As identified in Part 1 the following gas contaminants are particularly problematic when found in fuel gas for reciprocating engines (in no particular order): non-methane organic compounds (including long chain hydrocarbons); hydrogen sulphide; water; carbon dioxide; oxides of nitrogen; halogenated compounds; silicon; and solids.

One of the issues regarding contaminant removal is the size of the plant in question. A given technique that may be suitable and/or economic for a small plant may not be appropriate for a much larger plant, and vice versa.

For example consider water removal, one of the most common processes in fuel gas conditioning. A single gas engine may have a fuel consumption of 250-500 m3/h of gas per MWe output, depending on the fuel in use. Compare this fuel flow with oil/gas field processing units, which, at the very least, process around 12 000 m3/h. The result is both a change in scale and a change in technology, as illustrated by comparing Figure 1, showing a cartridge type coalescing filter, with Figure 2, which shows a gas field dehydration plant of the absorption type.

However, this is not the whole story since typical engine coalescers are often using treated gas with very low water content, typically 5 to 60 mg/Nm3 and, hence, have a low duty compared to in-field process plant that are usually handling water saturated gas with a water content circa 2500 mg/Nm3.

Plant size considerations apply to most process technologies used to treat fuel gas. Notwithstanding this two technologies are frequently found in both small and large plant and are being used for removal of a range of contaminants. These technologies are absorption and adsorption, the principles of which are outlined below in the context of water removal from fuel gas, often called dehydration.


For most small scale systems water is removed using coalescing type filters frequently combined with a particulate filter. If natural gas (gas H in Table 1 of Part 1) is being used as the fuel water is not likely to be significant since pipeline specifications limit water content. However, this should not be taken for granted since such specifications vary around the world. A particulate filter should always be used.

If the scheme is a large one, with significant gas flows, then water should be removed by one of the established processes, such as absorption, adsorption, refrigeration and membranes, although this list is by no means exhaustive.

Absorption systems most commonly uses triethylene glycol (TEG) as the absorbent, although diethylene glycol (DEG) is used where ambient temperatures are low. The glycol is recovered in the process. Standard glycol systems can achieve dew point temperatures anywhere between zero and -70˚C depending on requirements and system parameters although some proprietary technologies can go lower than -70. Glycol systems are very suited to very large gas flows.

The glycol absorbent, in liquid form, is fed to the dehydrator at high level where it falls through a rising stream of fuel gas containing water. As it falls through the gas the glycol absorbs water, leaving dry gas to exit the dehydrator at the top. The glycol is recycled through a regenerator that drives off the water using heat. Figure 3 shows the basic scheme.

As mentioned above absorption technology is used for removal of other contaminants by using different absorbents.

Adsorption specifically refers to physical adsorption where the water is held by van der Waals forces on the surface of the adsorbent. Molecular sieves are the most common form of adsorbent. These are porous crystalline substances with regular spaced cavities providing a very high surface area for the adsorption of water molecules. Silica gel is a common form of molecular sieve and is widely used as a desiccant. Commercial adsorbents include "molecular sieve 4A", which has a pore size ranging from 0.3 to 1 nm, the surfaces of which "hold" the water in the pore. Once the molecular sieve has reached its limit of water adsorption it has to be dried, usually using heated gas as the drying agent. Two molecular sieve vessels are deployed on a batch basis to achieve this, with one vessel in use while the other is being regenerated for re-use. See Figure 4.

Depending on the molecular sieve in use dew points between -60 and -100˚C are possible. Molecular sieve systems can remove carbon dioxide at the same time as water.

Thus adsorption is a physical process, unlike absorption, which is chemical. As with absorption, adsorption can be used for the removal of contaminants other than water through the use of different adsorbents.

Adsorption is usually more suited to lower flows than absorption but CAPEX/OPEX tends to be higher.

Membranes and coalescing filters both provide dehydration capability for low flows at relatively low cost, coalescing filters being the cheapest and coping with the lowest flows.

Non-methane organic compounds (NMOC)

These typically occur as ethane, propane, butane, pentane, etc, ie long chain or heavy hydrocarbons. These compounds increase the risk of engine knock necessitating a reduction in engine compression ratio for successful operation. This reduces output/efficiency and is obviously undesirable.

Removal of NMOC usually forms part of the overall upstream processing plant and is usually done in field. Processes to achieve this are complex and involve many other considerations relating to the whole upstream operation and for this reason will not be discussed. However, in the event the fuel gas contains NMOC that must be removed then the simplest and most cost effective approach is most likely to be a basic refrigeration system to cool the gas and knock out most of the NMOCs, if not all of them. Coalescers will remove NMOCs provided the local gas conditions are conducive to the NMOCs forming liquids that can be removed by coalescing filters.

Since the heat of combustion increases with the number of carbon atoms, and NMOCs have a larger number of carbon atoms than methane, it may be thought they would be beneficial to the engine. However, methane number is approximately proportional to the reactive hydrogen/carbon (H/C) ratio of the fuel. The H/C ratio decreases the heavier the hydrocarbon, ie as carbon number increases, leading to an overall reduction in fuel H/C ratio, corresponding to a lower methane number. Put simply, adding heavy hydrocarbons reduces the knock limit of the engine. As pointed out in Part 1 of this article, a lower methane number leads to knock thus heavier hydrocarbons beyond a certain limit reduce permissible engine compression ratio and in turn efficiency.

Heavy hydrocarbons are valuable and for this reason alone are frequently removed from natural gases by one of many possible processes. However, if the engine in question is stand alone or one of a small number then the fuel gas quantity will be small making many of the removal processes uneconomic.

Hydrogen sulphide

This is a very troublesome gas, which unfortunately occurs all too frequently. As pointed out in Part 1 it is both highly toxic and corrosive, particularly when in contact with water, the latter characteristic frequently requiring the use of special materials. Whilst the concentrations found in fuel gases are usually low (see Table 1 of Part 1) such levels are high enough to cause problems in engines and manufacturers set limits for hydrogen sulphide concentrations at very low levels.

There are many processes for removal of hydrogen sulphide, many of which are proprietary, and some of which can remove carbon dioxide at the same time. However, these are complex systems and aimed at large and continuous gas flows way beyond that of engine fuel gas requirements. However, all is not lost and several processes exist aimed at low flows, operating the plant on a batch basis. Examples of three such processes are:

  • Iron sponge. This is a very well established process but is limited to treating gas streams with H2S at low concentrations. It employs hydrated iron oxide on impregnated wood chips and is offered in its basic form by many suppliers.
  • Puraspec™ (Johnson Matthey). This process utilises fixed beds of granular metal oxide based chemical absorbents to achieve a high rate of reaction between hydrogen sulphide and the metal oxides thus ensuring complete removal of the H2S.
  • SulfaTreat/Select™ (M-I SWACO Schlumberger). This technology consists of two primary product groups for removing hydrogen sulphide from gas and light liquid hydrocarbon streams. These are SulfaTreat absorbent, a blend of iron oxides, and the Select absorbent series which is based on mixed metal oxides. Both technologies utilise fixed-bed technology. While SulfaTreat absorbent requires the gas stream to be water saturated, Select absorbent will work in both dry and wet gas applications. It is a mixture of non-hazardous granular products based on proprietary iron oxide chemistry and is engineered for the removal of hydrogen sulphide from gas streams.

A molecular sieve approach can be also be used (as per adsorption above), while there are processes employing membranes, making it possible to convert the sulphur in the H2S directly into saleable sulphur in one step. The economics of such processes need careful examination to ensure that the value of the sulphur exceeds the additional CAPEX of the plant relative to a non-recovery approach.

Other larger scale processes, include:

  • LO-CAT® (Merichem). The LO-CAT process is a wet scrubbing, liquid redox system that uses a chelated iron solution to convert H2S to elemental sulphur. It does not use any toxic chemicals and does not produce any hazardous waste byproducts.
  • SulFerox® (Shell Global Solutions). SulFerox is a redox-based process that converts hydrogen sulphide in sour gas to elemental sulphur through reaction with aqueous ferric iron. The process forms solid sulphur particles that are easily filtered out. There are three steps in the process: absorption; regeneration; and sulphur recovery. In most cases, filtration and melting produce yellow sulphur of a quality comparable to that of conventional Claus unit sulphur.
  • Activated carbon. This is a versatile adsorbent used for many gas treatment processes, one of which is hydrogen sulphide removal. It is an adsorbent and can be used in a flow scheme similar to that shown in Figure 4. Activated carbon and its associated housings are available from many suppliers.
  • Membranes. These are also used in the cleaning up of fuel gas to remove carbon dioxide, hydrogen sulphide and siloxanes (see below). They have not yet been widely adopted but are being used successfully.

Carbon dioxide

Whilst this is an acid forming gas it does not have the serious consequences (for the engine) of hydrogen sulphide. Although some engine manufacturers express concern about high CO2 content it is not normal to remove it unless it is removed along with another contaminant, such as hydrogen sulphide.

Halogenated compounds

Of concern here are fluorine and chlorine. In theory they can both be removed through water scrubbing using the water as an absorbent but such processes are not as straightforward as they may seem and are high in CAPEX. They also present a problem in disposing of the effluent.

An alternative solution is to capitalise on the absorbent properties of water by cooling the gas to below the dew point of water, which when removed by a coalescer, takes the dissolved halogenated compounds with it.


Siloxanes are a family of man-made organic compounds that contain silicon, oxygen and methyl groups. They are widely used in the manufacture of hygiene, health care and industrial products, a large proportion of which find their way via sewage and effluent to landfill and sewage treatment plant and thence to the gas produced. There are many forms of siloxanes but during combustion, and in the presence of other elements and lubricating oil, they convert to silicon dioxide that is deposited on surfaces in the combustion chamber as a hard, often white, solid on pistons (Figure 5) or a hard glass like finish on cylinder walls. Both/either of these raise reliability issues. The colour is a reflection of other minerals in the fuel.

Siloxanes shorten maintenance intervals between head and valve maintenance, spark plug replacement and oil changes. Removal of siloxanes from the fuel gas is the obvious way to minimise engine reliability/increase maintenance intervals. Various processes exist to do this but their comparative economics require detailed evaluation to see which is most appropriate for the level of siloxanes in the gas under consideration. There is the added complication that the level of siloxanes in the fuel gas is not going to be constant over time. Three main processes can be considered: refrigeration; adsorption; and absorption:

  • Chilling the gas to about 4˚C by refrigeration condenses out about 50% of the siloxanes but "deep chill" refrigeration to -30˚C can improve upon this, giving up to 95% removal.
  • When adsorption is used to remove siloxanes the adsorbent is usually activated carbon. This is an effective technique but requires replacement of the carbon since its regeneration is not usually economic on-site. Other adsorbents such as silica gel and proprietary media are in use and under investigation. However, these do require heat for regeneration of the adsorbent, which has to be taken in to account when considering the economics.
  • Absorption using Selexol (polyethylene glycol dimethyl ether) as the absorbent is in use in large plant.

Of these three processes none can be said to be the "best" and each has its place depending on the relative CAPEX and OPEX for the particular case under consideration; there is no right answer.


Last, but by no means least, solids must be removed if damage to the engine is to be averted. Fortunately this can be relatively easily achieved by using an in-line particulate filter.

About the Author

Roger Allen, Maple Engineering & Project Management Ltd, UK (