地下水模式工具於污染控制場址範疇界定之研析

Analysis of Groundwater Models for Defining the Contaminant Category of the Control Site

黃佳雯(CHIA-WEN HUANG)

指導教授 ： 駱尙廉

共同指導教授 ： 林正芳

國立臺灣大學/工學院/環境工程學研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.01031

摘要

根據國內之需求，參考國內專家學者之意見，研選出適合的三套地下水模式MODFLOW+MT3DMS、AT123D、FEMWATER，作為後續現場模擬之工具。地下水模式可以配合場址的水文地質資料及污染物檢測資料進行模擬，藉著模式參數的輸入與簡化實際情況，更了解模式本身的優點與限制。就本研究所建議之三套模式，探討參數敏感度與假定之參數是否合理，來評估模式之適用性。並以地下水污染管制標準濃度為基礎，劃定控制場址污染之範圍，以作為該地區後續整治、復育參考之用。
以適用情況來看，此三套模式各有其特色。MODFLOW僅可模擬水流狀態，污染物傳輸狀況需搭配其他模式。MT3DMS只限於模擬污染物傳輸，以污染物平流-延散的傳輸機制為主，對於油品污染具有較高的準確性。AT123D可在穩態水平流與簡單水文地質條件下執行，著重於污染物的傳輸，可粗估污染範圍。FEMWATER可模擬不同的密度流，對於不規則邊界有較高的準確度，但缺乏較複雜的化學與生物動力，且操作較為繁複。
在本研究中，控制場址為一加油站，研究區域為1400 m 900 m，「苯」為主要污染物質，主要之參數設定：水力傳導係數2.55 m/d，有效孔隙率0.3，總體密度1700 kg/m3，縱向延散係數100 m，污染質量負荷率0.088 kg/d(污染源濃度8800 mg/l)。三套模式模擬結果：於模擬時間四年時最接近實場，污染源位置主要在加油站內，以線或面的方式洩漏，污染源較嚴重的地方為加油站站區中間偏北。
MODFLOW+MT3DMS污染物傳輸南北向擴散；AT123D污染物向東擴散；FEMWATER則向北擴散。造成此三套模式傳輸方向差異的主要原因為MODFLOW+MT3DMS可設定異質水力傳導係數；FEMWATER在同一含水層下只能設定均質之水力傳導係數；AT123D則受限於穩態水平流的基本假設。以苯第二類地下水管制標準0.05 mg/l為界限，劃定其污染範圍，則FEMWATER所模擬的污染範圍最小，其次為MODFLOW+MT3DMS、AT123D。
三套模式在本研究場址條件下之敏感度分析顯示：總體密度對輸出結果沒有貢獻；水力傳導係數是MODFLOW最重要的敏感參數；有效孔隙率對MODFLOW+MT3DMS的輸出結果影響很大；水力傳導係數與縱向延散係數對FEMWATER的輸出結果影響很大；有效孔隙率亦對AT123D的輸出結果影響很大。三套模式之輸出結果皆隨污染質量負荷率(污染源濃度)變動成等比例、正比變動。
將模擬值比對觀測值，以相對誤差與累積誤差的方式檢驗其模擬結果，發現MODFLOW+MT3DMS的模擬結果最接近實場。從MODFLOW+MT3DMS、AT123D、FEMWATER之模式適用情況、限制條件、污染範圍、使用性來看，MODFLOW+MT3DMS在本研究中最適用。

關鍵字

地下水模式 ； MODFLOW ； MT3D ； AT123D ； FEMWATER ； 控制場址污染範圍

並列摘要

According to the domestic demand and the suggestions by local experts, three sets of groundwater models, MODFLOW+MT3DMS, AT123D and FEMWATER, were selected and used to simulate pollution conditions. The groundwater models should be able to comprehend hydrogeological characteristics and pollutant data for parameter inputs. During the process of simplifying the actual complex conditions reasonably, the advantages and restrictions of models can be assessed. The sensitivity analysis of parameters was used to assess the suitability and accuracy of these models. Considering the control standards for groundwater as a baseline, the research delimited the contaminated area of the control site to help the following remediation work on the site.
For comparison, MODFLOW can be only used to simulate the groundwater flow, and other models should be used to simulate the pollutant transport. MT3DMS can only be applied to simulate the transport process, which the transport mechanism is advection-dispersion and have higher accuracy on the simulation for oil pollution. AT123D assumes that the groundwater is stable and horizontal, so that it can be used to simulate the transport of the pollutants in a simple hydrogeological condition, inducing a rough estimation for the polluted area. FEMWATER can simulate various density-drawn flow conditions and has higher accuracy on the irregular boundary, but most of advanced chemical and biological dynamics is not available, and it is relatively complicated to operate.
In this research, the area of the control site is 1400 m 900 m and the main pollutant is “benzene”. At this control site, hydraulic conductivity is 2.55 m/d, effective porosity 0.3, bulk density 1700 kg/m3, longitudinal dispersivity 100 m, and mass-loading rate 0.088 kg/d (or source concentration 8800 mg/l). All of the simulation results had a better fit to the true condition when simulation period was four years. The position of the source was speculated to be in the gas station and the most serious pollution occurred in the northern area of the gas station.
The simulation by MODFLOW+MT3DMS indicated that pollutants transported from the north to south. Whereas AT123D indicated that pollutants spread eastward. FEMWATER indicated that pollutants spread northward. The main reason causing the differences of transport direction is that MODFLOW +MT3DMS can establish heterogeneous hydraulic conductivity, but FEMWATER and AT123D can only establish hydraulic conductivity homogeneously, especially AT123D only with the ability of simulating steady state and horizontal flow. When the Type II Groundwater Control Standard for benzene, 0.05 mg/l, is used as the baseline, the polluted area simulated by FEMWATER is the smallest, compared with the results simulated by MODFLOW+MT3DMS and AT123D.
The sensitivity analyses by three models show that bulk density does not affect the output. Hydraulic conductivity is the most important sensitive parameter for MODFLOW. Effective porosity influences tremendously on the simulation of MODFLOW+MT3DMS, and both Hydraulic conductivity and longitudinal dispersivity affect more the simulation results of FEMWATER. The output results by AT123D also indicate that effective porosity is a tremendous influential parameter. The output results of three sets of models all vary with the change of mass-loading rate (sink/source concentration) proportionally.
By comparing the simulation results with the observed values, and examining the relative errors and accumulating errors, it is concluded that MODFLOW+MT3DMS gives the closest simulation results to the real field. Considering the applicability, limiting conditions, simulation results of contaminated area and friendly operation, MODFLOW+MT3DMS is the most suitable simulation tool in this research.

並列關鍵字

Groundwater Model ； MODFLOW ； MT3D ； AT123D ； FEMWATER ； the Contaminated Area of the Control Site

參考文獻

張興亞(2001)，「灌溉水滲漏之數值模擬」，碩士論文，中原大學土木工程研究所。

翁琮哲(2001)，「地下水人工補注功效之探討」，碩士論文，中原大學土木工程學系研究所。

曾琮愷(2003)，「隧道開挖滲流現象之模擬」，碩士論文，中原大學土木工程研究所。

郭麗娥(2003)，「未飽和紅土水力傳導係數之研究」，碩士論文，中原大學土木工程研究所。

李昆性(2004)，「桃園台地紅土入滲行為之探討」，碩士論文，中原大學土木工程學系研究所。