Document Type : Research Paper

Authors

1 M.S. Student, Faculty of Petroleum and Natural Gas Engineering, Sahand University of Technology, Tabriz, Iran

2 Professor, Faculty of Petroleum and Natural Gas Engineering, Sahand University of Technology, Tabriz, Iran

3 Associate Professor, Faculty of Petroleum and Natural Gas Engineering, Sahand University of Technology, Tabriz, Iran

Abstract

In sour gas flares,  content like any other components in inlet gas influences adiabatic flame temperature, which, in turn, impacts on the pollutant emission. Wherever flame temperature increases, the endothermic reaction between  and  is accelerated, which means higher  emission to the atmosphere. In this work, we developed an in-house MATLAB code to provide an environment for combustion calculations. Then, this written code was used to perform sensitivity analyses on  content, air temperature, and excess air ratio in sour gas flares. We used Environmental Protection Agency (EPA) reports to assign weighting indexes to each air contaminant according to its harmfulness to environment; thereafter, sour gas flaring conditions were optimized for two real field case studies, namely Ahwaz (AMAK) and South Pars, to reach the minimum integrated pollutant concentrations. The results show that each 2% increase in the  content of the entrance feed may produce 0.3% additional  in the exhaust. The results also confirm that decreases of 20 °F and 50 °F in the oxidant temperature cause  emission to reduce by 0.5% to 1% respectively. Finally, to verify and validate our results acquired from the written MATLAB code, FRNC 2012 industrial software was used to duplicate the oxidation results for the two sour flare case studies.

Keywords

Main Subjects

In 2010, ten countries were accounted for 72% of the total oil and gas flare stacks worldwide. Top five leading countries were Russia (27%), Nigeria (11%), Iran (8%), Iraq (7%), and the United States (5%). Satellite data show that global gas flaring has decreased by about 20% in recent years. The most significant reductions in terms of volume were made in Russia (up to 40%) and Nigeria (up to 29%). One can expect Iran to rise to top of the list in the next few years. The existence of  and  within the produced oil or natural gas is called sourness, which gives a high corrosive trait to flow, especially in the presence of . Whenever possible, it is preferred to install sweetening units to reduce  contents in production lines. However, processes are not 100% effective, and some tail gases with high  contents are always flared. The formation of the sulfur oxides like  and  and the nitrogen oxides like  and during sour gas combustion causes air pollutant emissions to the atmosphere. Various air pollutants and greenhouse gases (GHG’s) show different levels of harmfulness to the environment, while these effects can change yearly depending on several factors such as local climates and the ozone layer condition. Hence, several attempts and progresses were made in achieving air quality standards. Organizations such as Environmental Protection Agency (EPA) gather information and process data and publish yearly reports on the current conditions of the atmosphere and local environments. Several calculations and experimental studies were performed in the literature to simulate combustion reaction in flares and to investigate the effects of different operational parameters on the process efficiency. McMahon et al. (1994) studied several flare stacks and reported that the type of gas, average heating values, and ambient temperature are effective in performance and pollutant emission from flares. Al Rubai et al. (1999) used mathematical kinetic models to express oxidation reaction formulas for sour feed in flares located at Al Daura refinery and estimated , , and  concentrations within the flue gas. Gzar et al. (2009) employed mathematical formulas which used EPA impact factors to develop codes in Visual Basic programming language for simulating combustion in flares. The concentrations of  and  compounds in flue gas were calculated and compared with the experimental data from some real field stacks in Basra state, southern Iraq. Talebi et al. (2014) performed an experimental study over intake gas composition and excess air using pilot scale flare. It was shown that whenever the contents of light hydrocarbon decrease, more  emits from flare, while the amount of  emission remains constant. Kahforoushan et al. (2010) measured  concentrations in the outlet flow of several gas flares which contain various composition gases. The measured values were then compared with the EPA reports on flare emission factors. The results showed high difference wherever  existed in feed. Kahforoushan et al. (2014) performed an experimental study on the effects of parameters such as fuel and oxidant flow conditions. It was confirmed that  formation is greatly dependent on oxidant temperature. The results also showed that  and  concentrations in flue gas were only affected by the inlet composition. Umukoro et al. (2015) developed computer programs in MATLAB to simulate the combustion of oil and gas by considering several types of reactions, including complete and incomplete combustion of sweet and sour gases. Their code calculated the  emission in types of , , and  formation in type of  only. Then, they compared the calculated concentrations of emitted GHG and pollutants in twelve samples from all around the world and illustrated that the entrance composition impacted on the formation of various air contaminants. Ismail et al. (2014) wrote MATLAB codes and performed calculations to show the influence of various parameters such as excess air ratios, natural gas type, carbon mass contents, fuel impurities, and flare systems on the composition of the flame outputs. The calculations were founded on the consideration of various combustion reaction types, including , , , , and  formations. Finally, several comparisons were made among the predicted emissions in some real field flare stacks from all over the world to investigate the effects of the entrance composition on the dispersion rates of air contaminants. Abdulkareem et al. (2009) studied environmental influences of the various flue gas components by accounting on complete and incomplete fuel oxidation and  and  formations. The dispersion of the air pollutant components such as , , and  against the distance from flare stacks was experimentally measured in some oil fields in Nigeria. The illustrated results showed that  and  concentrations were reduced by an increase in distance, while conversely  concentrations were measured higher farther from the flare stacks. Kahforoushan et al. (2008) used MATLAB to model sour gas flares in petroleum industry which emits pollutants like , , , and . The impact of operating parameters such as atmospheric conditions on long term reactions of air contaminants was also estimated. Susu et al. (2005) utilized mathematical models to calculate pollutant concentrations in the flue gas and the neighbor environment considering sulfur, nitrogen, and carbon oxides. The late reaction of air contaminants with excess air during dispersion was taken into account. Abdulkareem et al. (2005) used Visual Basic to simulate , , and  emission from flares. It was concluded that the air temperature and sour gas composition influence the pollutant emission. In this study, we firstly developed some codes in MATLAB R2015 to provide an environment for combustion calculations. In order to do so, we considered the independent oxidation of the fuel components such as , , and . The written code was then used to perform sensitivity analyses on  content, excess air temperature, and air to fuel ratio (AFR)[M1] , to obtain an optimum condition in which the pollutant emission ( + ) is minimized. The novelty of this study compared to previous works in the literature is that, in our viewpoint, GHG’s like  and  are not directly dangerous to the ecological systems, so we presume one may need to minimize  and  in flares. The other difference is that we used Environmental Protection Agency (EPA) reports (2012-2013-1999) to weigh the detriments caused by each pollutant, and we proposed some optimum excess air ratios for sour gas flares as a function of their  contents.

Computational methodology

In this paper, our focus is on the magnification of  existence within the flare entrance and its impact on flame temperature and pollutant emission. In sour gas flares,  is oxidized to  and  compounds which participate in acidic rains. Furthermore, the presence of  affects  emission via influencing adiabatic flame temperature. In addition to fuel composition, air temperatures and air-to-fuel proportions, which are also presented in excess air ratios, determine the adiabatic flame temperature of the flares. In order to calculate the amount of greenhouse gases and the air pollutants emitting from flare stacks, we need to simulate the combustion reaction of the fuel mixture and determine the composition of the flue gas. Each hydrocarbon component within the inlet sour gaseous phase participates in an exothermic reaction, which, in the presence of enough excess air, will go through complete combustion yielding only carbon dioxide and water vapor. The general form of the complete combustion we adjusted in our written MATLAB code is given by Equation 1:

 

(1)

Nitrogen which participates in  formation comes from two main sources, firstly from the oxidant mixture and secondly from  contents within the gaseous fuel. Similarly, the GHG products of the complete oxidation, namely  and , which appear in the flue gas may come from fuel oxidation or from the entrance. Conversely, in situations with limited oxidant, hydrocarbon components will go through an incomplete combustion, in which hydrogen and carbon monoxide are produced in addition to carbon dioxide and water vapor. The general form of the incomplete hydrocarbon oxidation we used in our written MATLAB code reads as:

 

(2)

Reaction yields and enthalpy changes relevant to fuel component combustions were calculated from thermodynamics, in which the difference among the total Gibbs free energy of the products and those of the reactants determines the exothermic or endothermic condition of each oxidation. Conversely, the rate of the conversion of each reactant was calculated using kinetics via considering especial orders and constants for each fuel compound within its combustion reaction. The enthalpies calculated for all the hydrocarbon combustions showed negative values, which exhibit exothermic oxidations, while the calculated rates represented the instant conversion of reactants to products. On the contrary, , , and  oxidations to  and  disobey the above rules. Nitrogen oxides, which are also called , are emitted from flares and are mainly consisted of  (about 90% to 95%) and much less of  (about 5% to 10%). Other nitrogen oxides like nitrous oxide appear only in negligible amounts. Nitrogen oxides which are formed in combustion reactions also came from two sources, namely air and fuel. Two main formation mechanisms which are called prompt and thermal contribute in the formation of nitric oxide from the oxidant resource. The prompt mechanism presents that firstly some available radicals react with nitrogen molecule to form , , and  molecules, and then these compounds are easily oxidized to nitric oxide. Oppositely, the thermal mechanism presents that some high energy molecules dissociate oxygen molecule to form oxygen radical, and this radical then attacks nitrogen molecule to form nitric oxide. Regardless of what extent each of the mentioned mechanisms participates in  and  oxidation, we adjusted  endothermic formation equation in our code in the form of Equation 3:

 

(3)

However, the slow spontaneous conversion of nitric oxide to nitrogen dioxide was considered to take place in small amounts according to the exothermic equation described by:

 

(4)

It is noticeable that the overall  formation is definitely endothermic, so wherever adiabatic flame temperature is high, the oxidation is accelerated, leading to more  emission.          

Sour flares contain  component that is oxidized into sulfur dioxide and trioxide. If oxygen is present in high amounts, the dominant product is , while limited oxygen tends to form more . Both reactions are exothermic, and while  participates in  formation in acidic rains,  conspires against  formation which shows more corrosive characteristics. The oxidation reactions are similar with the only difference that  formation consumes more oxygen. We considered adjusting  formation equations in our MATLAB code in the forms of below equations:

 

(5)

 

(6)

To provide an environment for calculations, we developed an in-house code in MATLAB R2015 which considers identical oxidant distribution among all the components. Under the existence of excess air, the written code automatically enables complete combustion option and calculates the enthalpy change within each independent reaction with the aid of the reference Gibbs free energies associated to each product or reactant component. The rate of reactions were defined as power law equations in which each component concentration influences conversion rate with a power factor equal to its stoichiometric coefficient within the reaction equation, i.e. its reaction order. The adiabatic flame temperature was then calculated according to total enthalpy change and intake air temperature. Finally, the conversion of reactants to  and to  was calculated. Almost the same progress happens when the air resources are limited. In such conditions, the written code automatically activates the incomplete combustion option, which replaces the equations relevant to the complete reaction with those of incomplete combustion. Total enthalpy change and adiabatic flame temperature were calculated in the same manner. Only difference is that the code coincidentally calculates the amounts of the additional products like  and  which appear only in the incomplete combustion.

Afterwards, to investigate the effects of parameters affecting contaminants emission from sour flares, we performed several sensitivity analyses on  content within the entrance stream, which its higher concentration is expected to accelerate pollutant emission, on excess air ratio which determines adiabatic flame temperature and consequently impacts on  and  emission and also on the inlet air temperature. One must notice that since in combustion reactions, oxidant is consumed in much larger volumes than fuel, inlet stream temperature is less affected with fuel rather than air.[M2]  Based on the results of the analyses, we proposed some optimal excess air ratios as functions of  content and intake air temperature to minimize pollutant emission from sour flares. Figures 1 and 2 show the schematic flowchart of the calculations and the developed MATLAB code. The calculated results were then verified and validated with FRNC 2012 industrial software using its Combustion Simulator Module.

 

Figure 1

A schematic flowchart explaining calculational procedure within the MATLAB code.

 

Figure 2

A screenshot of the developed MATLAB GUI code to calculate flue gas compositions.

Results and discussion

In order to perform the sensitivity analyses, we firstly selected two real field sour gas case studies from southern Iran, namely AMAK resource in Abadan refinery, which originally comes from Asmari field in Ahwaz, and another resource coming from South Pars field in the Persian Gulf. Next, we performed the sensitivity analyses on  content within the flare entrance via imaginarily changing  to  ratio while keeping the overall composition of the sour components constant. In this way, one can assure that the adiabatic flame temperature is least affected with changes in the inlet fuel composition. Then, we fed all the defined cases into the written MATLAB code and calculated the optimum excess air ratios and the relevant flue gas compositions. The exhaust gas compositions determined at the optimum excess air ratios, including  and , , and  are presented in Table 1.

Table 1

Exhaust gas composition for South Pars case study at the optimum excess air condition.

Intake sour compositions

   

 

 

 

 = 2.5%

 = 7.5%

13.9%

1.5%

0.3%

1.1%

0.1%

 = 4.5%

 = 5.5%

13.7%

1.7%

0.4%

1.1%

0.1%

 = 6.5%

 = 3.5%

13.5%

1.9%

0.5%

1.1%

0.1%

 = 8.5%

 = 1.5%

13.3%

2.2%

0.6%

1.1%

0.1%

The calculated values of the GHG ( ) compositions within the flue gas are almost constant. Only some small increases are seen in GHG emissions mainly due to the increment in  contents in the inlet feed. Conversely,  and  contents within the exhaust gas show unavoidable continuous rise with an  increase in the inlet feed. It is noticeable that the calculated optimum excess air within each case study was different from the others, which shows that one optimum air fuel ratio (AFR) value cannot be adapted to all sour flares. As expected, the amounts of the  and  in the flue gas show no change with any variation in  to  proportions. We then conducted several sensitivity analyses on the intake air temperature. Air temperature can extremely be lowered via a humidification process using cheap air coolers, in which air goes through some adiabatic water saturation hypothetical lines. Low temperature oxidant mixtures, particularly with up to 80% inert nitrogen, play an important role in lowering the adiabatic flame temperatures, which consequently decrease  productions. Table 2 tabulates the effects of the intake air temperature on each contaminant composition within the exhaust gas in the South Pars sour flare case study.

Table 2

Effects of intake air temperature on GHG and pollutant emission in the South PARS case study.

Intake temperatures

Intake temperature 90 °F

Intake temperature 80 °F

Intake temperature 70 °F

Intake temperature 60 °F

Intake temperature 50 °F

Intake temperature 40 °F

Air contaminant

 %

1.4%

1.1%

0.9%

0.7%

0.6%

0.5%

 %

0.2%

0.1%

0.07%

0.05%

0.04%

0.03%

( + )%

2.4%

2.4%

2.4%

2.4%

2.4%

2.4%

( + )%

13.5%

13.5%

13.5%

13.5%

13.5%

13.5%

The impact of the intake air temperature on the endothermic formation of  and  shows that with a 50 °F decrease in the intake oxidant temperature, one can expect  emission to increase threefold. The same trend appears in  emission with the difference that the effect is even larger. The production of the  compound was not influenced by the intake air temperature perhaps due to the fact that  formation reaction is only slightly exothermic. Also, GHG emissions were not affected with the intake air temperature and were only sensitive to the amount of excess air. Moreover, it is noticeable that the inlet fuel temperature influences the flame temperature, but its effect is negligible. Therefore, one can offer to cool the air in sour gas flares as well as sweet flares using some throttling valves or air humidifiers prior to the premix entrance. Thereafter, we investigated the role of excess air in determining the adiabatic flame temperature and its effect on pollutant emission from sour flares. Table 3 illustrates the calculated results of flame temperatures and  and  emissions from two flare case studies. 

Table 3

Effects of excess air on flame temperature and pollutant emission in AMAK case study.

Excess air

10.0%

15.0%

20.0%

25.0%

30.0%

South Pars

Adiabatic flame temperature

2750

2830

2800

2780

2760

Air pollutant (SOx +NOx)

3.8%

3.7%

3.6%

3.8%

4.0%

Ahwaz (AMAK)

Adiabatic flame temperature

3050

3100

3080

3050

3030

Air pollutant (SOx +NOx)

3.1%

3.0%

2.9%

2.8%

3.1%

While theory states that exact stoichiometric oxidant ratios maximize the flame temperatures, the experiences show that some excess air is always needed to achieve this goal. Our calculated flame temperatures versus the excess air show increases in exhaust gas temperatures when excess air quantities change from 0.0% to 15.0%. This value appears to be the critical excess air above which the adiabatic flame temperatures decrease with a further increase in excess air. Similarly, in the two case studies of South Pars and AMAK (Ahwaz), 15% excess air maximizes the flame temperatures to values of 2830 °F and 3100 °F respectively. In lower oxidant quantities, the incomplete combustion of the fuel components occurs to some extent, which causes the flame temperature to drop. On the other hand, in higher excess air quantities, more energy consuming components appear in the exhaust, which lowers the adiabatic flame temperature. The harmfulness of each air contaminant, including , , , and  was weighed according to EPA environmental reports and was used in calculating the optimum excess air conditions. The oxidation of  to  is extremely endothermic, so it shows high yields at high flame temperatures and falls by increasing excess air quantities. On the contrary, the presence of the high amounts of  within the reaction causes more  to be produced instead of . The EPA reports that the impact of  is several times more destructive than that of  in acid rains. It goes through a slow equilibrium with water vapor present in the atmosphere and forms , which is known as a highly corrosive compound within acidic rains. However, in an opposite way,  participates in the formation of , which is several times weaker than . We calculated optimum excess air quantities with respect to these effects reported by EPA (United States Environmental Protection Agency). In the South Pars case study, a 20% excess air with a tolerance of 5% was calculated as the optimum, while in the case of AMAK (Ahwaz) a value of 25% with a tolerance of 5% was determined as the optimum excess air to minimize the pollutants emission. In order to verify data resulted from the written MATLAB code, we used FRNC 2012 software to simulate flaring processes in the two previously selected field case studies, namely AMAK (Ahwaz) and South Pars. Inlet gas compositions and oxidant mixture conditions were defined in FRNC 2012, and the options for incomplete reaction, i.e.  and  formation, were activated; then, the simulation was performed to give the resultant flue gas compositions. In each case study, several simulation runs at various air temperatures and excess air ratios were performed to find the conditions which minimize pollutant emission. In addition, we compared the results with the experimental data acquired from the gas chromatography (GC) tests of the exhaust gases in the two selected case studies. A comparison among the calculated composition of air pollutants and GHG’s, the simulated results, and the experimental data is tabulated in Table 4.

 

Table 4

Validation of the written code results with FRNC Industrial Software and experimental data.

Flue gas compositions

CO+CO2 (%)

SO2 (%)

SO3 (%)

NO (%)

NO2 (%)

South Pars

via Code

13.50%

1.90%

0.50%

1.10%

0.10%

via FRNC

13.80%

1.80%

0.60%

1.00%

0.20%

Experimental data

13.95%

1.77%

0.53%

1.15%

0.15%

Asmari (AMAK)

via Code

15.40%

1.30%

0.30%

0.80%

0.10%

via FRNC

15.60%

1.20%

0.30%

0.70%

0.10%

Experimental data

15.73%

1.15%

0.28%

0.65%

0.15%

Conclusions

Several key findings of the present work can be summarized as follows:

1-        An increase in  content of the flare entrance by 2.0%, 4.0%, and 6.0% causes  emission to raise 0.3%, 0.6%, and 0.9% respectively.

2-        Low  contents in sour gases do not significantly affect  and  emissions. Hydrocarbons consume much larger volumes of oxygen in a way that greenhouse gas emissions are not affected with the presence of hydrogen sulfide.

3-        A decrease in the temperature of the entrance air significantly reduces the  formation. A drop of 20 °F and 50 °F in the oxidant temperature causes  emission to respectively fall by 0.5% to 1%.

4-        Pollutant emissions in sour flares are not necessarily minimized at optimum AFR’s by flame temperature maximization. The independent optimum AFR’s were calculated regarding the harmfulness of air contaminants.

5-        The optimum AFR values to minimize pollutant emission in Ahwaz AMAK and South Pars case studies are calculated 25% and 20% respectively.

6-        As EPA annually reports different indexes for the local harmfulness of each GHG and pollutant, the optimum flaring conditions vary locally and with respect to time.

7-        Each sour flare seems to have a particular optimum excess air ratio which depends on  content and other flow characteristics. One can expect the optimum condition to differ per case.

Acknowledgements

Authors acknowledge National Iranian South Oil Company (NISOC) and South Pars Gas Company (SPGC) for freely providing the technical data.

Nomenclatures

AFR

Air to fuel ratio

EPA

Environmental Protection Agency

FRNC

Heater simulation program

GC

Gas chromatography

GHG

Greenhouse gas


 [M1]Reply to Comment2 Editor: We replaced mentioned word with this phrase.

 [M2]Reply to Comment2 Editor: We changed this phrase to make it more clear.

Abdulkareem, A. S., Urban Air Pollution Evaluation by Computer Simulation: A Case Study of Petroleum Refining Company Nigeria, Leonardo Journal of Sciences, Vol. 4, No. 6, p. 17-28, 2005.
Abdulkareem, A. S., Odigure, J. O., and Abenege, S., Predictive Model for Pollutant Dispersion from Gas Flaring: A Case Study of Oil Producing Area of Nigeria, Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, Vol. 31, No. 12, p. 1004-1015, 2009.
Al Rubai, H. A. G., Mathematical Modeling for Dispersion of Air Pollutants Emitted from Al Daura Oil Refinery Stacks, Master Thesis, University of Baghdad, 1999.
EPA, Enforcement Targets Flaring Efficiency Violations, Enforcement Alert, US Environmental Protection Agency, 2012.
EPA, Needs to Improve Air Emissions Data for the Oil and Natural Gas Production Sector, Office of Inspector General, US Environmental Protection Agency, 2013.
EPA, Proffered and Alternative Methods for Estimating Air Emission from Oil and Gas Field Production and Processing Operations, 1999.
Gzar, H. A. and Kseer, K. M., Pollutants Emission and Dispersion from Flares: A Gaussian Case-study in Iraq, Journal of Al-Nahrain University, Vol. 12, No. 4, p. 38-57, 2009.
Ismail, O. S. and Umukoro, G. E., Modelling Combustion Reactions for Gas Flaring and its Resulting Emissions, Journal of King Saud University, Engineering Sciences, Vol. 28, No. 2, p. 130-140, 2014.
Kahforoushan, D., Bezaatpour, J., and Fatehifar, E., Effect of Various Parameters on Emission Factors of Gas Flares, Iranian Journal of Chemical Engineering, Vol. 11, No. 3, p. 59-66, 2014.
Kahforoushan, D., Fatehifar, E., Babalou, A. A., Ebrahimin, A. R., Soltanmohammadzadeh, J. S., and Elkamel, A., Modeling and Evaluation of Air Pollution from a Gaseous Flare in an Oil and Gas Processing Area, WSEAS Conference, Spain, 2008.
Kahforoushan, D., Fatehifar E., and Soltan, J., The Estimation of  Emission Factors for Combustion Sources in Oil and Gas Processing Plants, Energy Sources, Vol. 33, No. 3, p. 202-210, 2010.
McMahon, M., Estimating the Atmospheric Emission from Elevated Flares, BP Amoco Annual Report, 1994.
Susu, A. A., Abhulimen, K. E., and Adereti, A. B., Modeling of Air Pollution Systems with Chemical Reactions: Application to Gas Flares in Nigeria, International Journal for Computational Methods in Engineering, Vol. 6, No. 3, p. 201-213, 2005.
Talebi, A., Fatehifar, E., Alizadeh, R., and Kahforoushan, D., The Estimation and Evaluation of New , , and  Emission Factors for Gas Flares Using Pilot Scale Flare, Energy Resources, Vol. 36, No. 7, 2014.
Umukoro, G. E. and Ismail, O. S., Modelling Emissions from Natural Gas Flaring, Journal of King Saud University, Engineering Sciences, Vol. 29, No. 2, p. 178-182, 2015.