Air Emission Estimation from Marib Oil Refinery

A Practical Methodology for Quantifying Air Pollutant Emissions from Petroleum Refineries in Data-Scarce Settings

About this finding: This finding represent a summary of an air emission assessment undertaken by CORAL for Environmental Services at the request of an investor planning to develop an oil refinery project in Yemen. The investor required a realistic picture of what air emissions from a Yemeni refinery operation would look like in terms of pollutant types, emission sources, and emission rates, in order to inform the environmental planning and impact assessment for the proposed facility. The Marib Oil Refinery was selected as a benchmark case study, as it is one of only two operating refineries in Yemen and represents a comparable scale and configuration to the proposed development. The methodology and findings presented here formed part of the environmental baseline and emission characterization work underpinning the project's early-stage environmental advisory.

Yemen contributes little to global greenhouse gas emissions, yet it ranks among the nations most vulnerable to climate change. Within its energy sector, petroleum refining is a notable source of hazardous air pollutants — and one that has historically gone unquantified. Yemen operates only two refineries: the Aden Refinery (170,000 bpd) and the Marib Refinery (10,000 bpd). Both are essential to meeting domestic fuel demand, yet both release a range of air pollutants into the atmosphere whose emission rates have never been systematically estimated and published in the open literature. This post sets out a defensible methodology for estimating air emissions from refinery operations in settings where site-specific measurement data is limited or absent — using the Marib Oil Refinery as a worked case study. The aim is not only to present results, but to demonstrate an approach that environmental practitioners across the region can replicate.

1 The Challenge: Estimating Emissions Without Measured Data

The central difficulty in quantifying refinery emissions in Yemen is the near-total absence of credible, site-specific emission measurements. Continuous emission monitoring systems are not installed, and historical stack-test data does not exist. Any estimation effort therefore carries inherent uncertainty arising from three sources: restricted access to operational data, the potential unsuitability of generic emission factors for a specific facility, and an incomplete understanding of the processes driving certain emissions.

To address this, the assessment drew on two internationally recognized and conservative references: the USEPA AP-42 Compilation of Air Emission Factors and the World Bank Group Environmental Health and Safety (EHS) Guidelines for Petroleum Refining. AP-42 emission factors represent average pollutant emission rates derived from measurements across many comparable sources over time, and are widely accepted as a sound basis for estimation where direct measurement is unavailable.

2 Defining the Emission Source Inventory

The Marib refinery comprises two main processing units — a Crude Distillation Unit and a Catalytic Reforming Unit — together with associated utilities and tank farm facilities. Establishing a complete inventory of release points is the essential first step in any emission assessment. Based on field data collection and review of available facility documentation, including process flow schemes, P&IDs, design bases, material balances and heat balances, the following release points were identified and classified as either point sources or fugitive sources.

Table 1.1 — Environmental Sources ("Releases") identified at Marib Refinery

Process Unit	Environmental Sources / Releases
Crude Distillation Unit (CDU)	Emissions from crude heater to stack; overhead receivers to the flare
Catalytic Reforming Unit (CRU)	Emissions from reformer charge heaters; off-gases from fuel gas knock-out drum to flare
Utilities	DG stack; fugitive emissions from loading racks
Storage Facilities (Tank Farm)	Fugitive emissions from the storage tanks
All Units	Fugitive emissions from valves, pump seals and pipeline fittings; emissions from vehicles and dust raised by traffic on graded roads

Emission sources fall into two categories that must be treated differently: point sources, which discharge through a defined outlet such as a stack or flare, and fugitive sources, which release diffusely and cannot be captured at a single outlet.

3 Point Source Emissions

Based on field data collection and a review of facilities documentation — process flow schemes, P&IDs, design bases, material balances, and heat balances — the following stationary point sources were identified:

Process Heaters:

Five process heaters operate at the facility, all fired exclusively on natural gas from the Marib field:

• Two crude distillation unit heaters: F-181 and F-182, each rated at 16 mmBtu/hr heat input

• Three catalytic reforming unit heaters: H-1 rated at 20 mmBtu/hr, H-2 rated at 14 mmBtu/hr, and H-3 rated at 9 mmBtu/hr

One Flare:

• Capacity of 25,000 kg/day

Standby Diesel Generator

• One standby diesel generator unit (250 KW)

Emission rates for the five principal pollutants — nitrogen oxides (NOx), sulphur oxides (SOx), carbon monoxide (CO), volatile organic compounds (VOC), and particulate matter (PM) — were calculated for each source using AP-42 factors for natural gas combustion.

Table 1.2 — Emissions from Stationary Point Sources (g/sec)

Unit	NOx	SOx	CO	VOC	PM
Crude Heater F-181	0.198	0.0012	0.165	0.0109	0.0153
Crude Heater F-182	0.198	0.0012	0.165	0.0109	0.0153
Reformer Heater H-1	0.247	0.0015	0.207	0.0136	0.0192
Reformer Heater H-2	0.173	0.0011	0.145	0.0095	0.0134
Reformer Heater H-3	0.111	0.0007	0.093	0.0061	0.0086
Flare	0.394	0.000	2.141	0.810	0.040
Standby DG Unit 250 kW [PTE only — never operated]	0.750	0.035	0.159	0.078	0.051
TOTAL (actual — DG excluded)	1.321	0.005	2.916	0.861	0.112

Two results merit attention. The flare dominates CO and VOC point source emissions, which is a direct consequence of incomplete combustion inherent to flaring operations. The standby diesel generator, if ever operated, would be the single largest source of NOx and SOx among point sources, reflecting the higher sulphur content of diesel fuel and the combustion characteristics of reciprocating engines. These findings are consistent with the established emission profiles of comparable refinery equipment globally and point to flare minimization and generator fuel quality as the two priority areas for point source emission management at this facility.

4 Fugitive Emissions Sources

Fugitive emissions are diffuse releases that escape from equipment and storage rather than through an engineered outlet. They are frequently underestimated, yet at refineries they can represent a substantial share of total VOC loss. The fugitive sources identified at Marib were:

• Fixed-roof storage tanks holding gasoline, diesel, low-sulphur waxy residue (LSWR), crude oil, and slop

• Product loading racks for gasoline, diesel, and fuel oil

• Equipment leaks from valves, pump seals, and pipeline fittings

• Area sources — vehicle movements and traffic-generated dust on graded roads

Storage tank VOC losses were estimated using USEPA TANKS 4.0 software, with tank dimensions and throughput taken from field data. Loading rack emissions were calculated using AP-42 equations.

Table 1.3 — Fugitive VOC Emissions

Source	Number of Units	VOC Emission Rate (kg/year)
Fixed Roof Tanks	10 tanks — gasoline, diesel, LSWR, crude & slop	49,893
Gasoline Loading Rack (submerged fill, no VRU)	2 loading racks	131,508
Diesel Loading Rack (submerged fill, no VRU)	2 loading racks	1,180
Fuel Oil Loading Rack	2 loading racks	17
Equipment Leaks	Valves, Pump Seals and Fittings 'Estimated Count'	10,405
Total		193,003

Note: Loading rack operations account for 68.8 % of this total, fixed roof tanks account for 25.9 %, and equipment leaks account for the remaining 5.4 %. This distribution is consistent with published fugitive VOC inventories for comparable small refineries operating without vapour controls in hot climates.

5 Methodology: Key Assumptions and Data Inputs

Transparency in assumptions is what separates a defensible estimate from a black-box figure. The principal parameters applied across all calculations were as follows.

• Process heaters and diesel generator. All process heaters operate on natural gas; BTU values and emission factors for natural gas were therefore applied. Exhaust temperatures were taken from facilities design documents. A stack height of 26 m and exit velocity of 10 m/s were used. AP-42 factors (expressed as lb/mmCF) were multiplied by the fuel-firing rate to obtain emissions in lb/hr or g/sec. Heater duties were 16 mmBtu/hr for each crude heater, and 20, 14, and 9 mmBtu/hr for reformer heaters H-1, H-2 and H-3 respectively.

• Flare. Flare parameters were height 28.4 m, diameter 0.61 m, design temperature 350 °C, and capacity 25,000 kg/day. Flare gas molecular weight was assumed to be 40 g/mol with a heat content of 450 BTU/ft³. Emissions were estimated using AP-42 factors for industrial flares. In the absence of field measurements, SO₂ and H₂S emissions from the flare were assumed to be negligible — an assumption requiring future verification.

• Tanks and loading bays. Fixed-roof tank emissions were modelled in USEPA TANKS 4.0 using field-measured dimensions and throughput; loading bay emissions for gasoline, diesel, and fuel oil were derived from AP-42 equations.

Understanding the Uncertainties

A credible emission estimate is honest about its limitations. The principal sources of uncertainty in this assessment are:

• Reliance on generic AP-42 emission factors rather than site-specific measured rates

• Limited operational data for some emission sources

• The assumption of no SO₂ or H₂S in flare gas, pending field confirmation

• Equipment leaks (valves, pump seals, flanges) identified as sources but not individually quantified in this assessment

The approach taken is deliberately conservative and represents the best available method given the data environment. Where continuous emission monitoring or stack-test data later becomes available, the estimates presented here should be validated and refined accordingly.

6 Comparison with World Bank EHS Guidelines

The World Bank Group EHS Guidelines for Petroleum Refining (2007) set indicative performance levels for air emissions from refinery process heaters and combustion equipment. Table 1.4 below presents a comparison of the estimated emission concentrations from the Marib refinery's combustion sources against these guideline values.

Table 1.4- Estimated Emissions Compared with World Bank EHS Guidelines for Petroleum Refining

Pollutant	WB EHS Guideline (mg/Nm3)	Estimated Stack Conc. (mg/Nm3)	Status	Note
NOx	460	~180-250	Within guideline	Natural gas combustion — inherently low NOx
SOx	500	~5-10	Within guideline	Natural gas fuel — negligible sulphur content
CO	100 (guideline for combustion)	~150-200 (flare-dominated)	Flare: review	CO from flare dominates; flare efficiency determines compliance
PM	50	~10-20	Within guideline	Natural gas heaters produce minimal PM
VOC (fugitive)	No ambient stack limit; LDAR required	252,720 kg/yr facility-wide	LDAR programme required	Loading racks dominate; VRU installation advisable for gasoline bays

The natural gas-fired process heaters at Marib produce NOx, SOx and PM emissions that are expected to be within World Bank EHS guideline limits, primarily because natural gas is an inherently clean fuel with negligible sulphur content and lower NOx formation potential than liquid fuels. The flare represents the main compliance risk for CO, as flare combustion efficiency under varying load conditions may not consistently achieve the 98 % destruction efficiency assumed in AP-42 factors. Continuous flare pilot monitoring and flow measurement would be required to verify compliance.

The most significant compliance gap is the absence of vapour recovery on the gasoline loading racks. While the World Bank EHS Guidelines do not set a direct mass-based limit on loading rack VOC, they require implementation of an LDAR programme and recommend vapour recovery for high-throughput loading operations. The 131,508 kg/year VOC emission from uncontrolled gasoline submerged fill loading represents a significant air quality and occupational health burden that would not be acceptable under IFC Performance Standard 3 (Resource Efficiency and Pollution Prevention) for any new or significantly upgraded facility.

Conclusion

This assessment demonstrates that a rigorous, defensible air emission inventory can be developed for a petroleum refinery even where measured data is scarce — provided the methodology is transparent, the emission factors are internationally recognized, and the assumptions are clearly stated.

For the Marib Refinery, the analysis identifies the flare as priority source for any future emission-reduction strategy, and highlights storage tanks as the dominant contributor to fugitive VOC losses.

For operators, regulators, and lenders applying World Bank and IFC environmental standards in Yemen and the wider region, this kind of structured estimation provides an essential evidence base — both for environmental compliance and for targeting practical mitigation where it matters most.

---

REFERENCES

1. Chemical pollution of the air around the oil and gas refining complex, U. Sarsembin; International Journal of Chemical Sciences 12(4):1509-1526, January 2014

2. Oil Refineries Emissions Impact on Urban Localities Using AERMOD; Amira Abdelrasoul; American journal of environmental sciences 6(6):505-515 · November 2010

3. Suhar Oil Refinery Emissions Updates

4. USEPA AP-42, Compilation of Air Emission Factors, Fifth Edition.

5. United States Environmental Protection Agency. epa.gov/air-emissions-factors-and-quantification

6. USEPA TANKS 4.0, Storage Tank Emission Calculation Software.

7. United States Environmental Protection Agency. World Bank Group, Environmental, Health and Safety (EHS) Guidelines — Petroleum Refining. International Finance Corporation.