Effect of operation conditions on particulate matter removal by a packed-bed wet scrubber for a small-scale biofuel boiler

.


Introduction
Biofuel combustion has played an important role in producing heat and power in the past few decades.Due to rapid urbanization and high energy demand, the use of biofuels in heating for domestic space has been promoted.Because of the advantages of biofuels, such as availability, low CO 2 emission, and renewability, they can be considered a promising alternative to fossil fuels [1].In 2020, the contribution of renewable energy sources (such as biofuels) for heating and cooling purposes was around 23 % in the EU.The share of the heat produced using renewable energy in the Nordic countries is higher than the corresponding average share in EU countries; for example, in Finland, Denmark, Norway, and Sweden, 58 %, 51 %, 40 %, and 66 % of the heat produced, respectively, is derived from renewable energy sources [2][3][4].Despite the advantages of biofuels, the effect of the toxic substances and micro-particles released from biofuel combustion on public health should be considered.According to WHO, every year, numerous people (3.8 million in 2016) die as a consequence of pollution exposure due to household cooking and heating activities.This represents a higher mortality per annum than that caused by car accidents in developed countries [5].Among the various indicators of air pollution from biofuel combustion, particulate matter (PM) can be considered one of the main hazardous substances.PM can cause various diseases, such as pneumonia, stroke, heart attacks, chronic obstructive pulmonary disease, and lung cancer [6].
According to the Clean Air Policy Package, established in 2013 by the European Commission, it is the aim of the EU to reduce air pollution to half by 2030 compared to the level in 2005.In 2015, the emitted PM concentration was reported as 14.5 µg/m 3 , which is still higher than the WHO standard (10 µg/m 3 ) for the annual mean.Globally, domestic heating systems have contributed approximately 20 % of the PM, whereas in the European Nordic countries, domestic heating systems have been reported to contribute around 25 % of the PM [7].
Fuel gas particulate removal techniques can be grouped into three categories: inertial separation, barrier filtration, and electrostatic separation [8].PM removal methods such as electrostatic precipitators (EPS), scrubbers and filters are widely used in large-scale burners due to their high efficiency [9].Meanwhile, there is a shortage of studies performed to ascertain a cost-effective method for cleaning flue gas from small-scale (<500 kW) biofuel burners [10], even though small-scale boilers make a noticeable contribution to PM emissions.Thus, it is necessary to investigate whether flue gas cleaning satisfies the limitations on PM emission which have been established by the Clean Air Policy Package and WHO standards [7].
A wet scrubber with packed-bed material is one of the most common and economical methods for PM control [11].The simultaneous ability of this method to achieve PM removal and high heat recovery makes it an attractive method for small-scale boiler flue gas purification (<500 kW).In addition, wet scrubbers have the potential to be considered a multi-pollutant control technology due to the high gas-liquid mass transfer which they achieve [12].
The different forces dominating the particle movement toward the wet surface in the collection tower are affected by the operating conditions.Among the different forces, the diffusiophoresis force and Brownian force control the movement of small particles (0.1-10 µm), whereas the gravity force governs the movement of larger particles (>40 µm) [13].Thus, the PM 1-10 is mostly controlled by diffusion, which is conspicuously a function of the concentration gradient, temperature gradient, scrubber tower height, and flue gas velocity [14,15].Lee et al. found that the diffusion effect increased with a decreasing particle size as the influence of interception was decreased [16].
A pilot plant setup was constructed at Luleå University of Technology in Sweden to investigate an efficient small-scale flue gas cleaning method.The setup contains a boiler, heat exchangers, and a countercurrent packed-bed wet scrubber.The wet scrubber was selected for its simple procedure and maintenance, low cost, high efficiency in particle collection, its heat recovery of up to 70 %, and its ability to cool and clean the flue gas simultaneously [17,18].The scrubber is a tower that contains packed-bed material, which increases the liquid-gas contact surface area and represents a promising alternative method for flue gas purification for the small-scale boiler [19].Instead of water, an absorption solution consisting of salt (potassium acetate) and water is used, which is a rare method compared to using only water and has only been studied before by Westerlund et al [17] and Darbandi et al [20].The solution is hygroscopic and absorbs water vapour from the gas, creating streams towards the solution surface, which improves the particle collection.
The efficiency of the packed-bed wet scrubber for the particle size r > 5 µm could rise to 90 %, although, for smaller particle sizes, the efficiency decreased to around 50 % [18].The current study aimed to determine the impact of different operation conditions on particle collection efficiency.The results could be used to optimise wet scrubbers designed for small-scale boiler flue gas purification.Different investigated operation conditions were selected according to computational fluid dynamic (CFD) simulation results from a previous study [15].Thus, the obtained conditions from CFD simulations were implemented for the domestic heating pilot plant.

Experimental setup
Fig. 1 provides a schematic diagram of the experimental setup.The equipment is made of stainless steel, and all the parts are insulated to reduce the unwanted heat transfer from the system to the surroundings.The combustion tests were performed with a small pellet burner (an Ariterm BeQuem 20D, 20 kW) and a boiler (wood boiler Arimax 240, BeQuem 20).The chosen boiler size is typical for one-family houses and suitable for the laboratory setup.The technique is applicable to larger systems (<500 kW).Above this size, regulations require more exclusive cleaning methods.The fuel used was regular stem wood pellets, and the burner was operated to achieve 10 vol% O 2 in the flue gases.
A countercurrent packed-bed wet scrubber was installed, consisting of a column with an internal diameter of 114 mm and filled with steel pall rings as the packed-bed material (i.e.steel pall rings with the dimensions 25 × 25 × 0.6 mm and giving a specific surface area of 219 m 2 /m 3 and porosity of 0.950 m 3 /m 3 ) to provide a larger surface area between the solution and flue gas.The absorption solution was supplied from the top of the column through 19 holes with a diameter of 3 mm, while the flue gas entered from the bottom.The liquid also occupies a small portion (<2%) of the free volume, which reduces the available volume for gas flow.A pump was set to circulate the absorption solution flow through the absorber and to the generator.The generator was a column consisting of seven vertical tubes (d = 21.3 mm), and a part of T. Darbandi et al. the hot flue gas from the boiler was led through the generator and passed inside the tubes while liquid flowed outside the tubes.
In the boiler, thermal energy was transferred from the hot flue gas to water in the convection part of the boiler and gave off the heat to the air through a heat distributor (an Aerotemper).The flue gas was sent to the absorber, entering from the bottom and passing upward through the column filled with stainless steel packing material.Simultaneously, the flue gas contacted the absorption solution, which was introduced from the top of the absorber and distributed through the packed-bed material.The flue gas was dried, and its temperature was reduced.Finally, the cleaned, cooled, and dried flue gas flowed out of the chimney.
The diluted absorption solution from the absorber (with water taken up from the flue gas) was pumped to the generator.It was preheated when passing through the heat exchanger (HEX 2) by the concentrated solution leaving the generator.A part of the flue gas from the boiler was led to the generator.In the generator, the diluted solution flowed through the surrounding area outside the tubes.Thus, the heat was transferred from the flue gas to the absorption solution, and the concentrated absorption solution was returned to the absorber.The absorbed water in the absorber was evaporated from the solution in the generator, the steam from the absorption solution flowed through the condenser (HEX 3) and was drained, and the heat was given off to the cooling system.The temperature of the absorption solution was increased in contact with the flue gas.At the same time, a part of the PM was collected from the flue gas by the absorption solution in the tower.The PM was separated by continuous flow through a filter cartridge (3 µm).The pressure drop for a clean cartridge with a liquid flow of 0.6 m 3 / h is 8 kPa, with time the pressure drop increases and recommended change according to manufacture is 0.25 Mpa.The circulating solution in the absorber passed through HEX 1, and the heat taken up from the flue gas was transferred to the cooling system.

Instrumentation
As presented in Fig. 1, the temperature and flow rate of the system were controlled by thermometers and flowmeters.The flue gas humidity and temperature after the absorber were measured manually by a psychrometer (a Testo 635).An automatic control valve controlled the flue gas flow through the generator to keep the setpoint value of the solution temperature in the generator, i.e. the concentration of the solution.Table 1 presents the inaccuracy of the equipment.
A flue gas analyser continuously measured the flue gas composition after the absorber (a Testo 330).The efficiency of the system in PM purification was evaluated by a high-temperature, low-pressure cascade impactor (an HT-DLPI plus from Dekati).The apparatus collected and classified the particles based on the effective cut-off aerodynamic diameter range D 50 (0.01-10 µm) in different stages on 25 mm greased aluminium foil substrates.In addition, a cyclone operated as a pre-cut stage and collected particle sizes larger than 10 µm.Particle collection measurements were conducted under isokinetic conditions for 1 min before and after the absorber.The efficiency of the system in particle reduction was obtained according to the weight difference: where C i, before Abs and C i, after Abs represent the mass concentration of the PM (mg/Nm 3 ) before and after the absorber, respectively, and n denotes the number of stages (14 stages).The n for the PM 1 ranged from 2 to 10, the n for PM 2.5 ranged from 2 to 12, and that for PM 10 ranged from 2 to 15.

Design of the experiment
The aim of this study was to measure the effect of different operation conditions on wet scrubber particle collection efficiency for a domestic boiler size.Table 2 shows the different studied operation conditions, with E1 constituting the reference condition.The flow rates of the solution and the flue gas were kept in a stable range.The feed of fuel and air to the boiler was held constant through all the experiments.The flow rate of the solution entering the absorber was 25 m 3 /m 2 .h,which is a default rate commonly used for absorbers with fillings, and the temperature was set at 40 • C. The flue gas velocity was 0.7 m/s at the bottom of the scrubber and 0.5 m/s at the top due to a temperature decrease.The solution stability had been investigated in a previous study, whose results did not show any solution deterioration over a period of 8 months [20].The operation conditions were selected considering the CFD results from previous work and an analysis of the governing equations.In the CFD simulation, just one force effect was studied at a time, while in the empirical investigations, all the forces acted simultaneously.Therefore, distinguishing the forces from one another was impossible by changing the condition, and presenting the effect of other forces was inevitable [15].The maximum concentration of the solution was limited to 75±2 % from the generator to avoid solidification.

Results and discussion
In the study presented in this paper, the operation conditions of the scrubber (Table 2) were investigated to obtain the efficient operating conditions that could help to develop the pilot plant.Table 3 shows the temperature and relative humidity of the flue gas exiting from the absorber in different experiments.The flue gas conditions before the absorber were set to average values from experiments E1, E2, E4, E5, E6, and E7 when the efficiency was calculated.

Effect of the absorption solution concentration
The measurements were carried out for four different absorption solution concentrations to examine the impact of the concentration gradient on the system efficiency for particle removal.
An increased concentration level of the absorption solution (the remaining part referring to the water content) results in more water being absorbed from the flue gas.
In wet scrubbers for PM collection, the separation mechanisms are divided into three groups according to the size of the particles: a) inertial impaction (a particle size (D p ) greater than 5-10 µm), b) interception (0.5 < D p < 5 µm), and c) Brownian diffusion (D p < 0.5 µm).In real

Table 1
Equipment inaccuracy according to the manufacturer.

Equipment Maximum inaccuracy
Flowmeter (FM1) ± 5 % Flowmeter (FM2-3) 2 % Relative humidity (RH) ± 0.1 • C, ± 2 % RH RADWAG (Wagi Elektronicze) <0.00001 kg Thermometer (TT1-6, TT8-10) conditions, small particles follow the streamline of the flue gas, but their mass causes them to be separated from the flow and impact on the solution surface.Interception can affect smaller particles, which are subjected to Brownian motion and random collisions with gas molecules when the particles are suitably close to the wet surface.Diffusion is the dominant mechanism in the absorber system for particle sizes smaller than 1 µm [18].The effect of the concentration gradient and molecular motion can be explained by the diffusiophoresis force [21].Diffusiophoresis is caused by the effect of a concentration gradient in the system, as a result of which the particles tend to move toward the less concentrated area [22].This force is dominated by the concentration gradient of water vapour, and a higher concentration gradient increases the system's efficiency and particle collection [23].The diffusiophoresis force is strongest where gradients are largest.This gradient decreases as the gas travels through the packing, and as a result, it is very small close to the top of the absorber.
As presented in Fig. 3, the water vapour concentration gradient considerably affects the collection efficiency.In the case of particle sizes larger than 1 µm, the efficiency for lower absorption concentrations was less prominent.However, Reducing the solution concentration reduced the efficiency for all particle sizes.
The CFD results concerning the prediction of particle collection are consistent with the experimental results since, when the diffusiophoresis force was raised, the particles were attached faster [1].An increased liquid concentration leads to an increased water transport from the flue gas to the absorption solution and, therefore, an improved particle collection.The flue gas leaving the absorber has an increased relative humidity with a lower concentration level of the absorption solution, indicating a decreased absorption rate of the water vapour from the flue gas in the absorber, see Tables 2 and 3.The results show that an increased concentration level of the absorption solution improved the particle collection efficiency.Crystallisation occurs for higher salt concentrations at temperatures of practical levels, and therefore a concentration of 75 %± 2 is assumed as an upper limit.

Effect of the temperature gradient in the absorber
The effect of the temperature gradient was evaluated in the second step by setting the temperature of the absorption solution injected into the absorber at different temperatures, namely 30 • C, 40 • C and 50 • C.This was accomplished by adjusting the temperature of the cooling system in experiments E1, E5 and E6.The results from the particle distribution measurement at these different solution temperatures are presented in Fig. 4.
Fig. 4 shows the particle distribution before and after the absorber with different temperatures of the solution injected into the absorber, while the inlet flue gas temperature was constant at around 120℃.The plotted lines represent the particle distribution before and after the absorber.The particle collection efficiency was measured based on the results from a particle distribution evaluation by a cascade impactor.
Fig. 5 illustrates the efficiency of the particle collection with the three different temperatures for the circulating solution.The particle sizes were divided into three groups: PM 1 , PM 2.5 and PM 10 .
The results indicate that a higher temperature gradient (with an incoming flue gas temperature of 120 • C) increases the particle collection efficiency due to the thermophoresis/diffusiophoresis force effect.Moreover, when the solution temperature is decreased, the water vapour gradient increases, which results in more water being absorbed by the

Table 2
Studied operation conditions.E1-8 represent the experiment numbers; each experiment had different operating conditions.In this table, the full height of the packedbed material equals 0.75 m of packed-bed material in the absorber column.Half-height refers to a 0.37 m height of the packed bed, and a 0.19 m height of the packedbed material corresponds to a quarter of the column.solution.As a consequence, the diffusiophoresis force increases and the particle collection is thereby enhanced.The thermophoresis effect can be explained as the tendency of a particle to move toward a lower temperature area due to momentum transfer from gas molecules to particulate matter [24].In this case, the lower temperature of the solution liquid compared to the inlet flue gas temperature increases the number of collected particles, with the surface area of the absorption solution playing the role of a cold surface.According to the results, for PM 10 , the difference in the particle collection efficiency between inlet solutions with a temperature of 40℃ and those with a temperature of 50℃ is around 4 percentage points (pp).The collection efficiency difference for PM 10 between a solution temperature of 30℃ and a solution temperature of 50℃ is around 15 pp.When the temperature of the absorption solution was decreased from 50℃ to 30℃, the particle collection efficiency in the particle size range PM 2.5 was increased by around 15 pp.Totally, an inlet solution temperature of 30℃ showed the most efficient performance concerning particle collection.The results from the experiments showed the same trend as that obtained by the simulation [15].In the simulation, all the other forces were neglected, and only thermophoresis was investigated.As discussed earlier, in the experiments, the effect of the other forces cannot be removed, and therefore, the effect of the temperature on the concentration gradient influenced the results and improved the particle collection efficiency.

Effect of the height of the packed-bed material in the absorber
Three different heights for the packed-bed material were examined to study the effect of the height of the packed-bed material in wet scrubbers on the particle collection efficiency.The absorber tower, which had a constant height, was filled randomly with packed-bed material.It was considered full when the height of the packed-bed material in the wet scrubber column was 0.75 m (E1).The height of the packed bed in the first step was reduced by extracting half of the packed-bed material, which gave a height of about 0.37 m (experiment E7).Since the solution was injected at the top of the tower, in the empty part of the chamber, streams of liquid flowed down to the packed bed; therefore, there was still contact between the flue gas and the liquid in this part.Consequently, particles were still collected by the solution in this part, but the contact area was reduced.In experiment number E8, the height of the packed-bed material was decreased to a quarter of a full height, i.e., to about 0.19 m.In all the cases, the bottom was constant, and the filling height changed from the top.Fig. 6 presents the particle distribution after the absorber for different heights of packed-bed material and compares it with the particle distribution before the absorber.Fig. 7 displays the efficiency of the flue gas purification with different heights of the packed-bed material in the absorber column.
The height of the packed-bed material affects the contact surface between the flue gas and the absorption solution in the absorber.Fig. 7 shows that the contact surface area has a notable effect on the efficiency measured by the cascade impactor [11].The effect was significant when  the height of the packed-bed material decreased to a quarter of full height, with the particle collection efficiency for the particle size less than 10 µm being decreased by around half compared to the efficiency with the material at full height.According to the results, larger particles (D p > 5 µm) experience a greater effect than particles of smaller sizes (see Fig. 6(b)).This is explained by the interception and impaction mechanisms in particle collection.The packed-bed material acted as a collection obstacle for particles that deviated from the flue gas streamline.Therefore, the effect of the height of the packed-bed material is stronger for a particle size larger than 5 µm [12].
The relative humidity of the flue gas leaving the absorber is only slightly higher for a packed bed which is a quarter of full height than for a packed bed of full height, which is why the diffusiophoresis force was almost the same in both cases.Therefore, the difference in collection efficiency for PM 1 is smaller than that for the larger particles.Since the results showed an increased efficiency for the full bed height, we have not reached the optimum level.

Conclusions
In this study, a combustion system equipped with a packed-bed wet scrubber was employed to investigate the effect of the operating conditions on the particle removal efficiency for particles with an effective cut-off aerodynamic diameter ranging from 0.01 to 10 µm.The aim of this study was to improve the particulate matter collection efficiency of small-scale biofuel boilers in an experimental setup.
The effect of the temperature and the concentration of the solution and the height of the packed-bed material in the absorber column was studied.These three parameters form important variables influencing the efficiency of fine particulate matter removal due to their effect on small particle collection mechanisms.The concentration of the absorption solution had the most significant effect on the efficiency of the particulate matter removal, especially in the case of PM 1 .The diffusiophoresis effect can explain this phenomenon; diffusiophoresis was found to be the dominant mechanism for particle sizes smaller than 1 µm.With an absorption solution concentration of 75 %± 2, the particle collection efficiency obtained for the effective cut-off particle sizes PM 1 , PM 2.5 , and PM 10 was around 61 %± 1.
In addition, the height of the packed-bed material had a considerable effect, which was shown when the packed-bed height was reduced to a quarter of full height.The height of the packed-bed material has a stronger effect on larger particulate matter because, by decreasing the height of the packed bed, the obstacle for the particulate matter was reduced.
The absorption solution temperature also has an effect on the particle collection efficiency.With a decreased solution temperature, the flue gas is cleaned to a greater extent.
Overall, it can be concluded that the system efficiency will be improved by keeping the absorption solution concentration at the highest possible level practically.In addition, according to the results, it is beneficial to keep the solution temperature at as low a level as possible since this will lead to an improvement of particulate matter removal and heat recovery.
Furthermore, considering the results from our investigation of the height of the packed-bed material on particle collection efficiency, the optimum level of efficiency was not achieved due to the limitations concerning the absorber size.In addition, the results from the experiments and earlier simulations are in good agreement.Consequently, it can be confirmed that diffusiophoresis has the highest effect on the removal of small particulate matter.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .Fig. 3 .
Fig. 2. Effect of the absorber solution concentration on the particle distribution before and after the absorber; the blue line shows the particle distribution before the absorber.D P refers to the aerodynamic diameter of the particles.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4 .Fig. 5 .
Fig. 4. Effect of the absorption solution temperature on the particle distribution after the absorber and a comparison of the results with the particle distribution before the absorber.The blue line shows the particle distribution before the absorber.D P refers to the aerodynamic diameter of the particles.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6 .
Fig.6.(a) Effect of the height of the packed-bed material in the absorber column on the particle distribution after the absorber; the blue line shows the particle distribution before the absorber.(b) Effect of the height of the packedbed material in the absorber column on the particle distribution after the absorber for particle size Dp = 1-10.3µm.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7 .
Fig. 7. Particle collection efficiency with different packed-bed material heights in the absorber column.

Table 3
The temperature and relative humidity of the flue gas exiting from the absorber.