*2.3. DTF Experiments*

To examine the ash deposition tendencies according to the type of coal and biomass co-combustion, a DTF apparatus (600 mm long with an internal diameter of 70 mm) was used [29]. Figure 2 shows schematics of the DTF apparatus and deposition probe. The DTF apparatus used in this study was designed to simulate operation variables such as heating rate and gas temperature applied to the pulverized fuel boiler in actual coal-fired thermal power plants and to mount an ash deposition probe to indicate the ash deposition characteristics of the solid fuels. The deposited ash particles were captured in an alumina tube (60 mm long with an outer diameter of 25 mm) with a hollow cylindrical shape mounted at the end of the probe and inserted perpendicular to the gas flow inside the DTF apparatus. The ash deposition probe was positioned 10 cm upstream from the back of the DTF apparatus. The particles captured by the probe had relatively low carbon contents (<1% unburned carbon) and were protected from heat by the coolant supplied externally to the water jacket inside the probe. The temperature inside the DTF was maintained at approximately 1300 ◦C, and the total gas flow rate, comprising a mixture of O2 and N2, was 5 L/min. Furthermore, the feeding rate of the sample into the DTF during deposition experiments was 0.2 g/min over 1 h, and this rate was chosen to compensate for supply errors made by the screw feeder mounted on top of the DTF. These errors are caused by two phenomena. The first is supply retention by the screw feeder at the beginning of the experiment due to the densely packed samples, and the second is rate reduction at the end of the experiment due to the loosely packed samples, leading to decreased self-loading of fuel particles. The rate of air transport was adjusted so that the excess O2 content was maintained at 1.16 (*v*/*<sup>v</sup>*, dry), the typical furnace exit value for pulverized fuel boilers [29]. The mass of deposited ash particles was obtained by calculating the di fference between the clean and fouled samples, and this value was used to derive the deposit growth rates.

**Figure 2.** Schematic of the drop-tube furnace (DTF) apparatus and deposit probe (MFC: Mass flow controller; SUS: Steel use stainless).

In particular, to evaluate the ash deposition by biomass co-combustion at the same power output as that achieved with coal alone, the experiments were conducted based on the calorific value of T coal, which is similar to that of the coal typically used in actual domestic power generation boilers. The relevant conditions are illustrated in Figure 3, and the blend conditions for coal and biomass were customized using the thermal fraction of the biomass, as seen in Equations (1)–(3).

.

.

$$T\_{fB} = \frac{m\_B Q\_B}{\dot{m}\_B Q\_B + \dot{m}\_C Q\_C} ; \left( T\_f = T\_{fB} + T\_{fC} \right) . \tag{1}$$

$$M\_{fB} = \frac{m\_B}{\dot{m}\_B + \dot{m}\_C}; \left(M\_f = M\_{fB} + M\_{fC}\right), \text{ and} \tag{2}$$

$$A\_{fB} = \frac{\dot{m}\_B A\_B}{\dot{m}\_B A\_B + \dot{m}\_C A\_C} ; \left(A\_f = A\_{fB} + A\_{fC}\right) . \tag{3}$$

**Figure 3.** Blending conditions by thermal fractions of blends.

Here, *Tf*, *Mf*, and *Af* denote the thermal, mass, and ash fractions by the proportions of the blends, respectively. *Q*, *A*, and *m˙* are the low heating value, ash content on dry basis, and mass flow (subscripts B and C denote biomass and coal), respectively.

Furthermore, the capture efficiency (CE) and energy-based growth rate (GRE) were derived to compare the ash deposition characteristics for the single and blended conditions from the deposition results obtained through five or more repeated experiments. The equations used for the calculations are as follows [30,31]:

$$CE = \frac{m\_D}{m\_A \times \left(\frac{A\_C}{A\_R}\right)} \times 100 \text{ and} \tag{4}$$

$$GRE = \frac{m\_D}{LHV \times m\_F} \,\text{\,\,\,}\tag{5}$$

where *mD*, *mA*, and *mF* denote the weight (g) of the deposit accumulated in the tube, total fly ash flowing towards the tube in the reactor, and blends fed during the test, respectively. *AC* and *AR* are the projected areas of the coupon and the cross-section of the DTF, respectively. These indices were derived from the weight of the collected particles acquired during the deposition experiment, and the measured weight was normalized to explain the difference in the fuel feeding rate.

### **3. Results and Discussion**

#### *3.1. Ashes from the Laboratory Experiments and Actual Combustion*

The chemical composition of the ashes was analyzed to predict their deposition behaviors. In particular, to prevent the volatilization of the minerals generated from the initial biomass combustion process, separate analyses were conducted for the laboratory ash fabricated at a low temperature (575 ◦C) and the combustion ash obtained through the DTF experiment at a high temperature (1300 ◦C). For all the fuels fired, the char burnout in the combustion ash is always higher than 99% when the carbon content of the ash does not exceed 5%. The XRF results in Table 2 and Figure 4 show the chemical compositions of the laboratory/combustion ash and the propensity for slagging/fouling derived from several traditional indices. Table 3 presents the formulae and criteria pertaining to traditional prediction indices for ash deposition.

**Figure 4.** Trends of the traditional indices by X-ray fluorescence (XRF) results; (**a**) T coal + wood pellets (WP) and (**b**) T coal + empty fruit bunches (EFB) (B/A ratio, Si\_R, TA, and Fu refer to base/acid ratio, silica percentage, total alkali, and fouling factor, respectively).


**Table 2.** Results of oxide analyses of laboratory and combustion ashes.

a n.a.: Not analyzed.

**Table 3.** Summary of traditional ash deposition indices and associated criteria.


a Ap: mass percentage of ash in fuel (or blends).

For the laboratory ash, the basic oxides are enriched as the blending ratio is increased, as shown in Table 2. This result is attributed to the unreacted mineral components remaining as-is (un-volatilized) at the low temperature. The effects of basic oxides from both biomass ashes are more evident using the traditional predictive indices and consequently affect the base/acid (B/A) ratio, total alkali (TA), and fouling factor (Fu) of the laboratory ash. As shown in Figure 4, the trends of indices derived from the XRF analysis using laboratory ash show that the slagging/fouling propensity continuously increases as the blending ratio of the biomass is increased.

Conversely, for the combustion ash obtained from the DTF experiments, acid oxides of the deposited ash samples are increased slightly compared to those of the laboratory ash. Particularly, the trends in B/A ratio and Fu identified in Figure 4 have very similar values within the criteria of Table 3, regardless of ash type. This is thought to correspond to the characteristics of the coal ash imparted by the minute fraction of biomass ash in the blending sample, although there is also an effect from the basic oxide of the biomass ash. The combustion ash is a good representation of a fly ash generated by a boiler. The combustion ash of the T coal blended with biomass mainly contains silica and alumina oxides (more than 90 wt%). Nevertheless, the chemical compositions of the laboratory ash and combustion ash vary only slightly with increasing biomass blending ratio, and neither ash type is significantly different from the T coal ash under the given blend conditions. This finding is in agreemen<sup>t</sup> with that of Kupka et al. [21], who observed that, excepting sulfur oxides, fly ash and crucible ash compositions, expressed as oxides, are generally consistent. Thus, Pronobis [12] proposed that the chemical composition of biomass during the co-combustion process is not clearly distinguishable from that of coal when the blending ratio for the thermal fraction of the biomass does not exceed 20%. In other words, the result implies that the chemical composition of biomass ash does

not influence that of the blend when the amount of biomass supplied is sufficiently small compared with that of the coal in terms of thermal input.

Tables 2 and 3 show that the deposition tendencies for both types of ashes meet the low-level criterion within each prediction index range. However, as the addition of biomass increases, the evaluations of the traditional prediction indices gradually increase, especially the TA index for the WP 20% blend. This implies that the blending ratio of coal and biomass whose basic oxides content reaches 90%, such as in the case of WP, should be carefully selected. The XRF results of this study show a difference between the chemical compositions of the ash types, which were generated at different temperatures according to the biomass co-combustion conditions, while the traditional prediction index results do not show a clear difference. Thus, a distinct deposition tendency cannot be confirmed. It is found, however, that as the biomass blending ratio increases, the deposition tendency is slightly increased regardless of ash type because the basic oxides are derived from the biomass ash.

As remarked by Kazagic and Smajevic [20], ash deposition evaluation based on laboratory ash analysis, including the calculation of traditional indices, shows poor reliability for biomasses, and laboratory results cannot be accepted as sufficiently reliable for evaluating the slagging and fouling propensities of biomasses. Therefore, this study performed further experiments using improved methods such as the TMA and DTF, as explained in the following sections.

#### *3.2. Ash Melting Characteristics from TMA Experiments*

To characterize the melting behavior of the laboratory ash, TMA experiments were performed for single-coal and blended samples. Figure 5 shows the shrinkage curves as a function of temperature for each raw sample.

**Figure 5.** Shrinkage curves as a function of temperature for each raw sample.

The shrinkage curves shown in Figure 5 provide a good representation of the melting characteristics of each raw sample according to temperature. T coal is not completely melted at 1600 ◦C, which is the upper limit of the temperature in the TMA experiments performed. This is because T coal is a typical bituminous coal with acid oxide contents exceeding 90%, as seen from the XRF results [33].

Conversely, as demonstrated by the melting traces for both raw biomass samples, their melting reactions are completed at lower temperatures compared to that of T coal. For WP, the basic oxides content exceeds 85%, and T90% appears at a temperature lower than 1150 ◦C. For EFB, T90% appears at approximately 1300 ◦C, i.e., at a higher temperature than that of WP. This result is attributed to the ash components of WP, which have higher alkali contents than EFB, as shown in Table 2 and Figure 4.

When ash samples of blends are heated according to the thermal fraction, the melting trace of each blend follows that of raw T coal, which begins to melt at approximately 900 ◦C, as shown in Figure 6. The melting temperatures for blended biomass are slightly lower or similar to that of raw T coal when the shrinkage is lower than T50% at all blending ratios. Conversely, when the shrinkage is higher than T50%, the melting temperatures are lower than that of raw T coal under all experimental conditions. It appears that the shrinkage is increased because the melting of raw biomasses is almost complete at temperatures lower than approximately 1200 ◦C, at which they are converted to the liquid phase. Moreover, this tendency is conspicuous in the results for EFB and when comparing the T75% value for the blend conditions; the decreased temperature for EFB blending is approximately 100 ◦C greater than that of the WP blends because EFB has higher ash contents despite its lower alkali content.

**Figure 6.** Shrinkage curves as a function of temperature for different blending ratios. (**a**) T coal + WP; (**b**) T coal + EFB.

Figure 7 illustrates the derivative trace of the results seen in Figure 6a,b. These graphs show the melting peak at which rapid melting occurs identified from the first derivative of the melting results derived from the shrinkage of biomass blending and the change in the peak temperature. The melting peaks of the biomass blends revealed in Figure 7a,b occur before approximately 1050 ◦C, which corresponds to a value between T25% and T50% of T coal.

**Figure 7.** Derivative traces of the blend conditions derived from the TMA results; (**a**) T coal + WP; (**b**) T coal + EFB.

As shown in Figure 7, the melting peaks for the blends of both biomasses shift towards lower temperatures as the blending ratio increases based on the peak of T coal (at approximately 1050 ◦C). These peak temperatures are distributed between T25% and T50%, which is the shrinkage characteristic temperature of T coal. The melting is accelerated for the biomass as the peak height is gradually increased. Furthermore, in Figure 7a,b, the yellow peak distribution boxes indicate the melting intensity under a given blending ratio. The range of peak temperatures for both blends are not significantly different. However, the peak heights for EFB blends are higher than those of WP blends in a similar range of peak temperatures. Herein, the peak temperature is associated with the chemical composition of a sample and the height of the peak indicates a specific melting intensity. Since there is no difference in the chemical compositions of the ash of biomass blends and laboratory ash by XRF analysis, the melting intensity is affected by ash content according to blending conditions. As a result, the melting traces for biomass blending obtained with the TMA are confirmed. The melting temperature steadily decreases as the blending ratio of biomass increases, and this phenomenon is not reversed beyond a certain blending ratio. Therefore, this indicates that, as the biomass content in the blend increases, the melting temperature decreases and the slagging/fouling tendency can be expected to increase gradually.

#### *3.3. Ash Deposition Characteristics from DTF Experiments*

The DTF was used to examine the ash deposition of coal and biomass according to the experimental conditions. The deposited ash particles collected on the deposition probe during 1 h of experimentation and the deposition tendencies derived by the CE and GRE were used as the existing coal deposition indices. Figure 8 shows that co-combustion with WP and EFB generate relatively lower ash depositions than that of T coal alone. The single-coal deposition is evenly distributed over the entire area of the coupon and the deposited particles are stacked in a convex shape. In the case of biomass co-combustion, however, there is less deposition than in the case of single coal, and a narrow deposition distribution, with relatively low deposition on the side of the coupon, is observed along with a flatter deposit. Although these deposition images appear to show that the deposition is decreased compared to that of T coal as the blending ratio of the biomass is increased, it is difficult to confirm distinct trends.

**Figure 8.** Images of ash collected on the deposit probe for different blend conditions and singlecombustion.

Figure 9 shows all the CE and GRE values obtained by co-combustion the biomass for all the blend conditions mentioned in Figure 3. The results show that, as the blending ratio of biomass (WP and EFB) increases, the deposition rate decreases for the 10% and 15% blends, but it begins to increase again with additional blending. In general, it is known that the ash in biomass contains significant amounts of alkali matter, which leads to a lower melting temperature, as shown in the TMA results. Most previous studies [34–36] reported that the addition of biomass increases ash deposition and attribute this result to the lower melting temperature and higher stickiness of biomass. Namkung et al. [37] also remarked that sticky particles tend to adhere to each other because of the partial melting characteristics of the particle surface and that increased biomass blending enhances ash deposition and agglomeration behavior. However, even though the blending ratio is increased, the deposition is initially decreased to a certain extent. This is why the ash contents of the biomasses (0.29% for WP and 2.77% for EFB) are much lower than that of T coal ash (17.11%). This finding is in agreemen<sup>t</sup> with those of previous studies. For example, Abreu et al. [15] reported that the co-combustion of coal with lower ash biomass does not pose operational problems related to the occurrence of slagging and fouling because the deposition rate decreases with increases in the biomass thermal input in the blend. In addition, similar effects have been reported by Kupka et al., who indicated that the ash deposition rate is influenced by ash input [21]. Therefore, these results point to the existence of additive and non-additive phenomena caused by lower ash contents and higher agglomeration effects during the co-combustion of coal with the biomasses.

**Figure 9.** Values of capture efficiency (CE) and energy-based growth rate (GRE) for the blend conditions in the DTF; (**a**) T coal + WP; (**b**) T coal + EFB.

Figure 10 shows the normalized GRE values for the blend conditions in the DTF. Herein, GREmax represents the ratio of the GRE values of other samples to the maximum GRE value. The GREmax range of the WP blends is approximately 0.35–0.60, and that of the EFB blends is approximately 0.50–0.80. While the deposition of WP blends is less than half the deposition by the single combustion of T coal for all blending ratios except 5%, the deposition of EFB blends is higher than half the T coal deposition under all the co-combustion conditions. These results show that the deposition tendency with the co-combustion of EFB is higher than that of WP, unlike the predicted results using the chemical composition of the biomass revealed by XRF. In other words, the ash amounts of EFB, according to the feeding conditions, are approximately 10 times higher than those of WP, which demonstrates the influence of the deposition tendency. In addition, the reversal tendency of the EFB blend, i.e., the increase in deposition after the decrease, is changed at a higher blending ratio. This indicates that the deposition is more influenced by the feeding rate of the EFB than by its chemical composition when the amount of the biomass supplied is sufficiently small.

**Figure 10.** Normalized GRE values for blend conditions in the DTF; (**a**) T coal + WP; (**b**) T coal + EFB.
