It is estimated that about 150 billion cubic metres of natural gas is flared or vented globally every year – the equivalent of about 30% of the gas consumption in the European Union (Source: Worldbank Global Gas Flaring Reduction (GGFR) partnership).
Apart from being a major waste of a valuable energy resources, gas flaring also emits greenhouse gases, i.e. CO2 and methane, and other harmful pollutants into the atmosphere, such as nitrous oxides (NOx), sulphur oxides (SOx), Volatile Organic Compounds (VOCs) and particulates – smoke.
Venting releases VOCs, and other harmful gas constituents, directly to the atmosphere and is only an option where flaring is not feasible. Noise, noxious odours, radiation, and ignition hazards can also result from flaring and venting. These unwanted side effects are being increasingly frowned upon by regulators, NGOs (non governmental organisations) and the general public.
Governments around the world are therefore taking progressive action to reduce gas flaring and venting. The most prominent regulations currently exist in the EU, Norway, Canada and the US. Developing countries are gradually adopting best practises and knowledge gained from their participation in schemes such as the UNFCC (United Nations Framework Convention on Climate Change) Clean Development Mechanism (CDM).
The integrity of flare gas data is crucial if reductions are to be genuine and transparent. This is reflected in the measurement and reporting guidelines introduced by some countries in support of legislation and burgeoning emissions reduction schemes.
In the EU, greenhouse gas emissions from large combustion plants are covered by the EU Emissions Trading Scheme (EU ETS) (2003/87/EC), Phase III which commenced in January this year. This introduces a number of key changes such as the extension of auctioning of allowances for many activities, the inclusion of new sectors and gases and a drive towards obtaining harmonisation across Member States.
The accompanying Monitoring and Reporting Regulation (MRR) ((EU) No. 601/2012)1 effectively replaces the previous Monitoring and Reporting Guidelines (MRG 2007), and strengthens the need to demonstrate compliance with uncertainty targets, setting more rigorous verification requirements and guidance to auditors. Phase III also requires the inclusion of flaring from chemical production sites, and the allocation for routine and maintenance flaring, to be purchased through auction.
The EU ETS applies a tiered approach to measurement uncertainty which remains largely unchanged for flaring in Phase III for a calculation based monitoring approach2. The calculation approach involves the measurement of the quantity of gas flared (pre-combustion) and a calculation of the total amount of CO2e emitted using emission and oxidation factors.
The largest emitters must meet the uncertainty requirement for the highest tier which, for flaring, is Tier 3 for flow and Tier 2b for emission factor. Tier 3 requires ±7.5% on volume (referred to normal conditions) for the largest facilities emitting a total of 500 ktCO2e (500,000 tCO2e) or more per year. The uncertainty requirement relaxes to ±12.5 and ±17.5% for progressively smaller emitters. Reported uncertainty figures must be verified annually by an accredited body. The ultimate goal is that installations meet the top tier uncertainty requirements, however, unless it is proven to be “technically or economically unfeasible” to achieve them. Greater clarification of how to determine this is given in the accompanying Guidance Documents to the MRR.
In contrast to MRG 2007, the situation regarding the flare emissions factor (i.e. tonnes of CO2e released per quantity flared) is now clarified in a separate section in the MRR. For top tier (Tier 2b), the emission factor can be determined by process modelling simulation. No uncertainty requirements are specified for this method, however, since the uncertainty in process modelling software is largely undocumented.
Some regulators may therefore choose to impose tighter regulation on the emission factor. In some instances it may be possible to determine the emission factor more accurately using sampling and/or measurement methods, thus reducing the risk of exposure to bias that may occur using process modelling alone.
The EU ETS guidelines are not prescriptive in terms of specifying any one flow measurement technology over another, nor do they require a dedicated flow meter to be installed in the flare line. Thus, a number of approaches and methods may be used to derive the amount of gas flared. As a result, there may be variations in the level of stringency enforced by the regulators.
Some countries are specifying even lower uncertainty targets for flare. For example, the Norwegian regulator (NPD) requires flare gas volume to be measured to within ± 5% and includes flare within their fiscal measurement standards, e.g. NORSOK I-104. The main driver of this was the CO2 tax introduced in 1991 and, more recently, a tax on Nox – Norway voluntarily joined the EU ETS in 2008. A separate NOx fund has been set up whereby the oil and gas industry contributes to the development of emissions reduction measures across the Norwegian industry in return for a reduction in tax on NOx from its gas turbines and flares.
In Alberta, Canada, all facilities flaring and venting in excess of 500 m3/day, and all sour gas flares, must be metered. Alberta also has some of the most detailed guidance for the measurement and estimation of flare gas arising, primarily, from the high instance of sour gas wells in the region (Energy Resources Conservation Board Directive D017). USA perspectives
The recent regulatory measures in the United States, and pressure from initiatives such as the World Bank’s Global Gas Flaring Reduction scheme, indicate that the ‘easy ride’ oil and gas operators have had regarding gas flaring may well be coming to an end.
There is currently a drive in the USA to gather accurate flare and vent gas data in order to provide a baseline for developing future emissions reduction schemes. The Environmental Protection Agency (EPA) is largely responsible for enforcing federal laws protecting human health, and the environment, across a large proportion of the US and surrounding federal waters. The relevant legislation for gas emissions is the Clean Air Act (CAA) and Federal Rules pursuant to this Act.
Federal Rule 30 CFR 250 Subpart K targets offshore facilities processing more than 2,000 barrels of oil per day, with operators now being required to measure volumes of gas to flare and vent to within ± 5%. The American Petroleum Institute (API) has also produced flare measurement guidelines with a view towards helping operators achieve these targets (API MPMS 14.10). To date, operators of onshore facilities in the US are not required to measure their emissions to a specified uncertainty, but this may well change in the future.
Federal, state and district law is complex on the issue of licensing and regulation for onshore facilities. Some states have produced their own regulations covering emissions from flares at refineries (e.g. Rule 1118 in California).
Significantly, the EPA reached a settlement with the Marathon Petroleum Company (Marathon) in April 2012 for violation of the Clean Air Act at its flares at refineries in Texas City and Detroit. This was under the remit of its National Petroleum Initiative. The EPA claimed that the oversight in the VOC emissions reported by Marathon – resulting from inefficient flaring – came to a total of about 5,000 tons per year.
In what is seen as an industry first in the US, Marathon has agreed to invest more than $50 million in upgrades to its flaring systems in order to better control the release of VOCs. Marathon were also fined $450,000 by EPA.
These upgrades include installation of flow metering and gas analysis equipment, and measures to improve the combustion efficiency through the automation of steam and supplemental gas in order to ensure a high enough heating value of the flare gas during conditions such as high winds. It is likely that many more settlements will follow in the wake of this agreement.
The accurate measurement of flare or vent gas is key to ensuring both compliance with regulatory standards and to producing real and verifiable emissions reductions. The unpredictable nature of flaring makes flow measurement particularly challenging and difficult, however.
During routine flaring, velocities of around 0.05 m/s (or even less) can exist, rising rapidly to 50 – 100 m/s (or beyond) during emergency blow-downs. These conditions require flow meters that can operate adequately over a very wide range, have a fast enough response to cope with sudden changes in flow and are sensitive to very low flows.
The changing operating conditions can also induce large variations in temperature and composition. Liquids and solids entrained in the flare line may cause measurement errors and can temporarily knock out meter signals. Moreover, the flare meter must be robust enough to survive these testing conditions.
The primary purpose of a flare is to dispose of gas quickly and efficiently, and it is therefore imperative that any installed equipment in the flare line does not incur a significant pressure loss. The diameter of a flare line is typically large, which rules out most meter technologies on the grounds of cost and operability. Most flare meters therefore tend to be of the insertion design that can be retrofitted to existing flare lines.
Environmental factors also have to be considered when selecting and configuring measurement equipment, including the risks posed by explosive and corrosive elements within the gas stream. Health and safety considerations mean that access to flare meters is often limited and installation and maintenance must generally be carried out during shutdowns.
Flow meters ‘hot-tapped’ onto existing pipe cannot be removed for calibration in a flow laboratory. The difficulties accessing, removing and replacing even spool-pieced meters and difficulties accessing and breaking into parts of the pipework means that the only options are generally in-situ flow calibration – e.g. nucleonic gas tracer techniques – or spot checks of the zero reading from the measurement transducers.
On onshore refineries and chemical processing facilities, a key issue is often the widely variable composition. During some process outages the gas may contain, among other things: hydrogen, unsaturated hydrocarbons (e.g. ethylene or acetylene), hydrogen sulphide and aromatic hydrocarbons (e.g. benzene or toluene) in addition to the gas used to purge the system – a mixture of alkanes (typically 80-90% methane and ethane), CO2, nitrogen and water vapour. Large quantities of some components can cause significant measurement errors or a complete failure of the flow metering system.
Changes in gas viscosity and density also affect the shape of the velocity profile flowing through the flare line. At very low flow, stratification, unsteadiness and localised recirculation regions are common – potentially leading to high errors, and even falsely negative flows if the meter happens to sit within a recirculation zone.
The low flow region will often be important as it tends to make up a large proportion of the gas flared and can be one of the biggest measurement challenges to be overcome. Meter manufacturers have recognised this issue and are developing newer sensors designed to cope with both high and low flow; however, measurement still remains difficult when the flow does not meet the criteria of fully developed flow.
Operators are often required to determine flare gas composition in order to calculate emissions factors or to determine mass flow from a volumetric meter. This can be done by sampling the flare gas –pre-combustion – or by carrying out remote sensing of the gases produced, post-combustion. Both approaches are advocated in the EU ETS, in theory, but the latter is not used for monitoring flares owing to the costs and complexities involved.
Sampling from flare lines is generally prohibited on health and safety grounds, with periodic samples being taken from feed lines, in conjunction with process modelling, as a proxy. Gas Chromatographs (GCs), densitometers and analysers tend to be used more on onshore facilities.
Operators have the option of using a combination of measurement approaches and this often entails techniques such as ‘by-difference’ calculation, making use of other measurements on the platform to estimate the gas flared, e.g. the by-difference method whereby gas flared = gas in – gas out.
When flaring is relatively high, e.g. during an emergency blow-down or compressor trip, such methods can offer the only feasible way to plug gaps in the data where flare meters may over-range or fail to read a signal. The major drawback with the technique is its inherent inaccuracy when flaring is low and, therefore, calculated as the difference of two much larger numbers. In such cases the uncertainty has been demonstrated to be well in excess of 100%, with the result that the flow can, in fact, be calculated as negative, on occasions – which is impossible!
A more pragmatic approach might be to set up one or more flare gas meters to cover the low-to-medium flow range, with estimation techniques being reserved for high flaring episodes, which are normally infrequent and do not require the same level of accuracy. At low flow, consideration can be given to using meters on smaller pipes feeding into the flare header. This approach is more common on onshore facilities than offshore owing to the technical challenges and costs involved with installing such equipment.
From laboratory to real world
The ultrasonic time-of-flight flow meter (USFM) is one of the most widely used dedicated flare gas flow meters. The main advantages of USFMs are their very wide measurement range, absence of moving parts and low pressure drop. USFMs can also output molecular weight calculated from speed of sound, temperature and pressure measurements to determine density and, potentially, an online estimate of CO2 emission factor as has been used in some instances in Norway.
Other technologies that are being used are thermal mass probes and, more recently, optical correlation meters. Insertion vortex meters and averaging pitots have a limited measurement range and sensitivity to low flows.
There is often a misconception that the measurement uncertainty quoted by the meter manufacturer is tantamount to the uncertainty of the meter as installed, but this is very rarely the case in actual flare installations. These uncertainty figures are only strictly applicable under ideal flow conditions – free from liquids and solids, and with all critical dimensions accurately measured. In order to get the true installed measurement uncertainty, the influence of any additional uncertainty sources must be included in the analysis
One of the key issues is that flare gas flow meters are set up to ‘expect’ fully developed flow. In the absence of flow conditioning devices (not an option due to pressure drop), this requires long straight lengths of pipe upstream – and some downstream – of the flow meter. These requirements are rarely met due to space constraints, however. Fittings such as bends, tees and reducers, produce asymmetric flow profiles that may cause the flow to swirl down the pipe thus affecting the meter reading. This is especially true when the meter is in close proximity to the fitting, causing a misreading known as an ‘installation error’ that will not reduce on repeated measurement, i.e. a positive or negative bias in the reading. This means that there is a requirement to correct the meter to remove any errors in the reported data.
Computational Fluid Dynamics (CFD) simulation
CFD simulation offers a cost effective solution for determining installation error compared with expensive in-situ gas tracer or bespoke laboratory tests. CFD involves modelling the flow through the installation in three dimensions in order to get the flow profile. It can also be carried out before the meter is installed to determine the optimum location and configuration to minimise the error from the offset. The flow may also be modelled transiently where the flow is deemed to be unstable.
It is essential that the user has a firm understanding of both CFD simulation techniques and flow metering principles in order to select the correct models and boundary conditions to apply. It is also important to validate CFD results against experimental data where possible, but this has tended to be limited over the range of flow rates for flare gas applications.
To address these issues, NEL has undertaken a number of projects supported by the UK government’s National Measurement System which improve the knowledge of flare measurement through testing, CFD simulation and the development of rigorous uncertainty models with a view to providing input to measurement standards covering the application of flow meters in flare gas lines.
The most recent of these projects has seen the drafting of a comprehensive study into flare gas measurement from oil and gas facilities within the context of ever tightening regulatory regimes worldwide.
As flare legislation becomes more rigorous, and emission trading schemes are developed and rolled out worldwide, there will be a pressing need for operators to invest in technologies that can overcome the many challenges of flare gas measurement. This is not an easy task, but one which can be overcome provided care is taken to ensure the most suitable techniques are correctly employed and maintained.
Published: 31st May 2013 in AWE International