We are testing a new beta website for gov.scot go to new site

Measurement and Modelling of Fine Particulate Emissions (PM10 and PM2.5) from Wood-Burning Biomass Boilers

Listen

2 Development of emissions factors for biomass combustion

This section details the rationale behind the selection of boilers for emissions testing, details the results of the tests and discusses the choice of emissions factors for use in the modelling phase. These monitoring results are also compared to emissions data from some recent studies and test reports.

2.1 Emissions measurements

A total of six boilers were tested within this study to determine real-life emissions of PM 10 and PM 2.5 of a range of boilers in Scotland. The results from these tests were used to ascertain the most representative emission factors to be used in the modelling work. This section describes the process undertaken to select the boilers, the testing methodology employed and presents the results.

2.1.1 Boiler selection

A range of boilers were selected for testing based on four main criteria:

1) Boiler rating expressed as heat output in kW;
2) Fuel types (pellets, wood chips or logs);
3) Boiler make and model;
4) If external particulate abatement was fitted.

Boiler testing was carried out at six separate sites taking into account a variety of examples within the above criteria. Recognition also had to be made on the availability of the boilers for independent testing. All boilers were tested with the aim of getting real-world operational PM 10 and PM 2.5 emissions data to provide a reflection of current on-site performance. All boilers were therefore tested under normal operational conditions and under supervision of owners/operators.

2.1.1.1 Boiler Rating

It was important that the boilers selected for emission testing were representative of what is currently operating within Scotland. The amount of information available on the size and types of boilers operating was very limited. As part of this study a list of existing plant was compiled to identify what technology was currently being used. The list was generated from accurate information provided by a number of grant funds such as Scottish Biomass Support Scheme, Scottish Enterprise and other grant schemes such as Scottish Community and Householder Renewables Initiative ( SCHRI) and Highlands and Island Enterprise ( HIE).

This information was analysed to determine the location of the boilers, their size and type of fuels used. A total of 128 units were identified as part of this process. The total installed capacity in Scotland was calculate to be 18,428 kW. The existing applications were almost all hot water boilers that ranged from small domestic heating plants at 20 kW to larger commercial boilers at 2000 kW. The average installed capacity was found to be 222 kW and the most common units fitted ranged between 100 and 120 kW.

2.1.1.2 Fuel supply

The three main fuel types commonly used are Wood Chip, Wood Pellets and Log Wood.

Wood pellets are compressed wood generally made from sawdust and wood shavings and are typically 6-12 mm in diameter and 6-20 mm long. Pellets have the main advantages of being dry and consistent in shape and composition. This allows them to be easily handled and they flow freely from hoppers and through fuel feed systems to the boiler and hence make them ideal as a fuel for automatic feed systems. Pellets however are more expensive than wood chip or log but they have a higher density (600-700 kg/m 3) and so can be transported in larger volumes reducing transport costs.

Wood chip can be produced from a number of wood sources and are normally produced by industrial chippers that consist of rotating blade that reduce the larger wood stems to various sizes but typically 15 - 30 mm by 5 -10 mm. The final chip will have the same moisture content as the wood stems and require air drying by either natural means or a drying air floor system. The density of the wood chips is much lower than pellets at 200 - 350 kg/m 3 and therefore transport costs are higher. Wood chip quality can be problematic to automatic boiler plant and can cause blockages to feed systems.

Log wood, like chips, can be produced from a number of sources but generally is produced from smaller round woods that are of much lower value to forestry and sawmill operations. Good quality log wood will be cut, split and stored in the winter for use the following winter. Logs for automatic boilers are 300 -500 mm long and a maximum of 70 mm diameter and they should be stored under cover with free ventilation to promote good drying. The density of log wood is again much lower than pellets at 300 - 550 kg/m 3 and therefore transport cost are higher.

2.1.1.3 Boiler Designs

Biomass boiler manufactures offer a very wide and diverse range of units that are specifically designed to meet various efficiency and emission criteria predominantly. Because of this wide diversity of boiler size, design criteria and fuel type it was important to consider testing boilers which were representative of equipment used in Scotland.

The range of boilers available can loosely be characterised into three main technology groups. These are type of heating application; fuel type (discussed above) and combustion technology.

Boilers can be used to produce hot water or steam. Steam boilers are normally used by the industrial sector for process heating application. Other instances where steam boilers would be used are for larger heating applications where steam to hot water heat exchangers can be employed. These types of boilers can be found, for example, as larger centralised Hospital boiler plants. Hot water boilers can be supplied as low ( LTHW), medium ( MTHW) or high temperature ( HTHW) hot water boilers. This classification is determined by the type of heating system to which the boilers are applied. In nearly all instances the units in Scotland found were of the LTHW type and therefore these were put forward for testing.

The biomass boilers in Scotland were comprised of three key types of combustion technology: overfeed; underfeed and moving grates.

  • Overfeed stokers are generally used for larger sizes of fuel such as logs. These boilers operate on a downdraft principle where air is forced down through holes in the combustion chamber to a secondary chamber where more air is introduced for final combustion. Logs are automatically fed in from the top in the first combustion chamber. These boilers are generally up to 70 -100 kW and because of the size, feed system and lack of abatement fitted they tend to have the highest emission levels.
  • Underfeed stokers are designed generally for smaller wood chips or pellets less than 50 mm where the fuel is fed into the combustion chamber of the boiler by screw conveyors. This type of design covers boiler up to 2 MW capacity. Since fuel feed is controlled and abatement technology is used this type of combustion results in lower emissions than the overfeed stokers.
  • Moving grates are identified by the way that the grate moves such as vibrating, inclined, horizontal or travelling. This design is a further improvement on the underfeed stoker as it allows for more control of the air and fuel mixing at various stages of the combustion process. This gives the advantages of being able to handle a wider specification of fuel types and moisture contents and is generally applied to larger Industrial applications in excess of 2 MW.

2.1.1.4 Abatement technology

Smaller biomass boilers are generally not fitted with any pollution abatement devices as these are generally not required to meet current CAA requirements for emissions. However, most larger new automatic boilers are fitted with some form of flue gas cleaning device to remove particle (dust) from the flue gas before release to the atmosphere.

The dust collection system has to be chosen with respect to the required emission level and the actual operating conditions. For many small grate boilers and some underfeed boilers, single or multi cyclones are sufficient to meet the required emissions. The cyclone removes the coarse fraction but does not remove smaller particles. The cyclones are in some instances fitted internally as an integral part of the boiler plant, however the principle of operation is basically the same.

All single cyclone designs apply the same basic principle of inducing the particulate laden flue gases to swirl around inside the cyclone body for sufficient time and with sufficient vigour that grit and dust is centrifuged to the inside wall surface and is carried downwards to the dust outlet. The point of separation of the flue gases from the particulates occurs at the base of the cyclone body, where the gases reverse direction and vortex back upwards to the central clean gas outlet tube. Multi cyclone arrestors perform in a similar way to a single cell but use a number of smaller cells contained in a chamber that is directly fitted to the boiler flue.

Other Devices

More stringent particle emission requirements can easily be met with an electrostatic precipitator ( ESP). The ESP has usually high particle removal efficiency in the complete particle size range. The removal efficiency has a minimum for particles having an aerodynamic diameter around 0.2 mm. ESP's are widely used for this type of application in Sweden, when the requirements are more stringent or the multi cyclone is unable to meet the requirements.

Fabric filter is an alternative when high removal efficiencies are required. It is not widely used for grate boilers firing wood fuels due to the risk for fire. However, fabric filter is used after straw fired boilers, after fluidised bed boilers and after multi cyclones.

2.1.1.5 Final selections made

A selection of boilers was included in this study to represent examples of each type of fuel, abatement, boiler rating and combustion technology. The proportion of fuels used was also reflected. The majority of biomass boilers in Scotland use wood chip with only a small number of pellet and log boilers and so only one example each of pellet and log boilers was tested.

The selection of boilers for testing was reviewed and discussed by the project Steering Group and a final list agreed. We believe that the resulting range of boilers tested covers an optimum range of currently available boiler sizes and technologies and is a representative sample of boilers currently installed in Scotland, mainly in rural areas. However, it should be noted that it may not be representative of boilers included in the scenario which are outwith the size range and scale of those tested.

Table 2.1 lists the critical technical specifications of the final 6 boilers selected for the emission measurement programme.

Table 2.1: Specifications of the 6 Boilers Selected for the Emissions Testing Programme.

Site code

Output (kW)

Boiler Type

Abatement

Approved Appliance

Fuel Type

A

120

Underfeed

none

No

Chip

B

600

Underfeed

single cell cyclone

Yes

Chip

C

300

Underfeed

single cell cyclone

Yes

Chip

D

220

Moving Grate

Internal cyclone

Yes

Pellet

E

70

Downdraft

none

No

Log

F

400

Moving Grate

multi cell cyclone

No

Chip

2.1.2 Testing methodology

The testing phase of this study was subcontracted to TUVNEL who provide ISO9001 and UKAS accredited emissions monitoring. As well as determining PM 10 and PM 2.5 concentrations the flue gases were also tested for other pollutants, namely oxides of nitrogen, carbon monoxide and carbon dioxide.

It is important to note that all of the tests were conducted in the same manner, i.e. as they were found in a live operational situation with no involvement from installers or manufacturers to ensure data collected was true reflection of current on site performance.

PM 10 and PM 2.5 were sampled in accordance with USEPA Method 201A. A fixed flowrate, as near as possible to the isokinetic rate, was selected and the sample was drawn from the centre point of the stack through a filter via a cyclone. The cyclone removed matter with a nominal aerodynamic diameter of 10µm and 2.5µm depending on the test required. The remaining material (<10µm or <2.5µm) was collected on a filter and the mass determined gravimetrically at a laboratory according to TUVNEL's internal procedure WI/ PE2/962 which references BSEN 13284.

Flue gas concentrations of CO and CO 2 were determined using a non-dispersive infra-red analyser. Flue gas was sampled according to TUVNEL's internal procedure Work Instruction WI/ PE2/971 which references ISO 12039. The flue gases were extracted, dried and passed through a measuring cell. Infra-red radiation tuned to a frequency absorbed by the gas was transmitted through the cell and the attenuation of the beam was recorded.

Flue gas concentration of oxides of nitrogen, NOx was determined by following TUVNEL's internal procedure Work Instruction WI/ PE2/971 which references ISO 10849. The flue gases were extracted, dried and passed through an analyser operating on the chemiluminescence principle where ozone is added to the sample gas which oxidises the NO contained in the sample into NO2. A NOx converter reduces any NO2 in the initial sample to NO. The portion of the NO2 in an excited state radiates light when it returns to normal state. The light emitted from this reaction is detected and amplified by a photomultiplier tube.

The stack conditions were also measured. The flue gas concentration of oxygen was determined using a Zirconia cell oxygen analyser following TUVNEL's internal procedure Work Instruction WI/ PE2/971 which references ISO 12039. These devices make use of the fact that oxygen ions become highly mobile in Zirconia (ZrO2) heated to temperatures above 600°C. It is therefore possible to use Zirconia as a solid electrolyte for an oxygen sensor provided it is heated (typically 750°C). The stack gas velocity, flow rate and moisture content are integral to the PM 10 and PM 2.5 measurements and the measurement techniques are based on TUVNEL's internal procedure WI/ PE2/961 which references BSEN 13284. Velocity was determined by means of a pitot tube and manometer, temperature by means of a thermocouple and the moisture content was determined gravimetrically.

2.1.3 Results

The results from the monitoring are provided in detail in Appendix 1. Table 2.2 presents the particulate emissions data in g/ GJ for each test and the average and median for each boiler. The third PM 10 test for boiler B was removed from the dataset because there was an uncharacteristically low flow rate during that period of testing compared to the rest of the tests. The PM 10 result from the last test of boiler E was not removed as there were no unusual characteristics noted during this test and so it was deemed to be a real example of emissions from the boiler (which had no abatement equipment fitted). However, we have no information on how frequently or infrequently such events may occur.

Table 2.2: Particulate emissions by test and averaged over all tests (g/ GJ).

Site

Test No.

Mean

Median

1

2

3

4

5

6

7

PM 10

A

26.6

30.9

27.9

-

-

-

-

28.5

27.9

B

55.8

54.3

-

-

-

-

-

55.1

55.1

C

19.3

10.9

14.6

-

-

-

-

14.9

14.6

D

45.2

50.3

92.8

59.8

66.7

45.2

62.3

60.3

59.8

E

3.1

25.2

28.3

32.5

18.6

18.0

355.1

68.7

25.2

F

22.1

34.4

16.1

23.4

20.9

30.5

31.2

25.5

23.4

PM 2.5

A

22.5

5.6

19.9

-

-

-

-

16.0

19.9

B

47.1

39.1

40.5

-

-

-

-

42.2

40.5

C

22.0

14.4

11.7

-

-

-

-

16.0

14.4

D

28.2

34.1

46.0

69.6

87.2

47.9

37.1

50.0

46.0

E

2.4

17.2

24.1

32.4

19.2

18.8

15.3

18.5

18.8

F

19.2

18.5

20.0

20.1

17.5

15.7

19.9

18.7

19.2

Using the fuel test data and the operational data recorded for each site it was possible to estimate the firing rate of each boiler during the tests. This was calculated for each hour of operation that the tests were carried out and Table 2.3 shows the range of firing rate for each boiler. Boilers C, D and F were operating at lower loads than the appliance rating. These data are still valid for this study, however, as it was the aim to monitor emissions from boilers in their normal operational state. This perhaps a good indication that boilers are not often operated at their full potential. There was no correlation between boiler rating and emission rates or indeed the firing rate and emissions rates.

Table 2.3: Energy rating of each boiler versus firing rate (input range) during testing.

Site

Boiler rating (kW)

Calculated input range (kW net)

A

120

154-176

B

600

545-645

C

300

102-184

D

220

35-92

E

70

55-100

F

400

133-200

2.1.4 Derivation of emission factors for use in the model

The results from the six boilers show a range in median emission rate for particulate (both PM 10 and PM 2.5) of 14.4 g/ GJ to 59.8g/ GJ and a range in the mean emissions rate of 14.9 g/ GJ to 68.7 g/ GJ. It was decided to model for a best and worst case scenario - i.e. making two assumptions: firstly, that all boilers installed would be operating close to the lower emission rate and secondly that they would all be operating close to the higher emission rate. To be broadly consistent with the test figures gathered an upper level of 60 g/ GJ and a lower value of 20 g/ GJ was taken forward to the modelling phase. It is recognised that this is a relatively small sample and restricted to a specific boiler capacity range and, hence, extrapolation of the results to other cities with potentially different biomass boiler installations needs to be undertaken with caution.

To help to put these emissions into context Table 2.4 shows some approximate emission factors for other fuels. These were derived from the UK National Atmospheric Emission Inventory ( NAEI) emission factors and making some assumptions about the efficiency of plant for each fuel.

Table 2.4: Approximate emission factors (energy input) from other sectors within the combustion industry.

Fuel

Emission factor (g/ GJ)

coal

120

fuel oil

12

gas oil

5

natural gas

1

Based on the emission factors for biomass derived for this study, particulate emissions from biomass are typically lower than small coal fired plant but higher than oil and gas. Hence, where a biomass boiler is a replacement unit an important consideration is what type of combustion appliance is being replaced.

2.2 Comparison with existing emission factors

In order to put the measured emission rates into context they have been compared to emissions data from some recent studies and test reports

A very comprehensive particulate emissions study was undertaken recently by the International Energy Agency ( IEA) Bioenergy Task 32 (Nussbaumer et al, 2008). This study collected emission data directly from research institutes and universities, and considered data available from literature sources. Input was provided from the 17 institutions in the seven member countries of the IEA Task 32 (Austria, Denmark, Germany, Norway, The Netherlands, Sweden, and Switzerland). A range of data were analysed to include results from ideal operation, typical results at in-service operation and worst results at very bad handling. This therefore provided a good comparison dataset to the in-service testing that was carried out in Scotland.

The study looked at all biomass combustion equipment from a wide range of applications. The main conclusions from the report relevant to this study were the typical measured particulate emissions for logwood, pellets and chip. Table 2.5 shows a summary of the results.

Table 2.5: Average particulate emissions reported in Nussbaumer et al (2008).

Boiler type

Typical PM emission factors (g/ GJ)

Log boiler

105

Underfeed boiler using wood chip

80

Grate boilers using wood chip

60

Pellet boilers

30

Table 2.6 compares the emissions results from the monitoring of the six Scottish boilers with the typical emission rates reported in the IEA Bioenergy Task 32 report. It can be seen that all cases the Scottish emissions data were found to be lower than the typical IEA figures. This is not unexpected as the IEA study reported emissions figures from a much wider range of boilers encompassing installations of a much wider age range than those monitored in Scotland. It also specifically looked at "worse" case emissions as well as usual in-operation and ideal operation data. This provides perhaps a clearer picture of real emissions from boiler plant currently operational across Europe but is less relevant to this study where the objective it to look at emissions from more modern boilers with a view to reporting future trends with the installation of new, more efficient equipment.

Table 2.6: Comparison of monitoring data in Scotland with typical emissions reported from the IEA (Nussbaumer et al, 2008).

Site code

Boiler Type

Fuel Type

Average PM 10 emission factors (g/ GJ)

Typical IEA PM emission factor

A

Underfeed

Chip

28.5

80

B

Underfeed

Chip

55.1

80

C

Underfeed

Chip

14.9

80

D

Moving Grate

Pellet

60.3

60

E

Downdraft

Log

68.7

105

F

Moving Grate

Chip

25.5

60

The data from the Scottish tests were also compared to both the specific manufacturers test data, where available, and also some average statistics derived from over 200 type-approval test reports from the Austrian test-house Bundesanstalt für Landtechnik ( BLT (Federal Institution for Agricultural Engineering)) which were averaged for fuel use as opposed to boiler type. These are presented in Table 2.7.

A comparison with both the test data for the specific model monitored in Scotland and the average of results from BLT show that type-approval test results are generally below the real-world emissions as measured in Scotland and presented by Nussbaumer and co-workers (2008). Again this is not unexpected as the type-approval tests are undertaken under controlled conditions such as steady heat load and uniform fuel feed. In contrast boilers in real-world situations are subject to fluctuating heat demand and are unlikely to be used constantly at their optimum running criteria.

Table 2.7: Comparison of monitoring data from Scotland with typical emissions reported test reports.

Site code

Fuel Type

Average PM 10 emission factor
(g/ GJ)

Model- specific PM test data
(g/ GJ)

Average of BLT PM test data
(g/ GJ)

A

Chip

28.5

19

20

B

Chip

55.1

n/a

20

C

Chip

14.9

37

20

D

Pellet

60.3

18

17

E

Log

68.7

n/a

17

F

Chip

25.5

n/a

20

This exercise has shown the wide range of emission rates that are available in the literature. These reflect the wide range of factors that can affect the results of such testing. These include the boiler model and combustion technology used, the presence or absence of abatement technologies and the operation mode during testing. The use of real-world monitoring data from existing boilers in Scotland has gone some way to control for all of these factors by undertaking modelling work based on real emissions from boilers in operation.

2.3 Comparison with the London biomass study

The work carried out for the Scottish Government follows earlier studies of the impact of biomass combustion on air quality in London (Abbott et al, 2007). The London study was an initial assessment of the potential impact. It made various simplifying assumptions:

1) that the spatial distribution of new biomass combustion throughout London would be similar to that for gas;

2) that sources of biomass combustion could be represented in the modelling as low level volume sources throughout the city;

3) the rates of emission were based on default emission factors from the CORINAIR database.

The London study predicted that concentrations of particulate matter in the centre of the city would be substantially increased by biomass combustion.

A much more detailed approach has been adopted for this Scottish study. The key differences are:

1) new biomass emissions sources have been assumed to be installed where land is available for development, often on the outskirts of the city;

2) emission sources for large new developments have been represented as elevated point sources to take account of the dispersion of pollutants from chimneys;

3) the rates of emission were based on measurements from biomass combustion sources recently installed in Scotland. The emissions rate for PM 10 used in the London study was 66 g/ GJ. The testing undertaken in Scotland has illustrated that this was a very conservative estimate with average boiler emissions of PM 10 ranging from 14.9 g/ GJ to 68.7 g/ GJ with an overall mean of 34.3 g/ GJ.

These key changes to the methodology are designed to provide a more accurate and realistic estimation of the future impact of biomass uptake in the study areas of Dundee and Edinburgh.