Using inside knowledge to reduce NOx

20 August 1999

Swedish utility Vattenfall has developed and implemented a number of measures for improving combustion processes at existing power plants. The aim is to minimize environmental impact and optimize boiler performance. Based on modifications done so far at a range of boilers, good results have been achieved at low cost relative to alternative methods.

After the oil crises Swedish authorities introduced incentives for the utilization of biomass and waste fuels. In parallel Sweden has introduced taxation systems on energy, partly based on environmental considerations. At present there exist taxes on nuclear energy, carbon dioxide, sulphur and a system for NOx limitation. In recent years a vigorous public discussion has developed on the problems of global warming, ie carbon dioxide limitation.

As a result of the ever-changing political environment and fuel tax regime, plant owners were forced to make many investments in both new and old plants. Fuel flexibility has been essential as the incentives are often changed. Also meeting ever more stringent emission requirements has been a major preoccupation.

Many plants in Sweden have been modified, sometimes in a number of steps, to meet the new constraints, taxes and political expectations. A number of plants have been switched from oil to solid fuels over the last 20 years, while in the last decade a NOx tariff has made it economic to decrease emission of nitrogen oxides to low levels for all kinds of fuels.

In response to this, Vattenfall has developed a method for meeting the various demands placed on plant owners, with the focus on cost-effective decrease in NOx emissions.

The Vattenfall concept

In the case of operating power stations, the improvements needed and the possible ways of carrying them out always vary from plant to plant. A standard solution or a standardized method will often not be optimal for a particular problem. In fact standardized procedures can be very expensive and time consuming. Instead Vattenfall has developed a range of schemes for a range of applications by combining a basic problem-solving approach with the combined experience amassed from many projects. Essentially, up-to-date knowledge of the different basic technologies of combustion and emission reduction is combined with modern computational tools and traditional modelling of fluid dynamics.

Boiler furnace fluid dynamics is generally not sufficiently focused on boiler design or on boiler modifications. The reason is usually that fluid dynamics is considered costly and time consuming. But looked at in the context of an even costlier boiler modification, it might be a very sensible thing to do.

Another very important consideration is that it is only recently that computational techniques have attained the level where full blown 3 D modelling of a combustion chamber is accurate enough to be relied upon.

Modification vs cleaning

Is it best to reduce emissions by cleaning flue gases or by modifying combustion systems? This has always been a matter of discussion. The argument has traditionally been settled by the fact that, in general, combustion modifications have been much cheaper, although somewhat less effective. So in a case where emissions requirements were particularly onerous the flue gas cleaning option has been chosen, otherwise the less expensive combustion modification route has been preferred.

However, combustion modification has tended to be standardized and not always as effective as expected. Further, there were several companies active in the market who really did not know what was happening inside the combustion chamber. In larger boilers this is not easy to establish, and in some cases it is not even possible to take a look inside. Consequently the results were generally poor and when success was achieved it was considered to be more or less a matter of witchcraft. On the other hand several companies were serious and developed a good knowledge of a certain type of boiler, eg tangentially fired or some other type, and gained good results with a minimum of effort.

Good technologies develop quickly, but not always in the predicted way. For example, some years ago selective catalytic reduction (SCR) was considered to be very effective but too expensive for many plants. However, the investment costs decreased substantially when competition between manufacturers increased. At the same time alarming test results showed that the catalysts were badly contaminated if something other than gas or a good coal was fired. A few years later, procedures to retain or maintain the activity of the catalyst even when using waste fuel were developed. This illustrates the dynamic world we operate in when choosing between combustion modification methods, selective non-catalytic reduction (SNCR) and high or low dust selective catalytic reduction (SCR).

In principle, a well controlled combustion process will have several positive effects relative to a less well controlled process, such as higher capacity and higher efficiency in combination with lower emissions, but flue gas cleaning methods might be more efficient and less sensitive to changes in the process parameters. Usually the two in combination are very powerful. The choice depends on the situation at hand, the boiler configuration, the environmental constraints and fuel supply.


The concept adopted at Vattenfall goes back to a decision we faced regarding a substantial investment in one of our largest oil-fired plants, Stenungsund. The choice was between adding flue gas cleaning or rebuilding large parts of the four boilers, each of which has 18 burners. Both alternatives were very costly. We decided to try a novel technique.

Mathematical modelling

Mathematical modelling does not have the status of exact science. To model a whole boiler demands large computing capacity and a clever approach. The chemistry, the heat transfer and the fluid dynamics have to be included simultaneously. Further, the results must be of such quality that one can rely on them. Such a method was not to hand a couple of years ago. The solution we adopted lies in adopting a stepwise approach. We modelled the existing boiler using a high performance Cray computer. The computational work was done at SINTEF in Trondheim. Using a technique called table chemistry they included the chemical reactions. With this technique reactions are calculated in advance and the results called upon when the fluid dynamic and thermal modelling is performed.

Measurement to validate modelling

Local measurements cannot be performed completely in large boilers. They are too large to reach inside with a suction pyrometer for example. However one can measure at some places. A temperature in one spot, a chemical analysis somewhere else, etc. Thus one can do calculations at these points and check whether the model gives right answers. Then some changes are made in the combustion parameters, such as excess air, load, or numbers of burners in operation. The calculation is checked again at the points where measurements can be taken. In this way the model can be verified and validated.

Even then we did not rely on the modelling, so some inexpensive physical changes were made, such as rebuild of some air inlets and change of oil. Again the measured values were compared with modelling results. All the time the model was refined and became more and more accurate.

Physical modifications

At this stage we did dare to rely on the results, and some extrapolation was done to find the optimal configuration for this particular boiler giving least NOx emissions. It was concluded that with some relatively inexpensive modifications very good results could be achieved.

The least expensive rebuild was done and again the results were measured and checked with the modelling results. We then went ahead with the full rebuild, with great success.

Physical modelling for screening

Obviously this approach was very time consuming, the calculations were very large and expensive, but the methodology was very attractive. However, the development cost for the calculation and other costs were much less than the money saved relative to the alternatives. To speed up the process and to make the modelling more effective it was discovered that physical modelling could be used for initial screening of alternative options for boiler rebuild. Physical modelling is very well developed and cost effective. Thus the mathematical modelling could be concentrated on alternatives that looked promising. However, the physical models can only model some isothermal mixing and the fluid dynamic behaviour of the gas. A skilled person who can translate the flow and mixing pattern to the creation of undesired species must do the rest. Figures 5-6 give examples of physical modelling results.

The necessity of a skilled human brain

It must therefore be stressed emphatically that computers or physical models cannot do the job without the judgement of skilled humans. The computer cannot come up with good ideas as to how to alter the design in the right direction. It cannot evaluate the results, since different parameters act in different directions. A human brain is also essential when it comes to designing a process which actually reduces NOx emissions. The models can only calculate results for a chosen solution, which is always a multi dimensional compromise between many factors.

However, with the right people available, and the necessary modelling tools to hand the task is very easy, the method very effective and financial savings substantial.

Limitations of the method

The method described of course has several limitations. The most obvious is that if emission levels have to be very low combustion modifications alone cannot achieve them. It is not possible to get down lower than say 20 mg/MJ with light oils, 40 mg for heavy oils and for solid fuels the emissions depend on the nitrogen content in the fuel.

Other limitations lie in the boiler design in the form of physical limitations. A low temperature staged combustion flame is by definition larger than an intense flame with high temperature, and thus the combustion chamber size might get troublesome. Other limitations can lie in air channel areas, in fans, in flue gas passes and the induced draft fan, typically when flue gas recirculation is used. However, care in the analysis and in the choice of modifications usually can overcome most of these limitations.

Another limitation at present is limited experience with solid fuels. Here the mathematical models for the NOx chemistry and combustion process are not very well developed. The chemistry is much more complicated and the heterogeneous reactions governing char combustion more difficult to model. A complete model of a turbulent multiple pulverized fuel flame has not yet been developed anywhere in the world, but rapid progress is being made. Work is underway at Vattenfall and the results so far are promising.

Additional positive results

When analyzing boiler internals, one realizes that usually the present situation is far from optimized. Even the fundamentals, eg that the presence of combustibles and an oxidizing agent, plus the local heat transfer, govern the local temperature and reactions, do not apply anywhere. The reason is of course that it is often not possible to even look inside a large boiler. Further the designer will typically have based the design on a previous design but with alterations. Also the people operating the boiler very often, over the years, deviate from the way the design was originally intended to be operated. In a way the situation is like the blind leading the blind. When information is gleaned through modelling and measurement activity it often reveals that what is actually going on inside is not what was believed to be the case.

Very often the mixing between combustibles and oxygen is imperfect. Stratification and large scale standing vortices prevent fuel from meeting the oxygen. This creates the familiar situation that the excess air ratio cannot be lowered below a certain point, because the carbon monoxide level rises. When modelling the situation one realizes that this is usually caused by combustibles reaching the superheater before they mix with air, the gases are cooled down, and reactions are retarded or stopped. In other regions of the cross-sectional area of the radiative superheater, very oxygen rich gas is passing out to the stack. When this situation is changed in the computer model and in the real world, increased efficiency and higher capacity is achieved. Also emissions of several undesirable substances are avoided, such as unburned hydrocarbons and carbon monoxide.

The temperature is also often uneven due to mixing problems. This can cause the obvious situation that at the very local level nitrogen oxides are formed, while unburned matter is left elsewhere in the combustion chamber. Further the superheater may not be functioning well due to the absence of any real temperature control. The thermal stresses are not what they were believed to be because the temperatures are not at all what was assumed in the design. Knowing what is happening inside the boiler can often cure this as well.

The extent of deposits in a boiler often depend on one of two things. Either luminous flames reach a cooled surface and the combustion is quenched, with the unreacted matter forming a deposit including ashes and soot, or the ash in the flame is partially or completely melted and reaches a cooled surface. The ash solidifies and forms a layer of deposits. These two possibilities can be avoided if one has control of the flame pattern and the local gas temperatures. But both of these are difficult in large boilers since one can neither take measurements everywhere nor even look inside.

Furthermore, availability of the boiler plays a very important role in the economics of power plant operation. If the temperature field is controlled better, there will be less thermal stress, thus longer lifetime can be achieved. Less deposits, less corrosion and better utilization of available surface will result in better reliability and higher availability.

All in all, a relatively good knowledge of the internal situation in the combustion chamber can solve several problems. Unwanted emissions can be lessened, efficiency can be increased as can the capacity, deposits can be avoided, corrosion can be inhibited, higher availability will result, and the whole situation can be much more easily controlled and handled.

These "add-on" benefits of our work have in some projects become in some ways the primary goal and can by themselves more than justify the effort put into the analysis.

Modifying the boiler

To obtain good results of course some alterations need to be made to the boiler. Usually they are done in order of ascending cost, with easiest and cheapest changes first. One can usually divide the modification options into four groups:


  • Change in burner setting;
  • Optimize the air preheat;
  • Clean the boiler;
  • Get control and measurement systems in order.

    Group 2.

  • Make burners individually controllable for fuel and air;
  • Optimize atomization, burner nozzles;
  • Test other fuels;
  • Control grate processes.

    Group 3.

  • Install modified preheat systems;
  • Introduce staged combustion;
  • Install flue gas recirculation;
  • Install overfire air;
  • Modify burners or grates;
  • Modify heat transfer surfaces such as primary superheaters or wall configuration;
  • Install injection systems.

    Group 4.

  • Change burners;
  • Install SNCR;
  • Install SCR;
  • Rebuild major parts of the boiler.

    As a general rule the measures in group 1 and 2 are usually performed in all cases, whereas those in group 3 are the critical ones. If measures in group 4 are necessary the project has turned into a more or less standard project. In general we have noticed that this is not necessary. If one invites a tender from manufacturers, their standard solutions are the ones in group 4.

    In general the standard solutions are the ones in group 4. When they are not combined with the previous ones they are less effective and more expensive than they otherwise might be. Further, many of the built-in imperfections are still there, such as bad mixing.

    The reason that most companies offer solutions from group 4 is of course that this is where the measure is a saleable physical product. This is where they earn money. In group 3 and above we are talking about in-depth knowledge of the processes going on inside the boiler as the basis for changes and this is not considered a product. But we think it is the most valuable product of all.

    Examples of results achieved with the Vattenfall NOx minimization concept


    Fuel Favoured
    Plants where Vattenfall has applied its NOx minimization concept

    Linkedin Linkedin   
    Privacy Policy
    We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.