Scholarlink

<p>The continued occurrence of groundwater fluorosis poses a severe threat to the international community; therefore, high-performance, economically feasible remediation technologies are urgently needed. The study assesses the defluoridation capability of four common Indian agro-residues, namely Sugarcane Husk (SH), Garbanzo Husk (GH), Black Mustard Husk (BMH), and Cotton Husk (CH), via a coordinated experimental and computational approach. Extensive batch adsorption experiments were carried out in order to optimise five critical operation parameters, namely: the contact time (30–210&#xa0;min), the agitation rate (50–350&#xa0;rpm), the weight of the adsorbent material (0.25–2.5&#xa0;g), the solution pH (2–12), and the initial fluoride load (3–11&#xa0;mg/L). Empirical evidence revealed a distinct efficiency gradient: BMH &gt; GH &gt; SH &gt; CH, with Black Mustard Husk achieving a peak sequestration efficiency of 87.0%. The adsorption behaviour observed under acidic conditions suggests that electrostatic interactions may contribute significantly to fluoride uptake, although direct surface charge characterisation was not performed. To overcome the nonlinearity of the adsorption matrix, an advanced Multi-Output Artificial Neural Network (MANN) was developed. It demonstrated impressive predictive performance (R &gt; 0.98; MSE &lt; 0.04), successfully mapping the thermodynamic relationship between equilibrium concentration (Ce) and solid-phase capacity (Q) using the MANN. SEM, BET, and XRD analyses revealed that the agro-waste adsorbents possessed heterogeneous porous morphologies, mesoporous structures, and characteristic lignocellulosic crystalline phases, which are favourable for fluoride adsorption, among the investigated materials. Characterisation results indicated that the adsorbents exhibited favourable morphological and textural properties, including porous surface structures and mesoporous characteristics, which facilitated fluoride uptake. Black Mustard Husk demonstrated the most advantageous microstructural features and the highest adsorption performance. The global sensitivity analysis established that the concentration gradient and adsorbent dosage are the major kinetic drivers. This research combines the ideas of the Circular Economy and machine learning to provide a sound, AI-based framework for decentralised water purification that would transform agro-waste into a valuable strategic resource.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Experimental and multi-output artificial neural network modelling to compare agro-waste adsorbents in fluoride removal

  • 期刊论文

  • 发表日期：

    2026年7月27日

查看全文
  • Manoj Wagh,
  • Swapnil Kurhade,
  • Anirudh Shirke,
  • Navnath Raut,
  • Dnyaneshwar Wadkar

摘要

The continued occurrence of groundwater fluorosis poses a severe threat to the international community; therefore, high-performance, economically feasible remediation technologies are urgently needed. The study assesses the defluoridation capability of four common Indian agro-residues, namely Sugarcane Husk (SH), Garbanzo Husk (GH), Black Mustard Husk (BMH), and Cotton Husk (CH), via a coordinated experimental and computational approach. Extensive batch adsorption experiments were carried out in order to optimise five critical operation parameters, namely: the contact time (30–210 min), the agitation rate (50–350 rpm), the weight of the adsorbent material (0.25–2.5 g), the solution pH (2–12), and the initial fluoride load (3–11 mg/L). Empirical evidence revealed a distinct efficiency gradient: BMH > GH > SH > CH, with Black Mustard Husk achieving a peak sequestration efficiency of 87.0%. The adsorption behaviour observed under acidic conditions suggests that electrostatic interactions may contribute significantly to fluoride uptake, although direct surface charge characterisation was not performed. To overcome the nonlinearity of the adsorption matrix, an advanced Multi-Output Artificial Neural Network (MANN) was developed. It demonstrated impressive predictive performance (R > 0.98; MSE < 0.04), successfully mapping the thermodynamic relationship between equilibrium concentration (Ce) and solid-phase capacity (Q) using the MANN. SEM, BET, and XRD analyses revealed that the agro-waste adsorbents possessed heterogeneous porous morphologies, mesoporous structures, and characteristic lignocellulosic crystalline phases, which are favourable for fluoride adsorption, among the investigated materials. Characterisation results indicated that the adsorbents exhibited favourable morphological and textural properties, including porous surface structures and mesoporous characteristics, which facilitated fluoride uptake. Black Mustard Husk demonstrated the most advantageous microstructural features and the highest adsorption performance. The global sensitivity analysis established that the concentration gradient and adsorbent dosage are the major kinetic drivers. This research combines the ideas of the Circular Economy and machine learning to provide a sound, AI-based framework for decentralised water purification that would transform agro-waste into a valuable strategic resource.