|Abstract: ||本研究除了建立 UASB 反應器處理市鎮污水之動力模式(incorporating intrinsic kinetics)及經驗模式(incorporating apparent kinetics)外，亦使用兩組 UASB反應器(操作溫度 = 30℃)處理實際低濃度市鎮污水以獲得實驗數據。兩組UASB反應器皆採固定進流基質濃度(COD = 120 mg/L)及1.0 m/h之表面流速，反應器A及反應器B陸續改變三種水力停留時間(5.2、3.6及2.8 h )亦即改變三種有機負荷率(0.5、0.8及1.0 kg COD/m3-d )；為了比較UASB反應器處理一般市鎮污水與處理截流污水(含漲潮海水成分)之差異性，反應器A為進流一般市鎮污水，反應器B為進流模擬含海水成分之截流污水(加入NaCl，使污水導電度至4000 µmho/cm)，並在第三試程(有機負荷率1.0 kg COD/m3-d )時，為避免污泥顆粒流失，將表面流速調降為0.5 m/h，各試程並於穩定操作狀態下取出污泥顆粒以批次實驗求取市鎮污水厭氧降解之intrinsic及apparent動力常數，據此，不僅可瞭解UASB 反應器處理低濃度市鎮污水之效能及污泥顆粒特性，亦可釐清基質降解動力及質傳效應，最後以實驗數據驗證動力模式與經驗模式之適用性。反應器A及B 皆在有機負荷0.5~1.0 kg COD/m3-d及之操作條件下，兩組反應器皆無揮發酸累積(VFAs小於7 mg/L)且內污泥顆粒化及沉降性皆佳，反應器A及B去除率分別達75%及67%以上。此外，反應器A及B之污泥顆粒比重(1.06~1.13; 1.07~1.14)及污泥顆粒平均粒徑(1.10 ~1.18; 1.28~1.42 mm)皆隨有機負荷率之增加而增大，但反應器B之污泥顆粒平均粒徑略大於反應器A者;反應器A及B之生物密度分別為94370~125000 mg VSS/L及89800~117150 mg VSS/L。由批次實驗結果得知反應器A之intrinsic k 及k ' (0.26~0.33 mg COD/mg VSS-d; 0.24~0.30 mg COD/mg VSS-d)皆略大於反應器B者(0.25~0.31 mg COD/mg VSS-d; 0.22~0.28 mg COD/mg VSS-d)，反應器A之intrinsic Ks及K's (27~33 mg COD/L；32~45 mg COD/L ) 則明顯小於反應器B者(44~47 mg COD/L; 45~53 mg COD/L)。藉由動力模式求得反應器A及B之質傳參數2 (13~14; 10~14)及Bi (3.1~3.8; 4.0~4.8)之變化都不太大，惟質傳參數η (0.33~0.40; 0.27~0.30)隨著有機負荷率之增加而下降。上述結果意味著整體基質去除速率雖受內部質傳阻抗之影響，亦受到基質解速率去除速率之影響。本研究建立之動力模式及經驗模式模擬之COD去除率與UASB反應器處理低濃度市鎮污水實驗值誤差在±20%範圍內，而動力模式與經驗模式模擬值之間差異百分比則僅在13%範圍內。|
A kinetic model (incorporating intrinsic kinetics) and an empirical model (incorporating apparent kinetics) that can be used for simulating variations in substrate residual concentration with different operating conditions in the upflow anaerobic sludge bed (UASB) reactor are formulated. Two UASB reactors (reactors A and B; us = 1.0 m/h; operating temperature = 30℃) were also used to treat low-strength municipal wastewater by maintaining a fixed influent COD concentration of 120 mg/L but with three different hydraulic retention time (HRT) of 5.2, 3.6, and 2.8 h [organic loading rates (OLR) = 0.5, 0.8, and 1.0 kg COD/m3-d] to generate experimental data. Reactors A and B were fed with municipal wastewater and municipal wastewater plus sodium chloride (i.e., to simulate sea water-containing municipal wastewater by maintaining conductivity at 4000 μmho/cm), respectively. Thus, not only the performance of UASB reactors, granule characteristics, mass transfer, and reaction kinetics can be evaluated but the kinetic and empirical models can also be validated by experiments.When reactors A and B were maintained at the OLR of 0.5–1.0 kg COD/m3-d, not only an accumulation of volatile fatty acids was not observed (above 67% of COD removal) but fairly good sludge granulation/settling can also be achieved. With an increase in OLR in reactors A and B, the granule’s specific gravity (1.06–1.13; 1.07–1.14) and the average granule diameter (1.10–1.18 mm; 1.28–1.42 mm) increased, but the average granule diameter of reactor B was slightly larger than that of reactor A. In addition, the microbial density of reactors A and B were 94370–125000 mg VSS/L and 89800–117150 mg VSS/L), respectively. From the batch experiments, the Monod intrinsic k and apparent k' of reactor A (0.26–0.33 mg COD/mg VSS-d; 0.24–0.30 mg COD/mg VSS-d) are slightly larger than those of reactor B (0.25–0.31 mg COD/mg VSS-d; 0.22–0.28 mg COD/mg VSS-d). The Monod intrinsic Ks and apparent K's of reactor A (27–33 mg COD/L; 32–45 mg COD/L) are significantly smaller than those of reactor B (44–47 mg COD/L; 45–53 mg COD/L), indicating that the affinity of bacteria to substrate in reactor A is higher than that in reactor B. By using the validated kinetic model, the calculated mass transfer parameter 2 and Bi of reactors A (13–14; 3.1–3.8) varied slightly with B (10–14; 3.3–4.8). However, the calculated mass transfer parameter of reactor A (0.33–0.40) was higher than that of reactor B (0.27–0.30). The afore-said findings revealed that the overall substrate removal rate was not only influenced by internal mass transfer resistance but also influenced by substrate degradation rate. The calculated COD removal efficiencies using kinetic and empirical models are 20% deviated from the experimental results. The variations of the simulated results using kinetic and empirical models are within 13%.