跳到主要內容 :::

年度	113	專案性質	實驗性質
專案類別	模場試驗	研究主題	整治
申請機構	國立中央大學	申請系所	應用地質研究所
專案主持人	倪春發	職等/職稱	教授
專案中文名稱	整治場址數位監測與高精度汙染團層析技術於抽出處理及淋洗策略之應用
中文關鍵字	數位監測, 高精度層析技術, 抽出處理, 淋洗策略
專案英文名稱	Application of Digital Monitoring and High-Precision Contaminant Layer Profiling Technology for Pumping Treatment and Leaching Strategies in Remediation Sites
英文關鍵字	Digital Monitoring, High-Precision Contaminant Layer Profiling, Pumping Treatment Technology, Leaching Strategies
執行金額	2,270,000元
執行期間	2024/12/1 至 2025/11/30
計畫中文摘要	本計畫以佶鼎科技股份有限公司廠址作為模場試驗場域，驗證與優化土壤及地下水污染整治技術。調查結果顯示，地下水中銅（Cu）與鎳（Ni）污染分布面積分別為 507 及 811.2 平方公尺，另有多口監測井檢出鉛（Pb）濃度超標，顯示場址具多重重金屬污染特性。主要採用抽出處理法（Pump and Treat）控制污染，雖整體濃度呈下降趨勢，仍有局部監測井超標。為提升整治效能，輔以現地土壤淋洗法（In-situ Soil Flushing），促進土壤中金屬脫附並抽取處理，並已完成模場試驗。場址含水層由多層砂質與黏土交互組成，具高度異質性，7～8 公尺處之黏土層形成滯留屏障，使污染物分佈不均，抽出與灌注效率受限。此外，地下水流向受降雨與鄰近抽水行為影響，流場變異大，增加整治操作與成效評估之挑戰。因此，本模場計畫提出解決此挑戰之技術需求，包括：（1）即時地下水位與基礎水質觀測以強化整治策略規劃；（2）評估含水層水力傳導係數空間分布以增進抽注作業效率；（3）運用光纖高解析監測技術評估垂直向導水特性及水流通量。為強化地質分層與材料分布之空間解析，提高藥劑注入與抽出過程效率，本計畫採用水力層析成像技術（Hydraulic Tomography）推估導水係數二維空間分布場。相較於傳統抽水試驗僅進行單點分析並以數值內插模式進行特徵空間化分析，此方法可透過多次迭代與更新，根據現場量測（如水位、流速）調整模型參數，並以最小化觀測與預測差異為準則進行校正，更適用於大型、複雜含水層系統之多井參數估計。於整治過程中，污染團之清除狀況與殘餘熱區將作為抽出處理與淋洗作業修正依據。為解析地下水通量垂直分布，本計畫將進行井下熱示蹤試驗（Thermal Tracer Test），於井中主動施加微量熱源，並以高解析光纖感測技術監測溫度變化，藉由溫度偏移特徵推估地下水通量。第二年期將根據第一年所建立之觀測資料與試驗成果，導入機器學習模型進行大數據分析與污染行為預測，透過歷史水質與水位資料預測未來變化趨勢，協助識別污染源、高污染區及其影響範圍，並藉由模擬不同整治策略提供最佳化方案，以降低整治成本並提升處理效率。
計畫英文摘要	This project uses the Gi Ding Technology Co., Ltd. site as a pilot field to verify and optimize soil and groundwater contamination remediation technologies. Investigation results indicate that groundwater contamination by copper (Cu) and nickel (Ni) covers areas of 507 m² and 811.2 m², respectively. Several monitoring wells also showed lead (Pb) concentrations exceeding regulatory limits, suggesting multiple heavy-metal contamination at the site. The main remediation method is the Pump-and-Treat system; although overall concentrations show a declining trend, some wells remain above standards. To enhance remediation efficiency, in-situ soil flushing has been adopted to promote metal desorption and extraction from soil, and a pilot-scale test has been completed.The aquifer consists of alternating sand and clay layers with high heterogeneity; a clay layer at depths of 7–8 m forms a retention barrier, causing uneven contaminant distribution and limiting extraction and injection efficiency. Moreover, groundwater flow is influenced by rainfall and nearby pumping activities, resulting in significant flow variability and challenges for process optimization and performance evaluation. To address these issues, this pilot project proposes the following technical requirements: (1) real-time groundwater-level and basic water-quality monitoring to strengthen remediation planning; (2) spatial assessment of hydraulic conductivity to improve pumping and injection efficiency; and (3) application of high-resolution fiber-optic monitoring to evaluate vertical flow characteristics and groundwater flux.To refine geological layering and material distribution for more efficient chemical injection and extraction, hydraulic tomography will be applied to estimate two-dimensional spatial distributions of hydraulic conductivity. Compared with conventional single-point pumping tests and interpolation analysis, this method iteratively updates model parameters based on field data (e.g., hydraulic head and flow velocity) and minimizes discrepancies between measured and predicted values, making it suitable for large, complex multi-well systems.During remediation, the removal progress of contaminant plumes and residual hot spots will guide the adjustment of Pump-and-Treat and soil-flushing operations. To analyze vertical groundwater flux, the project will conduct in-well thermal tracer tests, introducing a controlled heat source and monitoring temperature changes with high-resolution fiber-optic sensors to infer groundwater flux from thermal responses. In the second project year, based on the first year’s monitoring and experimental data, machine-learning models will be employed for big-data analysis and contaminant-behavior prediction. By analyzing historical water-quality and water-level data, the models will forecast future trends, identify contamination sources and high-concentration zones, and simulate alternative remediation strategies to develop optimized solutions that reduce costs and enhance treatment efficiency.