The six baseload two-stroke low speed diesel engines of Coloane A power station, totalling 232 MWe, are the backbone of electricity supply in Macau, a Special Administrative Region (SAR) of China. Installed on the island of Coloane over the period 1987-96, the engines, each with an HRSG, are extremely efficient. But the downside of this has been their tendency (typical of such machines) to produce a brownish yellow plume, characteristic of the emission of NO2 formed during high temperature combustion. There is particular sensitivity to such a visible plume in Macau, where the economy is crucially dependent on tourism.

Other major contributors to the visible plume emanating from the power station were identified to be SOx, namely SO3, which is characterised by a bluish white hue, and particulates, in the form of soot and inorganics, adding black to the plume.

It was determined that all three major contributors, NOx, SOx, and particulates, would have to be reduced significantly to achieve an almost invisible plume. The efforts to achieve this were described in an excellent paper at the CEPSI conference in Shanghai (18-22 October) by Jason S.S. Leng of Companhia de Electricidade de Macau (CEM) (presenter) and Lou Meylemans of Argillon (with Karl-Eugen Meier and Raimund Mueller, also of Argillon as co-authors).

For primary reduction of emissions at source it was decided to install slide type fuel valves and an electronically controlled Alpha cylinder oil lubrication system, which cuts the oil feed rate and thus UHC and particulates.

For secondary reduction a sophisticated HRSG sonic soot removal system was installed, while SCR (selective catalytic reduction) was identified as the best option for NOx reduction – even though the technology “had not been applied to a 2-stroke, heavy fuel oil diesel engine of this power rating, especially to retrofit an existing installation”, according to Leng and Meylemans et al.

Spray tower technology was determined to be the best secondary measure for SO3 reduction and an electrostatic precipitator/bag house filter for secondary particulates reduction.

Scrubbers reduce SO2 very efficiently, however the SO3 is not captured to the same extent. As SO3 is the fraction responsible for visible plume, scrubber technology “seemed not to be the right choice”, Leng and Meylemans et al pointed out. Better was a dry process for SOx emission reduction in which fine milled CaO or CaOH is injected into the exhaust gas in a spray tower, where it captures the SO3. The solid Ca-SO3 particulate matter formed is then extracted from the exhaust gas with very high efficiency in an electrostatic precipitator or bag house filter. The filter also removes the particulates coming from the engine.

However, it soon became apparent that limited space on the power plant site, which is bounded by ocean shore line and a mountainous area, would make installation of a large spray tower, bag house or electrostatic precipitator very difficult.

It was therefore decided to concentrate the emission reduction programme on NOx and particulate control.

SOx emission reduction, which is directly proportional to the sulphur content in fuel used, would be reviewed in the light of the results of the emission control programme executed. As a potential primary measure to reduce SOx emissions, heavy fuel oil with lower sulphur content or diesel fuel oil could be used, however, that would increase the operating cost of the facility significantly.

According to Leng and Meylemans et al, the maximum emission values per pollutant, which the suppliers had to guarantee, were defined by the target of minimising visible plume. Therefore the acceptable emissions are well below the guidelines of the Macau government, World Bank, US-EPA and German TA-Luft for large oil fired combustion engines.

Because plume visibility is very difficult to define quantitatively, the practical approach is to agree on a measurable emission concentration which is low enough to reasonably expect an almost invisible plume, say Leng and Meylemans et al. Published regulatory guidelines stipulate that the threshold limit for visible plume depends on the pollutant, the concentration and the stack diameter. For multiple adjacent stacks and the pollutant NO2 the not-to-be-exceeded emission level is, according to Leng and Meylemans et al:

50/(stack diameter [in meters] x (sq rt 2).

So for Coloane A units G7/8, with stack diameter 2.4m, the NO2 should be kept below 14 ppm, while for G5/6 the figure is 16 ppm and for G3/4 it is 20 ppm. The resulting maximum total NOx concentrations, as presented by Leng and Meylemans et al, were calculated to be (@ 15% O2): 290 mg/Nm3), 340 mg/Nm3, and 300 mg/Nm3, respectively; and for NO2, 25 mg/Nm3, 30 mg/Nm3 and 35 mg/Nm3 respectively.

A consortium of Burmeister & Wain Scandinavian Contractors A/S (BWSC), Mitsui Engineering and Shipbuilding Co Ltd (MES) and Argillon GmbH was formed to execute the project, with a scope of supply consisting of: engine modifications, including slide valves, atomisers, electronic lubricators and cooling water bypass; sonic cleaning system for each HRSG; and the SCR backfit itself.

SCR design

The SCR system was designed for installation in humid and elevated ambient temperatures. For this reason, 25% aqueous ammonia was chosen as the reagent, rather than urea, explained Leng and Meylemans et al.

As part of the project, an aqueous ammonia ship unloading facility together with storage tanks were supplied. The resulting total storage capacity of 800m3 is sufficient for five days operation.

A compressed air system supplies the dosing equipment, soot blowing system and sonic cleaning system with a total consumption of almost 2000Nm3/h, provided by three screw compressors, with 100% backup.

Each engine is further equipped with a dosing panel, individual soot blower system for the reactor and an exhaust gas analyser to give performance feedback from the SCR system.

As described in the Leng and Meylemans et al paper, the SCR reactor is installed between the turbo charger and engine. For this reason, turbo re-matching is necessary. Only with this arrangement does the operating temperature of the exhaust gas allow a continuous NOx reduction process. The high sulphur content of the fuel used by CEM for the engines requires a minimum temperature for the exhaust gases to allow the physical and chemical processes of the SCR system to proceed.

The high operating temperature of the catalyst coupled with the high pressure of the SINOx SCR system results in maximum safety and reliability, with minimum catalytic volume. The reactor, some 4 m in diameter and 10 m high, from flange to flange, has an operating absolute pressure of 3.5 bar and an operating temperature of 490 degrees C.

The full weight of one reactor varies from 45 t to 60 t depending on engine size. Reactor wall thickness varies from 8 to 16mm. Because of the operating conditions, the reactors had to be designed as a round vessel, with the square catalyst elements packed in each layer and arranged to minimise the amount of exhaust gas bypassing the catalyst elements. Each reactor was installed with 3 catalyst layers and one spare layer for future use. For maintenance purposes, the various layers were equipped with manholes. These manholes are also used to take the catalyst elements in and out.

Engines G3, G4, G5 and G6 are equipped with one reactor per engine, while engine G7 and G8 are both equipped with two reactors.

The heavy fuel oil used by CEM has a sulphur content between 3 and 3.5%. The appropriate catalyst for this application is of the honeycomb type, according to Leng and Meylemans et al.

The homogeneous extruded base metal honeycomb SINOx catalyst (a TiO2-V2O5 type), a proprietary product of Argillon GmbH, was manufactured and specifically optimised for the project operating conditions. Parameters for optimisation are composition (active component concentration, composition modifications to enhance poisoning resistance and adjust operating temperature window) and cell density/channel diameter. Depending on engine type, exhaust gas flow varied between 137 000Nm3/h and 339 200Nm3/h. The total calculated volume of catalyst for each reactor was based on a required catalyst performance of 20 000 operating hours.

The NO2 levels in the raw gas emissions at each engine varied from 2900 to 3400mg/Nm3 depending on the engine size. The catalyst was designed for a NOx conversion up to 90%, and maximum 2.2mg/Nm3 ammonia slip.

Large amounts of soot were expected at the reactor. For this reason, each layer was equipped with sootblowers. Initial operating experience showed that the sootblowers maintained good cleaning performance, which was confirmed by laboratory results from a catalyst inspection after 4 months operation.

Catalyst life

Although the catalyst is not consumed in the process, the effective operating performance lifetime of catalyst in the environment of the exhaust gases is limited. The lifetime depends mainly on the composition of the exhaust gas. The exhaust gas of a combustion process consists, besides air, of CO2, NO2, SO2,

CO and unburned hydrocarbons, which all are regarded as non-poisoning to the SINOx SCR catalyst. However, in addition to these components, certain byproducts of fuel and lubrication oil are present in the exhaust gas, and some of them can influence the performance of the catalyst, for example reducing its effectiveness or changing its selectivity (which is the ability of a catalyst to promote certain desired chemical reactions and to suppress or limit undesired side reactions).

In the exhaust gas of heavy fuel oil combustion a variety of chemical elements are present in significant quantities, notably heavy metals and vanadium, and inorganic components such as alkaline and earth alkaline elements.

Some of them derive from fuel oil or lubrication oil additives designed to limit corrosion in the engine. The catalyst has to be designed to withstand the poisoning affect of these components. The formulation of the SINOx catalyst was adjusted to provide increased poisoning resistance to the greatest extent possible.

Nevertheless, because of the high concentrations in the exhaust gas of components damaging to catalyst performance, the catalyst lifetime will be limited. Typically, this is compensated for in the SCR design by increasing the initial catalyst volume installed, theoretically allowing higher NOx reduction performance at the beginning.

However, when firing with heavy fuel oil it has been found that the fairly high concentrations of vanadium in the fuel are transferred to the exhaust gas and will precipitate on the exhaust gas duct surfaces and the SCR catalyst. Vanadium is one of the active components of the SCR catalyst, ie additional vanadium deposition increases catalyst activity and increases NOx reduction performance. Experience has shown that the effect of vanadium deposition can balance the impact of catalyst poisoning constituents in the exhaust gas of heavy-fuel oil fired installations.

The negative aspect of vanadium deposition on the catalyst and other exhaust gas surfaces is the promotion of undesired side reactions, such as the oxidation of SO2 to SO3, Leng and Meylemans et al pointed out. Therefore a change of catalyst after a certain service life might be required not due to NOx reduction performance out of compliance, but due to SO3 being beyond acceptable limits. To cope with the ash content in the exhaust gas the cell density (pitch) of the catalyst has been selected to achieve for an optimum of high plugging resistance and low space/volume requirement. Semi-continuously operated pulsejet cleaning devices provide additional countermeasure against plugging.

Meeting the challenges

Due to the size of engines and the use of heavy fuel oil, there were various challenges in the design of steel structure, ducting, reactor, dosing system and analyser system, explained Leng and Meylemans et al.

One challenge regarding the design of the structure lay in the fact that for two engine units, a 10 m high reactor had to be constructed on the 20 m high rooftop of an existing building. Taking into consideration that Macau is situated in a moderate typhoon and earthquake area, existing roof structures had to be strengthened for such loads.

Due to space limitations, the aqueous ammonia storage tanks were of the vertical design. Five tanks, each 4 metres in diameter and 160m3 in volume had to be designed for same typhoon and earthquake loads.

In the case of the two largest engine units, the reactors had to be installed inside the engine hall. One reactor per engine would have been too big to transport. For this reason the design called for two reactors per engine. Each reactor had its own ammonia injection nozzle.

With the help of additional flowmeters to measure and regulators to control the balance of the air and ammonia injection system, good results were obtained during commissioning.

At a 90% connversion rate for the NOx, all the engines had less than 2ppm ammonia slip. Although up to 95% conversion would have been possible, the 90% was maintained to conserve the catalyst for a guaranteed 20 000h lifespan without reduced performance, Leng and Meylemans et al note. The ammonia storage and pump system showed rapid filter clogging during the initial months of operation. Samples taken from the filters exhibited high concentrations of lime. These high lime concentrations suggested that, most probably, the aqueous ammonia had not prepared with demineralised water. This situation improved but high iron content continued to cause clogging of filters. The high iron content was probably caused by the fact that the aqueous ammonia had been prepared and transported in carbon steel tanks.

At the plant itself the entire ammonia delivery and storage system, from receiving to the point of injection is made of stainless steel.


At the conclusion of their CEPSI 2004 presentation the authors summarised the achievements of the project in the following table (which gives emissions before and after in mg/Nm3 at 15% O2 dry): (see table a)

The SCR system has proved to be reliable, performing well within the contractual requirements, with, as already noted, ammonia slip at all reactors less than 2 ppm at a 90% NOx conversion. But as the table, and photograph, suggest, while NOx reduction has been successful, SO3 production has increased (a not untypical side effect of such an SCR backfit) as has water vapour in the plume. The result is that the level of plume visibility is still regarded as unacceptable.

Reduction in the sulphur content of the fuel oil, by blending with lower sulphur oil, has proved effective. Another possibility is lowering of the engine operating temperature by modification of the turbochargers.

Secondary abatement measures under consideration include revisiting the possibility of installing SOx scrubbing equipment, as well as modification to the catalyst.


Diesel engine generating sets at Coloane A (all Mitsui-MAN B&W)
table a