Adsorptive removal of Cd2+ ions using dolochar at an industrial-scale process optimization by response surface methodology

In this work, performance of laboratory-synthesized dolochar has been investigated for adsorption of Cd2+ ions in a large-scale process with the application of Aspen Adsorption. Moreover, the optimum values of the operating parameters (namely, flow rate, bed height, and inlet metal ion concentration) that would result into maximum amount of cadmium ion adsorption (high exhaustion capacity) in minimum time (less exhaustion time) for a fixed mass of dolochar have been calculated via the application of response surface methodology. It was found that, at optimum values of bed height (3.48 m), flow rate (76.31 m3/day), and inlet concentration (10 ppm), the optimized value of exhaustion capacity and exhaustion time for cadmium ion adsorption in dolochar packed bed is equal to 1.85 mg/g and 11.39 h, respectively. The validity of these simulation experiments can be proven by the fact that the obtained exhaustion capacity of dolochar packed bed always remained in close proximity of the experimentally obtained value of adsorption capacity of the dolochar in batch process mode (equal to 2.1 mg/g).


Introduction
Heavy metal ions are one of the major pollutants that possess serious threat to the environment. Release of effluents contaminated with high concentrations of metal ions via the industries can have very toxic effect on nature as these released metal ions can enter the food chain with the help of plants and fish. Metals ions dissolved in water get accumulated at the roots of plants and later get translocated to the edible parts of the plant (like fruits) and hence eventually enter the food chain (Ahmad et al. 2020). These heavy metal ions can be classified into three categories: toxic metals (Hg, Cr, Sn, Co, As, Cd, Ni, Cu, Zn, Pb), radioactive metals (Am, Ra, Th, U), and valuable and worthy metals (Pd, Ru, Au, Ag, Pt) (Shafiee et al. 2019). Table 1 summarizes the toxic effects and permissible discharge levels of toxic metal ions (Agarwal et al. 2020;Foroutan et al. 2021e;Peighambardoust et al. 2021). As evident from Table 1, heavy metal ions, if present in high concentration inside the human body, can lead to very serious health conditions which might eventually result in death. Given the high concentrations in which these heavy metal ions are present in the industrial effluents, humans will be more prone to the hazardous effects of toxic metal ions, if these industries discharge their effluents into the environment without prior treatment. For example, concentrations of cadmium ions found in the few effluent samples from electroplating (Bankole et al. 2019), mining (Fu et al. 2020), smelting (Xu et al. 2018), and automobile industries (Akpomie and Dawodu 2016) have been reported to be equal to 3.02 ppm, 21.3 ppm, 60 ppm, and 6.09 ppm, respectively, which is significantly higher than the WHO permissible limits, and hence, it is extremely important to treat the effluent before discharging it to the environment.
To achieve the separation of metal ions from the industrial effluents, several techniques like ion exchange, precipitation, solvent extraction, and reverse osmosis are available. However, most of these techniques are tedious, expensive, and known to cause secondary pollution. On the other hand, adsorption has been one of the most popular techniques for removal of metal ions due to its low cost, effectiveness, and simplicity (Liao and Huang 2019;Foroutan et al. 2021c). Easy regeneration of the adsorbent and recovery of the adsorbate further make adsorption one of the most attractive options. Most of the industries, today, use either activated carbon, zeolites, or polymeric adsorbents in their adsorption process setup. In spite of achieving high metal removal efficiencies with these adsorbents, the industries are aiming to shift towards another class of adsorbents called biosorbents (Wang and Chen 2014). Biosorbents are those low-cost adsorbents which are usually produced from commercially worthless products, i.e., from industrial waste (such as dolochar, slag, and fly ash) or forestry/agricultural waste (such as neem bark, banana peels, and coconut husk) (Ogata et al. 2020). Many different types of biosorbents have shown high efficiency on a lab scale, and dolochar is one of them. Dolochar, a commercially worthless product, can be obtained from the waste of sponge iron industries and has been tested multiple times for adsorption of heavy metal ions in a laboratory setup where it has yielded satisfactory results (Panda et al. 2011(Panda et al. , 2017. Dolochar is a carbonaceous material with extremely porous structure as desired for an ideal adsorbent. In addition, dolochar also possess several electron-donating groups like hydroxyl and amine groups which substantially contribute towards the adsorption of heavy metal ions. Hence, it is expected that, with dolochar, it is possible to achieve high removal percentages for cadmium, which is one of the toxic heavy metal ions. Therefore, this work focuses on industrialization of dolochar as an adsorbent by predicting its behavior, if it is applied for cadmium removal on a large industrial scale. While potential of many biosorbents have been explored in laboratory setup, very few studies have tried to scale up the biosorption process and predict the performance of these biosorbents in a large-scale industrial setup. Since ultimate goal is commercialization of these biosorbents, in this work, we design a dynamic large-scale adsorption process whose parameters (namely flow rate, bed height, and inlet metal ion concentration) have been optimized with the application of response surface methodology (RSM) in order to get maximum value of bed exhaustion capacity in minimum time (and hence ensuring the maximum removal of metal ions in minimum time). This work can help in predicting and optimizing the expected behavior of any adsorbent, which is synthesized in lab, for a real industrial setup.

Materials and methods
The system of cadmium ion-contaminated stream flowing through the dolochar packed fixed bed adsorption column was simulated on the software Aspen Adsorption V11.0. Physical properties that are crucial for execution of these simulation-based experiments were taken from the literature. Bulk density, particle density, and particle size of dolochar, as reported in literature, are equal to 1203 kg/m 3 , 2250 kg/ m 3 , and 63 µ, respectively (Panda et al. 2011(Panda et al. , 2014. For the application of response surface methodology, software Design Expert V11.0 was used.

Aspen Adsorption
Under Aspen Adsorption, liquid adsorption model was used to conduct the experiments. The general assumptions made in this model were (Zhang et al. 2019): • The behavior of fluid across the column was assumed to be plug flow with axial dispersion. The value of axial dispersion was estimated with Slater correlation (Slater 1991) as represented in Eq. (1), where E z denotes axial dispersion coefficient, v l denotes velocity of the liquid flowing through the bed, ε i is the bed porosity, r p is the radius of dolochar particles, and Re is the Reynold's number. The liquid phase pressure and superficial velocity are assumed to have a constant value throughout the bed. These assumptions are valid since the quantity of metal being adsorbed is negligible in comparison to volume of liquid flowing through the bed, and hence, metal ion adsorption will have negligible effect on mass balance equation.
• Molar concentrations are calculated from molar volumes. Ideal mixing is assumed to occur in the liquid phase, so molar volume is a linear function of composition. • Isothermal conditions are applicable. Figure 1 shows the adsorption process as designed in Aspen Adsorption.
The discretization method used in all the simulations was upwind differencing scheme 1 (UDS 1) which is based upon the first-order Taylor series expansion. Equations (2) and (3) show the general differencing scheme made under the UDS 1 method.
Δz 2 UDS 1 was chosen in this work because of its unconditionally non-oscillatory, unconditionally stable nature, minimum simulation time, and good accuracy. The accuracy can be further increased with increase in number of nodes. In this work, number of nodes was selected to be equal to 30.

Mass balance
Equation (4) presents the partial differential mass balance equation which was used to express the metal ion concentration in a small control volume inside the adsorbent bed, where ε i is the bed porosity, E z is the axial dispersion coefficient (m 2 /s), z is the distance along the bed (m), q is the concentration of adsorbed cadmium ions onto dolochar (mg/g), C is the concentration of cadmium ions in the liquid phase, ρ a is the bulk density, and v i is the interstitial velocity of the fluid through the adsorbent bed (Ahmed et al. 2020).
The first term of Eq. (4) indicate dispersion forces, whereas the second term denote the convective forces that are responsible for the biosorption process. The third term is to account for accumulation of the cadmium ions onto dolochar, and finally, the fourth term is the mass transfer term to account for the transfer of cadmium ions from liquid to a solid phase.
The bed porosity, in Eq. (4), was calculated from Eq. (5), where ρ b is the bulk density and ρ p is the particle density.

Kinetic model
In this work, linear lumped resistance kinetic model was assumed. This model assumes that driving force for mass transfer of the components is a linear function of the component concentration in liquid phase or solid phase and can be mathematically represented by Eq. (6) (Ahmed et al. 2020).
where w k is the instantaneous equilibrium loading of adsorbate onto adsorbent (mg/g) and MTC is the overall mass transfer coefficient (m/s). MTC accounts for mass transfer resistance between the liquid and external adsorbent surface, and the mass transfer resistance due to pore structure of adsorbent surface. The value of MTC can be calculated from Katoka correlation (Eq. (7)) (Zulfadhly et al. 2001).
Along with linear lumped resistance model, solid film assumption was considered in this work since it expresses the mass transfer driving force in terms of the solid-phase concentrations of the components (here, cadmium ions).
The adsorption of cadmium ions onto dolochar process can be best described by Langmuir isotherm as it gives a much higher R 2 value (equal to 0.99) in comparison to Freundlich and Dubinin-Radushkevitch isotherm models (Panda et al. 2011). Therefore, this work assumes that adsorption process is being governed by Langmuir isotherm (Eq. (8) (9) lnq e = 1 n lnC e + lnk f (10) lnq e = lnq s − 2 which makes three basic assumptions-(i) adsorption on one site is independent of its other neighboring sites, (ii) adsorption sites are homogenous, (iii) monolayer adsorption takes place (Upadhyay et al. 2020). Based upon the convention of Aspen Adsorption, Eq. (11) and Eq. (12) can be used to calculate the isotherm parameters, where q max is the adsorbate concentration on the adsorbent surface (mg/g) and k L is the Langmuir constant (L/mg), and IP 1 and IP 2 are input parameters to Aspen Adsorption. Values of Langmuir isotherm parameters, namely, q max and k L , are taken to be equal to 2.02 mg/g and 0.33 L/mg, respectively. Also, another important parameter, equilibrium factor (R L ), was found equal to 0.13, which indicates the favorability of the adsorption process as the values lie between 0 and 1 (Panda et al. 2011).

Evaluation of bed
One of the most widely used parameters for the evaluation of an adsorbent bed is its exhaustion capacity (Kavand et al. 2018). In order to calculate exhaustion capacity, Eq. (13) can be used, where, q tot e is the exhaustion capacity of the bed (mg/g), m is the total mass of adsorbent packed in the bed, C t is the metal ion concentration in the exiting stream, C 0 is the inlet metal ion concentration, and Q is the flow rate of effluent through the bed.
The evaluation of integral term in Eq. (13) has been done with the help of Origin Pro software in this work. The data points obtained from simulation experiments were plotted on Origin Pro followed by required area calculation in the graph using the Integrator gadget of the Origin Pro software.

Response surface methodology
One of the aims of this work is to find the optimum set of exhaustion capacity (maximum) and exhaustion time (minimum) from a set of varying values of inlet metal ion concentration, bed height, and flow rate. The working range of these parameters were taken to be between 1 to 5 m for bed height, 10 to 100 ppm for cadmium ion inlet concentration, and 50 to 100 m 3 /day for flow rate through the bed. To achieve the desired task, response surface methodology has been applied which is a method based on the (11) fitting of mathematical models (linear, square polynomial functions, and others) to the experimental results generated from the designed experiment and the verification of the obtained model by means of statistical techniques. The following steps were followed in order to reach at the set of optimum parameters (Witek-Krowiak et al. 2014): 1. Deciding upon the independent variables, working range of the independent variables, and responses to be considered in the study. 2. Designing of experiments. 3. Carrying out the experiments designed in step (2) in order to get results, i.e., getting responses. 4. Fitting a model to the obtained set of variable-response data. 5. Validation of the model with analysis of variance (ANOVA). 6. Determination of the optimum parameters set.
For designing of experiments, various experimental strategies are available such as factorial design (Cestari et al. 2008), central composite design (Oden and Sari-Erkan 2018), Box-Behnken approach (Jaafari and Yaghmaeian 2019), and Plackett-Burman design. In this work, the experimental route has been mapped with help of central composite design (CCD) strategy due to its ability to analyze the complex interaction between the parameters with low number of experimental run requirements (Lingamdinne et al. 2018;Oden and Sari-Erkan 2018). Design Expert software has been used for applying RSM in this work.

Adsorbent dolochar and simulation experiments
Dolochar, along with being porous, is also rich in electrondonating groups like -OH, -NH, -SiO, and -C≡N. These electron-donating groups interact with cadmium ions via complexation mechanism which results in high performance of dolochar as shown by Panda et al. (2011). Besides being rich in metal-binding groups, dolochar is extremely cheap as well. Its cost is around only Rs.1000 per ton, and hence, it can significantly lower the cost of the effluent treatment plants that use expensive conventional adsorbents. Another added benefit of using dolochar for industrial effluents lies in its way of interaction with other negatively charged species like nitrate and phosphate. Dolochar contains iron, calcium, aluminum, and magnesium compounds which interact only with these negatively charged species (and not metal ions) via complexation and precipitation mechanisms and remove them from the effluent (Rout et al. 2015). Hence, with dolochar, it is possible to achieve simultaneous removal of metal ions (via complexation with electron-donating groups) and negatively charged species like nitrate and phosphate (via complexation with iron, calcium, aluminum, and magnesium compounds). To summarize, based upon the metalbinding capability, extremely low cost, and ability to adsorb both positively charged and negatively charged ions simultaneously, it can be said that dolochar has great potential as an adsorbent and hence can be considered for usage in treatment of real industrial effluents.
All the simulation experiments, conducted on Aspen Adsorption, were conducted for a fixed mass of 1200 kg of dolochar (and hence 3.94 m 3 volume of the packed bed reactor). In one of the conducted experiments, bed height, flow rate, and inlet cadmium ion concentration were set to be equal to 3 m, 75 m 3 /day, and 55 ppm, respectively. Since the mass of adsorbent is fixed to 1200 kg and the bulk density of the adsorbent is taken to be 1203 kg/m 3 , the diameter of the bed was calculated to be 0.65 m. The overall mass transfer coefficient, as calculated from Eq. (7), was equal to 0.000137 for this case.
Post running the simulation for this system, it was found that the bed reached its breakthrough point (C/C 0 = 0.05) at 11.11 h and eventually exhausted after 19 h (C/C 0 = 0.95). The concentration profile (C/C 0 ) was plotted against time, and the resulting graph obtained has been shown in Fig. 2. From the application of Eq. (13), the exhaustion capacity of the bed was calculated to be 1.71 mg/g.

Response surface methodology
For an industrial setup, it is preferable that maximum amount of effluent is treated (hence, high exhaustion Fig. 2 Concentration profile of the stream exiting from the dolochar packed bed when inlet metal ion concentration, flow rate, and bed height were set equal to 75 m 3 /day, 55 ppm, and 3 m, respectively capacity is required) in minimum time (hence, low exhaustion time is preferred). This ensures that the industry will have to bear the minimum cost for the metal ion separation process since the run time of the equipment and adsorbent quantity requirement are being minimized. Therefore, RSM is being applied here to optimize the values of exhaustion capacity and exhaustion time as the independent variables flow rate, bed height, and inlet metal ion concentration vary (Zinatizadeh et al. 2006).

Central composite design
The experiments were designed with the use of CCD strategy (Jiryaei Sharahi and Shahbazi 2017;Foroutan et al. 2020b). The independent variables and their input range are presented in Table 2.
CCD approach designed a total of 17 experiments out of which, at central point, 3 were replicates so as to measure the accuracy level of results (Zhang et al. 2016).
Simulation experiments were conducted for all the designed 17 experiments in Aspen Adsorption, and the response variables, namely exhaustion time and exhaustion capacity, were obtained. Table 3 shows the CCD matrix and results obtained.

Analysis of variance and examination of the relationship between factors and responses
The regression analysis of experimental data obtained in Table 3 led to the conclusion that exhaustion time of the bed for this system can be expressed as a function of bed height, flow rate, and inlet metal ion concentration via Eq. (14). The same, for the case of exhaustion capacity of the bed, can be modeled by Eq. (15). Both the models obtained were analyzed and were found to be significant with the P-value less than 0.0001 in both cases. Moreover, for both the models, the lack of fit is not significant (relative to the pure error) which further proves the suitability of these models (P-value equal to 0.19 for exhaustion time model and equal to 0.57 for exhaustion capacity model) (Foroutan et al. 2021a). The R 2 , predicted R 2 , and adjusted R 2 values were found to be very close to 1 in case of both the models which indicates that the suggested models can be good predictors of the experiment results (refer to Table 4 and Fig. 3). Moreover, the signalto-noise ratio (adeq. ratio) was also much greater than the minimum required value (i.e., 4) for both the models which where R1 is exhaustion time (hours), R2 is exhaustion capacity (mg/g), A is bed height (m), B is flow rate (m 3 /day), and C is inlet concentration of cadmium ions (kmol/m 3 ). Table 5 and Table 6 show the analysis of variance (Foroutan et al. 2021d) for exhaustion time and exhaustion capacity models. While it was observed that, for exhaustion time, all three factors, i.e., height, flow rate, and inlet concentration, were significant factors (with P-value equal to 0.0004, < 0.0001, and < 0.0001, respectively), for exhaustion capacity, only inlet concentration (P-value < 0.0001) and flow rate (P-value = 0.002) were significant factors and bed height was less significant (P-value = 0.1). The reason behind this less significant behavior of height factor could be the fact that the same mass of dolochar biosorbent is being used in all experiments (1200 kg). Therefore, with increase in bed height, diameter decreases and vice versa which leads to minimal change in exhaustion capacity of the bed in response to the change in bed height. With respect to flow rate, it was found that increasing the flow rate results in the decrease in both exhaustion time and exhaustion capacity. The decrease in exhaustion capacity with the increase in flow rate is the result of decrease in residence time of the cadmium ions inside the bed which ceases the opportunity for the cadmium ions to reach the micro-and mesopores of the dolochar particles (Simate and Ndlovu 2015). However, the decrease in exhaustion time is a favorable factor which results from the increase in the volume of contaminated stream being treated per unit time (Hernández-Hernández et al. 2017). With respect to inlet metal ion concentration, it was found that increasing the inlet metal ion concentration results in increase in exhaustion capacity and decrease in exhaustion time.
Since the lumped linear resistance kinetic model has been assumed in this work (which assumes that rate of metal uptake is directly proportional to its concentration gradient), it was expected to observe that increase in metal ion concentration in the incoming stream would result in increase in exhaustion capacity. Therefore, decrease in mass transfer resistance and increase in concentration gradient with increase in inlet metal ion concentration can be attributed for the observed increase in exhaustion capacity (Sarin et al. 2006). Moreover, with the increase in the metal ion concentration, the exhaustion time decreases. This is the result of adsorption sites being exhausted early since more number of ions are entering into the system per unit time in addition to more number of ions being adsorbed due to lower mass transfer resistance (Popovic et al. 2020). Figure 4 and Fig. 5 help to visually represent the effect of varying factors onto the response variables via help of 2D and 3D contours (Foroutan et al. 2021b).

Optimization of variables
Based upon the models, the optimum values that would maximize the exhaustion capacity of bed and minimize the exhaustion time (hence achieve maximum cadmium ion adsorption in minimum time) were calculated with the help of a multiple response method called desirability function (Zaferani et al. 2019). The optimum values of bed height, flow rate, and inlet metal ion concentration were found to be equal to 3.48 m, 76.31 m 3 /day, and 10 ppm, respectively. For these set of values, the predicted exhaustion time and exhaustion capacity from the model were equal to 11.55 h and 1.87 mg/g, respectively, and the desirability level was 0.835. When experiments were conducted with the same optimum values as input, the values of exhaustion time and exhaustion capacity were found to be equal to 11.39 h and 1.85 mg/g, respectively, which is in a very close approximation with the predicted values from the model and hence further justifies the preciseness of the models.

Validation of simulation with the experimental results
As can be inferred from the Table 3, the exhaustion capacity of the dolochar packed bed lies in the range of 1.3 to 1.9 mg/g. In this work, the values of dolochar bulk density, particle density, particle size, and isotherm

Conclusions
This work was aimed at providing motivation for the industrialization of biosorbents for the application of heavy metal ion adsorption. Through the application of Aspen Adsorption, this work showed how the performance of a biosorbent in a real industrial process can be predicted. With the observed proximity between the exhaustion capacity of dolochar obtained from Aspen Adsorption and the maximum adsorption capacity obtained from laboratory experiments, it was confirmed that Aspen Adsorption can be used as a good predictor of an adsorbent performance. Furthermore, through the application of response surface methodology, the effect of variation in column bed height, flow rate, and inlet metal concentration onto the exhaustion time and exhaustion capacity of the dolochar packed bed was examined. Moreover, it was shown that how the optimum working parameters can be calculated for the dynamic adsorption process that would maximize the performance of the corresponding biosorbent in the industrial process being considered. Finally, it was found out that, for a fixed mass of dolochar biosorbent and at optimum values of bed height (3.48 m), f low rate (76.31 m 3 /day), and inlet concentration (10 ppm), the optimized value of exhaustion capacity and exhaustion time for cadmium ion adsorption in a dolochar packed bed is equal to 1.85 mg/g and 11.39 h, respectively. It is high time to truly utilize the potential of biosorbents in the treatment of industrial eff luents, and hence, it is recommended that future works focus more on industrialization of biosorbents. Studies focused upon scaling up the synthesis process of successful biosorbents, designing of optimized continuous adsorption processes including regeneration cycles (and the management of waste stream generated from regeneration of adsorbents), and designing of metal recovery systems for the spent adsorbents are highly desirable, and researchers are encouraged to work in this direction.