Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody biomass

Energy Conversion and Management xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody biomass
gas a, L. Vorotinskiene˙ a,⇑, R. Paulauskas a, R. Navakas a, A. Dzˇiugys a, L. Narbutas b N. Striu a b

Laboratory of Combustion Processes, Lithuanian Energy Institute, Breslaujos Str. 3, LT-44403 Kaunas, Lithuania Technology and Innovation Department, JSC Axis Technologies, Kulautuvos Str. 45A, LT-47190 Kaunas, Lithuania

a r t i c l e

i n f o

Article history: Received 7 December 2016 Received in revised form 31 March 2017 Accepted 3 April 2017 Available online xxxx Keywords: Furnaces Biomass Wood chips Combustion Condensing economiser Moisture content

a b s t r a c t In small countries like Lithuania with a widespread district heating system, 5–10 MW moving grate biomass furnaces equipped with water boilers and condensing economisers are widely used. Such systems are designed for firing biomass fuels; however, varying fuel moisture, mostly in the range from 30% to 60%, complicates the automated operation. Without manual adjustment of the grate motion mode and other parameters, unstable operation or even extinction of the furnace is possible. To ensure stable furnace operation with moist fuel, the indirect method to estimate the fuel moisture content was developed based on the heat balance of the flue gas condensing economiser. The developed method was implemented into the automatic control unit of the furnace to estimate the moisture content in the feedstock and predictively adjust the furnace parameters for optimal fuel combustion. The indirect method based on the economiser heat balance was experimentally validated in a 6 MW grate-fired furnace fuelled by biomass with moisture contents of 37, 46, 50, 54 and 60%. The analysis shows that the estimated and manually measured values of the fuel moisture content do not differ by more than 3%. This deviation indicates that the indirect fuel moisture calculation method is sufficiently precise and the calculated moisture content varies proportionally to changes in the thermal capacity of the economiser. By smoothing the data using sliding weighted averaging, the oscillations of the fuel moisture content were identified. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction As environmental regulations are becoming more stringent, renewable fuel sources that substitute fossil fuels and are more widely being used for electricity and heat production [1]. For heat production, woody biomass is mostly used in the form of chips, logs, chunks and coppice stems. However, freshly sawn wood contains over 50 wt.% of moisture. This makes direct combustion difficult due to the extra energy required to evaporate the moisture from the wet fuel. According to the literature [2], the harvested wood loses 1–2% of its moisture content per month in uncovered storage, and producers of biofuels are inclined to store the harvested wood to reduce the moisture content before supplying it to heating plants. Sometimes, when there is a high demand for heat production, freshly harvested biomass is delivered. During the cold season, biomass with 30–60% moisture content is delivered to heating plants, where it is combusted in a furnace. Depending on the heat demand, different types of furnaces, such as fixed ⇑ Corresponding author. E-mail address: [email protected] (L. Vorotinskiene˙).

bed combustors (up to 5 MWth), moving grate (up to 100 MWth), fluidised bed (up to 500 MWth), and co-firing (up to 900 MWth) are used [3]. Nowadays, the moving grate or fluidised bed combustors are the most popular. Fluidised bed combustors are designed for various forms of biomass with a moisture content up to 60% [4]. However, moving-grate furnaces have difficulty burning chips with moisture content higher than 55% [5]. Regardless, combustion on grates is the most widespread heat production method in small and medium-scale heating plants, where produced heat is used for district heating in small countries such as Lithuania. To meet the requirements of district heating systems, the modern movinggrate combustors are designed for high thermal efficiency with low emissions of gaseous pollutants (CO, NOx, etc.) firing biomass with a 30–55% moisture content [6]. Even though the combustors are equipped with advanced parameter measuring instrumentation, significantly fluctuating biomass moisture levels can cause operational problems in biomass combustors, such as the lower burning stability and insufficient heat supply under conditions of elevated heat production [7]. These problems can be avoided by adjusting the boiler operating regime if the fuel properties or, at minimum, the moisture content is known. Typically, the moisture

http://dx.doi.org/10.1016/j.enconman.2017.04.014 0196-8904/Ó 2017 Elsevier Ltd. All rights reserved.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

2

Nomenclature Abbreviations FGCE flue gas condensing economiser FT-IR fourier transform infrared spectroscopy MC moisture content NIR near infrared spectroscopy PLC programmable logic controller RH relative humidity SCADA supervisory control and data acquisition Notations AFE flue gas blower reference capacity, m3/h cp1 specific heat of the dry flue gas, kJ/kg K specific heat of water evaporation (latent heat of water cp2 vapour), kJ/kg K cp,wv specific heat of water vapour, kJ/kg K moisture content in the dry flue gas at FGCE inlet hD hD1 moisture content in the dry flue gas at FGCE outlet LHV lower heating value of the completely dry feedstock, kJ/kg mD mass of combustion products, kg mF reference mass of fuel, kg mO2 mass of consumed dry air, kg mRH moisture content in flue gas obtained from the RH mfg method, kg/h

content of fuel supplied to the heating plant is measured manually by drying the sample in an oven [8], but this method takes from 4 h to 16 h to estimate the moisture content. Also, multiple samples must be taken to determine the bulk moisture content. Besides, fuel from various heaps is mixed before it enters a furnace and adjustments to the boiler operations for manually measured fuel moisture content could cause insufficient heat supply. To ensure stable furnace operation with moist fuel, automatic control of the furnace with continuous operation should respond to changing fuel parameters obtained by real-time measurements. Fuel moisture content can be measured online before it enters the furnace or estimated from flue gas analysis. Methods of fuel analysis include nuclear magnetic resonance [9], microwave [10], near-infrared spectroscopy (NIR) [11], radio frequency [12] and X-ray spectroscopy [13]. Microwave and nuclear magnetic resonance methods are more adaptable to a laboratory than to industrial applications and are unreasonably expensive to use in a small or medium-scale heating plant [14]. Using the NIR method, the fuel surface is illuminated with wavelengths from 800 to 2500 nm and the three absorption maximums are associated with moisture in fuel [15]. According to Sweden scientists [16], NIR is a suitable method for measuring the moisture in moving bark and wood chips. However, variations of fuel characteristics such as density, composition and size requires re-calibrating the NIR. Daassi-Gnaba et al. [12] proposed a moisture estimation method based on radio frequency measurements with an antenna that is fully buried into the wood chips. The results showed that this method is suitable for real-time measurements of fuel moisture content in the heating plant, but an industrial prototype is still in the testing phase. Niedermayr et al. [17] presented an innovative fuel feeding system: a crane and a sensor-equipped gripper, featuring different sensors, such as ultrasound, moisture, and image recognition with a camera. The sensor-equipped gripper analyses the fuel in the hopper and feeds the fuel with the required moisture at the required feeding rate according the boiler operating regime. This approach helps to avoid fluctuations in the boiler operation regime due to differing moisture content, as well as to

mHB mfg Patm QB Q dry fg QFGCE Tfg Tfg1 Tr YRH YHB

Dt

moisture content in flue gas obtained from the heat balance of the economizer, kg/h atmospheric pressure, Pa boiler capacity, kJ amount of heat extracted in the FGCE, kJ thermal capacity of the economizer, kJ inlet flue gas temperature, °C outlet flue gas temperature, °C reference temperature, °C fuel moisture content of fuel calculated from the RH method, % fuel moisture content of fuel calculated from the FGCE heat balance method, % time period of data recording in the SCADA system

Greek letters u measured relative humidity in the flue gas, % flue gas blower reference frequency, Hz fr tE flue gas blower frequency, Hz qssT density of superheated steam according to the flue gas temperature, kg/m3 g boiler efficiency

reduce CO2 emissions, though the feeding system is only at a lab scale and has only been tested in a thermal power plant with a capacity below 500 kW [17]. The reviewed methods for measuring the fuel moisture content [9–17] are financially disadvantageous to use in small and medium size biofuel fired heating plants. Besides, these methods require adjustments and calibration of the instruments and handling skills for the measurements [18]. In this case, indirect methods that estimate the fuel moisture content from the flue gas analysis are more suitable and economically attractive. These methods include Fourier-transformed infrared spectroscopy (FT-IR) [19], tunable diode laser spectroscopy [20], relative humidity (RH) [21] and soft-sensor measurements [22]. Kortela and Jämsä-Jounela [23] analysed moisture-saturated gas using an FT-IR gas analyser. The results revealed that the measured moisture content differs from that present in the fuel by up to 4.1%, but this result was obtained by analysing clean gas. Measuring moisture content in a flue gas duct in a boiler house with this instrument produces imprecise results because of the presence of solid particles in the flue gas, as well as the effects of pressure and temperature. More precise results can be obtained from tunable diode laser spectroscopy, but the measuring instrumentation is more expensive than that of FT–IR, and continuous cleaning of diodes and handling skills for measurements are necessary [24]. Another indirect moisture measurement method is the relative moisture sensor [25]. Hermansson et al. [21] from Sweden performed analysis to improve the accuracy (<4% error) by reducing the signal delay and thereby expanding the application capabilities of the RH sensor to measure the moisture content in flue gas from biomass combustion to determine the moisture content of the fuel. Since the sensor can operate only at a temperature of up to 200 °C, the flue gas duct was cooled before performing the measurements. The method is able to detect variations in the moisture content within seconds; however, new devices, measurements, and calibration are needed to apply this method in wood fired boilers. Kortela and Jämsä-Jounela [26] presented a soft-sensor for online monitoring of fuel moisture in a BioPower 5 CHP plant, where


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

the heat used for steam generation is obtained by burning solid biomass fuel (bark, sawdust, and pellets). The method is based on the combustion power estimation by a combustion model and a dynamic model of a secondary superheater where the fuel moisture content is estimated from the difference of the measured and estimated output enthalpy of the secondary superheater. The authors also performed experiments with spruce bark (moisture content of 54%) and dry woodchips (moisture content of 20%) using an FT-IR sensor to measure the fuel moisture to test the fuel moisture soft-sensor. The fuel moisture soft-sensor predicted the moisture content in a furnace with an error of 3.6% at a
22 min delay for changes in fuel parameters to influence the flue gas. The reviewed methods of the measured fuel moisture content have their own advantages and disadvantages. The methods developed so far require additional implementation, calibration, and maintenance, and are too expensive for use in the small and medium-size biofuel fired heating plants equipped with a condensing economizer. Therefore, a more cost-effective method to measure the fuel moisture in a small-scale heating plant is needed. This paper presents a cheap and accurate indirect method to estimation the fuel moisture content in a grate-fired furnace with a boiler and a flue gas condensing economiser (FGCE). To develop the method, the moisture content in the fuel and in the flue gas exiting the smokestack were measured and all the parameters of the furnace were recorded. Based on thermodynamic equations, the heat balance of the economizer was established and an algorithm was developed to automatically control the furnace unit that estimates the moisture content in the feedstock and predictively adjusts the furnace parameters for the optimal fuel combustion.

3

grate, which consists of two separately moving parts with 13 rows of movable grate bars in total. The combustion air is fed to the furnace at three locations. The primary air is supplied under the grate and enters the fuel layer from three separated wind boxes through the air openings installed between the grate bars. The secondary air ports are distributed over the layer of fuel and assembled in the side walls along the furnace. Finally, to ensure complete combustion, the tertiary air is supplied via the air ports located in the cylindrical duct connecting the furnace with the boiler. All the air distribution ports are equipped with individually controlled air blowers. The temperature of flue gas exiting the furnace depends upon the load of furnace and varies in the range from 900 to 1100 °C. Downstream in the boiler, the flue gas is cooled and exits with a temperature of approximately 160 °C. To control the bed and the furnace temperature, the flue gas is recirculated into the primary and the tertiary air flow. The remaining flue gas is fed into a multicyclone to remove the solid particles. From the multicyclone, the flue gas is fed into a direct contact flue gas condensing economiser. In the FGCE, water is sprayed into the flue gas to cool it to the temperature below the dew point. The formed condensate flows into a chamber installed at the bottom of the economiser, where it is pumped through plate heat exchangers. The returned water from the district heating network with a temperature of approximately 50 °C cools the condensate to approximately 51 °C from 58 °C. The flue gas blower ejects the flue gas with the temperature of 46–56 °C from the economiser into the atmosphere. The technical data of the furnace are presented in Table 1.

3. Characterisation of fuel and analysis of furnace operation 2. Description of the experimental combustion system The experiments were performed at a biomass-fired water heating plant located in Garliava (Lithuania) for heat supply to the local district heating system. The plant consists of a reciprocating grate furnace XILO AX WOOD (JSC Axis Industries, Lithuania), a 5.2 MWth multi-pass vertical convection boiler DANSTOKER (Denmark), a flue gas cleaning system, and a 1.2 MWth direct contact flue gas condensing economiser (JSC Axis Technologies, Lithuania). The schematic view of the heat plant is given in Fig. 1. The fuel is forced into the furnace by hydraulic feeders. Then, the fuel travels along the furnace on the inclined reciprocating

For each experimental run, the fuel supplier delivers fuel with a specified moisture content to the plant, which is later used for the experiments (see Table 2 and Table 3 in Section 5). Samples of the fuel were taken from the fuel feeder hopper. The fuel characteristics (moisture, ash content, and calorific value) were analysed according to the standards LST EN 14774-1, LST EN 14775 and LST EN 14918. To determine the fuel moisture correctly, a foreground analysis was performed. During the experiments, the furnace was operating in the automatic mode, depending on the actual heat demand. For performance evaluation, three experimental runs with different fuel moisture contents of 37, 46 and 50 wt.% were performed over

Fig. 1. Scheme of the reciprocating grate furnace XILO AX WOOD with the 5.2 MWth multi-pass vertical convection boiler DANSTOKER, the flue gas cleaning system and the 1.2 MWth direct contact flue gas condensing economiser.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

4

To estimate and compare the measurement data, the moisture content in the flue gas is recalculated into the fuel moisture by the following equation:

Table 1 Parameters of the reciprocating grate furnace. Parameter

Measuring unit

Value

Nominal capacity Fuel hopper volume Averaged size of fuel particles Feedstock moisture content Load range Air excess coefficient Temperature of flue gas exiting the furnace Fuel amount for 1 MW LHV at 40% moisture Wood chips Mixture of wood chips and peat Efficiency of boiler with a furnace at feedstock moisture of 40%, at least Ash content in fuel

MW m3 mm % % °C kg/h (m3/h)

6 5 5  20  50 30 . . . 55 30 . . . 100 1.4 900 . . . 1100 530 (1.74)

kJ/kg kJ/kg %

10,215 7327 89

%

1.5 . . . 5.5

Y RH ¼

mRH mfg QB mRH mfg þ LVHg

 100

ð2Þ

where YRH is the fuel moisture content calculated from the RH method, %. The scheme of input and output data used for calculations of moisture content based on RH method in fuel is shown in Fig 2A.  Estimation of fuel moisture content from the FGCE heat balance.

a period of several days. The plant was fully equipped with temperature, pressure, and differential pressure sensors for complete measurement while recording all of the input and output parameters of interest. Additionally, to estimate the air, flue gas and flue gas recirculation flows, the velocity was measured with a measurement unit Testo 454 using connected Pitot and thermal anemometer probes. The flue gas composition was measured downstream from the boiler using a portable flue gas analyser Testo 350XL. The solid particle concentration in the flue gas was measured with the gravimetric method in accordance with the international standard EN 13284-1: 2002

The second indirect method was based on the heat balance of the FGCE. To calculate the heat balance, the parameters, such as the flue gas and water temperatures upstream and downstream of the condensing economizer, frequency of the flue gas exhaust fans and boiler capacity, were used from the SCADA system. To estimate the moisture content in the flue gas, the thermal capacity of the FGCE is split into two components: the amount of heat extracted from the physical heat of the dry flue gas and that extracted from the condensation of moisture contained in the flue gas. The calculations of the first component were based on the combustion reaction of ideal (perfectly dry and free of incombustible mineral premixes) biomass described by the equation [27]:

CH1:44 O0;66 ðBiomassÞ þ 1:03  a ðO2 þ 3:76N2 Þ ðAirÞ ! CO2 ðCarbondioxideÞ þ 0:72H2 O ðWaterÞ

4. Indirect methods for real-time estimation of fuel moisture The moisture content in biomass feedstock is estimated using two indirect methods:  Estimation of fuel moisture content from the measured RH of the flue gases. To obtain the relation between the humidity in the flue gases and the fuel moisture content, experiments consisting of six runs with different fuel moisture content (see Table 3) were performed over a period of several days. In parallel, fuel samples were taken to measure the humidity directly in accordance with the LST EN 14774-1 standard. The relative humidity of the flue gas was measured in the flue gas duct downstream of the boiler by a Testo 454 equipped with an RH sensor with an accuracy of 2% RH in the range from +2 to +98% RH. It was determined that the relative humidity in the flue gas depends on the flue gas temperature, which in turn depends on the fuel moisture content and the thermal load of the boiler. The measured relative humidity of the flue gas was recalculated into the moisture content of flue gas (mRH mfg ) using the follow-

þ ð/ 1ÞO2 ðOxygenÞ þ a  3:87N2 ðNitrogenÞ

where CH1.44O0.66 is the average composition of the dry biomass used for combustion and a is the excess air ratio. Using the atomic masses of the chemical elements involved in the combustion reaction (3), it is calculated that the stoichiometric combustion of 1 kg of biomass (mF) consumes 5.89 kg of dry air (mO2) and produces 6.89 kg of combustion products (mD) containing 1.83 kg of carbon dioxide, 0.54 kg of water vapour (mH2O) and 4.52 kg of nitrogen. In this case, the moisture content in the dry flue gas hD is calculated by the following equation:

hD ¼

mH2 O mD  mH2 O

Q dry fg ¼

Q B mD LHV gmF     cp1 T fg þ ðcp2 þ cp;wv T fg ÞhD  cp1 T fg1  hD1 cp2 þ cp;wv T fg1

where

Q dry fg

ð5Þ

 
 
4 T fg 3 AF E tE Tr u  10   f T r þT 100


¼

r

ð4Þ

Using the calculated parameters from the combustion reaction (3), the heat extracted from the dry flue gas is calculated by multiplying the produced amount of dry flue gas by the difference of enthalpies of the flue gas entering and leaving the economizer:

ing equation:

mRH mfg

ð3Þ

fg

Patm

 qSST

ð1Þ

where mRH mfg is the calculated moisture content in flue gas, kg/h; AFE is the flue gas blower capacity of 22,750 m3/h at the motor rotation rate fr = 50 Hz; tE is the flue gas blower frequency at the time of measurement, Hz; u is the measured relative humidity in flue gas, %; Patm is the atmospheric pressure, 101,325 Pa; Tfg is the flue gas temperature after the boiler, °C; Tr = 273 °C is the reference temperature for conversion between the values expressed in Kelvin and Celsius degrees; and qSST is the density of the superheated steam according to the temperature of flue gas, kg/m3.

is the amount of heat extracted in the FGCE in kJ/kg, cal-

culated from the amount of flue gas which, in turn, is calculated from the amount of fuel required to maintain the given boiler capacity QB, by cooling down the inlet flue gas at the temperature of Tfg, °C, to the temperature of the outlet flue gas Tfg1, °C, when the moisture content in dry flue gas decreases from hD to hD1; cp1 is the specific heat of the dry flue gas, kJ/kg°C; cp2 = 2500 kJ/kg is the specific heat of water evaporation (latent heat of water vapour), cp,wv = 1.8 kJ/kg K is the specific heat of water vapour, kJ/kg °C; g is the boiler efficiency equal to 0.89 according to the boiler specification; LHV is the lower heating value of the completely dry feedstock, 18,000 kJ/kg; and hD1 is the amount of residual moisture, expressed as the mass fraction of the dry flue gas, calculated as [28]:


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

5

Fig. 2. Scheme of input and output data used to calculate the moisture content in the fuel: (A) based on the RH method and (B) based on the heat balance of the FGCE method.

hD1 ¼ 0:004  10T fg1 =37:6

ð6Þ

The moisture content in flue gas, produced by the combustion of moist fuel, is estimated by subtracting the calculated amount of heat in the dry flue gas from the thermal capacity of the economizer indicated in the control panel, and then dividing the difference by the latent heat of the water vapour:

mHB mfg ¼

 3600
Q FGCE  Q dry fg cp2

ð7Þ

where mHB mfg is the moisture content in the flue gas obtained from the heat balance of the economizer, in kg/h; QFGCE is the thermal capacity of the economizer, kW; and the factor 3600 appears for conversion between seconds and hours, to obtain the moisture content per hour. The calculated moisture content in the flue gas is recalculated into the fuel moisture content with the following equation:

Y HB ¼

mHB mfg QB mHB mfg þ LVHg

 100

ð8Þ

where YHB is the fuel moisture content calculated from the FGCE heat balance method, %. The calculation scheme using this method is presented in Fig. 2B.

Both developed methods were integrated into the SCADA system of the plant. The main aim of this trial was to predict the changes in the fuel moisture, and understand how promptly these methods would react to changes and select the correct method to determine the fuel moisture. The experiments were carried out over several days of feeding the fuel (wood chips) with different moisture contents (54 and 60 wt.%) into the furnace. The fuel used for the experiments was divided into two separate piles, depending on its moisture content, and manually supplied to the fuel transportation system in selected time periods. As the fuel was changed, the samples were taken simultaneously to analyse the moisture content. During the experiments, the furnace operation was manually corrected to adjust the parameters of the selected regime. 5. Performance of automatic furnace operation with time varying fuel moisture content In this study, the furnace was automatically controlled with programmable logic controllers (PLC), and the technological process was adjusted in response to changes in the varying parameters, such as temperature, pressure, water flow rate and oxygen concentration in the flue gas. However, registering these parameters is not sufficient to react completely to the changes in the fuel parameters. Changes in the fuel characteristics made it necessary to manually adjust the character of the motion of the reciprocating grate for efficient feedstock combustion After the furnace opera-


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

6

Table 2 Average operational data of the boiler and furnace at different fuel moisture content (different test runs T_11, T_12, T_13). Parameter

T_11

T_12

T_13

Fuel properties Fuel moisture, wt.% HHV, kJ/kg LHV (wet basis), kJ/kg Ash (dry basis), wt.% C (dry basis), wt.% H (dry basis), wt.% N (dry basis), wt.% S (dry basis), wt.% Cl (dry basis), wt.%

37 12,471 10,799 1.7 47.95 6.26 0.18 <0.08 0.002

46 10,766 9018 0.74 47.02 6.13 0.20 <0.08 0.003

50 10,038 8218 1 46.98 6.1 0.18 <0.06 0.01

1:1 5.2 1900 229 7780 (4.1) 2108 756 4914 8850 (4.66) 1220 410

2:1 4.4 1950 239 9460 (4.8) 2530 1173 5756 9915 (5.08) 704 474

6:1 4.7 2260 253 9770 (4.3) 2719 878 6177 11,170 (4.94) 370 517

820 156 4.8/5.4 16 14.9 217 530 91.5 8.5 10.4 90.3 0.009 0.03 0.65 1.5 7.5

850 178 6/7 7 13.1 287 900 91.8 8.2 5.5 84.6 0.006 0.06 0.35 1.8 13.1

780 176 5.7/6.9 7 13.6 300 640 92.1 7.9 1.21 87.0 0.006 0.04 0.04 1.6 11.3

Boiler and furnace operational data Ratio of 1st and 2nd grate motion Boiler load MW Fuel feeding rate, kg/h Fuel (dry basis) feeding rate, kg/MWh Total air flow, m3/h (for 1 kg of fuel) Primary air, m3/h Secondary air, m3/h Tertiary air, m3/h Flue gas flow, m3/h (for 1 kg of fuel) Flue gas recirculation, m3/h Temperature in the furnace above the secondary air inlets, °C Flue gas temperature after furnace, °C Flue gas temperature after boiler, °C O2, vol.% (local device/Testo analyser) CO, mg/m3 CO2, vol.% NOx, mg/m3 PM before cyclone, mg/m3 Ash in the PM, wt.% Combustibles in PM, wt.% Combustibles in ash, wt.% ‘‘Brutto” efficiency, % Heat loss due to combustible in flue gas Heat loss due to combustible in PM Heat loss due to combustible in ash Heat loss due to radiation from boiler walls Heat loss due to latent heat of flue gas

tion became stable, the main parameters of the process were evaluated. Table 2 presents the data of the main operational parameters of the plant after the correction of the reciprocating grate motion for the fuel with a moisture content of 37, 46 and 50 wt.%.

As the moisture content changed, so did the main technical parameters of the furnace. One of the observed changes was the fuel consumption rate required per one MWh of heat production. The highest fuel consumption calculated from the baseline dry case was in the case when the wettest fuel was fired in case T_13 where 253 kgdry/MWh was required, as compared to 229 kgdry/MWh for T_11 and 239.3 kgdry/MWh for T_12 (see Table 2). As dry fuel composition is comparable for all the three cases, the main factor influencing the increase in the fuel consumption might be the heat demand for the evaporation of moisture. With the change in the fuel moisture, the flow rates of the primary, secondary and tertiary air also varied. The airflow rate for 1 kg/h of fuel was 4.3 m3/kg for the feedstock with the MC of 50 wt.%. A higher flowrate of 4.8 m3/kg was observed for the feedstock with an MC of 46 wt.%, and the lowest airflow was 4.1 m3/kg for the driest feedstock (MC-37 wt.%). The theoretical amount of air for complete combustion of fuel should be the same to generate 1 MWh of heat despite the variation in the fuel moisture. However, the obtained results show that a higher amount of air for combustion of moist fuel was required. Due to increase of fuel moisture content, the water content in flue gas and the flue gas volume increased as well. Therefore, the oxygen concentration in the flue gas decreased. The latter leads to the automatic adjustment of the airflow according to the pre-set oxygen content (6 vol.%) in the flue gas. The oxygen sensor worked without humidity compensation, leading to the increase in the real oxygen content recalculated to the dry baseline. This relation was proven by the measured O2 concentration in the flue gas with a gas analyser equipped with a cooling system. The concentration of O2 in the dry exhaust gas increased with the increase of fuel moisture content (Table 2). Finally, the existing automatic control algorithm of the furnace lead to insufficient air supply for the moist feedstock at maximum loads, even though the air blowers were selected with a certain excess capacity. As the moisture content increased, the air flow rate also increased; therefore, an increase in the flue gas flow rate was detected accordingly (Table 2). The increase in the flue gas flow rate led to increased flue gas temperatures at the exit of the boiler and concomitant heat loss due to the latent heat of flue gas, which decreased the total efficiency of the system. As seen from the results presented in Table 2, the emissions of gaseous pollutants were sufficiently low. Meanwhile, the concentration of solid particles in the boiler before entering the treatment device varied between 530 and 900 mg/m3 at different fuel moisture content levels. It was determined that solid particles contain approxi-

Table 3 Moisture content measurement data during biomass boiler operation. Parameter

T_21

T_22

T_23

T_24

T_25

T_26

Fuel moisture, wt.%

52.1

54.2

49.1

44.3

47.2

53.0

Boiler parameters Boiler capacity, MW FGCE capacity, MW Flue gas temperature, °C Flue gas blower motor frequency, Hz Calculated fuel (daf) amount, kg/h

3.32 0.70 137.3 24 720

3.34 0.75 141.0 27 736

3.35 0.68 144.1 30 749

4.09 0.79 153.1 32 863

4.16 0.80 152.6 32 889

3.96 0.81 151.2 33 790

Estimation of fuel moisture with RH sensor Relative humidity of the flue gas, % Calculated moisture content in the flue gas, kg/h Calculated fuel moisture, wt.% Deviation from the measured moisture, wt.%

4.4 699 49.3 2.8

3.7 716 49.0 5.2

3.4 692 48.0 1.1

2.2 647 42.1 2.2

2.0 671 44.1 3.1

2.4 700 47.0 6.0

Estimation of fuel moisture from FGCE heat balance Calculated moisture content in the flue gas, kg/h Calculated fuel moisture, wt.% Deviation from the measured moisture, wt.%

724 50.2 1.9

846 53.5 0.7

655 46.7 2.4

715 45.3 1

785 46.9 0.3

831 51.3 1.7


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

mately 8% of the residual combustible mass. Consequently, a higher concentration of solid particles in the flue gas leads to a higher heat loss due to combustibles contained in the particles. During the combustion of the wood chips with various moisture content, the motion ratio between the 1st and 2nd grates was adjusted manually to avoid incomplete combustion or even combustion extinction. The ratio was changed from the predetermined 1:1 (this ratio shows that the fuel motion velocity on the 1st grate is equal to that on the 2nd grate) to 2:1 and 6:1 (Table 2). To better understand the effect of the grate motion ratios on the behaviour of the fuel particles and the reciprocating grate, computer simulations were performed (Fig. 3). It was determined that the thickness of the fuel bed layer on the 2nd grate increased as the grate motion ratio was increased. This was confirmed by measuring the primary air pressures (Fig. 3). The primary air pressure in wind box 1 under the 2nd grate (see Fig. 1) increased with increasing fuel bed thickness. It was observed in wind box 2 that the air pressure begun to decrease as the grate motion ratio increased. In this wind box, the coupling of both grates was intended. Due to different motion velocities, a fracture in the fuel bed appeared. A thick layer of fuel covered half of wind box 2 with a larger aerodynamic resistance, while the other half was a thin layer with a rather low air resistance. Consequently, air from this wind box permeated through the thin layer

7

of fuel with the lower aerodynamic resistance, was distributed unevenly and increased the spatial velocity through the layer. The increased velocity through the layer could result in a pseudo-fluidization of the layer, arching, tunnels or similar effects. Because of these effects, light particles might be entrained and carried away into the flue gas duct of the boiler, as a larger quantity of solid particles in the flue gas was detected at an increased grate motion ratio. The fact that the air from the primary flow in wind box 2 reacted incompletely in the fuel layer and was partially involved in the secondary combustion was evidenced by the increasing furnace temperature with the increasing grate motion ratio (Table 2). In all cases, negative pressure in wind box 3 of the primary air flow was detected. As the flue gas fan automatically maintained the draught of 90 Pa in the furnace, the pressure was transferred to wind box 3 through a thin layer of char and ash. At the grate motion ratio of 6:1, the pressure of 76 Pa in the wind box was close to the pressure in the furnace. Therefore, it can be concluded that the fuel layer becomes the thinnest in this case of all the analysed cases. This is also proven by the smallest quantity of organic residues of unburnt feedstock in the ash (Table 3). The analysis of performance of the automatic furnace operation with a time varying fuel moisture content implies two main conclusions. First, due to specific properties of the oxygen measuring

Fuel feeding

Motion ratio 1:1 2nd grate

y, m

1st grate

x, m 310 Pa

200 Pa

-25 Pa

Motion ratio 2:1 2nd grate

y, m

1st grate

x, m 370 Pa

38 Pa

-29 Pa

Motion ratio 6:1 2nd grate

y, m

1st grate

x, m 415 Pa

-2 Pa

-76 Pa

Fig. 3. Distribution of the feedstock particles on the grate at different grate motion ratios and the measured air pressure under each grate.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014

8


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

instrument installed in the boiler, the measured oxygen excess changed only insignificantly in all the measured cases and led to an excessive supply of combustion air. Second, to ensure the quality combustion of the feedstock with increasing moisture content, automatic adjustments are required not only for the airflow distribution, but also for the grate motion ratio.

6. Estimation of fuel moisture content during biomass boiler operation To estimate the accuracy of the developed methods during the biomass furnace operation, simultaneous measurements of the fuel samples and the relative moisture content in the flue gas were performed. Additionally, the developed algorithm was integrated in the SCADA system and recorded. Fuel was sampled and the integrated data were recorded for several days (Table 3). Processing the collected data showed that the calculated fuel moisture content was close to that determined by direct measurement in the laboratory. However, larger deviations of the moisture content of up to 6 wt.% compared to the directly measured values were obtained when using an RH sensor, whereas the calculations from the heat balance produced the maximum deviation from the directly measured values of up to 2.4 wt.% (Table 3). The main cause of the high inaccuracy is the dependence of the relative humidity in flue gas from the temperature. When operating at high temperatures, the RH approaches the instrumental measuring limit that falls

between 2 and 98%. In this case, higher uncertainties of the results were observed. Hermansson et al. [6] obtained similar results in their work and also noticed the low accuracy of the RH method. To determine the real fuel moisture content at each experimental run, at least five fuel samples were taken with the time interval of 10 min. The averaged values of the fuel moisture content (MC) are given in Table 3 and presented by the dashed line in Fig. 4. Graphs in Fig. 4 represent changes in the estimated fuel moisture content from the FGCE heat balance during the biomass furnace operation at selected hours of different days and different fuel moisture values. Since the calculated fuel moisture depends directly on the heat capacity of the FGCE, a larger deviation of the calculated fuel moisture content appears at larger oscillations of the economiser heat capacity. For instance, in the case of the measured MC of 52 wt.% or 49 wt.%, the FGCE load during the measurement was unstable. This leads to a larger deviation in the determined MC. In the case of the measured MC = 53 wt.%, the FGCE load was stable and the calculated MC values had only a negligible fluctuation. However, the calculated fuel moisture depends not only on the heat capacity of the economiser, but also on other indirect measurements, e.g. the frequency of the motor of the flue gas blower. If the latter works unevenly for some reasons (e.g., contamination of the furnace or the boiler and additional air suction in the flue gas duct), oscillations of the calculated MC are possible even though the FGCE works smoothly, as was the case of the measured MC = 53 wt.%. Therefore, the performed experiments imply that the indirect method for estimating the fuel moisture content

Fig. 4. Measured fuel moisture content (dashed line) versus the values estimated by the indirect method and FGCE heat capacity during the biomass boiler operation on different days.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

based on the FGCE heat balance is suitable for biomass furnaces operating in an assembly with the boiler. However, to avoid the influence of short term oscillations of the measured parameters, the calculated MC must be averaged over time. As the control algorithm is developed, the correction must be introduced to the furnace control automation based on the obtained values, depending on the increasing or decreasing fuel moisture content. 7. Prediction of fuel moisture for automatic furnace operation A correlation between the change in moisture content and the thermal load of the FGCE was established by performing the combustion tests described in Section 6. However, the performance and sensitivity of the furnace for automated operation during the change in the fuel moisture content must be estimated.

60

9

To evaluate the proposed fuel moisture prediction method, a two-day trial was performed. The data were collected and taken from the SCADA system for further analysis. Fig. 5 shows the variation in the calculated (fuel moisture, fuel water amount, and dry fuel amount) and measured (FGCE and boiler capacities) parameters over the period of experiment. Additionally, the fuel feeding periods are indicated in the figure. Experiments were started by feeding the biofuel with a 54 wt.% moisture content into the furnace over the period from 0 to 7 h. Over the period from 7 to 12 h, fuel with a moisture content of 60 wt.% was feed into the furnace. During the night period (12–24 h) the boiler was left to operate with fuel having a moisture content of 54 wt.%. Finally, from 26 to 30 h of testing, the fuel with a higher moisture content was again fed. It should be noted that after 30.5 h of testing, the water circulation pump of the economiser broke up and the economiser

60

Fig. 5. Calculated (fuel moisture, fuel water and dry fuel amount) and measured (FGCE and boiler capacity) temporal changes. The sliding weighted average is calculated for n = 101 measuring points.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

10

had to be stopped. Due to this fact, it was impossible to estimate the stabilization of the curves. In the SCADA system, the presented values were recorded every time period of Dt = 00:00:55. As seen from the plots (Fig. 5). The scatter of the values is very uneven, meaning that there is a rather large amplitude of oscillations. To eliminate the noise, signals were processed (smoothed) using a weighted sliding average:

Xn=2 i¼n=2
xj ¼ X n=2

xi xji

x i¼n=2 i

¼ t0 þ jDt;

;

j ¼ ½0; N  1;

Dt ¼ 55s

xj ¼ xðtj Þ;

tj ð9Þ

Data smoothing reveals the trends in the temporal changes of the parameters presented in the plot with varying fuel moisture, Fig. 5. The window width of the averaging filter presented in this plot is n = 101 measuring points, i.e. 46 min 17 sec to both sides from the current point. However, it is seen that the window width of the averaging filter is not sufficient as the obtained value of the fuel moisture oscillates for the same fuel type. Such type of data processing might be unfit to be used for automated correction because it may cause an instability in the combustion process. Increasing the window width to 151 measuring points (1 h 09 min. 12 sec) revealed the minimum and maximum values of the fuel moisture content (Fig. 6). The presented figures thus show

Fig. 6. Calculated (fuel moisture, fuel water and dry fuel amount) and measured (FGCE and boiler capacity) temporal changes. The sliding weighted average is calculated for n = 151 measuring points.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

that the system is rather inert and changes of the feedstock moisture from one value to another lead to a transition period of up to 3 h. This is influenced by the fuel feeding system and the furnace size: the fuel supply hopper installed upstream of the furnace holds up to 1 tonne of wood chips and the full combustion cycle on the grate takes up to 30 min. Besides, changing the feedstock in the storage area and its supply to the furnace hopper also take additional time. To introduce a correction factor for the automated control of the furnace due to the fuel moisture calculated from Eq. (8), it is necessary to analyse the MC changes over time. It is convenient to observe the intensity of the temporal increases or decreases of the moisture by calculating the time derivative of HB

the average feedstock moisture dYdt , Fig. 7. A positive derivative shows the increase of the moisture content in flue gas, a negative derivative shows its decrease, and the zero value shows no changes in moisture. This obtained curve shows the moment when the fuel moisture attained its maximum or minimum values, i.e. when the furnace switches operation to the other types of fuel. Fig. 7 demonstrates several characteristic points. The first maximum peak of the fuel moisture change rate was obtained after 3 h from starting the feed with a moisture content of 54 wt.%. Later, the derivative started to approach zero and, if the fuel remained the same, the values of this curve would oscillate around zero showing the stabilisation of the fuel moisture. However, after seven hours of work, the fuel with the higher moisture content was fed into the furnace again. This approach led to another peak of in the moisture change rate at the 10th h (measured MC of fuel was 60 wt.%). Overnight, the system was left to work on the drier feedstock and the fuel moisture change rate started to decrease and reached its minimum point at the 14th h after the start of the measurement (54 wt.%). The similar characteristics were also obtained on the next day when the wettest fuel was again fed (at the 26th h (60 wt.%)). From this analysis, it can be concluded that the most effective way for the process control is to use the time derivative of the averaged HB

feedstock moisture, i.e. when dYdt – 0, the system should react and start to change the motion of the reciprocating grates based on the adjusted regime for a certain type of fuel. Eventually, the real-time measurements make it impossible to smooth the data based on Eq. (9) because the future points that have not been measured yet cannot be introduced into the smoothing function. The real-time smoothing function is as follows:

Xn1

xi xji

i¼0
xj ¼ X n1 i¼0

xi

¼ t0 þ jDt;

;

j ¼ ½0; N  1;

Dt ¼ 55s

xj ¼ xðtj Þ;

tj ð10Þ

11

The smoothed feedstock moisture content from Eq. (10) will trail behind the real-time measurement by half the filter width. For the number of filter points n = 151, the total temporal width of the filter is 2 h 18 min and 24 sec, and half the width is 1 h 09 min and 12 sec. In this case, at every time moment, there is the average (smoothed) value of moisture that was present 1 h 09 min and 12 sec previous. It can be assumed that at
27.5 h, the change of the average
HB  moisture becomes positive dYdt > 0 . This finding would mean that the feedstock moisture started to increase at the time 1 h 09 min and 12 sec previous. Consequently, the automated control circuit must follow the changes in the moisture content and, from the moisture change rate, predict the time when it will attain the highest value for the selected regime. If the derivative decreases, it is possible to predict from the rate of decrease (i.e. the acceleration of the moisture change) when it will reach zero and the process will be stabilized, meaning there will be no need for additional changes. Having determined the nature of the smooth transition of the process parameters by observing the changes in the moisture content, the motion of the reciprocating grate could be changed gradually rather than stepwise. The control algorithm is proposed (Fig. 8) for the real application of the presented indirect fuel moisture prediction method. The correction must be introduced to the furnace control automation based on the obtained values, depending on the increasing or decreasing fuel moisture content. The control algorithm constitutes the computational set of fuel moisture prediction formulas presented in Section 4. After the fuel moisture content YHB from the FGCE heat balance is calculated (8), the averaged moisture using the real-time smoothing function (10) is content Y HB j estimated:

Xn1

Y HB j

Y HB xji i¼0 i ¼ X ; n1 HB Y i¼0 i ¼ t 0 þ jDt;

j ¼ ½0; N  1;

Dt ¼ 55s

xj ¼ xðt j Þ;

tj ð11Þ

For the following calculations in the algorithm, it is necessary to estimate the time step dt during the boiler adjustment. In the d/dt is calculated at the time loop, the temporal derivative of dY HB j moment t:

G2H O d dY HB ¼ ¼ 2 2 2 100=dt dt dt GH2 O þ BDS If the derivative is zero




dY HB dt

ð12Þ  ¼ 0 , the loop is started again; and

if the derivative is more than zero, the loop is continued and the furnace automated control tasks are corrected using the condition
HB  d/dt > 0 dYdt > 0 . If the value. meets the condition 65 > Y HB > 54, j

Fig. 7. Fuel moisture content (straight line) and the rate of moisture change (dashed line) during the boiler operation. The sliding weighted average is calculated for n = 151 measuring points.

the automatic control of the furnace adjusts to the parameters corresponding to the moisture content of 60 wt.%. Otherwise, the furnace control adjusts to the parameters corresponding to the moisture content of 54 wt.%. The proposed algorithm for the fuel moisture prediction provides an advanced control strategy for the efficient and stable combustion of the wood chips with a time-varying moisture content. Such a control strategy allows a stable fuel flow maintained along the grate by changing the grate motion ratios. In addition, calculating the fuel moisture content of flue gas makes it is possible to implement a correction factor to measure the oxygen content in the flue gases and to reduce the excess air. The latter enables the reduction of the heat loss and increases the net efficiency of the water heating plant.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014

12


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu

Fig. 8. Fuel moisture prediction algorithm to correct the automatic control of the furnace.

8. Conclusions The combustion of fuel with different properties and a moisture prediction algorithm was experimentally validated in a 6 MW grate-fired furnace. For the combustion experiments, biofuels with different moisture content were used. The method for indirect estimation of the fuel moisture content was integrated into the boiler SCADA system and the obtained data were recorded. The fuel moisture content was directly measured during the experiments and the main parameters of the boiler were recorded. As a result, the following conclusions were made:  The standard sensors to determine the oxygen content in the flue gas used for the automatic boiler control did not provide the proper moisture compensation. Therefore, the humidity and volume of the flue gas increased, and the oxygen content decreased as the feedstock moisture increased. Consequently, it was determined that the automatic control circuit supplies excessive amounts of air and thus reduces the net efficiency of the plant.  To ensure the stable combustion of the wood chips with a timevarying moisture content, not only does the distribution of the air supply have to be automatically adjusted, but the grate motion ratio must be automated as well.  The values of the fuel moisture content calculated by the method proposed in this work and those directly measured do not differ by more than 2.4%. Taking into account this deviation, the indirect fuel moisture calculation method is sufficiently precise and can be used to enhance the control strategy for fur-

naces firing wet biomass and operating in an assembly with a boiler and the FGCE.  A large scatter of values of the fuel moisture calculated in the SCADA system was detected. The obtained data were smoothed with a sliding weighted average. Therefore, it was determined that the wood chips combustion system is rather inert and the transition process can take up to 3 h to change fuel types from one moisture content to another.  The equation and the control algorithm to predict changes in the fuel moisture were proposed. Application of this method to predict the fuel moisture content will lead to a delay of data by half the filter width from the real-time measurement. In case of the presented in this work, the half width of the filter is 1 h 09 min and 12 s. Acknowledgements The authors acknowledge the technical and financial support of JSC Axis Technologies. References [1] Demirbas MF, Balat M, Balat H. Potential contribution of biomass to the sustainable energy development. Energy Convers Manage 2009;50:1746–60. http://dx.doi.org/10.1016/j.enconman.2009.03.013. [2] Krigstin S, Wetzel S. A review of mechanisms responsible for changes to stored woody biomass fuels. Fuel 2016;175:75–86. http://dx.doi.org/10.1016/ j.fuel.2016.02.014. [3] Williams A, Jones JM, Ma L, Pourkashanian M. Pollutants from the combustion of solid biomass fuels. Prog Energy Combust Sci 2012;38:113–37. http://dx. doi.org/10.1016/j.pecs.2011.10.001.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014


gas et al. / Energy Conversion and Management xxx (2017) xxx–xxx N. Striu [4] Zhang L, Xu C (Charles), Champagne P. Overview of recent advances in thermochemical conversion of biomass. Energy Convers Manage 2010;51:969–82. http://dx.doi.org/10.1016/j.enconman.2009.11.038. [5] Sefidari H, Razmjoo N, Strand M. An experimental study of combustion and emissions of two types of woody biomass in a 12-MW reciprocating-grate boiler. Fuel 2014;135:120–9. http://dx.doi.org/10.1016/j.fuel.2014.06.051. [6] Razmjoo N, Sefidari H, Strand M. Measurements of temperature and gas composition within the burning bed of wet woody residues in a 4 MW moving grate boiler. Fuel Process Technol 2016;152:438–45. http://dx.doi.org/ 10.1016/j.fuproc.2016.07.011. [7] Svoboda K, Martinec J, Poho/vrely´ M, Baxter D. Integration of biomass drying with combustion/gasification technologies and minimization of emissions of organic compounds. Chem Pap 2009;63:15–25. http://dx.doi.org/10.2478/ s11696-008-0080-5. [8] Samuelsson R, Burvall J, Jirjis R. Comparison of different methods for the determination of moisture content in biomass. Biomass Bioenerg 2006;30:929–34. http://dx.doi.org/10.1016/j.biombioe.2006.06.004. [9] Nyström J, Dahlquist E. Methods for determination of moisture content in woodchips for power plants – a review. Fuel 2004;83:773–9. http://dx.doi.org/ 10.1016/j.fuel.2003.11.002. [10] Trabelsi S, Paz AM, Nelson SO. Microwave dielectric method for the rapid, nondestructive determination of bulk density and moisture content of peanut hull pellets. Biosys Eng 2013;115:332–8. http://dx.doi.org/10.1016/j. biosystemseng.2013.04.003. [11] Lestander Ta, Rhén C. Multivariate NIR spectroscopy models for moisture, ash and calorific content in biofuels using bi-orthogonal partial least squares regression. Analyst 2005;130:1182–9. http://dx.doi.org/10.1039/b500103j. [12] Daassi-Gnaba H, Oussar Y, Merlan M, Ditchi T, Géron E, Holé S. Wood moisture content prediction using feature selection techniques and a kernel method. Neurocomputing 2016. http://dx.doi.org/10.1016/j.neucom.2016.09.005. [13] Dahlquist E. Biomass as energy source: resources, systems and applications. CRC Press; 2013. [14] Neonila S, Zibtsev S. Swedish University of Agricultural Sciences. Writing 2010;26:17–20. [15] Tsuchikawa S. A review of recent near infrared research for wood and paper. Appl Spectrosc Rev 2007;42:43–71. http://dx.doi.org/10.1080/ 05704920601036707. [16] Axrup L, Markides K, Nilsson T. Using miniature diode array NIR spectrometers for analysing wood chips and bark samples in motion. J Chemom

[17] [18] [19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

13

2000;14:561–72. http://dx.doi.org/10.1002/1099-128X(200009/12)14:5/ 6<561::AID-CEM608>3.0.CO;2-2. Niedermayr F, Waid S, Auñón DG, Riedl M, Matt D. Design of a scalable and robust woodchip handling control system. SDEWES 2015;2015–1012:1–9. Aulin R, Karlsson M. Standardisation of moisture measurement using NIR spectroscopy for delivery control. Stockholm: Värmeforsk Rapp; 2008. Bak J, Clausen S. FTIR emission spectroscopy methods and procedures for real time quantitative gas analysis in industrial environments. Meas Sci Technol 2001;13:150. http://dx.doi.org/10.1088/0957-0233/13/2/302. Bolshov MA, Kuritsyn YA, Romanovskii YV. Tunable diode laser spectroscopy as a technique for combustion diagnostics. Spectrochim Acta – Part B At Spectrosc 2015;106:45–66. http://dx.doi.org/10.1016/j.sab.2015.01.010. Hermansson S, Lind F, Thunman H. On-line monitoring of fuel moisturecontent in biomass-fired furnaces by measuring relative humidity of the flue gases. Chem Eng Res Des 2011;89:2470–6. http://dx.doi.org/10.1016/j. cherd.2011.03.018. Jaakkola PT, Vahlman TA, Roos AA, Saarinen PE, Kauppinen JK. On-line analysis of stack gas composition by a low resolution FT-IR gas analyzer. Water Air Soil Pollut 1998;101:79–92. http://dx.doi.org/10.1023/A:1004905521433. Kortela J, Jämsä-Jounela SL. Fuel-quality soft sensor using the dynamic superheater model for control strategy improvement of the BioPower 5 CHP plant. Int J Electr Power Energy Syst 2012;42:38–48. http://dx.doi.org/ 10.1016/j.ijepes.2012.03.001. Lackner M. Tunable diode laser absorption spectroscopy (TDLAS) in the process industries – a review. Rev Chem Eng 2007;23:65–147. http://dx.doi. org/10.1515/REVCE.2007.23.2.65. Ikonen S, Stormbom L, Ranta-aho T. Chemical interference test results of a novel humidity sensor. In: 5th int symp humidity moisture, Rio de Janeiro; 2006. Kortela J, Jämsä-Jounela SL. Fuel moisture soft-sensor and its validation for the industrial BioPower 5 CHP plant. Appl Energy 2013;105:66–74. http://dx.doi. org/10.1016/j.apenergy.2012.12.049. Nussbaumer T. Combustion and co-combustion of biomass: fundamentals, technologies, and primary measures for emission reduction. Energy Fuels 2003;17:1510–21. http://dx.doi.org/10.1021/ef030031q. Bukharking EN. Conditions ensuring reliable service of stack-gas paths in boiler houses using condensing economizers. Therm Eng 1997;44:724–9.


gas N et al. Estimating the fuel moisture content to control the reciprocating grate furnace firing wet woody bioPlease cite this article in press as: Striu mass. Energy Convers Manage (2017), http://dx.doi.org/10.1016/j.enconman.2017.04.014