Experimental investigation of sulfite oxidation enhancement in a micro-pore aeration system

Jun Zhang, Cheng-hang Zheng, Yong-xin Zhang, Zhe-wei Xu, Li Wang, Cun-jie Li and Xiang Gao*
State Key Laboratory of Clean Energy Utilization, State Environmental Protection Engineering Center for Coal-Fired Air Pollution Control, Zhejiang University, Hangzhou 310027, Zhejiang Province, China. E-mail: xgao1@zju.edu.cn; Tel: +86 571 87951335

Received 30th July 2016 , Accepted 12th October 2016

First published on 13th October 2016


Abstract

Seawater wet flue gas desulfurization is a promising process for coal-fired power plants. The relationship between sulfite oxidation and different parameters in the process of desulfurization seawater recovery was investigated in a laboratory-scale experimental system. The results suggest that the effect of pH on the KS(IV) shows a trough shape, and the best KS(IV) was achieved in the pH of 5.8. High temperature, salinity and alkalinity show a positive effect on the oxidation of S(IV), whereas the effect of flow rate is more complicated. It is found that micro-pore aeration has a better performance than coarse aeration with low flow rate, which is good for energy savings. In addition, a multi-layer perceptron model has been designed for prediction of the influence of different factors on sulfite oxidation by micro-pore aeration. The results demonstrated that the top five factors were pH, alkalinity, temperature, aeration depth and flow rate.


1. Introduction

Atmospheric pollution has serious effects on human health; consequently, it has gained wide attention from the government and public. The sulfur dioxide released in the atmosphere during the coal combustion process is one of the main causes of air pollution. In China, about 2.79 billion tons of standard coal were consumed in 2014, accounting for more than 50% of global consumption.1 About 90% of the sulfur dioxide emissions across the country were from burning coal. Furthermore, coal consumption intensity is much higher than the national average in eastern China, which leads to higher pollutant emission intensity in this area. Thus, high-efficiency desulfurization technology for coal-fired power plants is an urgent topic of environmental research.

In China, wet flue gas desulfurization (WFGD) accounts for more than 90% of the total FGD systems installed.2,3 Seawater WFGD provides many advantages, such as high desulfurization efficiency and no chemical absorbent (only seawater and air), for the treatment of flue gas containing SO2 from coal-fired power plants. The seawater WFGD process generally includes a flue gas system, an absorption system, a water recovery system, etc. In the absorption system, alkaline substances contained in the seawater will react with HSO3 or SO32− that come from the aqueous dissolution and hydrolysis of SO2. In the recovery zone, the air injected in the reaction tank by bottom sprinklers reacts with HSO3 or SO32−, which will be oxidized to SO42−. Moreover, the CO2 formed by the neutralization reaction is stripped by an aeration membrane and the pH value is raised as a result, thus meeting the legal requirements.4

The seawater recovery system is energy-intensive, and it is also the key to emission compliance for seawater WFGD. In recent years, oxidation-enhanced sulfite in the aeration system has gained increasing interest. Vidal et al.5–8 determined the oxidation kinetics of S(IV) in seawater, and the results demonstrated that the reaction was first order with respect to S(IV) and zero order with respect to oxygen. Vidal and Ollero8 also found that activated carbon had a significant catalytic effect on the S(IV) oxidation process, which was demonstrated in a pilot scale. Lan9 revealed that approximately 25% of absorbed S(IV) was oxidized to S(VI) by surplus oxygen in the flue gas during the absorption process. It has been reported that transition metal ions such as Fe2+, Mn2+, Co2, Se have great effect on the oxidation rate of S(IV).10–12 In addition, some studies focused on improving the performance of the aeration system. Chern et al.13 tried to enhance the capacity of a diffused aeration system using various diffusers. The results of this study showed that the volumetric mass-transfer coefficient and the surface reaeration zone increased with an increase of air flow rate and temperature, but decreased with an increase of water depth. S Gillot14 established relationships enabling prediction of oxygenation performances of fine bubble aeration systems. Several papers reported novel high-efficiency oxidation methods to convert S(IV) to S(VI), such as pulsed corona discharge,15 plate falling film corona-streamer discharge16 and cylindrical wetted-wall corona-streamer discharge.17

A multi-layer perceptron (MLP) neural network is a computational model based on the structure and operation of a biological neural network. It has a strong ability to implement approximating nonlinear functions by arbitrary accuracy through regulating variable weight connections. In recent years, the artificial neural network (ANN) has obtained certain achievement in different fields. The multi-layer perceptron (MLP) neural network is a powerful strategy in ANN. Due to their self-learning, modeling and prediction abilities, MLPs are in a position to reproduce the mapping relationships between input and output data based on limited experimental samples.18,19 A well-trained MLP neural network is able to approximate any continuous reasonable function mapping with a satisfactory level of accuracy.20 Han et al.21 applied the RBFN to the prediction of output-water quality in the wastewater treatment process. Gao et al.22 developed a three-layer back propagation ANN model to get a better understanding of the roles of different process parameters on methanol removal efficiency and energy efficiency in the post-plasma catalytic process.

Further study of seawater WFGD is needed to determine the effects of operational parameters, such as initial pH, temperature, aeration flow rate, aeration depth, alkalinity, and salinity, on sulfite oxidation in simulated seawater. The objective of this study is to investigate the oxidation-enhanced sulfite mechanism in a new micro-pore aeration system. The oxidation efficiency was investigated under different operating parameters, and the optimal operation conditions were obtained. A MLP ANN model has been developed for modeling and prediction of the effects and relative importance of different parameters on the sulfite oxidation performance. The results can be used for optimization of seawater recovery in WFGD systems.

2. Experimental approach

2.1 Experimental apparatus and materials

The schematic of the experimental system is illustrated in Fig. 1. Simulated seawater was prepared from water and sea salt to the desired concentration and temperature in the bath before being fed into the reactor. First, N2 gas was used to strip DO (dissolved oxygen) in the simulated seawater until its concentration was lower than 0.2 mg L−1. Then, SO2/N2 (5%) gas, or sodium sulfite and hydrochloric acid solution, was used to adjust the initial pH. The reactor was a cylinder with a volume of 107.4 L (380 mm in diameter and 2000 mm in height). The aeration plate was at the bottom of the reactor. The aeration flow rate was controlled by mass flow meters (SevenStar Huachuang Co., Ltd., China) with an accuracy of ±1% F.S.
image file: c6ra19337d-f1.tif
Fig. 1 Schematic of the experimental setup.

All the chemicals were used as received without further purification: sodium chloride for industrial use, potassium chloride (≥99.5%), sodium bicarbonate (≥99.5%), hydrochloric acid (36–38%), sodium sulfite, ammonium sulfamate (≥99%), ammonium sulfate, sodium thiosulfate standard solution (0.1 mol L−1), iodine standard (0.1 mol L−1), hydrochloric acid standard solution (0.1 mol L−1). Gases, SO2 (5%) and N2 (>99.999%), were stored in steel cylinders as provided by Hangzhou New Century Gas Co., Ltd.

2.2 Experimental procedure

The pH, DO, temperature and other parameters of the simulated seawater were subjected to real-time monitoring by an online monitoring system. The salinity content and alkalinity content in the solution were measured using a ATAGO PAL-ES3 and an acid-based indicator titration method, respectively. The solution samples containing sulfite were analyzed by Chinese standard, GB/T 14426-93. Sulfite analysis of water used in the boiler and cooling systems was performed in a similar manner. The concentration of sulfate ions in the samples was analyzed using ion chromatography (Dionex ICS-900). All the experimental data measurements were repeated at least three times. The removal efficiency of S(IV) was defined as follows:
 
image file: c6ra19337d-t1.tif(1)
where ct0,S(IV) and ct,S(IV) were the S(IV) concentrations in the solution at the beginning and after t minutes of the experiment, respectively.

Because the aeration reaction process can be considered as a Gaussian process, sulfite oxidation performance index (KS(IV)) was defined as follows:

 
KS(IV) = t90%t10% (2)
where t90% and t10% were the times at which the oxidation efficiency of S(IV) under analysis was 90% and 10%, respectively, of the steady-state value, measured in seconds.

2.3 Development and optimization of the neural network

In this study, an MLP neural network containing 3 layers was developed for the modeling and prediction of sulfite oxidation performance index using SPSS Statistics (v22). Six operating parameters, initial pH value, temperature, flow rate, alkalinity, aeration depth and salinity, were identified as the input variables for the neural network, whereas sulfite oxidation performance index was used as the output variable. Therefore, the input and output layers consisted of 6 and 1 neurons, respectively.

The optimization algorithm, activation function and neuron number in the hidden layer were optimized using a trial and error method. A combination of a scaled conjugate gradient optimization algorithm with a hyperbolic tangent activation function at the hidden layer and a sigmoid activation function in the output layer was chosen for the neural network due to its superior accuracy among the tested cases (see Table S1 in ESI). The optimal neuron number at the hidden layer for the effective MLP model was 15, with a minimum MSE of 8.34 × 10−4 (see Fig. S1 in ESI). Fig. 2 shows the structure of the optimized model. The weight matrix produced by the well-trained MLP model is listed in Table S2 in the ESI.


image file: c6ra19337d-f2.tif
Fig. 2 Optimized three-layer MLP neural network model.

3. Results and discussion

3.1 Effect of initial pH

The effect of initial pH on S(IV) oxidation efficiency was investigated, and the results are shown in Fig. 3. Experiments were operated at VG = 2 m3 h−1, SAL = 3%, T = 293.15 K, ALK = 130 mg L−1, and H = 1.5 m with various initial pH values. The results suggested that the initial pH value does play a critical role in sulfite oxidation. The oxidation efficiency reached a maximum of 82.5% when the initial pH was 5.8.
image file: c6ra19337d-f3.tif
Fig. 3 Effect of initial pH on the S(IV) oxidation efficiency.

As shown in Fig. 3, the oxidation of sulfite can be divided into two processes. In the fast response phase, which occurred during the first 3 min, the concentration of sulfite dropped rapidly and the oxidation efficiency of sulfite rose. Equilibrium, or slow phase, was achieved after 3 minutes, at which point the reaction entered the aeration system.

As indicted in Fig. 4, the initial pH value played a critical role in KS(IV). The definition of KS(IV) is given in the formula (2). The V-shaped curve, with the best KS(IV) value in the region of around pH 5.8, means that sulfite was most easily oxidized when the pH value was 5.8. It is evident that the experimental data fit the reference results well.23 Wilkinson et al.23 studied the uncatalyzed S(IV) oxidation reaction and found that the reaction rate increased with increasing pH to a value between 5.7 and 5.9, but it was not influenced by pH above this range.


image file: c6ra19337d-f4.tif
Fig. 4 The performance index KS(IV) vs. pH.

It has been reported that sulfite is much easier to oxidize than bisulfite. Zhang9 and Vidal7 pointed out that the kinetic constant of S(IV) oxidation ​reaction increases as the pH increases up to an approximate value of 6, and for values greater than this, the kinetic constant decreases. The main reactions of the S(IV) oxidation in aeration process can be summarized as

 
O2(g) ↔ O2(aq) (3)
 
HSO3(aq) ↔ SO32−(aq) + H+(aq), ΔfG0 = 41.2 kJ mol−1 (4)
 
image file: c6ra19337d-t2.tif(5)
 
image file: c6ra19337d-t3.tif(6)

The pK1 and pK2 for the dissociation of H2SO3 in sodium chloride solutions from potentiometric measurements are 2.63 × 10−2 and 7.41 × 10−7, respectively.24,25 Zhang et al.9 observed the maximum oxidation rate at pH 6.5 in the presence of Mn and Fe ions and found that the concentrations of SO32− and HSO3 were 50% each. Thus, the mechanism can be summarized as follows:

 
O2(aq) + SO32−(aq) + HSO3(aq) ↔ 2SO42−(aq) + H+, ΔfG0 = −474.8 kJ mol−1 (7)

3.2 Effect of reaction temperature

Fig. 5 shows the effect of slurry temperature on S(IV) oxidation efficiency. The initial pH was 5.5, VG was 3 m3 h−1, salinity was 3%, alkalinity was 130 mg L−1 and aeration depth was 1.5 m. According to the real reaction conditions of seawater WFGD, three different temperature gradients were set up: 293.15 K, 303.15 K and 313.15 K respectively.
image file: c6ra19337d-f5.tif
Fig. 5 Time vs. S(IV) oxidation efficiency in different temperature.

As illustrated in Fig. 5, reaction temperature has a positive effect on oxidation efficiency. The S(IV) oxidation efficiency increases from 75.5% to 99.5% as the temperature increases from 293.15 K to 313.15 K. As mentioned before, the oxidation reaction process includes two stages, i.e., fast response and slow response. In the first 3 minutes, the reaction was maintained in the fast response stage with different temperatures. Thus, the reaction temperature had no effect on the fast response stage time.

The result is probably due to the fact that the diffusion coefficient of O2 in gas–liquid two-phase and SO32− in the liquid phase will increase correspondingly with increasing temperature, promoting more oxygen dissolved in the liquid phase, thus improving the chemical reaction rate, which results in higher oxidation rate.

3.3 Effect of aeration flow rate

The effect of flow rate on S(IV) oxidation efficiency was investigated, and the results are shown in Fig. 6. Experiments were carried out at SAL = 3%, pH = 5.8, T = 303.15 K, ALK = 130 mg L−1, and H = 1.5 m, with various aeration flow rates. Different flow rates of aeration were set up to observe the effect of aeration intensity on the oxidation efficiency.
image file: c6ra19337d-f6.tif
Fig. 6 Effect of flow rate on the KS(IV).

As can be seen from Fig. 6, S(IV) oxidation efficiency first increases, and then decreases, with increasing flow rate. The oxidation efficiency reached a maximum of 98.2% at an aeration flow rate of 2.0 m3 h−1. This occurred because aeration enhanced the transfer rate of dissolved oxygen through drumming into a large number of tiny bubbles. However, bubble diameter increased and residence time decreased with increasing aeration flow rate. This resulted in decreased contact time and, to some extent, a smaller interface, causing the decline in efficiency. Under conditions of high aeration, the bubbles had the characteristics of high flow velocity, large diameter and streamlined shape. Visible particulate air bubbles could be observed in the aeration diffusor export when flow rate was 0.5 m3 h−1 and 1.0 m3 h−1. As flow rate increased from 1.0 m3 h−1 to 8 m3 h−1, bubbles slowly started to show streamlined shape.

Aeration flow rate had a great effect on the energy consumption of the process. As stated above, the performance of aeration is better with a flow of 1 and 2 m3 h−1. It is very important to adjust the aeration flow rate for overall system energy savings.

3.4 Effect of aeration depth

Aeration depth refers to the height difference between the aeration diffusor and the horizontal plane. The effect of aeration depth on the oxidation efficiency of S(IV) was observed with alkalinity, temperature, salinity, initial pH and flow rate of 130 mg L−1, 303.15 K, 3%, 6.0 and 2 m3 h−1, respectively.

The effect of aeration depth on oxidation efficiency is illustrated in Fig. 7. These results suggest that aeration depth had a relatively minor effect on S(IV) oxidation. The best aeration depth was 1.0 m under the experimental conditions. The appearance of the optimum operating point mainly resulted from the fact that deeper aeration simultaneously brings in more oxygen content and resistance. On one hand, increased aeration depth brought with it increased gas–liquid contact area, whereas, the aeration flow rate per unit of seawater decreased.


image file: c6ra19337d-f7.tif
Fig. 7 Effect of aeration depth on the S(IV) oxidation efficiency.

3.5 Effect of alkalinity and salinity

The effect of alkalinity on S(IV) oxidation efficiency is shown in Fig. 8. Experiments were operated at SAL = 3%, pH = 6.0, T = 293.15 K, VG = 2 m3 h−1, and H = 1.5 m with varying alkalinity. The results showed that the oxidation efficiency increased with increasing alkalinity. When the alkalinity increased from 100 mg L−1 to 130 mg L−1, the S(IV) oxidation efficiency increased slowly. However, it increased notably when the alkalinity was increased from 130 mg L−1 to 160 mg L−1. At an alkalinity of 160 mg L−1, the S(IV) oxidation efficiency was about 96.94% after 10 minutes, and lower alkalinity resulted in poorer performance. Sodium bicarbonate was used to adjust the alkalinity of the simulated seawater. The mechanism of the effect of alkalinity on sulfite oxidation can be summarized as follows:
 
CO32− + H+ ↔ HCO3, ΔfG0 = −59 kJ mol−1 (8)
 
HCO3 + H+ ↔ H2O(l) + CO2(g), ΔfG0 = −44.7 kJ mol−1 (9)
 
HSO3 ↔ H+ + SO32−, ΔfG0 = 41.2 kJ mol−1 (10)

image file: c6ra19337d-f8.tif
Fig. 8 Effect of alkalinity on the S(IV) oxidation efficiency.

The concentration of HCO3 in the simulated seawater was increased with the addition of sodium bicarbonate. The above equilibrium reactions indicate that H+ depletion by reaction (9) will promote reaction (10) to the right. Zhang et al.9 reported that sulfite is more easily oxidized by O2. This explains why alkalinity has a positive effect on sulfite oxidation.

The effects of different salinity (0%, 3% and 6%) on the oxidation efficiency of S(IV) were observed, with the other parameters being held constant at alkalinity of 130 mg L−1, temperature of 30 °C, initial pH of 6.0, aeration intensity of 2 m3 h−1, and aeration depth of 1.5 m.

Fig. 9 shows the effect of salinity on the oxidation efficiency of S(IV). The oxidation efficiency of S(IV) in simulated seawater with a certain salinity increased rapidly, and reached the maximum value in the first two minutes. The S(IV) oxidation efficiency at 6% salinity was slightly higher than that at 3% salinity after ten minutes. These results suggested that salinity does play a role in promoting sulfite oxidation. The possible reason is that a higher salinity meant the presence of more bicarbonate and carbonate in the simulated seawater, which resulted in a better pH buffering effect. Another possible reason is the greater quantity of metal and Cl ions in seawater of higher salinity; these types of elements were reported to have a positive catalytic effect on sulfite oxidation.


image file: c6ra19337d-f9.tif
Fig. 9 Effect of salinity on the S(IV) oxidation efficiency.

3.6 Comparison of micro-pore and coarse aeration systems

The performance of sulfite oxidation using a micro-pore aeration system has already been discussed in previous parts of this study; herein, it is compared to a coarse aeration system. It is evident from the data presented in Table 1 that the distribution uniformity was better and the bubble size was smaller in the micro-pore as compared with coarse aeration system.
Table 1 Comparison of different experimental conditions
Aeration mode Bubble size (mm) Flow rate Distribution uniformity
Coarse aeration >20 Fast Middle
Micro-pore 1–4 Middle Well


The performance of sulfite oxidation in seawater was greatly affected by aeration condition. As Fig. 10a and b show, the micro-pore aeration system has a better performance than coarse aeration system with a flow rate of 1 and 2 m3 h−1. Under the flow rate of 2 m3 h−1, the oxidation efficiency of the micro-pore system increased rapidly for the first 2 min, reaching 86.72% after 8 minutes, and the oxidation efficiency of the whole process was 95.31%. In comparison, the oxidation rate of the coarse aeration system for first two minutes was significantly less. The oxidation efficiency of coarse aeration at 10 min was just equivalent to that of micro-pore at 2 min.


image file: c6ra19337d-f10.tif
Fig. 10 Comparison of micro-pore results with coarse aeration results data at (a) 1 m3 h−1 (b) 2 m3 h−1 (c) 4 m3 h−1, and (d) 8 m3 h−1 flow rate.

As Fig. 10c and d show, micro-pore and coarse aeration systems produced almost the same results when aeration flow rate was increased from 2 m3 h−1 to 4 m3 h−1 and 8 m3 h−1. That's probably due to the fact that the mass transfer of the oxidation reaction was enhanced by the micro-pore aeration system. However, the strengthening effect has been undermined with a flow rate of 4 m3 h−1 to 8 m3 h−1. Furthermore, the results showed that the micro-pore aeration has great advantages in reducing system power consumption.

DO value is one of the criterions to evaluate the recovery of seawater and a higher DO value can promote the oxidation of sulfite. Fig. 11 shows the effect of aeration condition on DO value, with seawater pH of 5.8, temperature of 30 °C, flow rate of 2 m3 h−1 and aeration depth of 1.5 m. The results showed that the simulated seawater had a rapid increase in DO value to 6 mg L−1 in the first 4 min. On the contrary, DO value did not increase in the first 8 min, but slowly increased after 8 min. The micro-pore aeration system had a significant advantage in the rapid increase concentration of dissolved oxygen.


image file: c6ra19337d-f11.tif
Fig. 11 Comparison of micro-pore results with coarse aeration results data.

3.7 Sensitivity analysis of different process parameters

A multi-layer perceptron (MLP) model has been developed for modeling and prediction of the influence of different factors on sulfite oxidation by micro-pore aeration. Fig. 12 shows the differences in the experimental and predicted values of KS(IV) for the ANN patterns used. A good agreement has been observed, for which the average mean square error (MSE) is 8.34 × 10−4. The developed MLP model was applied to carry out a sensitivity analysis, and the sensitivity analysis results are receivable.
image file: c6ra19337d-f12.tif
Fig. 12 The comparison between the predicted and experimental values of KS(IV).

In this study, the mean impact value (MIV) was used to evaluate the relative importance of each parameter on seawater flue gas desulfurization. The relative importance of each processing parameter is shown in Table 2. It shows that initial pH has a significant impact on the property index with a relative weight of 62.39%; the relative importance of the other parameters on the property index is similar. The results suggest that efficiency of S(IV) oxidation can be effectively improved, and aeration time can be decreased by controlling the initial pH value.

Table 2 Relative importance (%) of parameters on KS(IV)
Parameter Relative importance (%)
Initial pH 62.39
Temperature 19.23
Aeration flow rate 10.17
Alkalinity 6.09
Aeration depth 1.76
Salinity 0.36


The mean values of each of the control variables, when applied to the model, yield an output taken as the base or reference value for the sensitivity analysis. The base values were pH = 5.74, T = 303.15 K, VG = 2.23 m3 h−1, SAL = 132.6 mg L−1, H = 1.50 m, and ALK = 3%. Using this approach, the sensitivities of the output of each of the control input variables were obtained, ranging over a ±15% variation. Fig. 13 shows the sensitivities graphically.


image file: c6ra19337d-f13.tif
Fig. 13 Sensitivity analysis results for the control variables.

It was observed by the variation indicated in Fig. 13 that the initial pH has the most significant impact. It is represented as a V-shaped curve, which is similar to the results above, since the basic pH value was 5.74. The change in the percentage output was almost linear with respect to temperature, aeration flow rate, salinity and aeration depth. The association of these parameters with KS(IV) was positive, having approximate gradients of −2.48, −0.97, −0.85, −2.23, respectively. It was noticed that the alkalinity does not have a strong influence on the KS(IV). The rise in KS(IV) for a 15% decrement in the alkalinity was 5.07%.

4. Conclusions

In this study, the effect of pH, temperature, alkalinity, aeration depth, and aeration flow rate on sulfite oxidation in the process of desulfurization seawater recovery was investigated. The results indicated that the effect of pH on the KS(IV) shows a trough shape, and the best KS(IV) was achieved in the pH of 5.8. High temperature, salinity and alkalinity of seawater and aeration flow rate all had a positive effect on the oxidation of S(IV). The effect of aeration depth was more complex, and the best aeration depth was 1.0 m under the experimental conditions. Because of the increase of aeration depth, the gas–liquid contact area increased and the aeration flow rate per unit of seawater decreased.

A comparison of micro-pore and coarse aeration system was also discussed. The micro-pore aeration system had a great advantage under the condition of low aeration flow rate. This was explained by the fact that micro-pores are good for oxygen dissolution, which is one of the criterions to evaluate the recovery of seawater.

A multi-layer perceptron (MLP) artificial neutral network (ANN) model has been developed for the modeling and prediction of the influence of different factors on sulfite oxidation by micro-pore aeration. The predicted results of the ANN model were in good agreement with the experimental results. Initial pH was found to be the most important factor affecting KS(IV), with a relative importance of 62.39%. The top five factors were initial pH, alkalinity, temperature, aeration depth and flow rate.

Nomenclature

KS(IV)Sulfite oxidation performance index(s)
DODissolved oxygen (mg L−1)
TTemperature (K)
ALKAlkalinity (mg L−1)
VGAeration flow rate (m3 s−1)
HAeration depth (m)
SALSalinity (%)
ηSulfite oxidation efficiency (%)
ct0,S(IV)S(IV) concentrations in the solution at the beginning of the experiment (mg L−1)
ct,S(IV)S(IV) concentrations in the solution after t min of the experiment (mg L−1)

Acknowledgements

This work was financially supported by the National Science and Technology Support Program (grant no. 2014BAC22B06, 2015BAA05B02), the Key Consulting Project of China Academy of Engineering (grant no. 2014-ZD-12), the Environmental Welfare Project of Ministry of Environmental Protection of China (grant no. 201509021), and the Key Innovation Team for Science and Technology of Zhejiang Province (grant no. 2011R50017) for their support for this work.

References

  1. China, N. B. o. S. P. s. R. o., China Energy Statistical Yearbook 2015, China Statistics Press, 2015 Search PubMed.
  2. C. H. Zheng, C. R. Xu, X. Gao, Y. X. Zhang, J. Zhang, Z. Y. Luo and K. F. Cen, Simultaneous absorption of NOx and SO2 in oxidant-enhanced limestone slurry, Environ. Prog. Sustainable Energy, 2014, 33(4), 1171–1179 CAS.
  3. C. H. Zheng, C. R. Xu, Y. X. Zhang, J. Zhang, X. Gao, Z. Y. Luo and K. F. Cen, Nitrogen oxide absorption and nitrite/nitrate formation in limestone slurry for WFGD system, Appl. Energy, 2014, 129, 187–194 CrossRef CAS.
  4. J. Gong, X. Zhang, X. Wang and L. Lei, Oxidation of S(IV) in seawater by pulsed high voltage discharge plasma with TiO2/Ti electrode as catalyst, Plasma Sci. Technol., 2013, 15(12), 1209–1214 CrossRef.
  5. B. F. Vidal, P. Ollero, F. Gutiérrez Ortiz and A. Villanueva, Catalytic seawater flue gas desulfurization process: an experimental pilot plant study, Environ. Sci. Technol., 2007, 41(20), 7114–7119 CrossRef.
  6. F. Vidal Barrero, P. Ollero, A. L. Villanueva Perales and A. Gomez-Barea, Catalytic seawater flue gas desulfurization model, Environ. Sci. Technol., 2009, 43(24), 9393–9399 CrossRef CAS PubMed.
  7. F. Vidal and P. Ollero, A kinetic study of the oxidation of S(IV) in seawater, Environ. Sci. Technol., 2001, 35(13), 2792–2796 CrossRef CAS PubMed.
  8. F. Vidal, P. Ollero, F. J. G. Ortiz and R. Arjona, Catalytic oxidation of S(IV) in seawater slurries of activated carbon, Environ. Sci. Technol., 2005, 39(13), 5031–5036 CrossRef PubMed.
  9. T. Lan, X. Zhang, Q. Yu and L. Lei, Study on the relationship between absorbed S(IV) and pH in the seawater flue gas desulfurization process, Ind. Eng. Chem. Res., 2012, 51(12), 4478–4484 CrossRef CAS.
  10. T. Lan, L. Lei, B. Yang, X. Zhang and Z. Li, Kinetics of the iron(II)- and manganese(II)-catalyzed oxidation of S(IV) in seawater with acetic buffer: a study of seawater desulfurization process, Ind. Eng. Chem. Res., 2013, 52(13), 4740–4746 CrossRef CAS.
  11. Y. Zhao, S. C. Ma, X. M. Wang and Q. Zhang, Experimental and mechanism studies on seawater flue gas desulfurization, J. Environ. Sci., 2003, 15(1), 123–128 CAS.
  12. L. Chai, B. Yang, H. Liu, K. Xiang, S. Yang and C. Peng, Se-catalyzed process of sodium bisulfite disproportionation, J. Chem. Technol. Biotechnol., 2016, 91, 1023–1030 CrossRef CAS.
  13. J. M. Chern and S. P. Yang, Oxygen transfer rate in a coarse-bubble diffused aeration system, Ind. Eng. Chem. Res., 2003, 42(25), 6653–6660 CrossRef CAS.
  14. S. Gillot, S. Capela-Marsal, M. Roustan and A. Heduit, Predicting oxygen transfer of fine bubble diffused aeration systems-model issued from dimensional analysis, Water Res., 2005, 39(7), 1379–1387 CrossRef CAS PubMed.
  15. N. Shi, X. Zhang and L. Lei, Sulfite oxidation in seawater flue gas desulfurization by a pulsed corona discharge process, Sep. Purif. Technol., 2009, 70(2), 212–218 CrossRef CAS.
  16. X. Wang, Z. Li, T. Lan and L. Lei, Sulfite oxidation in seawater flue gas desulfurization by plate falling film corona-streamer discharge, Chem. Eng. J., 2013, 225, 16–24 CrossRef CAS.
  17. X. Wang, Z. Li and L. Lei, Model study of sulfite oxidation in seawater flue gas desulfurization by cylindrical wetted-wall corona-streamer discharge, Chem. Eng. Sci., 2013, 97, 7–15 CrossRef CAS.
  18. D. S. Broomhead and D. Lowe, Multivariable functional interpolation and adaptive networks, Complex Systems, 1988, 2(3), 321–355 Search PubMed.
  19. M. Kubat, Neural networks: a comprehensive foundation by Simon Haykin, Knowl. Eng. Rev., 1999, 13(4), 409–412 CrossRef.
  20. J. Park and I. W. Sandberg, Universal approximation using radial-basis-function networks, Neural Comput., 1991, 3(2), 246–257 CrossRef.
  21. H.-G. Han, Q.-L. Chen and J.-F. Qiao, An efficient self-organizing RBF neural network for water quality prediction, Neural Networks, 2011, 24(7), 717–725 CrossRef PubMed.
  22. X. Zhu, S. Liu, Y. Cai, X. Gao, J. Zhou, C. Zheng and X. Tu, Post-plasma catalytic removal of methanol over Mn-Ce catalysts in an atmospheric dielectric barrier discharge, Appl. Catal., B, 2016, 183, 124–132 CrossRef CAS.
  23. P. M. Wilkinson, B. Doldersum, P. H. M. R. Cramers and L. L. Van Dierendonck, The kinetics of uncatalyzed sodium sulfite oxidation, Chem. Eng. Sci., 1993, 48(5), 933–941 CrossRef CAS.
  24. F. J. Millero, J. B. Hershey, G. Johnson and J.-Z. Zhang, The solubility of SO2 and the dissociation of H2SO3 in NaCl solutions, J. Atmos. Chem., 1989, 8(4), 377–389 CrossRef CAS.
  25. R. N. Roy, J.-Z. Zhang, M. A. Sibblies and F. J. Millero, The pK2* for the dissociation of H2SO3 in NaCl solutions with added Ni2+, Co2+, Mn2+, and Cd2+ at 25 °C, J. Solution Chem., 1991, 20(5), 467–478 CrossRef CAS.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra19337d

This journal is © The Royal Society of Chemistry 2016
Click here to see how this site uses Cookies. View our privacy policy here.