Pilot-scale Validation of the Modeling of NO x Reactive Absorption Process Using Aqueous Solutions Containing Nitric Acid and Hydrogen Peroxide

NO x are harmful pollutants emitted by many industries and must be removed before the gases are released into the atmosphere. Among the various techniques used for NO x removal, reactive absorption by hydrogen peroxide in a packed column has the advantage of being able to transform NO x into nitric acid which can be recycled and reused in the plants. This work aimed at modeling the NO x absorption by means of aqueous solutions of hydrogen peroxide and nitric acid, applied in a packed column operating under different representative conditions. The model established using the Aspen software considers the equilibrium and the kinetics of the reactions taking place both in the gas phase and in the liquid phase as well as the thermodynamic properties of the chemical (molecular and ionic) components estimated using the Electrolyte NRTL model. It allows predicting NO x absorption performances in terms of efficiency and selectivity. The model was validated by comparing simulation results with experimental ones obtained in a pilot-scale absorption column and published previously. The predicted values are in satisfactory agreement with the experimental data showing a deviation between 5 and 8%. Therefore, the model developed in this work could be advantageously used to design industrial-scale reactive NO x absorption columns.


Introduction
Nowadays, the unfavorable environmental and human health harms caused by emissions of NO x [1,2] make its reduction one of the most important challenges for industrial and scientific researchers.A variety of techniques for NO x abatement has been developed and can be divided into two categories: dry processes and wet scrubbing processes [3,4].Among the dry processes can be distinguished the selective catalytic reduction (SCR), the selective non-catalytic reduction (SNCR) and the NO x adsorption process.The selective catalytic reduction (SCR) uses ammonia or urea to promote the NO x reduction at the surface of a catalyst conventionally based on V 2 O 5 -WO 3 / TiO 2 or Cu-and Fe-zeolites, operating at low temperatures of 300 to 400 °C [5].It presents certain disadvantages such as the high investment cost, the poisoning (by SO x for example) and the limited lifetime of the catalyst, and the risk of ammonia slips [3,6].For its part the selective non-catalytic reduction (SNCR) using the same reagents operating under high temperature between 900 and 1000 °C [7,8], is limited by the quite low efficiency and higher operation cost due to the higher reagent consumption [9,10].The NO x adsorption process has been considered as one of the environmentally friendly technologies for the removal of NO x .However, it presents the disadvantages of high adsorbent materials cost and a quite low efficiency [11].On the other hand, wet scrubbing processes using water as absorbent liquid are very inefficient and strongly limited by the low solubility in water of nitric oxide NO which is usually the major constituent of NO x species [4,9].To overcome this inconvenient, some reagents are added to water like sodium hypochlorite (NaClO) [12,13], potassium permanganate [3,14], sodium chlorite ( NaClO 2 ) [6,15,16] and hydrogen peroxide.Numerous experiments using aqueous mixtures of nitric acid and hydrogen peroxide as scrubbing solvents and applied in hollow fiber membrane modules were achieved [5] to test the NO x reduction in gaseous effluents.Furthermore, our previous works [17] also reported that a scrubbing medium using dilute HNO 3 aqueous solutions containing H 2 O 2 as oxidant is efficient for the abatement of NO x gaseous.Experimental tests were performed in a cables-bundle contactor.In 2020, Ghriss et al. [18] present an experimental investigation of the NO x treatment process into aqueous hydrogen peroxide solutions, with or without the presence of nitric acid, applied specifically to the flue gases issued from the nitric acid manufacturing unit of the Tunisian Chemical Group (TCG).Hydrogen peroxide has been unanimously proven to be a very attractive reagent when added in scrubbing liquors containing nitric acid as it is efficient thanks to rapid oxidation reactions involved, and results in the production of valuable nitric acid without generating liquid wastes.Table 1 [3-6, 9-11, 17, 18] lists and compares the advantages and limitations of currently used dry and wet purification methods for NO x removal.
To apply at industrial scale the NO x removal technology by reactive absorption using aqueous solutions containing nitric acid and hydrogen peroxide, it is necessary to have an adequate predictive model.Aspen Plus ® includes tools for property methods estimation, thermodynamic calculations, and the ability to use electrolyte equilibriums and to handle chemical reactions in a wide range of unit operations.A rate-based unit operation model, in which rate-based mass and heat transfer calculations are possible, was selected in Aspen Plus ® .It was used by Loutet et al. (2011) [19] for the modeling of the NO x absorption into water and nitric acid solutions using a packed column.The built model was validated by comparison of predicted values with experimental data from industrial-scale plant in Redcar, United Kingdom.Later Laribi et al. (2019) [20] have also used Aspen Plus ® simulator to model the SO x and NO x absorption process emitted from both power and cement oxyfuel plants operating under pressure and using water as absorbent solution and a packed column as contactor.In their model, they included a chemical reaction mechanism for the SO x and NO x compounds in the gas/ liquid phases to evaluate the optimal conditions of the CO 2 capture process.More recently, Kurillová et al. (2019) [21] studied the modeling of the reactive absorption of nitrogen oxides generated during the production of calcium nitrate fertilizers with water as absorbent, using Aspen Plus ® as a simulator.The simulation results were found to be in good agreement with the experimental data with a relative error of 5%.In their work, they propose the addition of a H 2 O 2 solution into the second packed column of the calcium nitrate fertilizers process.An experimental and modeling study of desulphurization using sodium chlorite seawater solutions was presented by Flagiello et al. (2020) [22].Aspen Plus ® Process Simulation software, once properly integrated with appropriate equilibrium and mass transfer models, was used to provide a good

Wet techniques
Scrubbing process using water -Not expensive.-Easily available solvent.
-Very inefficient -Strongly limited by the low solubility in water of nitric oxide [4,9] Scrubbing process using dilute HNO 3 solutions containing an oxidant ( H 2 O 2 ) -Reagent attractivity when added in scrubbing liquors containing nitric acid -Production of valuable nitric acid without generating liquid wastes, with an economical advantage for processes producing initiallyHNO 3 -High NO x abatement efficiency -High cost of H 2 O 2 [5,17,18] prediction of SO 2 solubility and removal efficiency data.In other works [23], a predictive mass transfer model was specifically developed, implemented in Aspen Plus ® and successfully validated with desulphurization tests at different SO 2 concentrations, temperatures and for different sorbents.In [24] was reported an experimental and modeling study of the thermodynamics and kinetics of SO 2 absorption with H 2 O 2 oxidative solutions.
As far as the reactive absorption of NO x by the H 2 O-HNO 3 -H 2 O 2 system is concerned, it was the subject of a study carried out by [25] using a pilot-scale packed column.A mathematical model based on the two-film absorption theory and considering chemical reactions and equilibrium in gas and liquid phases as well as diffusive transport, has been established, determining overall kinetic parameters for tetravalent NO x ( NO 2 and N 2 O 4 ) and trivalent ( N 2 O 3 + HNO 2 ), varying with the acidity, but not reflecting the precise complex chemistry of the system.According to the bibliography listed above, the modeling of reactive NO x absorption was mainly studied and analyzed for water or dilute nitric acid solutions but without any additional oxidizing agent.Therefore, due to the lack of modeling developments on the reactive NO x absorption into H 2 O-HNO 3 -H 2 O 2 system the objective of this work was to develop a numerical simulation using Aspen Plus ® .The model was developed to predict NO x absorption performances (both efficiency and selectivity) reached in an absorption column for varied operating conditions.Its validation was adequately discussed by comparing the simulated results with the 40 experimental data obtained in a pilot-scale packed column and published in earlier works.

Modeling and Simulation of NO x absorption 2.1 Implementation of the absorption process
In this work, a model for NO x absorption into the chemical system H 2 O-HNO 3 -H 2 O 2 was developed using Aspen Plus ® as process simulator.This software includes tools for property methods estimation, thermodynamic calculations, and the ability to use electrolyte equilibriums and to handle chemical reactions in a wide range of unit operations [26].Process simulations of NO x reactive absorption were performed using a rate-based calculation module because it considers both chemical reactions and mass (and heat) transfer in gas and liquid phases [19].In this calculations module, the rate-based model used the correlation of Billet and Schultes (1993) [27] and the Chilton and Colburn method [28] to solve mass and heat transfers respectively.Fig. 1 presents the flowsheet of the NO x absorption process in packed column designed in Aspen Plus ® .A flue gas stream (GAS-IN) is introduced at the bottom of the absorber where it is contacted counter currently with the scrubbing solution flowing down from the top of the column (LIQ-IN).NO x are selectively absorbed into the scrubbing solution, the rich liquid leaving at the bottom of the absorber (LIQ-OUT) and the purified gas (GAS-OUT) being released.Flow rates and compositions are given for the incoming gas and liquid, while the same variables are computed for the gas and liquid at the outlets.

Description of chemical reactions and corresponding kinetic or equilibrium data
The simulation performances are extremely dependent on the chemical mechanisms selected and, on the equilibrium, and kinetic data used.The mechanism of NO x absorption into aqueous solutions containing H 2 O 2 and HNO 3 , shown in Fig. 2, has been described in several sources such as [29,30] and produce nitrous and nitric acids.Nitrous acid HNO 2 is unstable and can decompose with the release of NO while nitric acid is stable in the liquid phase.It is reported that the presence of H 2 O 2 in the liquid phase improves NO x absorption by oxidation of HNO 2 [31,32] and of NO [33] into HNO 3 according to reactions (R.7) and (R.8), respectively known to be fast and irreversible.It was shown that the presence of nitric acid promotes the oxidation reaction of HNO 2 (R.7) thanks to an auto-catalytic effect [29].Characteristics of the reactions included in the model are given in Table 2 [30,[32][33][34][35][36][37]. . 51 Pa [35] NO NO R.
N O m o N O H O R. HNO HNO 293 670 1

Selection of thermodynamic model for the liquid phase
In addition to reactions R.1 to R.8, three electrolyte dissociation reactions (R.9)-(R.11)giving ionic species are taken into consideration in our absorption system and are shown also in Fig. 2. Equilibrium constants for these reactions are available in Aspen Plus ® Database.The Electrolyte Non-Random Two Liquid (NRTL) property method in Aspen Plus ® was selected for thermodynamic properties estimations as the recommended option for systems containing electrolytes because of ionic interactions in the liquid phase requiring the use of an electrolytic equation of state.Its compatibility with the system NO x -HNO 3 -H 2 O 2 for both the liquid and gas phases was tested successfully by [38].The properties of the liquid phase are assessed from an activity-coefficient model based on the NRTL equations and the properties of the vapor phase from the Redlich-Kwong equation of state [19,39].
3 Application of experimental results at the pilot scale 3.1 Design and operating conditions NO x removal efficiencies were obtained from absorption tests achieved in a pilot packed column where the gas enters axially at the bottom of the column while the fresh scrubbing solution is at the top with a continuous countercurrent contact.The gaseous mixture, made of air and NO x , was analyzed at the inlet and outlet of the absorber by chemiluminescence.This technique gives the total NO x and the chemical NO (noticed NO * ), representing the low oxidation state NO x species given by chemiluminescence (sum of bivalent species and half of trivalent species).
The partial pressures of NO 2 * , the chemical NO 2 (high oxidation state NO x species, representing the sum of tetravalent species and half of trivalent species) and of total NO x can be defined as follows: The oxidation ratio OR in the gas phase is an important characteristic of the gas and is defined by As far as the pilot-scale absorber is concerned the characteristic design parameters are depicted in Table 3.
A screening series of continuous absorption tests was conducted at 293 K with a given gas flow rate but varying liquid flow rates and liquid phase compositions according to values in Table 4.
Each absorption experiment was evaluated by a quantitative performance, the overall NO x absorption efficiency (A NO x ) and by a qualitative one, the absorption selectivity of tetravalent nitrogen oxides (S NO 2 * ) with these definitions: (5) where "in" means inlet and "out" means outlet.Further details on the experimental set-up and procedure are given by [40].

Definition and distribution of NO x species in the gaseous phase
Aspen Plus ® simulator requires the partial pressures of the different NO x species ( NO, NO 2 , N 2 O 3 and N 2 O 4 ) present in the inlet gas to estimate their individual absorption fluxes.These can be calculated from P NO x and OR values (or P NO * and P NO 2 * ) according to the following approach, taking into account that the various nitrogen oxides are and The distribution of the different NO x species can be then determined by an iterative resolution.The obtained results were regrouped in Table 5 by considering the operating P NO x and OR values specific to the 40 experimental data published by [40].Fig. 3 gives the evolution of the distribution of the partial pressures of the various species versus the NO x oxidation ratio at T = 293 K, expressed on logarithmic scale, for values of P NO x = 105.9Pa and P NO x = 398.5 Pa.
According to Fig. 3, whatever the NO x partial pressure, it is found that the N 2 O 3 species is in low proportion in comparison with other species and reaches a maximum partial pressure for an oxidation ratio of about 50%.The partial pressure of N 2 O 4 increases with OR similarly to that of the NO 2 species, whereas the partial pressure of NO decreases when OR is increasing.Nevertheless, it was clearly shown in the literature [40] that due to their very high solubility and reactivity properties the N 2 O 3 and N 2 O 4 species, with low concentrations but rapidly produced (by equilibrium reactions), take part substantially in the overall NO x absorption mechanism.

Model results and validation
To check the validity of the model developed, a comparison of the results predicted by the Aspen rate-based model and the experimental data taken from the literature [38] is performed, by varying the solvents tested, the liquid flow rate, the NO x partial pressure and the oxidation ratio.The input data, operating conditions and absorption results of the 40 experimental pilot-scale tests are summarized in Table 5.
For each absorption test, from the inlet gas composition, whose speciation ( NO, NO 2 , N 2 O 3 , and N 2 O 4 ) is determined as discussed previously, the model allows to calculate successively for the different NO x species the absorbed quantities and the partial pressures in the outlet gas.Therefore, the NO x absorption efficiency (A NO x ,mod ) and the NO 2 * absorption selectively (S NO 2 * ,mod ) can be obtained according to expressions 18 and 19.Model results and experimental results for A NO x and S NO 2 * are presented in Fig. 4 through parity plots.The average deviation between experimental and calculated values of A NO x and S NO 2 * was determined using Eq. ( 9) given by [40].
In fact, since the absorption efficiency and selectivity play similarly important roles in characterizing the absorption performances of chemical NO and NO 2 , which reflects respectively the complementary aspects of quantity and quality of NO x absorbed, the same weightings were given to both parameters in the deviation definition: From the results shown in Fig. 4, it can be observed that the model built predicts the experimental results within an accuracy of 5% for Water + H 2 O 2 and for HNO 3 0.5 M + H 2 O 2 , and of 8% for HNO 3 1 M + H 2 O 2 .The rate-base model here developed is thus quite satisfactorily validated for the given set of experimental results.The effect of nitric acid with concentration varied from 0 to 1 M into aqueous solution of H 2 O 2 (0.02 M) was studied and presented in Fig. 5.As clearly given by Fig. 5, A NO x increases when nitric acid is gradually added to the hydrogen solution.This result is attributed to the fact that the oxidation reaction of HNO 2 is enhanced by the auto-catalytic effect of H + and particularly visible for intermediate oxidation ratios [29].concentrations as H + ions enable to catalyze the oxidation reaction of HNO 2 by H 2 O 2 in the liquid phase.The applicability of the model is then assessed in the operating range of HNO 3 (0-1 M) and H 2 O 2 (0.2 M) concentrations and of NO x partial pressures (~50 Pa-500 Pa) investigated to design in the future a new full-scale oxidative scrubbing system for deNO x , applied, as an interesting case study, to tail gases issued from a Tunisian nitric acid plant.
. The gas phase reactions are the kinetic oxidation of nitric oxide NO to nitrogen dioxide NO 2 by oxygen O 2 (R.1) and the equilibrium reactions producing dinitrogentetraoxideN 2 O 4 (R.2) and dinitrogen trioxide N 2 O 3 (R.3).The liquid phase reactions are considered as irreversible and kinetically controlled reactions.Via reactions (R.4) (R.5) and (R.6), N 2 O 4 , NO 2 and N 2 O 3 compounds react rapidly with water (hydrolyses reactions)

Fig. 1
Fig. 1 Aspen Plus ® flow-sheet for NO x reactive absorption column

17 .Fig. 2
Fig. 2 Absorption mechanisms of NO x into aqueous solutions containing H 2 O 2 and HNO 3 used in the model (adapted from [30])

Table 3
Main characteristics of the pilot-scale absorber

Table 4
Operating conditions for the pilot-scale column equilibrium in the gas phase according to reactions R.2 and R.3 whose constant values are given in Table2: 2 : 0.2 M Abbreviations: G: Volumetric gas flow rate (m 3 /s), L: Volumetric liquid flow rate (m 3 /s) in

Table 5
Operating conditions, gas phase compositions and experimental absorption performances