assessment
There has been a significant increase in awareness of environmental issues in recent years and there is a great concern among people over how their health is affected by environmental factors.Exposure assessment includes estimating the dose or concentration of the contaminant taken in by human or ecological receptors per unit of time.Characterization of the exposure setting, identification of exposure pathways and quantification of exposure are different steps in exposure assessment.
Gas dispersion modeling helps to predict the ground level concentration and deposition of air pollutants.One of the key elements of an effective dispersion modeling study is to choose an appropriate tool to match the scale of impact and complexity of a particular discharge [16].
In medium-complex atmospheric and topographical conditions as existed in the vicinity of Khangiran, Gaussianplume models can be used to produce reliable results.In more complex atmospheric and topographical conditions, advanced puff or particle models may be required to achieve a comparable degree of accuracy [16][17].Furthermore, plume models are usually applicable to near field within 10 km from the source calculations.It is not wise to assume the meteorology will be the same greater than 10 km away as at the source [17][18].Since the distance between the specified location (personnel dormitory), which should be considered as the critical point during flare flame out, from the flare stack is 1 km, the plume model would be the proper model in comparison with puff model.

Contaminants Transport Modeling using AERMOD
There are some Gaussian plume dispersion models such as AERMOD, ISCST3, AUSPLUME and CTDMPLUS.AERMOD is the most common Gaussian plume dispersion model which is recommended to use [16][17][18].AERMOD was developed in 1995, reviewed in 1998 and formally proposed by the USEPA as a replacement for ISCST3 in 2000 [18][19].AERMOD is a near field, steady state guideline model.It uses boundary layer similarity theory to define turbulence and dispersion coefficients as a continuum, rather than as a discrete set of stability classes.Also, dispersion coefficients for unstable conditions are non-Gaussian, to represent the high concentrations that can be observed close to a stack under convective conditions [20].The modeling system consists of AERMOD, as the main program, and AERMET and AERMAP as two pre-processors.AERMET is used to calculate boundary layer parameters.The meteorological interface uses these parameters to generate profiles of the needed meteorological variables.It passes all meteorological observations to AERMOD.In addition, AERMAP characterizes the terrain, and generates receptor grids for the dispersion model [16].BREEZE® AERMOD [21] was used to model gas dispersion during flare flame out in the present study.This model is an air quality modeling system used to support both regulatory and non-regulatory modeling requirements worldwide.

Contaminants Transport Modeling using UDM -PHAST
DNV PHAST [22] as a comprehensive hazard analysis software tool is also used for gas dispersion modeling to corroborate AERMOD results.PHAST uses a proprietary dispersion model called the Unified Dispersion Model (UDM).
It was formulated as a similarity model in which concentration and other variables are assumed to have a predefined profile.It also assumes generalized Gaussian profiles.Entrainment and spreading are calculated from a numerical (Runge-Kutta-Milne) solution to differential equations for mass, momentum and heat transfer between the cloud and its environment [23].The UDM is formulated as an integral model.A set of differential equations is integrated to give the key variables as a function of distance or time.A number of algebraic equations are then solved to obtain other variables describing the dispersing cloud.The set of differential equations are basically the same for instantaneous and continuous releases, although they are integrated with respect to time at first and then with respect to distance.The same differential equations apply throughout all phases of dispersion [22].Gas dispersion modeling was done for the flare flame out as well as normal flaring in this study with following considerations which can be seen in Table 2 and Table 3 (the stack diameter is 1.5 m and the stack length is 50 m).The inhalation rates were received from USEPA [24] for two different conditions of slow and fast activity levels.For the average condition, the workers spend 12 hours inside the site, and in the worst case they may spend the entire day within the site.Some of the workers are not at the site during weekends, which is 96 days of the entire year.The average exposure duration was estimated for 70 years for a lifetime cancerous effect [24].The dose received by a human due to inhalation of each metal (carcinogens intake) is calculated by the following equation: Dose-response assessment is one of the steps of the risk assessment process that connects the likelihood and severity of damage on human health from exposures to different levels of risk agents.The reference concentration (RFC) is used to assess inhalation risks, where concentration refers to levels of contaminants in the air.For carcinogens, the slope of this straight line between dose and response, called the Slope Factor (SF) or cancer slope factor is used to estimate the risk at exposure levels.The RFCs and SFs used in this modeling scenario are presented in Table 5.Previous investigations by NTP (National Toxicology Program), OSHA (Occupational Safety and Health Administration), or ACGIH (American Conference of Governmental Industrial Hygienists) did not classify SO 2 as a carcinogenic substance.Therefore, the non-carcinogenic risk is only considered for SO 2 in this scenario.

Risk Characterization
By integrating exposure assessment and toxicity assessment, which are discussed previously, the probability of negative effects is assumed.As sufficient information is available in the literatures to determine the toxicological benchmark (RFC and SF are available for the specific component), the quotient method for calculating the final risk value is used [27].
The data from Table 5 are used to calculate the Hazard Quotient (HQ) and the risk for carcinogens.The HQ is calculated from Equation (2), while the carcinogenic risk was calculated from Equation (3).HQ = Non-carcinogenic intake (mg/m 3 )/ Reference concentration (mg/m 3 ) (2) Risk = Carcinogenic intake (mg/kg/day) × Cancer slope factor (mg/kg/day) -1 (3) If the HQ is less than 1, it shows that the risk is slight and little or no action is required.If the HQ is near 1, it shows uncertainty in the risk estimate and additional data is required.Finally, if the HQ is more than 1, it shows that the risk is greater and regulatory action may be indicated [28].A single specified acceptable risk level, applicable for all carcinogens regulated by EPA is 10 -6 (one death per million people).Whenever the risk of carcinogenic substance exceeds this value, it shows that the component has a carcinogenic effect [27].

RESULTS AND DISCUSSION
Implementing the four steps of EPA framework, with the contaminant concentrations released in two different conditions of normal flaring and flare flame out (see Tab. 6) leads to the following results.

Dispersion Modeling Results using AERMOD
Figure 3 illustrates gas dispersion due to flare flame out condition in the study region.This figure shows that the personnel dormitory is covered by a plume with a concentration of more than 500 ppm.

Dispersion Modeling Results using PHAST
To study the gas dispersion conditions, two areas have been considered: the restricted and the impact area.The restricted area is the area within the boundaries of the installation and hence is under control of the company, either automatically through appropriate systems or manually.The impact area is the area that extended beyond the boundaries of the installation but is nevertheless affected permanently by normal The maximum concentration of H 2 S in the restricted area has been illustrated in Figure 4.As shown in this figure, the hazard distance for refinery employee inhalation is a circle with a 1470 m radius.Therefore, in the particular climate conditions in which the wind direction might be changed toward the working area and dormitory, the flare flame out would have the irremediable consequence and shall be con-sidered.Also, the flare flame out in this condition would be important for the public, as it is drawn in Figure 5.In Figure 6, the maximum concentration of SO 2 dispersion for the impact and the restricted area has been shown.As demonstrated in this figure, the hazard distance for SO 2 inhalation is more than 12 000 m, which includes the entirety of the working area as well as the personnel dormitory.Therefore, the continuous flaring in an emergency situation would endanger the employee and it shall be considered during the emergency response plan.

Risk Estimation
The concentrations of SO 2 and H 2 S in two different conditions of normal flaring and flare flame out are calculated, as can be seen in Table 7.The maximum concentration of H 2 S within the restricted area.
After finding the concentrations of these chemicals based on two different conditions considered in this modeling scenario, HQs are calculated to evaluate the hazardous effect of SO 2 in the normal flaring condition.As illustrated in Table 8, the HQs in both areas (NF1 & NF2) are grater than the criteria, which is 1.It shows the high non-carcinogenic risk of inhaling SO 2 in these areas.
Also, the values of HQs calculated based on the inhalation of H 2 S in different areas (FFO1 to FFO4) in the flare flame out condition are far above the criteria.This shows a great risk of inhalation of H 2 S in the flare flame out condition.
Comparing the results of carcinogenic health risk with the criteria (10 -6 ) shows that, there is a great carcinogenic health risk by inhaling H 2 S during flare flame out condition in both scenarios (normal and worst case) considered in this study.H 2 S release in the flare flame out condition can be so dangerous, that it could result in having the harmful effects on human health.Therefore, the following recommendations selected from previous investigations [29][30][31]8]  The maximum concentration of H 2 S within the impact area.
-Provision of a liquid Knock Out (KO) drum, which is equipped with high level alarms to warn of an excessive accumulation of liquid; -Prevention of the introduction vapors into the system when it is not operational; -Automate the ignition sequence using continuous pilot flame monitoring; -Use direct electrical ignition of pilots with flame front only as a back-up system to cover for electrical problems with inaccessible equipment; -Use pilot burners specially adapted for high inert atmospheres wherever high inert purge flows are likely; -Use Flare Gas Recovery System (FGRS).Typically, the gas is recovered from a vent header feeding a flare.Depending on flare gas composition, the recovered gas may be recycled back into the process for its material value or used as fuel gas.FGRS is commonly used in refineries to recover flammable gases for reuse as fuel for process heaters.Zadakbar et al. proposed an FGRS for the Khangiran gas facilities.It has a modular design and comprises three separate and parallel trains capable of handling various gas loads and compositions.Gas emissions will be decreased up to 70% by using such recovery system [1].Thus it could be a unique suggestion to reduce the hazardous consequences of flare flame out condition.

CONCLUSIONS
Flaring is the process of burning waste gases which creates emissions such as sulfur oxides (SO x ) and greenhouse gases (CO 2 and CO).In certain conditions i.e. storm, heavy rain, strong wind, long serviced pilot, etc., flare may lose the flame.Analysis of toxic effects of H 2 S during flare flame out as well as analysis of toxic effects of SO 2 which is released during normal flaring in Khangiran natural gas process facility has been evaluated.Since there is a large amount of sour gas which is burnt off in the Khangiran, flare flame out, during particular climate condition when wind blows from southeast to northwest, could have fatal effects on employees inside the site and dormitory.Environmental Risk Assessment is used as a tool to evaluate the hazardous effect of toxic   The maximum concentration of SO 2 for impact and restricted areas.
gases on human health.The fate and transport of the contaminants (H 2 S and SO 2 ) are modeled using AERMOD and UDM.In medium-complex atmospheric and topographical conditions, with relatively simple effects, as may be seen in vicinity of Khangiran, Gaussian-plume models can produce reliable results.Comparing the results of cancer and non-cancer risk values with the existing guidelines shows that the employees are at risk of inhaling carcinogenic and non-carcinogenic contaminants.Both values of carcinogenic and non-carcinogenic risk are much greater whenever the flare flame out condition happens.Moreover, living with such a condition for a long time leads to death.Therefore, evaluating the flare conditions regularly and assessing alternatives to preclude the flare flame out is required.
Intake = C air * Inhalation rate * Exposure time * Frequency (1) Body weight * Averaging time 1.4 Dose-Response Assessment

Figure 3
Figure 3 Gas dispersion due to flare flame out.

TABLE 1
[8]centrations of H 2 S and SO 2 in two different conditions considered in this modeling scenario[8]

O
Zadakbar et al. / Risk Analysis of Flare Flame-out Condition in a Gas Process Facility NF1: Normal Flaring area 1. ** FFO: Flare Flame Out area 1 *

TABLE 8
Calculating hazard quotients and cancer health risk for different scenario Inhalation of SO 2 in normal flaring