| 摘要: | 為了響應我國推動的2050 淨零排放目標，增強型地熱系統（Enhanced Geothermal System; EGS）成為乾淨能源的選項之一。由於EGS 相關技術涉及向地底注入流體，這可能導致影響地下孔隙水壓力以及斷層穩定性，進而造成不穩定滑移並增加地震誘發之風險。而過去研究指出地震誘發過程可以透過實驗模擬摩擦行為，其中斷層摩擦組成律（rate-and-state friction law; RSF）被廣泛應用於斷層在準靜態（quasi-static）滑移
速率的力學行為，對於瞭解斷層不穩定性與促進EGS 安全發電有高度價值。本研究旨在建立實驗室三軸岩石力學摩擦試驗的技術，使用中央大學的GCTS 自動伺服控制三軸岩石測試系統（GCTS RTX-1000）進行直剪式樣本組合的三軸摩擦實驗。本研究使用高嶺土粉末作為合成斷層泥之樣本，將其夾於兩個一英寸之L 形不鏽鋼塊之間進行摩擦實驗，由圍壓施加正應力，而軸向壓力施加剪應力。本研究共進行四種不同條件：高水壓與低水壓飽和排水實驗（saturated drained with pore pressure control）、飽和不排水實驗（saturated undrained without pore pressure control）、以及室內溼度乾粉（dry）實驗。所有條件之有效正向應力（effective normal stress）皆為10 MPa，水壓控制實驗的圍壓（confining pressure; Pc）及水壓（pore pressure; Pp）分別為20 及10 MPa或分別為10.5 及0.5 MPa；其餘兩種實驗則不使用水壓控制系統，圍壓為10 MPa。經過八至十二小時於目標圍壓和水壓等待斷層泥中的水壓達到平衡後，我們使用外部荷載差度變壓器（DCDT）控制滑移速率。摩擦實驗剛開始的速率設置為1 μm/s，當軸差應力（deviator stress; Sd）自背景值明顯增加後表示斷層泥開始剪切，此時速率會降低至0.1 μm/s，直到位移達1.5 mm。之後速率會以10 為倍數進行增減調整（例：0.15、0.015、0.15 μm/s），每一步驟設定滑動距離為0.3 mm。實驗結果顯示高水壓飽和排水實驗的有效摩擦係數在達到穩態後為0.27，而所有實驗均呈現速度強化行為。而RSF參數使用RSFit3000（Skarbek & Savage, 2019）並假設彈性勁度k = 0.02 μm⁻¹進行擬合。擬合結果顯示a-b 值為0.006 至0.011，而含水樣本高於室內溼度乾樣本。Dc 值（特徵滑移距離；Characteristic distance）範圍較廣，從4 至40 μm。而擬合結果中決定係數（R2）顯示從0.44 至0.95 不等，這表示部分數據過於分散，可能來自於Sd 的震盪較大。我們可以適當的調整擷取頻率及使用移動平均方法減少雜訊，以獲得更好的擬合結果。另外，儀器使用外部荷重元（External load cell）量測荷重，因此我們對儀器進行O-ring 摩擦校正實驗，觀察儀器本身對於數據量測的影響並校正實驗結果。結果顯示儀器本身僅對於速度步驟中慢到快（0.015 至0.15 μm/s）提升了至少0.1 kN，扣除儀器影響重新擬合後a-b 值降低為0.004 至0.008，這樣的數據也更接近前人研究結果（飽和水壓實驗中為0.002-0.006）。我們預計透過推動並完善這樣的實驗方法，未來將可運用於地熱能源發展中，作為針對斷層穩定性以及潛在風險的評估分析。;In response to the 2050 net-zero emission in Taiwan, identifying potential sites for
exploration of Enhanced Geothermal System (EGS) has become critical. The technology
involves fluid injections into the subsurface, which can cause an increase in pore-fluid
pressure and and potentially lead to unstable slip, consequently increasing the risk of inducing
seismicity. Recent studies have applied the rate-and-state friction law (RSF) in modeling fault
slip behavior under quasi-static slip conditions to better understand mechanisms of induced
seismicity during EGS operations. To develop experimental approaches to understand faultslip
behaviors, we conducted confined direct-shear tests using the GCTS servo-controlled
Triaxial Rock Testing System (GCTS RTX-1000) at National Central University, Taiwan. We
used kaolinite powder as synthetic gouge samples that were sandwiched between two 1-inch
L-shaped stainless-steel platens. We conducted four types of experiments: saturated drained
with pore pressure (Pp) control at either high or low Pp, saturated undrained without pore
pressure control, dry sample at room humidity. All experiments were deformed at room
temperature under a constant effective normal stress of 10 MPa. For experiments with Pp
control, confining pressure (Pc) and Pp were maintained at either 20 and 10 MPa or 10.5 and
0.5 MPa, respectively. The saturated undrained and dry experiments were conducted without
Pp control with Pc maintained at 10 MPa. All experiments were performed at the same
effective pressure (Pc - Pp). After 8 –12 hours of compaction to achieve equilibrium in Pp
within the fault gouge, we applied a load-point displacement rate of 1 μm/s controlled by the
load frame displacement transducer (DCDT). The deviator stress (Sd) would significantly
increase from the background level until the loading piston contacted with the L-shape platen,
marking the onset of fault gouge shearing. The velocity was then decreased to 0.1 μm/s until
1.5 mm of displacement was reached. The velocity was then adjusted in a 10-fold step
sequence (e.g., 0.15, 0.015, 0.15 μm/s), with each test targeted to reach 0.3 mm in sheardisplacement. Each experiment was calibrated for the effects of friction from the sealing
stacks on the external load measurement. The results show that the effective friction
coefficient reached a steady state of 0.27 under the experiments at Pp of 10 MPa, and all
experiments exhibit velocity-strengthening behavior. The RSF parameters were fit using
RSFit3000 (Skarbek & Savage, 2019) on datasets after moving averaging to reduce noises,
assuming system elastic stiffness of k = 0.02 μm⁻¹. The a-b values range from 0.006 to 0.011,
and characteristic distances (Dc) are approximately 4-40 μm. The coefficient of determination
(R²) of the fitting results varies between 0.44 and 0.95, where the lower R2 may be attributed
to oscillations in Sd. O-ring calibration experiments revealed a load increase of at least 0.1 kN
during velocity steps from 0.015 to 0.15 μm/s. After calibration, the recalculated a-b values
decreased to 0.004–0.008, which better aligns with previous studies (0.002–0.006 for
saturated conditions). We anticipate that our experimental approaches can be further applied
to provide a basis for evaluating fault-slip instability with geothermal development in Taiwan. |
