Transcription

1

2

3 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 REP-44-2 修 正 終 稿 213 年 2 月 奧 雅 納 - 莫 特 麥 克 唐 納 聯 營 公 司 又 一 城 5 樓 達 之 路 8 號 九 龍 塘 九 龍 香 港 本 報 告 涵 蓋 了 業 主 的 特 殊 指 示 和 要 求 它 並 非 且 不 應 被 其 他 任 何 第 三 方 所 引 用 若 引 用 而 導 致 的 一 切 後 果, 均 由 第 三 方 自 行 承 擔 奧 雅 納 - 莫 特 麥 克 唐 納 聯 營 公 司

4

5 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 目 錄 1 前 言 1 背 景 1 九 龍 灣 內 的 海 底 隧 道 1 保 護 海 港 條 例 1 報 告 結 構 2 2 凌 駕 性 公 眾 需 要 2 引 言 2 交 通 方 面 的 理 據 2 項 目 的 效 益 7 結 論 8 3 沒 有 合 理 的 不 涉 及 填 海 的 替 代 方 案 9 引 言 9 設 計 及 施 工 安 排 的 限 制 9 不 需 填 海 之 建 造 方 案 1 使 用 填 海 的 建 造 方 法 14 替 代 走 線 16 頁 6 結 論 23 附 錄 附 錄 一 有 關 臨 時 填 海 的 凌 駕 性 公 眾 需 要 23 沒 有 合 理 而 不 涉 及 填 海 工 程 的 建 造 方 式 23 最 小 的 填 海 範 圍 23 第 二 期 公 眾 參 與 活 動 23 獨 立 專 家 審 查 23 符 合 保 護 海 港 條 例 23 林 興 强 教 授 - 關 於 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 材 料 報 告 之 獨 立 專 家 審 查 附 錄 二 吳 宏 偉 教 授 - 關 於 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 材 料 報 告 之 獨 立 專 家 審 查 4 最 少 範 圍 的 臨 時 填 海 18 引 言 18 填 海 的 長 度 18 填 海 的 寬 度 18 臨 時 填 海 的 時 間 19 最 少 填 海 要 求 的 總 結 19 5 公 眾 諮 詢 2 前 階 段 的 公 眾 參 與 活 動 2 29 年 7 月 18 日 公 眾 論 壇 2 29 年 6 月 2 日 焦 點 小 組 會 議 2 公 衆 參 與 活 動 結 論 2 現 階 段 的 公 眾 參 與 活 動 2 現 時 有 關 中 九 龍 幹 線 的 公 眾 意 見 21 諮 詢 區 議 會 21 諮 詢 海 濱 事 務 委 員 會 21 九 龍 灣 臨 時 填 海 專 業 論 壇 21 九 龍 灣 臨 時 填 海 專 題 討 論 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX

6

7 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 1 前 言 背 景 1.1 中 九 龍 幹 線 全 長 4.7 公 里, 採 用 雙 程 三 線 分 隔 車 道 的 設 計, 連 接 西 九 龍 的 油 麻 地 交 匯 處 與 東 九 龍 的 啟 德 發 展 區 及 九 龍 灣 道 路 網, 成 為 貫 通 中 九 龍 的 快 速 道 路 中 九 龍 幹 線 總 平 面 圖 及 縱 剖 面 請 見 圖 於 27 年 至 29 年 間 的 勘 測 和 初 步 設 計 階 段 中, 我 們 曾 研 究 4 多 個 中 九 龍 幹 線 的 走 線 方 案 在 比 較 各 方 案 對 區 內 樓 宇 社 區 設 施 環 境 土 地 及 交 通 方 面 的 影 響, 參 考 公 眾 參 與 活 動 中 所 收 到 的 公 眾 意 見 後, 選 取 現 時 的 走 線 立 法 會 油 尖 旺 九 龍 城 和 觀 塘 區 議 會 均 普 遍 支 持 選 定 的 走 線 九 龍 灣 內 的 海 底 隧 道 1.3 幹 線 由 九 龍 城 碼 頭 至 啟 德 發 展 區 有 一 段 長 約 37 米 的 隧 道 會 經 過 九 龍 灣 海 床 由 於 現 場 的 種 種 限 制, 該 段 隧 道 需 以 臨 時 填 海 方 法 建 造 保 護 海 港 條 例 不 准 填 海 的 推 定 1.4 保 護 海 港 條 例 源 自 1996 年 保 護 海 港 協 會 以 私 人 法 案 形 式 呈 交 立 法 局, 該 草 案 於 1997 年 6 月 正 式 通 過 成 為 法 例, 並 經 法 律 程 序 一 再 修 訂, 保 護 海 港 條 例 藉 設 定 不 准 在 海 港 內 進 行 填 海 工 程 的 推 定, 以 達 致 保 護 和 保 存 海 港 的 目 的, 再 於 1999 年 12 月 修 訂 至 涵 蓋 整 個 維 港 的 範 圍 1.5 條 例 第 三 條 規 定 : (1) 海 港 須 作 為 香 港 人 的 特 別 公 有 資 產 和 天 然 財 產 而 受 到 保 護 和 保 存, 而 為 此 目 的, 現 設 定 一 個 不 准 進 行 海 港 填 海 工 程 的 推 定 [ 第 3 條 (1) 段 ] (2) 所 有 公 職 人 員 和 公 共 機 構 在 行 使 任 何 歸 屬 他 們 的 權 力 時, 須 顧 及 第 (1) 款 所 述 的 原 則 以 作 為 指 引 [ 第 3 條 (2) 段 ] 1.6 保 護 海 港 條 例 將 填 海 明 確 地 定 義 為 任 何 為 將 海 床 或 前 濱 海 形 成 為 土 地 而 進 行 或 擬 進 行 的 工 程, 條 例 中 填 海 的 釋 義 是 針 對 所 形 成 土 地, 意 思 可 推 斷 為 超 越 海 平 面 的 工 程 故 此, 縱 使 進 行 只 有 小 規 模 填 海 要 求 的 工 程, 例 如 興 建 碼 頭 登 岸 梯 級 等, 亦 需 要 符 合 保 護 海 港 條 例 城 市 規 劃 委 員 會 對 保 護 海 港 協 會 有 限 公 司 案 件 的 終 審 法 院 判 決 (24 年 1 月 9 日 ) 九 龍 灣 臨 時 填 海 1.7 城 市 規 劃 委 員 會 曾 於 22 年 12 月 及 23 年 2 月 就 灣 仔 北 分 區 計 劃 大 綱 草 圖 ( 編 號 5/H25/1) 作 出 決 定, 包 括 在 維 港 範 圍 內 填 海 的 建 議 保 護 海 港 協 會 就 該 決 定 經 司 法 程 序 提 出 反 對, 認 為 城 市 規 劃 委 員 會 作 出 不 合 法 不 合 理 及 不 合 乎 邏 輯 的 決 定, 該 反 對 意 見 在 高 等 法 院 的 聆 訊 中 獲 判 勝 訴 1.8 在 24 年 1 月 9 日 終 審 法 院 的 判 決 中, 上 訴 被 駁 回 終 審 法 院 裁 定 (a) 為 了 貫 徹 保 護 和 保 存 海 港 這 個 強 而 有 力 的 法 定 原 則, 不 准 許 進 行 海 港 填 海 工 程 的 推 定, 只 能 被 具 有 凌 駕 性 公 眾 需 要 的 填 海 工 程 所 推 翻 這 可 以 稱 為 凌 駕 性 公 眾 需 要 的 測 試 ; (b) 如 果 還 有 其 他 合 理 解 決 方 法, 就 不 能 把 填 海 工 程 說 成 有 凌 駕 性 需 要 因 為 當 其 他 合 理 解 決 方 法 可 以 滿 足 公 眾 需 要 的 時 候, 就 沒 有 所 謂 的 凌 駕 性 需 要 ; 以 及 圖 1-1 中 九 龍 幹 線 的 走 線 (c) 所 建 議 的 填 海 工 程 範 圍, 不 應 超 越 凌 駕 性 需 要 所 要 求 的 最 低 限 度 1.9 公 眾 需 要 將 是 社 區 的 需 要, 包 括 社 區 對 經 濟 環 境 和 社 會 的 需 要 只 有 迫 切 和 即 時 的 需 要 方 可 稱 為 凌 駕 性 需 要 這 個 需 要 必 須 令 人 信 服, 也 就 是 實 際 上 有 必 要 去 凌 駕 公 眾 保 護 和 保 存 海 港 的 強 烈 需 要 這 個 需 要 必 須 是 即 時 的 需 要, 並 且 在 規 劃 時 已 經 考 慮 到 的 一 個 在 明 確 和 合 理 的 時 間 內 將 出 現 的 需 要 如 果 這 個 需 要 不 會 在 一 定 時 間 內 出 現, 就 不 足 以 推 翻 不 准 填 海 的 推 定 了 當 前 迫 切 的 需 要, 是 遠 遠 超 越 乎 那 些 人 們 樂 於 擁 有 的 應 有 的 可 取 的 或 有 益 的 事 物 但 另 一 方 面, 亦 未 達 到 非 到 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 1 頁

8 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 最 後 才 會 需 要 或 公 眾 不 可 或 缺 的 地 步 當 前 的 需 要 是 顧 及 規 劃 工 作 的 時 間 表, 在 確 實 而 合 理 的 時 間 內 會 出 現 的 需 要 1.1 判 詞 亦 指 出, 當 有 另 一 個 合 理 的 方 案 可 取 代 填 海, 填 海 工 程 則 不 具 淩 駕 性 公 眾 需 要 所 有 情 況 均 應 一 併 考 慮, 包 括 每 個 方 案 對 經 濟 環 境 和 社 會 方 面 所 造 成 的 影 響, 並 要 顧 及 所 涉 及 的 成 本 時 間 和 引 致 的 延 誤 建 議 的 填 海 範 圍, 不 應 超 越 為 符 合 淩 駕 性 需 要 所 要 求 的 最 低 限 度, 每 個 建 議 的 填 海 範 圍 均 必 須 具 備 理 據 1.11 要 符 合 淩 駕 性 公 眾 需 要 測 試 的 準 則, 公 職 人 員 或 公 共 機 構 必 須 以 有 力 和 令 人 信 服 的 資 料 證 明 填 海 是 具 淩 駕 性 公 眾 需 要 保 護 海 港 協 會 對 律 政 司 司 長 案 件 的 原 訟 法 庭 一 審 判 決 (28 年 3 月 2 日 ) 1.12 政 府 擬 建 的 中 環 灣 仔 繞 道 是 一 條 沿 香 港 島 北 岸 而 建 的 幹 道, 以 疏 導 現 有 東 西 走 廊 的 交 通 興 建 中 環 灣 仔 繞 道 的 一 段 明 挖 回 填 式 隧 道, 需 要 在 灣 仔 前 公 眾 貨 物 裝 卸 區 及 銅 鑼 灣 避 風 塘 進 行 臨 時 填 海 臨 時 填 海 預 計 會 維 持 約 6 年 工 程 完 成 後, 將 臨 時 填 海 及 將 海 床 回 復 原 狀 政 府 當 時 認 為, 此 臨 時 填 海 工 程 毋 須 受 到 保 護 海 港 條 例 的 約 束 1.13 保 護 海 港 協 會 向 高 等 法 院 提 出 訴 訟 的 理 據 是 保 護 海 港 條 例 並 沒 有 明 確 分 辨 形 成 永 久 土 地 的 填 海 和 臨 時 填 海 的 分 別 保 護 海 港 協 會 尋 求 作 出 澄 清, 即 使 填 海 屬 過 渡 性 質, 甚 至 是 為 了 避 免 永 久 性 填 海, 填 海 工 程 仍 須 符 合 不 准 填 海 的 推 定, 除 非 能 通 過 淩 駕 性 公 眾 測 試 去 說 明 其 需 要 性, 否 則 不 可 合 法 地 進 行 1.14 原 訟 法 庭 判 保 護 海 港 協 會 勝 訴, 並 重 申 保 護 海 港 條 例 和 不 准 填 海 的 推 定 適 用 於 中 環 灣 仔 繞 道 計 劃 中 建 議 的 臨 時 填 海 工 程 報 告 結 構 1.15 本 報 告 包 括 具 有 力 和 令 人 信 服 的 資 料, 以 證 明 興 建 中 九 龍 幹 線 的 海 底 隧 道 所 建 議 在 九 龍 灣 的 臨 時 填 海 工 程, 符 合 保 護 海 港 條 例 的 要 求 本 報 告 根 據 24 年 8 月 19 日 房 屋 及 規 劃 地 政 局 發 出 的 技 術 通 告 第 1/4 號 而 準 備 技 術 通 告 是 為 公 職 人 員 和 公 共 機 構 在 考 慮 及 批 准 填 海 建 議 前 提 供 準 則, 包 括 對 臨 時 填 海 工 程 在 公 眾 諮 詢 過 程 裡 中 的 資 料 1.16 本 報 告 的 後 部 分 內 容 如 下 (a) 第 2 部 分 提 供 支 持 中 九 龍 幹 線 工 程 凌 駕 性 公 眾 需 要 的 資 料 和 評 估 ; (b) 第 3 部 分 說 明 建 議 在 九 龍 灣 的 臨 時 填 海 沒 有 合 理 的 替 代 方 案 ; (c) 第 4 部 分 闡 述 建 議 的 填 海 工 程 的 範 圍 將 會 是 最 低 限 度 ; (d) 第 5 部 分 描 述 為 臨 時 填 海 工 程 所 進 行 的 公 眾 諮 詢, 並 總 結 了 公 眾 的 意 見 ; 及 2 凌 駕 性 公 眾 需 要 引 言 2.1 房 屋 及 規 劃 地 政 局 技 術 通 告 第 1/4 的 準 則, 提 供 了 中 九 龍 幹 線 工 程 的 凌 駕 性 公 眾 需 要 之 評 估 方 法 技 術 通 告 指 出, 第 一 個 步 驟 是, 確 定 項 目 是 否 有 迫 切 和 即 時 的 公 眾 需 要 公 眾 需 要 被 定 義 為 社 區 的 經 濟 環 境 和 社 會 需 要 社 區 的 需 要 是 指 大 眾 至 需 要 的, 而 不 是 滿 足 小 部 分 人 的 特 別 需 求 和 利 益 2.2 根 據 房 屋 及 規 劃 地 政 局 技 術 通 告 1/4 中 所 說 令 人 信 服 的 是 指 具 有 必 要 的 力 量 去 壓 倒 公 眾 保 護 和 保 存 海 港 的 強 烈 需 要 這 是 需 要 被 有 力 和 令 人 信 服 的 資 料 支 持 ( 例 如 研 究 結 果, 預 測, 成 本 和 效 益 分 析 等 ), 以 支 持 填 海 工 程 項 目 的 凌 駕 性 需 要 2.3 目 前 需 要 這 個 詞 被 定 義 為 證 明 在 一 個 明 確 和 合 理 的 時 間 框 架 內 將 出 現 的 需 要 要 滿 足 這 一 點, 必 須 有 一 個 具 體 的 實 施 方 案 和 堅 定 承 諾 並 獲 得 有 關 政 府 部 門 的 與 相 關 部 門 的 認 可 ( 如 適 用 ) 2.4 下 面 介 紹 中 九 龍 幹 線 目 前 迫 切 的 公 共 需 要, 將 提 供 給 香 港 社 會 的 利 益 : 交 通 方 面 的 理 據 ( 一 ) 概 覽 2.5 現 時 連 接 九 龍 東 西 主 要 幹 道 的 交 通, 包 括 龍 翔 道, 界 限 街, 太 子 道 西, 亞 皆 老 街, 窩 打 老 道, 加 士 居 道 行 車 天 橋 及 漆 咸 道 北 已 接 近 飽 和, 交 通 擠 塞 時 有 發 生 該 區 政 府 已 實 施 交 通 管 理 及 改 善 措 施 然 而, 由 於 現 有 的 東 西 幹 道 兩 端 都 是 高 度 發 展 的 地 區, 近 乎 沒 有 可 改 善 交 通 的 空 間 因 此, 這 些 措 施 只 能 在 短 期 內 紓 緩 該 區 的 交 通 問 題 2.6 擬 建 的 中 九 龍 幹 線 是 一 條 貫 通 九 龍 半 島 的 三 線 雙 程 主 幹 路, 連 接 西 面 的 西 九 龍 與 東 面 的 啟 德 發 展 區 幹 線 的 西 面 出 口 連 接 西 九 龍 油 麻 地 交 匯 處 經 由 交 匯 處, 車 輛 就 可 以 分 別 前 往 西 區 海 底 隧 道 尖 沙 咀 西 九 龍 文 化 區 西 九 龍 公 路 8 號 幹 線 和 3 號 幹 線 東 面 出 口 在 啟 德 發 展 區, 然 後 分 別 連 接 到 九 龍 灣, 東 九 龍, 觀 塘 繞 道, 將 軍 澳 隧 道,T2 道 路 和 將 軍 澳 - 藍 田 隧 道 中 九 龍 幹 線 連 同 T2 道 路 及 將 軍 澳 - 藍 田 隧 道 將 形 成 一 個 策 略 性 道 路 網, 即 6 號 幹 線, 連 接 西 九 龍 和 將 軍 澳 2.7 中 九 龍 幹 線 西 面 的 西 九 龍 文 娛 藝 術 區 和 東 面 的 啟 德 發 展 區 均 已 取 得 長 足 發 展, 隨 著 這 些 發 展 的 階 段 性 完 成, 對 在 九 龍 區 內 連 接 東 西 策 略 性 道 路 和 關 鍵 匯 合 處 的 交 通 產 生 顯 著 影 響 2.8 早 在 二 十 年 前, 已 確 證 需 要 有 一 條 直 接 連 接 東 西 九 龍 的 交 通 路 線, 以 應 付 橫 跨 九 龍 的 交 通 需 求 及 紓 緩 現 時 九 龍 中 部 交 通 擠 塞 的 狀 況 往 後 的 幾 年 裡, 隨 著 西 九 龍, 九 龍 灣 及 將 軍 澳 的 發 展, 交 通 量 不 斷 增 長, 橫 貫 九 龍 的 交 通 需 求 也 有 所 增 加 為 廣 大 公 眾 和 道 路 使 用 者 提 供 一 條 可 以 紓 緩 擠 塞 的 東 西 方 向 道 路 的 需 求, 已 是 刻 不 容 緩 ; 而 興 建 中 九 龍 幹 線 便 能 滿 足 這 需 求 此 外, 如 圖 2-1 所 示, 擬 建 的 啟 德 發 展 區 及 西 九 龍 文 娛 藝 術 區 的 發 展 無 疑 進 一 步 加 大 了 對 幹 線 的 需 求 (e) 第 6 部 分 提 出 對 臨 時 填 海 工 程 的 結 論, 並 描 述 推 薦 的 方 案 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 2 頁

9 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 圖 2-2 九龍區的現有道路的服務水平 圖 2-1 未來東西九龍的發展項目 2.9 東西九龍之間的道路交通 目前是由龍翔道 界限街 太子道 亞皆老街 窩打老道和漆咸道北 2.1 現有的主要東西公路走廊 龍翔道 太子道西及界限街 一直被交通容量不足的問題所困擾 部分歸結為最近幾年九龍半島和新界的西部 如西九龍填海區 大嶼山 和東部 如將軍澳 的物業發 展迅速 而沒有提升的交通容量導致交通出現容量不足的情況 一些交通擠塞的實例顯示在圖 2-3 組成 除了龍翔道及加士居道天橋外 這些現有的東西方向道路 經常被頻密的道路交界處和信號控制 的路口限制了交通流量 根據在 211 年進行的綜合交通調查 對現有的路口和道路的表現作基線研究 以及現有的交通問題進行了匯總 我們還注意到在 211 年交通統計年報 超過 6 往返東西九龍的車 輛交通 圖 2-2 中的交通調查線 A-A 是商用車 即的士 公共小巴 貨車 旅遊車和巴士 44-2 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 第3頁

10 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 道 路 方 向 行 車 量 / 容 車 量 比 率 (V/C) 211 ( 現 有 道 路 ) 上 午 下 午 加 士 居 道 天 橋 東 行.8 >1.3 ( 彌 敦 道 東 側 ) 西 行 東 行 漆 咸 道 北 西 行 ( 從 蕪 湖 街 至 平 治 街 ) 西 行 ( 遠 側 的 自 由 流 通 車 道 ) 東 九 龍 走 廊 東 行 ( 從 馬 頭 角 道 至 漆 咸 道 北 ) 西 行.8.6 說 明 : 行 車 量 / 容 車 量 比 率 (V / C) 是 一 個 指 標, 它 反 映 的 道 路 性 能 一 個 V / C 比 等 於 或 低 於 1., 表 示 道 路 有 足 夠 的 能 力, 以 應 付 車 輛 交 通 量 的 考 慮, 所 產 生 的 交 通 暢 順 V / C 比 率 高 於 1., 則 表 示 交 通 開 始 擠 塞 ; 高 於 1.2 則 表 示 更 嚴 重 的 交 通 擠 塞 與 流 量 的 進 一 步 增 加, 速 度 會 逐 漸 減 慢 圖 2-3 九 龍 中 西 部 現 有 的 交 通 狀 況 2.11 其 中 許 多 道 路 走 廊 已 經 達 到 了 繁 忙 時 間 的 交 通 設 計 可 承 載 能 力, 例 如, 在 繁 忙 時 間 於 太 子 道 西 及 界 限 街 的 方 向 流 量, 達 到 的 行 車 量 / 容 車 量 比 率 分 別 為.8 和 1.1, 顯 示 已 接 近 或 超 過 承 載 的 能 力 211 年 西 九 龍 及 東 九 龍 的 行 車 量 / 容 車 量 的 比 率 總 結 在 下 列 表 2-1 表 2-1 在 211 年 東 西 九 龍 關 鍵 道 路 的 表 現 行 車 量 / 容 車 量 比 率 道 路 方 向 (V/C) 211 ( 現 有 道 路 ) 上 午 下 午 東 行.9.8 龍 翔 道 ( 從 獅 子 山 隧 道 路 至 竹 園 路 ) 西 行.9.9 西 行 ( 竹 園 道 支 路 ).9.6 界 限 街 ( 從 大 坑 東 道 至 基 堤 道 ) 東 行 太 子 道 西 ( 從 基 堤 道 至 嘉 道 理 道 ) 西 行.8.7 亞 皆 老 街 東 行.8.9 ( 從 嘉 齡 道 至 天 光 道 ) 西 行.8.8 窩 打 老 道 ( 碧 街 至 登 打 士 街 ) 北 行.9 1. 南 行 中 九 龍 幹 線 將 提 供 一 個 新 的 路 線, 繞 開 堵 塞 的 道 路 網 絡, 從 而 顯 著 地 減 少 行 車 時 間 據 估 計 在 221 年, 於 繁 忙 時 間 內, 使 用 中 九 龍 幹 線 從 西 九 龍 到 東 九 龍 的 行 車 時 間 只 需 要 約 5 分 鐘, 而 沒 有 中 九 龍 幹 線 的 行 車 時 間 則 需 要 3-35 分 鐘 中 九 龍 幹 線 也 將 大 大 減 少 在 主 要 東 西 走 廊 上 的 行 車 量, 紓 緩 交 通 擠 塞 因 路 面 的 交 通 情 況 顯 著 改 善, 鄰 近 地 區 包 括 黃 大 仙 何 文 田 及 九 龍 城 也 將 受 惠 2.13 中 九 龍 幹 線 也 將 與 東 西 九 龍 的 高 速 道 路 網 連 接, 成 為 策 略 性 道 路 網 的 重 要 組 成 部 分 在 這 方 面, 東 面 的 啟 德 交 匯 處 將 連 接 中 九 龍 幹 線 和 九 龍 灣 的 道 路 網 絡, 從 而 提 高 了 觀 塘 及 啟 德 發 展 區 與 西 九 龍 之 間 的 交 通 便 利, 以 及 為 政 府 提 出 的 起 動 九 龍 東 計 劃 提 供 基 礎 交 通 設 施 配 套 中 九 龍 幹 線 將 會 與 在 啟 德 發 展 區 擬 建 中 的 T2 主 幹 路, 以 及 將 軍 澳 - 藍 田 隧 道, 組 成 全 長 12.5 公 里 的 6 號 幹 線, 直 接 連 繫 西 九 龍 和 將 軍 澳 2.14 在 中 九 龍 幹 線 西 面 的 油 麻 地 交 匯 處 有 四 通 八 達 的 支 路 連 接 西 九 龍 公 路 及 連 翔 道 車 輛 可 經 西 九 龍 公 路, 南 往 香 港 島 西 部, 西 至 香 港 國 際 機 場 和 葵 涌 貨 櫃 碼 頭, 北 行 前 往 新 界 西 北 車 輛 亦 可 經 連 翔 道 進 出 西 九 龍 發 展 區 廣 深 港 高 速 鐵 路 (XRL) 西 九 龍 站 及 西 九 龍 文 化 區 2.15 以 下 段 落 總 結 了 在 九 龍 東 西 和 中 部 發 現 的 交 通 問 題 ( 二 ) 西 九 龍 的 交 通 情 況 2.16 在 西 九 龍 內 區 的 各 交 界 處 ( 如 柯 士 甸 道 西 / 柯 士 甸 道 / 廣 東 道, 連 翔 道 / 佐 敦 道, 廣 東 道 / 匯 翔 道 ) 觀 察 到 的 總 流 量 可 見, 交 通 主 要 由 往 返 住 宅 及 辦 公 樓 的 活 動 形 成, 故 此 一 般 在 週 一 至 週 五 的 繁 忙 時 間 最 為 明 顯 在 更 主 要 的 路 口, 一 些 車 龍 主 要 從 下 游 的 交 界 處 伸 延, 例 如, 在 佐 敦 道 東 行 進 入 佐 敦 道 / 渡 船 街 / 廣 東 道 交 界 處, 在 週 六 中 午 繁 忙 時 間 車 龍 已 延 伸 到 連 翔 道 / 佐 敦 道 交 界 處 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 4 頁

11 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 2.17 從 加 士 居 道 的 道 路 聯 絡 能 力 評 估 的 結 果 顯 示, 加 士 居 道 天 橋 東 行 和 西 行 的 長 長 車 龍 也 影 響 車 輛 通 過 該 路 段 的 實 際 需 求 從 觀 察 得 知, 在 加 士 居 道 天 橋 上, 東 行 和 西 行 的 車 龍 最 遠 分 別 可 及 窩 打 老 道 和 佛 光 街 ( 五 ) 道 路 網 絡 可 靠 性 2.25 如 圖 2-4 所 示, 於 25 年 5 月 9 日 在 太 子 道 東 發 生 了 一 宗 棚 架 倒 塌 的 重 大 意 外, 由 於 這 太 子 道 東 是 其 中 一 條 最 重 要 的 東 西 走 廊, 事 故 導 致 整 個 九 龍 區 的 交 通 嚴 重 癱 瘓 ( 三 ) 中 九 龍 的 交 通 情 況 2.18 九 龍 半 島 中 央 部 分 的 道 路 網 的 特 點 是 由 東 西 和 南 北 道 路 基 本 上 組 成 一 個 網 格 的 形 式 在 目 前 的 研 究 中 特 別 值 得 關 注 的 是 在 道 路 等 級 制 度 中 較 高 級 的 道 路, 即 幹 道 主 要 幹 路 和 地 區 支 路, 對 跨 區 交 通 提 供 高 吞 吐 能 力 具 有 重 要 的 作 用 2.19 雖 然 上 述 的 一 些 道 路 在 規 劃 時 被 列 為 主 要 支 路, 主 要 負 責 跨 區 的 交 通, 但 事 實 上, 由 於 某 些 原 因, 這 些 道 路 有 一 定 行 車 量 是 來 自 地 區 性 的 交 通 2.2 中 九 龍 區 的 路 口 一 般 都 在 平 日 高 峰 時 段 比 較 繁 忙 ( 例 如 柯 士 甸 道 / 漆 咸 道 南 / 暢 運 道 路 口, 亞 皆 老 街 / 彌 敦 道 路 口, 亞 皆 老 街 / 窩 打 老 道 / 公 主 道 路 口, 亞 皆 老 街 / 洗 衣 街 路 口 及 漆 咸 道 北 / 蕪 湖 街 路 口 ) 車 龍 的 長 度 還 顯 示, 大 部 分 東 西 向 道 路 的 關 鍵 路 口, 如 太 子 道 西 及 界 限 街, 平 日 的 車 龍 比 週 末 的 繁 忙 時 間 還 要 長 2.21 在 亞 皆 老 街 / 洗 衣 街 路 口 與 亞 皆 老 街 / 染 布 房 街 路 口 也 觀 察 到 長 車 龍 根 據 觀 察, 在 平 日 和 週 末 的 時 間, 洗 衣 街 北 行 線 / 旺 角 道 路 口 的 車 龍, 會 影 響 實 際 需 要 通 過 亞 皆 老 街 / 洗 衣 街 路 口 的 車 輛 因 為 車 龍 的 影 響, 該 路 口 在 週 末 計 算 出 的 剩 餘 容 車 輛 將 高 於 平 日 同 樣 在 亞 皆 老 街 / 染 布 房 街 路 口, 亞 皆 老 街 西 行 到 亞 皆 老 街 / 洗 衣 街 路 口 的 車 龍 也 將 影 響 到 實 際 需 要 通 過 在 亞 皆 老 街 / 染 布 房 街 路 口 的 車 輛 洗 衣 街 與 染 布 房 街 之 間 的 亞 皆 老 街 行 車 道 的 短 距 離 是 限 制 因 素 之 一, 也 加 劇 了 道 路 網 絡 的 車 龍 問 題 2.22 行 車 時 間 調 查 的 結 果 也 顯 示, 沿 中 九 龍 區 的 重 要 道 路 ( 即 界 限 街, 太 子 道 西, 佐 敦 道 ) 在 上 午 和 下 午 繁 忙 時 間, 平 均 的 行 駛 時 速 慢 於 其 他 重 要 道 路 這 主 要 是 受 交 界 處 停 車 ( 例 如 渡 船 街 / 窩 打 老 道 路 口 及 亞 皆 老 街 / 塘 尾 道 路 口 ) 進 出 大 廈 影 響 ( 四 ) 東 九 龍 的 交 通 情 況 2.23 東 九 龍 區 的 道 路 網 的 特 點 是 一 系 列 密 集 的 主 要 道 路, 呈 西 北 - 東 南 方 向 走 向 ( 啟 德 隧 道 - 啟 福 道 走 廊 觀 塘 繞 道 太 子 道 東 - 觀 塘 道 走 廊 和 清 水 灣 道 走 廊 ) 並 接 近 垂 直 相 交 ; 三 條 策 略 性 道 路, 即 龍 翔 路, 將 軍 澳 隧 道 及 東 區 海 底 隧 道, 以 及 一 些 支 路 目 前 的 研 究 特 別 關 注 的 T2 主 幹 路, 即 上 述 那 一 類 西 北 - 東 南 方 向 的 主 要 道 路, 因 為 T2 主 幹 路 亦 能 紓 緩 這 些 道 路 的 交 通 狀 況 另 外, 九 龍 灣 內 的 道 路 和 路 口 會 包 括 在 啟 德 機 場 南 停 機 坪 的 道 路 和 發 展 的 研 究 裡, 因 為 這 些 道 路 將 直 接 連 接 南 停 機 坪 與 啟 德 發 展 區 外 的 地 區 2.24 在 東 九 龍 區 的 道 路 交 匯 處, 由 於 交 通 來 自 工 業 和 商 業 的 活 動, 所 以 在 週 一 至 週 五 的 繁 忙 時 間 最 為 繁 忙 然 而, 在 路 口 的 通 行 能 力 評 估 結 果 顯 示, 這 些 關 鍵 的 路 口, 除 了 偉 業 街 / 偉 發 道 路 口 超 負 荷 運 作 之 外, 在 週 一 至 週 五 的 繁 忙 時 間 仍 然 在 可 接 受 的 水 平 運 作 ; 該 路 口 的 超 負 荷 運 作 源 於 通 過 該 路 口 觀 塘 繞 道 往 觀 塘 及 藍 田 區 的 繁 忙 交 通 流 量 而 成 圖 年 5 月 9 日 在 太 子 道 東 發 生 的 棚 架 倒 塌 事 故 2.26 擬 建 的 中 九 龍 幹 線, 可 以 作 為 九 龍 中 部 東 行 及 西 行 線 車 輛 的 緊 急 走 廊 25 年 6 月 公 布 的 緊 急 事 故 交 通 協 調 工 作 組 報 告 中 指 出, 由 於 惡 劣 天 氣 狀 況, 九 龍 中 部 東 行 及 西 行 的 交 通 受 到 嚴 重 影 響 工 作 組 還 表 示, 在 九 龍 南 北 方 向 的 交 通 較 為 發 達, 但 東 西 方 向 的 道 路 依 然 有 不 足 之 處 他 們 認 為 有 需 要 增 加 東 西 方 向 道 路 的 行 車 容 量 ( 六 ) 交 通 管 理 措 施 2.27 即 使 政 府 已 實 施 了 一 些 地 區 性 的 交 通 管 理 和 改 善 措 施, 然 而, 由 於 現 有 的 東 西 幹 道 兩 端 都 是 高 度 發 展 的 地 區, 幾 乎 沒 有 改 善 的 空 間 因 此, 這 些 措 施 只 能 在 短 期 內 紓 緩 該 區 的 交 通 問 題 為 了 有 效 解 決 九 龍 中 部 東 西 行 的 交 通 問 題, 應 盡 快 落 實 中 九 龍 幹 線 的 工 程 以 提 供 另 一 條 替 代 路 線, 繞 開 堵 塞 的 道 段 並 增 加 東 西 走 廊 的 容 車 量 ( 七 ) 交 通 預 測 2.28 交 通 研 究 預 測 在 沒 有 中 九 龍 幹 線 的 情 況 下 東 西 走 廊 行 車 量 增 加 的 後 果 221 年 的 交 通 量 預 測 已 考 慮 了 新 的 土 地 利 用 總 體 規 劃 假 設 及 人 口 預 測, 以 確 保 交 通 量 預 測 是 符 合 目 前 的 戰 略 和 地 方 的 規 劃 意 向, 以 上 種 種 證 實 了 對 幹 線 的 需 要 性 第 三 次 整 體 運 輸 研 究 (CTS 3), 基 礎 區 域 交 通 模 型 (BDTM) 及 區 域 性 人 口 及 就 業 矩 陣 (TPEDM) 都 用 來 參 考 和 審 視 有 和 沒 有 中 九 龍 幹 線 對 該 區 的 交 通 狀 況 2.29 如 圖 2-5 所 示, 在 沒 有 中 九 龍 幹 線 的 情 況 下, 未 來 的 交 通 狀 況 在 221 年 預 期 將 會 變 得 惡 劣 表 2-2 總 結 了 在 221 年 有 和 沒 有 中 九 龍 幹 線 的 情 況 下 東 西 走 廊 的 表 現 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 5 頁

12 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 行車量/容車量比率 (V/C) 道路 方向 漆咸道北 從蕪湖街至平治街 221(沒有中九龍幹線) 221(有中九龍幹線) 上午 下午 上午 下午 東行 西行 西行 (遠邊自由流動車道) 東九龍走廊 東行 從馬頭角道至漆咸道北 西行 東行 西行 中九龍幹線 說明: 行車量/容車量比率 V / C 是一個指標 它反映的道路性能 一個 V / C 比等於或低於 1. 表示道路有足夠的能力 以應 付車輛交通量的考慮 所產生的交通暢順 V / C 比率高於 1. 則表示交通開始擠塞; 高於 1.2 則表示更嚴重的交通擠塞與 流量的進一步增加 速度會逐漸減慢 2.3 在一般情況下 如果 221 年沒有中九龍幹線的話 現有的主要路線的服務水平預計保持在 F - 爬行駕駛速度 而如果 221 年有中九龍幹線 服務水平會提高到 D - 降低駕駛速度 參考圖 主要路口的剩餘容車量 表 2-3 及圖 2-6 簡介現有情況 221 年有中九龍幹線情況和 221 年 圖 年在東西九龍的交通狀況 沒有中九龍幹線情況的表現 表 2-3 主要路口的表現 表 2-2 在 221 年關鍵的行車量/容車量的有關東西道路總結 路口 行車量/容車量比率 (V/C) 道路 方向 221(沒有中九龍幹線) 21１年 221(有中九龍幹線) 221 年沒有中九 221 有中九龍 龍幹線 幹線 上午 下午 上午 下午 東行 西行 西行 (支路從竹園道) 東行 西行 在 221 年沒有中九龍幹線的情況下 全部的剩餘容車量數目都變負值 這表示路口沒有剩餘容 亞皆老街 東行 車能力 及交通擠塞的情況將會惡化 從嘉齡道至天光道 西行 亞皆老道 北行 從皮特街至登打士街 南行 加士居道天橋 東行 1.3 > 彌敦道東側 西行 龍翔道 從獅子山隧道路至竹園路 界限街 從大坑東道至基堤道 太子道西 從基堤道至嘉道理道 44-2 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 界限街 / 窩打老道 1% -1% 2% 太子道西 / 窩打老道 1% -2% 2% 太子道西 / 嘉道理道 2% -1% 1% 亞皆老街 / 彌敦道 1% -2% 1% 漆咸道北 / 蕪湖街 -1% -2% 5% 第6頁

13 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 項 目 的 效 益 ( 一 ) 經 濟 效 益 2.37 中 九 龍 幹 線 主 要 的 經 濟 效 益 是 節 省 行 駛 時 間 和 路 程, 令 乘 客 直 接 受 惠, 可 以 享 受 更 直 接 和 快 捷 的 旅 程 往 返 東 西 九 龍 中 九 龍 的 交 通 同 樣 間 接 受 惠 ( 例 如 在 窩 打 老 道 行 駛 ), 擠 塞 路 線 如 亞 皆 老 街, 窩 打 老 道 及 佐 敦 道 的 車 輛 將 分 流 到 中 九 龍 幹 線, 令 行 駛 的 速 度 提 高, 並 因 此 減 少 行 車 時 間 雖 然 公 共 巴 士 所 節 省 行 車 時 間 相 對 較 小, 但 較 高 的 車 輛 佔 用 率 意 味 著 公 共 交 通 節 省 的 行 車 時 間 是 總 節 省 時 間 的 一 個 重 要 組 成 部 分 2.38 預 計 在 221 年, 在 早 上 繁 忙 時 間, 用 中 九 龍 幹 線 從 西 面 到 東 面 行 車 時 間 只 需 要 約 5 分 鐘, 與 沒 有 中 九 龍 幹 線 的 情 況 下 節 省 約 25 至 3 分 鐘 如 果 任 何 東 西 走 廊 被 堵 塞 引 致 嚴 重 交 通 擠 塞, 中 九 龍 幹 線 的 重 要 性 更 高 2.39 節 省 時 間 值 是 以 相 對 每 組 乘 客 的 節 省 時 間 ( 以 小 時 ) 為 單 位, 乘 以 該 組 時 間 價 值 (VOT) VOT 因 此 反 映 乘 客 為 節 省 一 小 時 的 願 付 價 格 對 每 組 旅 客 VOT 的 評 估 是 基 於 22 年 的 出 行 特 徵 調 查, 並 折 算 成 當 前 的 價 格 預 計 在 23 年, 節 省 的 行 車 時 間 將 會 達 到 每 日 12 萬 旅 客 小 時, 每 年 帶 來 的 經 濟 價 值 約 為 26 億 港 元 2.4 乘 客 從 地 面 道 路 分 流 到 隧 道 會 減 少 交 通 意 外 根 據 運 輸 署 提 供 的 交 通 意 外 統 計, 在 26 年 至 21 年 隧 道 的 每 公 里 車 輛 意 外 率 比 地 面 道 路 低 85% 圖 2-6 九 龍 主 要 路 口 剩 餘 容 車 量 2.41 基 於 為 期 48 年 的 預 算, 包 括 建 設 和 營 運 期 間, 與 假 設 的 實 際 折 扣 率, 中 九 龍 幹 線 項 目 預 計 產 生 億 港 元 的 淨 現 值,7.5% 的 内 部 收 益 率 和 1.8 的 效 益 成 本 比 率 分 析 可 見, 發 展 中 九 龍 幹 線 的 經 濟 效 益 足 以 支 付 開 支 及 成 本, 因 此 中 九 龍 幹 線 項 目 是 經 濟 可 行 的 ( 八 ) 結 論 2.33 現 時 來 往 九 龍 東 西 的 主 要 道 路 交 通 已 接 近 或 已 超 出 這 些 道 路 的 容 車 輛, 所 以 經 常 出 現 較 交 通 擠 塞 情 況 上 期 及 近 期 的 交 通 研 究 一 直 對 東 西 走 廊 的 交 通 需 求 預 測 增 加, 並 確 認 需 要 一 條 額 外 的 東 西 主 幹 線, 以 避 免 道 路 網 會 受 到 更 廣 泛 更 頻 繁 的 交 通 堵 塞, 甚 至 造 成 交 通 癱 瘓 年 的 交 通 預 測 也 確 定 雙 程 三 線 主 幹 路 及 其 連 接 道 路, 從 現 有 的 東 西 走 廊 和 區 内 道 路 網 提 供 足 夠 的 分 流, 疏 導 交 通 2.35 中 九 龍 幹 缐 將 連 接 九 龍 東 西 兩 端 的 高 速 公 路, 從 而 形 成 策 略 性 道 路 網 絡 的 一 個 重 要 組 成 部 分 中 九 龍 幹 線 會 紓 緩 現 有 東 西 主 幹 道 的 交 通 擠 塞 在 221 年, 用 中 九 龍 幹 線 來 往 西 九 龍 與 九 龍 灣 的 行 車 時 間 會 從 約 3-35 分 鐘 減 至 約 5 分 鐘 2.36 通 過 交 通 及 運 輸 研 究, 建 設 中 九 龍 幹 線 的 需 求 已 被 確 立 研 究 亦 顯 示, 建 設 中 九 龍 幹 線 是 有 當 前 迫 切 需 要 的 ( 二 ) 環 境 效 益 2.42 各 條 橫 跨 九 龍 半 島 的 市 區 幹 道 接 近 或 已 超 過 其 設 計 負 荷 能 力, 造 成 交 通 阻 塞 和 長 長 車 龍 中 九 龍 幹 線 項 目 的 主 要 目 標 是 紓 緩 目 前 東 西 幹 道 在 繁 忙 時 間 的 交 通 擠 塞 ; 這 些 幹 道 包 括 龍 翔 道, 界 限 街, 太 子 道 西, 亞 皆 老 街, 窩 打 老 道, 加 士 居 道 行 車 天 橋 及 漆 咸 道 北 2.43 中 九 龍 幹 線 將 現 有 連 接 東 西 九 龍 主 要 道 路 的 交 通 分 流, 減 少 地 面 道 路 的 車 輛 數 目, 從 而 紓 緩 交 通 擠 塞 所 造 成 的 污 染, 改 善 環 境 2.44 中 九 龍 幹 線 通 車 後, 這 些 連 接 東 西 九 龍 主 要 道 路 的 交 通 情 況 ( 例 如 平 均 行 駛 速 度 ) 將 能 得 到 改 善, 這 將 導 致 減 少 污 染 物, 如 二 氧 化 碳, 氮 氧 化 物 和 可 吸 入 懸 浮 粒 子 等 2.45 沿 中 九 龍 幹 線 而 建 的 三 座 通 風 大 樓, 通 過 採 用 先 進 的 技 術, 如 空 氣 淨 化 系 統 (APS), 靜 電 除 塵 器 (ESP) 和 二 氧 化 氮 去 除 系 統, 令 廢 氣 濃 度 大 幅 下 降 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 7 頁

14 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 2.46 連 接 東 西 九 龍 主 要 道 路 的 221 年 交 通 量 預 測 已 輸 入 環 保 署 最 新 的 EmFAC-HK V2.1 軟 件 模 型, 它 已 反 映 了 最 新 排 放 控 制 措 施 中 九 龍 幹 線 的 環 境 影 響 評 估 (EIA) 已 使 用 同 一 個 的 EmFAC 模 型 EmFAC-HK 模 型 會 考 慮 多 項 因 素, 包 括 交 通 流 量, 行 駛 速 度, 區 域 性 車 輛 - 公 里 - 行 駛 (VKT), 車 輛 組 合 等, 以 估 計 有 關 路 段 每 年 的 二 氧 化 碳, 氮 氧 化 物 和 可 吸 入 懸 浮 粒 子 的 總 排 放 在 下 面 的 表 2-4 中 總 結 表 2-4 預 計 的 年 排 放 量 為 的 主 要 的 東 西 走 廊 221 年 物 質 的 廢 氣 排 放 中 九 龍 幹 線 帶 來 的 削 減 [1], 噸 / 年 二 氧 化 碳 約 2, 氮 氧 化 物 約 18 可 吸 入 懸 浮 粒 子 約 2 [1] 只 考 慮 九 龍 區 主 要 東 西 交 通 走 廊 ( 三 ) 社 會 效 益 2.47 未 來 中 九 龍 幹 線 完 工 後, 行 車 時 間 將 會 縮 短, 提 高 區 域 間 的 連 接 及 支 持 社 會 發 展 結 論 2.48 因 爲 建 設 中 九 龍 幹 線 以 紓 緩 現 時 的 交 通 擠 塞 有 迫 切 需 要, 及 因 爲 中 九 龍 幹 線 可 以 改 善 環 境, 也 可 以 帶 來 經 濟 及 社 會 效 益, 證 明 建 造 中 九 龍 幹 線 是 有 淩 駕 性 公 衆 需 要 的 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 8 頁

15 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 3 沒 有 合 理 的 不 涉 及 填 海 的 替 代 方 案 引 言 3.1 這 部 分 會 了 解 不 用 在 九 龍 灣 臨 時 填 海 建 造 海 底 隧 道 的 可 能 性 我 們 先 會 考 慮 選 定 走 線 可 不 可 以 用 不 需 臨 時 填 海 的 其 他 合 理 方 案 建 造 海 底 隧 道, 然 後 再 考 慮 其 他 不 需 要 填 海 的 走 線 設 計 及 施 工 安 排 的 限 制 東 端 佈 局 3.2 如 圖 3-1 所 示, 鑽 挖 隧 道 會 從 油 痲 地 開 始, 經 過 京 士 柏, 何 文 田 及 馬 頭 圍, 在 地 面 以 下 約 4 米 至 14 米 深 的 岩 石 層 建 造, 以 確 保 中 九 龍 幹 線 的 建 設 和 經 營 不 會 與 大 廈 地 基 的 樁 柱 相 衝 突, 及 確 保 走 線 上 的 大 廈 的 正 常 用 途 由 於 中 九 龍 幹 線 會 與 九 龍 灣 及 啟 德 的 道 路 網 連 接, 豎 向 走 線 會 從 馬 頭 圍 道 東 面 以 4% 的 坡 度 爬 升 在 九 龍 灣 下 的 中 九 龍 幹 線 縱 截 面 如 圖 3-1 所 示 3.3 因 爲 4% 坡 度 爬 升 的 一 段 路 超 過 5 米 長, 此 斜 度 根 據 交 通 規 劃 及 設 計 手 冊 要 求, 隧 道 有 4 條 東 向 行 車 道 以 提 供 一 條 給 重 型 車 輛 用 的 爬 坡 車 道 此 外, 由 於 在 九 龍 灣 的 平 面 半 徑 為 33 米, 東 西 線 行 車 道, 需 要 加 闊 最 大 3 米 以 符 合 設 計 手 冊, 並 給 駕 駛 者 一 個 足 夠 的 安 全 視 距 3.4 施 工 方 案 是 根 據 幾 何 功 能 性 要 求 和 已 知 的 地 質 條 件 對 於 一 個 非 常 深 及 闊 的 ( 大 約 47 至 58 米 ) 海 底 隧 道 結 構 來 說, 設 計 必 定 十 分 複 雜 及 要 求 有 一 個 完 善 的 設 計 方 案 圖 3-1 東 端 佈 局 圖 九 龍 城 渡 輪 碼 頭 3.5 此 碼 頭 供 市 民 乘 坐 渡 輪 來 往 馬 頭 角 及 北 角 使 用 工 程 會 在 碼 頭 的 北 面 進 行 為 了 保 持 正 常 的 渡 輪 活 動, 將 會 實 施 合 理 的 安 排 馬 頭 角 公 衆 碼 頭 3.6 由 於 建 造 中 九 龍 幹 線 的 海 底 隧 道, 需 要 拆 卸 馬 頭 角 公 衆 碼 頭 工 程 期 間, 公 衆 碼 頭 會 在 景 雲 街 附 近 的 海 濱 走 廊 臨 時 重 置 工 程 完 工 後, 碼 頭 會 於 原 位 重 置 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 9 頁

16 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 香 港 中 華 煤 氣 有 限 公 司 (HKCG) 石 腦 油 碼 頭 3.7 這 碼 頭 正 在 被 用 來 接 收 海 運 而 來 的 石 腦 油, 然 後 將 之 轉 送 到 附 近 的 香 港 中 華 煤 氣 有 限 公 司 馬 頭 角 廠 房 據 香 港 中 華 煤 氣 有 限 公 司 表 示, 目 前 每 年 會 輸 送 三 十 六 次 的 海 運 石 腦 油 在 施 工 期 間, 為 了 繼 續 正 常 裝 運, 海 上 通 道 將 要 維 持 不 需 填 海 之 建 造 方 案 3.8 我 們 已 考 慮 下 面 兩 種 不 需 填 海 的 建 造 方 案 來 建 設 海 底 隧 道 : (a) (b) 沉 管 式 隧 道 (IMT); 及 隧 道 鑽 挖 機 (TBM) ( 一 ) 沉 管 式 隧 道 (IMT) 3.9 根 據 這 方 案, 一 個 22 米 濶 及 3 米 深 的 坑 道 會 跟 著 隧 道 走 線 在 海 床 疏 浚 海 泥 ( 見 圖 3-2) 隧 道 組 件 ( 大 約 47 米 至 58 米 濶 及 16.5 米 高 ) 會 於 場 外 預 製 然 後 浮 拖 到 隧 道 走 線 及 沉 落 坑 道 指 定 的 位 置 隧 道 連 接 後, 坑 道 會 回 填 到 現 有 的 海 床 沉 管 隧 道 典 型 剖 面 和 典 型 回 填 安 排 如 圖 3-3 及 圖 3-4 所 示 圖 3-2 工 程 影 響 的 設 施 圖 3-3 沉 管 式 隧 道 方 案 的 典 型 剖 面 圖 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 1 頁

17 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 圖 3-4 (a) 第 一 階 段 在 海 床 進 行 大 面 積 疏 浚 圖 3-4 (c) 第 三 階 段 沉 放 及 接 駁 預 製 組 件 和 回 填 鎖 定 圖 3-4 (b) 第 二 階 段 將 預 製 隧 道 組 件 浮 運 到 指 定 地 點 圖 3-4 (d) 第 四 階 段 回 填 把 海 床 恢 復 原 狀 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 11 頁

18 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 3.1 浮 拖 沉 管 式 隧 道 預 製 件 的 例 子 如 圖 3-5 所 示 3.12 現 有 的 馬 頭 角 及 啟 德 海 堤, 及 旁 邊 私 人 樓 宇 地 基 結 構 的 完 整 性 都 會 被 坑 道 影 響 ( 見 圖 3-7) 中 華 煤 氣 公 司 的 碼 頭 及 九 龍 城 碼 頭 也 會 需 要 在 建 造 期 間 暫 時 搬 遷 圖 3-5 沉 管 式 隧 道 的 例 子 建 造 深 槽 要 拆 除 以 下 結 構 Deep trench would require the demolition of: 中 華 煤 氣 公 司 運 載 原 材 料 碼 頭 HKCG Gas Station Jetty 馬 頭 角 公 眾 碼 頭 Ma Tau Kok Public Pier 九 龍 城 渡 輪 碼 頭 Kowloon City Ferry Pier 車 輛 渡 輪 碼 頭 ( 已 停 用 ) Vehicular Ferry Pier (Disused) 疏 浚 槽 太 寬 ( 大 約 22 米 ) Large Dredged Channel Width (approx. 22 m) 啟 德 KAI TAK 中 華 煤 氣 公 司 運 載 原 材 料 碼 頭 HKCG Gas Station Jetty 3.11 挖 掘 沉 管 隧 道 之 疏 浚 槽 將 需 要 移 走 和 處 理 約 75 萬 立 方 米 的 海 泥 另 外, 由 於 九 龍 灣 只 有 6 米 至 九 龍 城 渡 輪 碼 頭 Kowloon City Ferry Pier 8 米 深, 一 條 1,3 米 長,15 米 濶 及 12 米 深 的 引 道 需 在 海 床 挖 掘 運 送 隧 道 組 件, 從 而 造 成 總 數 約 18 萬 立 方 米 的 海 泥 需 疏 浚 和 處 理 ( 見 圖 3-6) 土 瓜 灣 TO KWA WAN 九 龍 灣 KOWLOON BAY 圖 3-7 疏 浚 槽 會 影 響 的 結 構 挖 掘 海 床 以 建 造 沉 管 式 隧 道 Dredging for constructing Immersed Tube Tunnel 3.13 由 於 需 要 放 置 沉 管 節 段 和 挖 掘 引 道 而 要 疏 浚 的 大 量 海 泥 及 對 海 堤, 私 人 樓 宇, 中 華 煤 氣 公 司 碼 頭, 及 渡 輪 碼 頭 等 產 生 影 響, 沉 管 式 隧 道 不 是 一 個 合 理 的 方 案 ( 二 ) 隧 道 鑽 挖 機 3.14 此 方 法 涉 及 以 隧 道 鑽 挖 機 ( TBM ) 鑽 挖 圓 形 隧 道, 穿 越 現 有 海 床 下 的 土 壤 鑽 挖 後 的 隧 道 表 面, 會 以 預 製 混 凝 土 隧 道 襯 砌 作 保 護 隧 道 鑽 挖 機 的 如 圖 3-8 所 示 挖 掘 海 床 以 建 造 引 道 ( 深 約 12 米 ) Dredging for approach channel (approx. 12m depth) 圖 3-6 因 沉 管 隧 道 所 需 的 疏 浚 槽 及 航 道 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 12 頁

19 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 圖 3-9 隧 道 鑽 挖 機 所 需 的 覆 土 圖 3-8 隧 道 鑽 挖 機 3.15 在 建 造 混 凝 土 襯 砌 之 前, 隧 道 內 需 施 用 3 千 帕 至 5 千 帕 的 氣 壓 ( 或 3 至 5 倍 的 大 氣 壓 ) 以 支 撐 隧 道 挖 掘 面 及 防 止 隧 道 滲 水 因 此, 需 要 有 足 夠 的 覆 土 以 抵 受 隧 道 內 壓 力 覆 土 厚 度 將 取 決 於 地 質 條 件, 考 慮 到 九 龍 灣 海 床 土 壤 的 低 強 度 性 ( 不 排 水 抗 剪 強 度 低 至 4 千 帕 ), 如 圖 3-9 所 示, 所 需 覆 土 厚 度 約 為 隧 道 直 徑 的 1.5 倍 東 行 隧 道 (3 條 行 車 線 及 1 條 爬 坡 線 ) 的 直 徑 約 為 2.5 米 西 行 隧 道 (3 條 行 車 線 ) 的 直 徑 約 為 17 米 所 需 覆 土 厚 度 於 是 分 別 為 3.75 米 及 25.5 米 3.16 因 隧 道 需 徐 徐 爬 升 至 地 面 以 連 接 九 龍 灣 和 啟 德 的 道 路 網, 如 圖 3-1 所 示, 西 端 的 最 厚 覆 土 僅 有 17 米, 東 端 更 僅 有 2 米, 均 少 於 1.5 倍 隧 道 直 徑 馬 頭 角 Ma Tau Kok 17 米 < 隧 道 鑽 挖 機 直 徑 (2.5 米 ) 的 1.5 倍 17m < 1.5 times Tunnel Dia. (2.5m) 馬 頭 角 海 堤 Seawall at Ma Tau Kok 九 龍 灣 Kowloon Bay 海 水 水 位 Sea Level 2 米 < 隧 道 鑽 挖 機 直 徑 (2.5 米 ) 的 1.5 倍 2m < 1.5 times Tunnel Dia. (2.5m) 啟 德 Kai Tak 啟 德 海 堤 Seawall at Kai Tak 路 面 水 平 Road level 隧 道 鑽 挖 機 直 徑 需 達 2.5 米 TBM Diameter shall reach 2.5m 泥 層 Soil 石 層 Rock 圖 3-1 九 龍 灣 可 用 覆 土 4% 坡 度 Gradient of 4% 3.17 可 用 覆 土 不 足 以 維 持 支 持 隧 道 挖 掘 面 和 防 止 隧 道 滲 水 所 需 的 氣 壓, 因 此 會 引 致 圖 3-11 所 示 的 爆 裂 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 13 頁

20 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 3.19 綜合以上原因 採用隧道挖掘機建造海底隧道將對施工人員和公眾構成危險 隧道挖掘機因而不 土層覆蓋不足的情況(一) Scenario 1 with insufficient overburden above the tunnel 沒有足夠的土層覆蓋 Insufficient overburden above the tunnel 是合理的替代方案 海床水平 Seabed level 使用填海的建造方法 3.2 基於沉管隧道和隧道挖掘機都不是合理的替代方案 我們考慮是否可用圖 3-13 所示的臨時填海 土層 Soil 配合明挖回填方法建造海底隧道 西行方向隧道 Westbound Tunnel 東行方向隧道 Eastbound Tunnel 隧道直徑需達2.5米 Tunnel diameter shall reach 2.5m 海平面 Sea level 隧道直徑需達17米 Tunnel diameter shall reach 17m 隧道內壓力太大破壞周邊地層 的穩定性 Too high compressed air pressure within tunnel causing blowout failure 圖 3-11 爆裂情況 3.18 另一方面 如降低氣壓 氣壓將不足以支撐挖掘段及防止地下水滲入 因此 如圖 3-12 所示 隧道亦會因挖掘面坍塌和過量地下滲流而發生事故 土層覆蓋不足的情況(二) Scenario 2 with insufficient overburden above the tunnel 海平面 Sea level 海床水平 Seabed level 沒有足夠的土層覆蓋 Insufficient overburden above the tunnel 圖 3-13 臨時填海配合明挖回填法 3.21 在香港 多數的樓宇 地鐵及基建項目都使用明挖回填的方法建造隧道的 建造海底隧道或連接 海底隧道的下沉式道路 使用臨時填海提供一個乾燥的工作平台及明挖回填隧道的建造方法亦已廣泛運 土層 Soil 用 並被認爲是最實用的方法之一 圖 3-14 展示了在臨時填海範圍內使用明挖回填方法建造隧道的例 子 西行方向隧道 Westbound Tunnel 東行方向隧道 Eastbound Tunnel 隧道直徑需達2.5米 Tunnel diameter shall reach 2.5m 隧道內壓力太小會引致地下 水湧入隧道 Insufficient compressed air pressure within tunnel causing water inflow to the tunnel 隧道直徑需達17米 Tunnel diameter shall reach 17m 圖 3-12 滲流情況 44-2 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 第 14 頁

21 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 圖 3-15 (a) - 明挖回填法配以臨時填海 - 第一步 圖 3-14 臨時填海配合明挖回填法的例子 3.22 採用此方法 臨時海堤將使用大直徑管樁沿海底隧道走線建造 臨時海堤圍起的空間內將進行填 海 形成工作平台 隔牆將於平台上建造以構成圍堰 圍堰內將展開挖掘工程以配合隧道結構的建造 臨時填海及海堤將在隧道完工後移除 海床也將恢復至原本水平 圖 3-15 展示了供參考的建造次序 3.23 用臨時填海形成工作平台之後 將安裝臨時擋土結構來做大型挖掘 擋土結構類型取決於挖掘深 度 水平面及地質條件 對於需要愈 2 米深度挖掘的海底隧道 高潮位超過水平標高正 2 米 海平面 及大約 9 米的海相沉積軟土 臨時擋土牆需能支撐來自附近土壤和海水的強大壓力 地下連續牆因其有 較大的抗土壓和水壓的能力 被認為是擋土牆的首選結構類型 地下連續牆需穿透足夠的深度亦或深入 圖 3-15 (b) - 明挖回填法配以臨時填海 - 第二步 石層 若石層不深 以提供抗傾覆穩定性和止水保護 3.24 連續牆安裝後 爲了提供乾燥工作平台 大型挖掘及側向支撐可進行 隧道建造完成後 現有的 海床平面及移除剩下的臨時填海 最後 海床會被重建回原來的樣子 圖 3-15 (c) - 明挖回填法配以臨時填海 - 第三步 44-2 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 第 15 頁

22 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 3.25 臨時填海將分兩階段進行 以維持進入中華煤氣公司石腦油碼頭的海上通道 及避免影響位於前 啟德機場跑道與馬頭角之間 排放進九龍灣的現有箱形雨水渠 第一階段於九龍灣海面近啟德發展區處 進行約 1.8 公頃的填海 以建造需時 26 個月 18 米長的隧道 海床於完工後將恢復至原有水平 第 二階段於九龍城輪渡碼頭對開海域進行約 2. 公頃的填海 以建造需時 26 個月 19 米長的隧道 海 床亦會恢復至原有水平 九龍城輪渡碼頭之乘客服務於施工期間將保持正常 此兩階段臨時填海的安排 詳見於圖 3-16 及圖 3-17 圖 3-17 九龍灣兩階段臨時填海的安排 3.26 綜合以上理由 使用明挖回填法配合臨時填海以建造海底隧道是可行的並是唯一安全可行的建造 圖 3-16 九龍灣兩階段臨時填海的安排 方法 替代走線 3.27 我們在 27 年至 29 年進行勘測及初步設計研究亦曾探討有否其他替代走線 包括走線 A C 至 E 及選取的走線 走線 B 見於圖 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 第 16 頁

23 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 圖 3-18 勘 測 及 初 步 設 計 階 段 的 替 代 走 線 方 案 3.28 走 線 A 是 全 陸 地 走 線, 不 涉 及 填 海 唯 走 線 A 有 一 段 約 6 米 長 的 隧 道 會 經 土 瓜 灣 木 廠 街 一 帶 的 數 十 幢 私 人 樓 宇 受 影 響 的 私 人 樓 宇 需 要 被 收 回 及 拆 卸, 以 進 行 工 程 走 線 A 並 不 是 合 理 的 替 代 走 線 3.29 雖 然 走 線 C 至 E 皆 穿 過 現 有 道 路 及 沿 岸 未 發 展 地 帶, 再 經 海 路 連 接 啟 德 交 匯 處, 但 亦 會 涉 及 收 回 及 拆 卸 私 人 樓 宇, 而 它 們 所 需 的 填 海 範 圍 比 走 線 B 較 大, 所 以 走 線 C 至 E 亦 非 合 理 替 代 走 線 唯 有 走 線 B, 即 現 時 選 取 的 走 線, 只 需 臨 時 使 用 現 有 的 九 龍 城 碼 頭 公 共 交 通 交 匯 處 進 行 建 造 工 程, 但 不 涉 及 收 回 及 拆 卸 私 人 樓 宇, 而 且 所 需 填 海 範 圍 亦 最 小 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 17 頁

24 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 4 最 少 範 圍 的 臨 時 填 海 引 言 4.1 本 章 探 討 建 造 九 龍 灣 海 底 隧 道 所 需 臨 時 填 海 範 圍 的 因 素 並 確 立 最 小 範 圍 臨 時 填 海 的 方 案 填 海 的 長 度 4.2 如 下 圖 4-1 所 示, 海 底 隧 道 的 軸 線 為 一 條 半 徑 約 33 米 的 曲 線 這 是 位 於 九 龍 城 輪 渡 碼 頭 公 共 交 通 交 匯 處 與 啟 德 河 之 間 以 最 小 半 徑 連 接 的 最 短 反 轉 曲 線 海 底 隧 道 的 長 度 即 臨 時 填 海 的 長 度 亦 為 最 小 圖 4-2 視 距 要 求 4.4 如 下 圖 4-3 所 示, 臨 時 填 海 的 寬 度 由 87 米 至 98 米 不 等 並 由 以 下 構 成 : (a) 由 連 續 牆 構 成 的 圍 堰 寬 度 由 47 米 至 58 米 不 等 ; 及 (b) 連 續 牆 兩 邊 各 2 米 寬 的 工 作 平 台 圖 4-1 填 海 的 長 度 填 海 的 寬 度 4.3 隧 道 寬 度 受 約 33 米 的 曲 線 半 徑 控 制 如 圖 4-2 所 示, 這 是 符 合 交 通 規 劃 及 設 計 手 冊 第 二 卷 節 視 距 要 求 的 最 小 半 徑, 以 提 供 確 保 交 通 安 全 的 合 適 視 野 圖 4-3 臨 時 填 海 典 型 剖 面 4.5 海 底 隧 道 的 寬 度 及 臨 時 填 海 的 寬 度 都 已 為 所 需 最 小 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 18 頁

25 合約編號 CE 43/21 (HY) 中九龍幹線 - 設計及施工 九龍灣臨時填海工程具有力和令人信服的資料更新 路政署 4.6 對於 4.4(a)段落 第三章 3.25 段已解釋 隧道將分兩階段建造 (a) 第一階段於九龍灣海面近啟德發展區處進行約 1.8 公頃的填海 以建造需時 26 個月 約 18 米長的隧道 海床於完工後將恢復至原有水平 及 (b) 第二階段於九龍城輪渡碼頭對開海域進行約 2. 公頃的填海 以建造需時 26 個月 約 19 米長的隧道 海床亦會恢復至原有水平 4.7 對於第一階段需建的隧道 因要維持啟德發展區的海堤下的附近的最少水深 通風槽被安排在隧 道兩邊 此段長約 58 米 對於第二階段需建的隧道 因有足夠淨高 通風槽被安排在隧道上方以減少 填海的面積 此段因此長約 47 米 4.8 對於 4.4(b)段 需以工作平台提供建造機械(起重機 自動傾斜卡車 挖土機等)所需的操作空間 利用駁船輸送材料的裝卸區域 施工車輛的活動 根據以相同方法建造海底隧道的中環-灣仔繞道項目 的建造經驗 2 米的寬度剛好滿足這些用途 因此 擬臨時填海的寬度為最小的 圖 4-4 及 4-5 釋 臨時填海的時間 4.9 臨時填海的總週期約為 52 個月 石樁 海堤 地下連續牆 開挖及隧道箱涵的建造各需 5 至 6 圖 4-4 中環-灣仔繞道的臨時填海 個月完成 填海 回填 移除填料和恢復原狀各需 1 至 2.5 個月完成 因一些工作可同步進行 每一階 段需時約 26 個月完成 最少填海要求的總結 4.1 已就海底隧道的建造 填海 海堤 受影響設施的重置等工程要求作詳細檢討 以準確確立所需 臨時填海的最小範圍 4.11 兩個階段的填海各持續時間不會超過 26 個月 計劃於 215 年初至 219 年中/底進行 4.12 結論是 臨時填海面積已是最小的 臨時填海總面積不超過 3.8 公頃 任何時間的最大填海面積 不超過 2. 公頃 圖 4-5 中環-灣仔繞道的臨時填海 44-2 修正終稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\九龍灣臨時填海工程具有力和令人信服的資料.DOCX 第 19 頁

26 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 5 公 眾 諮 詢 前 階 段 的 公 眾 參 與 活 動 5.1 路 政 署 在 27 年 展 開 中 九 龍 幹 線 的 勘 測 及 初 步 設 計 工 作 由 於 工 程 項 目 會 對 交 通 土 地 運 用 及 環 境 有 重 大 影 響, 我 們 探 取 全 面 的 公 眾 參 與 策 略, 透 過 公 眾 論 壇 焦 點 小 組 會 議 外 展 活 動 採 訪 問 卷 調 查 實 地 考 察 及 規 劃 比 賽, 收 集 公 眾 對 各 主 要 範 疇 的 意 見 我 們 透 過 這 些 活 動 與 各 持 份 者 包 括 居 民 業 主 委 員 會 社 區 組 織 商 戶 及 區 議 會 建 立 良 好 的 溝 通 我 們 亦 就 中 九 龍 幹 線 項 目 諮 詢 油 尖 旺 九 龍 城 和 觀 塘 區 議 會, 以 及 立 法 會 交 通 事 務 委 員 會 5.2 勘 測 研 究 的 其 中 一 個 主 要 課 題 是 選 出 一 條 合 適 的 走 線 我 們 檢 討 了 過 往 在 中 九 龍 幹 線 研 究 過 的 4 多 個 方 案, 並 擬 定 出 14 個 新 的 方 案 我 們 比 較 各 方 案 對 社 區 設 施 重 置 環 境 土 地 及 交 通 方 面 的 影 響, 並 參 考 公 眾 與 過 程 中 所 收 集 到 的 意 見, 選 取 現 時 的 走 線 走 線 已 獲 立 法 會 油 尖 旺 九 龍 城 和 觀 塘 區 議 會 的 普 遍 支 持 5.3 透 過 公 眾 參 與 活 動, 我 們 了 解 到 公 眾 普 遍 認 為 需 要 建 造 中 九 龍 幹 線, 並 希 望 中 九 龍 幹 線 能 有 敁 接 駁 東 九 龍 和 西 九 龍 的 道 路 網, 以 紓 緩 交 通 擠 塞 同 時, 應 妥 善 保 育 油 麻 地 警 署 廟 街 夜 市 天 后 廟 及 榕 樹 頭 文 化 29 年 7 月 18 日 公 眾 論 壇 5.4 通 過 長 期 與 提 案 中 的 中 九 龍 幹 線 東 部 的 居 民 討 論 有 關 臨 時 性 填 海 的 事 宜, 最 終 對 提 案 中 的 臨 時 填 海 的 意 見 達 成 基 本 共 識, 丌 反 對 臨 時 填 海, 但 是 只 有 在 沒 有 合 理 的 備 選 方 案 的 時 候 才 可 以 選 擇 填 海, 並 在 期 間 需 要 考 慮 環 境 因 素 5.5 一 些 成 員 認 為, 其 實 市 民 的 願 望 是 可 以 有 小 規 模 永 久 九 龍 灣 填 海 工 程 以 提 供 新 的 社 區 設 施, 如 公 園, 圖 書 館, 休 憩 處 及 一 條 連 續 的 散 步 徑, 連 接 馬 頭 角 啟 德 機 場 發 展 區 少 數 公 衆 更 表 達 了 在 九 龍 灣 地 區 進 行 小 規 模 永 久 性 填 海 可 以 為 九 龍 灣 提 供 減 少 臭 味 問 題 的 機 會 5.6 設 計 小 組 已 經 收 到 了 公 衆 提 出 關 於 小 規 模 永 久 性 填 海 的 提 案, 但 是 根 據 對 相 關 條 例 的 理 解, 此 類 提 案 只 有 在 通 過 高 等 法 院 根 據 保 護 海 港 條 例 第 三 條 設 定 的 三 項 測 試 後 才 可 被 考 慮 由 於 中 九 龍 幹 線 的 提 案 中 沒 有 涉 及 永 久 性 填 海, 此 提 案 應 該 被 列 爲 單 獨 考 慮 項 目 29 年 6 月 2 日 焦 點 小 組 會 議 5.7 此 項 工 程 的 歷 叱 在 專 題 小 組 會 議 中 已 經 介 紹 給 各 個 出 席 者 丌 同 走 線 的 供 選 方 案 以 及 此 工 程 將 會 對 公 衆 設 施 及 環 境 的 影 響 已 在 會 議 中 討 論 過 5.9 會 議 提 出 馬 頭 圍 公 衆 碼 頭 的 拆 除 以 及 其 它 附 屬 的 碼 頭 重 置 方 案 將 會 為 加 強 海 港 前 線 形 象 及 新 置 九 龍 灣 公 衆 設 施 帶 來 機 會 公 衆 參 與 活 動 結 論 5.1 就 中 九 龍 幹 線 優 選 走 線 及 提 案 的 工 程, 我 們 已 經 吐 公 衆 成 員, 油 尖 旺 區 議 會, 九 龍 城 區 議 會, 觀 塘 區 議 會 以 及 立 法 會 運 輸 委 員 會 進 行 過 諮 詢 活 動 5.11 總 體 情 況 上, 大 部 分 的 公 衆 以 及 各 團 體 代 表 都 支 持 此 建 議 的 走 線 方 案 並 對 臨 時 填 海 提 案 沒 有 顯 著 反 對 現 階 段 的 公 眾 參 與 活 動 5.12 中 九 龍 幹 線 第 二 期 公 眾 參 與 活 動 於 212 年 12 月 5 日 正 式 展 開, 就 中 九 龍 幹 線 的 詳 細 設 計 及 施 工 安 排 收 集 公 眾 意 見 5.13 在 為 期 三 個 月 的 公 眾 參 與 活 動 期 間, 路 政 署 舉 行 了 丌 同 類 型 的 活 動, 如 焦 點 小 組 會 議 諮 詢 會 和 簡 介 會 包 括 與 幹 線 沿 途 居 民 專 業 學 會 環 保 團 體 玉 器 市 場 等 的 焦 點 小 組 會 議 和 諮 詢 會 並 吐 丌 同 的 區 議 會 和 海 濱 事 務 委 員 會 簡 介 和 諮 詢 他 們 的 意 見 5.14 由 於 隧 道 需 要 於 九 龍 灣 段 以 臨 時 填 海 方 式 建 造, 項 目 統 籌 機 構 必 須 按 保 護 海 港 條 例 的 要 求 提 出 其 有 力 和 令 人 信 服 的 資 料, 證 明 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要 就 此, 路 政 署 舉 辦 了 丌 同 類 型 的 公 眾 參 與 活 動, 包 括 一 個 海 濱 事 務 委 員 會 諮 詢 會 九 龍 灣 臨 時 填 海 專 業 論 壇 及 九 龍 灣 臨 時 填 海 專 題 討 論 的 公 眾 論 壇 5.15 為 方 便 公 眾 參 與 討 論, 第 二 期 公 眾 參 與 摘 要 分 別 以 中 文 及 英 文 出 版 此 外, 第 26 及 27 期 中 九 龍 幹 線 通 訊 則 以 中 英 文 出 版 及 廣 泛 派 發, 有 助 公 眾 了 解 中 九 龍 幹 線 項 目 及 項 目 進 度 摘 要 及 通 訊 詳 細 解 說 中 九 龍 幹 線 的 主 要 關 注 事 項, 包 括 中 九 龍 幹 線 的 敁 益 設 計 綠 化 及 景 觀 文 物 保 育 重 置 公 共 設 施 環 境 影 響 施 工 安 排 以 及 九 龍 灣 段 臨 時 填 海 的 需 要 公 眾 參 與 活 動 亦 包 括 巡 迴 展 覽, 於 油 麻 地 觀 塘 九 龍 城 及 何 文 田 地 區 展 出 隧 道 模 型 及 虛 擬 場 景 模 型 5.16 為 推 廣 第 二 期 公 眾 參 與 活 動, 路 政 署 吐 社 區 丌 同 的 持 份 者, 包 括 沿 線 及 就 近 幹 線 的 居 民 學 校 社 區 組 織 商 業 大 廈 商 戶 區 議 會 地 區 組 織 專 業 學 會 及 環 保 團 體, 發 出 約 5, 封 邀 請 信, 鼓 勵 各 界 人 士 踴 躍 參 與 第 二 期 公 眾 參 與 活 動 路 政 署 亦 於 兩 份 中 文 報 刋 ( 東 方 日 報 和 頭 條 日 報 ) 及 一 份 英 文 報 刋 (The Standard) 刋 登 廣 告, 於 213 年 1 月 4 日,11 日,18 日,25 日 及 2 月 1 日 刋 出, 大 力 宣 傳 及 鼓 勵 公 眾 參 加 公 眾 論 壇, 希 望 丌 同 背 景 的 持 份 者 就 中 九 龍 幹 線 的 詳 細 設 計 提 供 意 見 及 作 出 討 論 5.8 出 席 單 位 包 括 保 護 海 港 協 會, 海 港 之 友, 香 港 工 程 師 學 會, 香 港 建 築 師 學 會, 香 港 測 量 師 學 會 以 及 九 龍 城 區 議 會 的 成 員 44-2 修 正 終 稿 213 年 2 月 G:\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 2 頁

27 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 5.17 表 5-1 展 示 所 有 第 二 期 的 公 眾 參 與 活 動 表 5-1 第 二 期 的 公 眾 參 與 活 動 日 期 公 眾 參 與 活 動 212 年 12 月 11 日 諮 詢 會 ( 駿 發 花 園 居 民 ) 212 年 12 月 12 日 焦 點 小 組 會 議 ( 京 士 柏 ) 212 年 12 月 13 日 諮 詢 會 ( 油 尖 旺 區 議 會 ) 212 年 12 月 13 日 焦 點 小 組 會 議 ( 馬 坑 涌 ) 212 年 12 月 14 日 焦 點 小 組 會 議 ( 馬 頭 角 ) 212 年 12 月 15 日 焦 點 小 組 會 議 ( 佐 敦 西 ) 212 年 12 月 15 日 焦 點 小 組 會 議 ( 油 麻 地 ) 212 年 12 月 16 日 諮 詢 會 ( 翔 龍 灣 居 民 ) 212 年 12 月 17 日 焦 點 小 組 會 議 ( 樂 文 ) 212 年 12 月 18 日 焦 點 小 組 會 議 ( 海 心 ) 212 年 12 月 19 日 焦 點 小 組 會 議 ( 常 樂 ) 212 年 12 月 19 日 焦 點 小 組 會 議 ( 愛 俊 ) 212 年 12 月 2 日 焦 點 小 組 會 議 ( 愛 民 ) 213 年 1 月 2 日 焦 點 小 組 會 議 ( 玉 器 市 場 商 販 ) 213 年 1 月 4 日 焦 點 小 組 會 議 ( 環 保 團 體 ) 213 年 1 月 7 日 簡 介 會 ( 海 濱 事 務 委 員 會 ) 213 年 1 月 8 日 簡 介 會 ( 觀 塘 區 議 會 ) 213 年 1 月 8 日 簡 介 會 ( 黃 大 仙 區 議 會 ) 213 年 1 月 1 日 九 龍 灣 臨 時 填 海 專 業 論 壇 213 年 1 月 12 日 公 眾 論 壇 ( 油 尖 旺 區 議 會 ) 213 年 1 月 17 日 簡 介 會 ( 油 尖 旺 區 議 會 交 通 及 運 輸 委 員 會 ) 213 年 1 月 17 日 簡 介 會 ( 九 龍 城 區 議 會 ) 213 年 1 月 18 日 諮 詢 會 ( 偉 恆 昌 新 邨 ) 213 年 1 月 19 日 公 眾 論 壇 ( 九 龍 城 區 議 會 ) 213 年 1 月 22 日 諮 詢 會 ( 香 港 建 築 師 學 會 ) 213 年 1 月 26 日 諮 詢 會 ( 京 士 柏 居 民 ) 213 年 2 月 1 日 簡 介 會 ( 油 尖 旺 區 議 會 社 區 建 設 委 員 會 ) 213 年 2 月 2 日 九 龍 灣 臨 時 填 海 專 題 討 論 213 年 2 月 4 日 諮 詢 會 ( 香 港 規 劃 師 學 會 香 港 城 市 設 計 學 會 及 香 港 園 境 師 學 會 ) 現 時 有 關 中 九 龍 幹 線 的 公 眾 意 見 5.18 以 下 部 份 概 述 了 公 眾 及 不 同 持 份 者 就 有 關 九 龍 灣 臨 時 填 海 的 公 眾 參 與 活 動 中 發 表 的 重 點 意 見 5.19 公 眾 和 當 區 居 民 普 遍 認 為 中 九 龍 幹 線 能 有 效 接 駁 東 九 龍 及 西 九 龍 的 道 路 網 絡, 以 紓 緩 交 通 擠 塞 有 個 別 環 保 團 體 及 公 眾 卻 指 出, 興 建 更 多 道 路 設 施 並 不 能 有 效 地 改 善 交 通 擠 塞 問 題, 他 們 認 為 加 強 交 通 需 求 管 理, 才 是 解 決 交 通 擠 塞 的 方 法 然 而 這 意 見 亦 經 獨 立 運 輸 專 家 檢 閱, 他 指 出 香 港 私 家 車 擁 有 率 比 較 低, 而 且 道 路 使 用 者 主 要 為 商 用 車 輛, 所 以 有 需 要 建 造 中 九 龍 幹 線 5.2 大 多 數 參 加 者, 包 括 當 區 居 民 和 公 眾, 就 中 九 龍 幹 線 九 龍 灣 臨 時 填 海 建 議 沒 有 強 烈 意 見 一 些 居 民 更 建 議 把 臨 時 填 海 轉 為 永 久 填 海, 從 而 有 助 解 決 九 龍 灣 海 岸 的 臭 味 問 題 及 改 善 與 啟 德 地 區 的 行 人 接 駁 諮 詢 區 議 會 5.21 油 尖 旺 九 龍 城 黃 大 仙 及 觀 塘 區 議 會 均 認 同 中 九 龍 幹 線 能 有 效 接 駁 東 九 龍 及 西 九 龍 的 道 路 網, 以 紓 緩 交 通 擠 塞 他 們 促 請 政 府 盡 快 落 實 中 九 龍 幹 線 的 工 程 有 不 少 區 議 員 表 示, 中 九 龍 幹 線 通 車 後 會 有 助 減 少 於 油 麻 地 何 文 田 及 九 龍 城 的 路 面 行 車 量, 從 而 有 效 改 善 這 些 地 區 的 交 通 情 況 及 環 境 5.22 有 些 區 議 員 表 達 支 持 臨 時 填 海 的 建 議 及 認 同 顧 問 公 司 就 填 海 徵 詢 專 業 人 士 意 見 的 做 法 有 九 龍 城 區 區 議 員 進 一 步 建 議, 研 究 合 理 的 方 法, 有 助 控 制 九 龍 灣 海 岸 的 臭 味 問 題 諮 詢 海 濱 事 務 委 員 會 5.23 路 政 署 和 顧 問 公 司 向 海 濱 事 務 委 員 會 匯 報 其 有 力 和 令 人 信 服 的 資 料 以 證 明 九 龍 灣 段 的 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要, 以 及 解 釋 中 九 龍 幹 線 的 詳 細 設 計 及 施 工 安 排 5.24 海 濱 事 務 委 員 會 大 致 認 同 項 目 的 策 略 性 需 要 及 其 走 線, 並 且 認 可 因 建 造 中 九 龍 幹 線 而 需 進 行 臨 時 填 海 工 程 的 需 要 九 龍 灣 臨 時 填 海 專 業 論 壇 5.25 由 於 項 目 工 程 需 要 在 九 龍 灣 段 以 臨 時 填 海 方 式 建 造, 項 目 統 籌 機 構 須 要 按 保 護 海 港 條 例 的 要 求, 證 明 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要 就 此, 顧 問 公 司 於 專 業 論 壇 向 有 關 的 專 業 及 學 界 人 士, 匯 報 其 有 力 和 令 人 信 服 的 資 料 以 支 持 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要 匯 報 完 結 後, 路 政 署 邀 請 了 兩 名 獨 立 專 家 檢 閱 顧 問 公 司 所 提 供 的 資 料 5.26 專 業 及 學 界 人 士 對 顧 問 公 司 及 兩 名 獨 立 專 家 所 提 供 就 中 九 龍 幹 線 的 凌 駕 性 公 眾 需 要 的 理 據, 並 沒 有 異 議 5.27 專 業 及 學 界 人 士 普 遍 認 為 沒 有 其 他 合 理 的 建 造 方 式 及 其 他 幹 線 走 線 能 夠 取 代 填 海 工 程, 他 們 認 為 隧 道 鑽 挖 機 及 沉 管 式 隧 道 建 造 方 法 均 不 是 替 代 建 造 中 九 龍 幹 線 九 龍 灣 段 海 底 隧 道 的 合 理 建 造 方 案 5.28 專 業 及 學 界 人 士 普 遍 認 為 臨 時 填 海 工 程 的 建 議 比 其 他 建 造 方 式 對 海 床 造 成 較 少 損 害 及 滋 擾 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 21 頁

28 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 九 龍 灣 臨 時 填 海 專 題 討 論 5.29 按 保 護 海 港 條 例 的 要 求, 證 明 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要, 在 九 龍 灣 臨 時 填 海 專 題 討 論 的 公 眾 論 壇 上, 就 著 (1) 臨 時 填 海 有 沒 有 凌 駕 性 的 公 眾 需 要 (2) 沒 有 其 他 合 理 替 代 填 海 的 方 案 和 (3) 所 需 的 填 海 範 圍 是 最 小 顧 問 公 司 就 匯 報 其 有 力 和 令 人 信 服 的 資 料 以 支 持 臨 時 填 海 有 凌 駕 性 的 公 眾 需 要 匯 報 完 結 後, 路 政 署 邀 請 了 兩 名 獨 立 專 家 審 閱 顧 問 公 司 所 提 供 的 資 料 5.3 公 眾 人 士 同 意 解 决 交 通 擠 塞 的 迫 切 需 要 而 建 造 中 九 龍 幹 線 (1) 有 凌 駕 性 的 公 眾 需 要 ; 公 眾 人 士 並 同 意 (2) 沒 有 其 他 安 全 可 靠 的 合 理 替 代 填 海 的 方 案 和 (3) 現 時 建 議 所 需 的 填 海 範 圍 亦 是 最 小 的 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 22 頁

29 路 政 署 合 約 編 號 CE 43/21 (HY) 中 九 龍 幹 線 - 設 計 及 施 工 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料 更 新 6 結 論 有 關 臨 時 填 海 的 凌 駕 性 公 眾 需 要 6.1 大 多 數 參 加 者, 包 括 油 尖 旺 九 龍 城 及 觀 塘 區 議 會 幹 線 沿 途 居 民 和 公 眾, 認 同 中 九 龍 幹 線 能 有 效 接 駁 東 九 龍 及 西 九 龍 的 道 路 網 絡, 以 紓 緩 交 通 擠 塞 他 們 促 請 政 府 盡 快 落 實 中 九 龍 幹 線 的 工 程 海 濱 事 務 委 員 會 亦 表 示 認 同 興 建 中 九 龍 幹 線 有 策 略 性 需 要 不 同 的 專 業 學 會 亦 表 示 不 會 質 疑 興 建 中 九 龍 幹 線 的 需 要 6.2 多 個 區 議 員 表 示, 中 九 龍 幹 線 通 車 後 會 有 助 減 少 於 油 麻 地 何 文 田 及 九 龍 城 的 道 路 行 車 量, 從 而 有 效 改 善 當 區 的 交 通 情 況 及 環 境 6.3 在 九 龍 灣 臨 時 填 海 專 題 討 論 的 公 眾 論 壇 上, 公 眾 人 士 同 意 解 决 交 通 擠 塞 的 迫 切 需 要 而 建 造 中 九 龍 幹 線 有 凌 駕 性 的 公 眾 需 要 6.4 因 此, 公 眾 認 為 以 臨 時 填 海 方 式 建 造 中 九 龍 幹 線 是 有 凌 駕 性 的 公 眾 需 要 沒 有 合 理 而 不 涉 及 填 海 工 程 的 建 造 方 式 6.5 中 九 龍 幹 線 臨 時 填 海 的 需 要 性 是 由 顧 問 公 司 及 二 名 專 業 人 士 提 出, 而 海 濱 事 務 委 員 會 表 示 不 會 質 疑 臨 時 填 海 的 需 要 性, 包 括 沒 有 其 他 走 線 及 合 理 的 建 造 方 式 代 替 填 海 工 程 的 理 由 6.6 專 業 及 學 界 人 士 於 專 業 論 壇 都 認 同 沒 有 其 他 合 理 的 建 造 方 式 及 其 他 幹 線 走 線 能 夠 取 代 填 海 工 程, 他 們 認 為 隧 道 鑽 挖 機 及 沉 管 式 隧 道 建 造 方 法 均 不 是 合 理 的 替 代 建 造 中 九 龍 幹 線 九 龍 灣 段 的 海 底 隧 道 的 建 造 方 案 6.7 在 九 龍 灣 臨 時 填 海 專 題 討 論 的 公 眾 論 壇 上, 公 眾 人 士 同 意 沒 有 其 他 安 全 可 靠 的 合 理 取 代 填 海 的 方 案 6.8 雖 然 有 個 別 書 面 意 見 建 議 把 中 九 龍 幹 線 的 走 線 向 北 移 來 避 免 填 海 工 程, 但 公 眾 大 致 認 為 沒 有 其 他 合 理 的 建 造 方 式 取 代 填 海 工 程 最 小 的 填 海 範 圍 6.9 中 九 龍 幹 線 臨 時 填 海 的 需 要 性 是 由 顧 問 公 司 及 兩 位 專 業 人 士 提 出, 而 海 濱 事 務 委 員 會 表 示 不 會 質 疑 臨 時 填 海 的 需 要 性, 包 括 臨 時 填 海 會 對 海 岸 造 成 的 少 量 損 害 第 二 期 公 眾 參 與 活 動 6.13 我 們 在 212 年 12 月 初 展 開 中 九 龍 幹 線 的 第 二 期 公 眾 參 與 活 動, 以 收 集 公 眾 對 中 九 龍 幹 線 的 詳 細 設 計 及 施 工 安 排 的 意 見 到 目 前, 我 們 已 按 計 劃 舉 行 過 了 一 連 串 的 公 眾 參 與 活 動, 包 括 為 沿 線 居 民 環 保 團 體 專 業 學 會 及 其 他 持 分 者 舉 行 十 多 場 的 焦 點 小 組 會 議, 以 及 於 油 麻 地 何 文 田 土 瓜 灣 和 觀 塘 不 同 地 點 進 行 五 次 巡 迴 展 覽 6.14 我 們 亦 已 諮 詢 油 尖 旺 九 龍 城 黄 大 仙 及 觀 塘 的 區 議 會, 和 海 濱 事 務 委 員 會 此 外, 我 們 亦 在 213 年 1 月 12 日 及 1 月 19 日 分 別 在 油 尖 旺 區 及 九 龍 城 區 舉 行 兩 場 公 眾 論 壇, 廣 泛 探 討 中 九 龍 幹 線 項 目 我 們 藉 此 機 會 感 謝 各 界 人 士 積 極 參 與 並 提 供 寶 貴 意 見 獨 立 專 家 審 查 6.15 因 應 保 護 海 港 條 例 及 為 進 行 與 此 報 告 相 關 的 公 眾 參 與 活 動, 我 們 聘 請 了 兩 位 獨 立 專 家 審 評 員, 他 們 需 要 : a) 提 供 嚴 謹 的 審 查 報 告, 考 慮 中 九 龍 幹 線 在 九 龍 灣 填 海 的 必 要 性 b) 確 認 提 議 中 臨 時 填 海 的 論 據 是 有 力 的 和 令 人 信 服 的 c) 確 定 項 目 通 過 凌 駕 性 公 眾 需 要 測 試 的 要 求 d) 確 認 並 無 其 他 切 實 可 行 性 選 擇 替 代 的 填 海 工 程 e) 確 認 擬 填 海 的 範 圍 是 最 小 的 6.16 獨 立 專 家 的 審 查 請 參 閱 附 錄 一 和 附 錄 二 符 合 保 護 海 港 條 例 6.17 總 括 而 言, 保 護 海 港 條 例 中 關 於 禁 止 填 海 的 三 項 測 試 準 則 均 已 通 過 : 為 配 合 中 九 龍 幹 線 的 建 造 及 符 合 公 眾 對 此 道 路 網 絡 的 凌 駕 性 需 要, 在 九 龍 灣 填 海 有 當 前 迫 切 的 需 要 所 有 填 海 都 是 臨 時 的 且 在 項 目 完 成 後 會 被 移 除, 海 床 也 將 恢 復 至 原 有 水 平 沒 有 除 臨 時 填 海 以 外 的 合 理 性 替 代 方 案 可 用 於 建 造 中 九 龍 幹 線 九 龍 灣 段 的 海 底 隧 道 填 海 範 圍 為 最 小 所 需 範 圍 6.1 在 專 業 論 壇 期 間, 專 業 及 學 界 人 士 都 認 為 臨 時 填 海 會 對 海 岸 造 成 少 量 損 害 6.11 在 九 龍 灣 臨 時 填 海 專 題 討 論 的 公 眾 論 壇 上, 公 眾 人 士 同 意 現 時 建 議 所 需 的 填 海 範 圍 是 最 小 的 6.12 因 此, 路 政 署 就 最 小 填 海 範 圍 已 與 公 眾 達 成 共 識 44-2 修 正 終 稿 213 年 2 月 \\HKGNTS19\CIVIL\+CURRENT JOBS\ CENTRAL KOWLOON ROUTE\2 PROJECT ADMINISTRATION\FILING\4.3 OUTGOING REPORTS\REP-44-2 REVISED FINAL CCM REPORT (REF. 795)_MAR 213\CHINESE\RAW\ 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 資 料.DOCX 第 23 頁

30

31 附 錄 一 林 興 强 教 授 - 關 於 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 材 料 報 告 之 獨 立 專 家 審 查

32

33 PolyU Technology & Consultancy Company Limited 1. PolyU Technology & Consultancy Company Ltd ( Hy(S)Q/62/212) : P , (HyD) (PTeC) ( Hy(S)Q/62/212) (CKR) 2. ( 3.9 ) 8 3 T2 - T ( : REP-81-) ( : HMW 1/21 (TT)) ( REP-77-1) Arup-Mott MacDonald William H.K. LAM 18 February ( ) QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 1

34 PolyU Technology & Consultancy Company Limited PolyU Technology & Consultancy Company Limited - ( ) : / ; / ; / / ; / / ; / ; / ; / / ; / ; / ; / ; / ; (MegaBox )/ ; / ; / QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page ( ) ( ) 5 1 : ( 6 ); S221/A 6 (Route 6) ; /C (CKR) - (TKO-LTT) T2 ; S226/A 6 ; S226/G 6 1 : 5 S216 S221/A S221/C S226/A S226G 6 T2 ( /, /, / / ) (R.C.) V/C RC v/c 1.. RCs RCs (6 RC < -15%). v/c 1.. RCs ( RC < -15%). v/c v/c V/C = ; RC = 18 RCs ( S221A RC ). RCs (3 RC < -15%). QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 3

35 PolyU Technology & Consultancy Company Limited PolyU Technology & Consultancy Company Limited 3.4 1% 3% ( ) / - ( REP-77-1) % 29% 7%,,, ( ) : ( REP-77-1) S S (1), T2 6 (2) 1 (3) (4) 15 (S ) QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 4 QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 5

36 PolyU Technology & Consultancy Company Limited PolyU Technology & Consultancy Company Limited (a) 8 7% Rmin = V 2 /2.822 (7%) V = 8 / ; Rmin = V 2 /2.822 (7%) = 324 < S 4.2 S ( S ) 8 / S 33 4% 9 S S (S) (b), (TPDM) (a) QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 6 QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 7

37 PolyU Technology & Consultancy Company Limited PolyU Technology & Consultancy Company Limited 8 /, (S) (R) s 8RT (3.65/2) = R=33 T=6.525 S 131 (S) / (c) 4% 1% ( CTS-3) ( ) ( ) 211, ( A-A) A-A (LOS) ( ) 221 D F S QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 8 QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 9

38 PolyU Technology & Consultancy Company Limited PolyU Technology & Consultancy Company Limited 4 (S ) S ( 8 /, 33 4%) F, (JTIS) (ERP), 211 ( A-A) 6% (,,, ) 3 (1) V/C.9 1% (2) (3) S S ( ) ********** END ********** 5.2 QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 1 QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 11

39 PolyU Technology & Consultancy Company Limited Prepared by Ir Professor William H.K. Lam BSc, MSc, PhD, CEng Chair Professor of Civil & Transportation Engineering Department of Civil and Environmental Engineering The Hong Kong Polytechnic University Yuk Choi Road Hung Hom, Kowloon HONG KONG Tel : (852) Fax : (852) [email protected] Web Site: QR 6 QR63 QR63, 6/F, QR Core, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. [email protected] Tel: (852) / Fax: (852) Page 12

40

41 附 錄 二 吳 宏 偉 教 授 - 關 於 九 龍 灣 臨 時 填 海 工 程 具 有 力 和 令 人 信 服 的 材 料 報 告 之 獨 立 專 家 審 查

42

43 a. Ng, C. W. W., Rigby, D., Lei, G. & Ng, S. W.L. (1999) b. Ng, C. W. W., Rigby, D., Ng, S. W.L & Lei, G. (2) c. Ng, C. W. W., Lu, H. & Peng, S.Y. (212)... d. Wong, K. S., Ng, C. W. W., Chen, Y.M. & Bian, X.C. (212). [J]

44 (Docklands)

45 (Docklands) 1. ICE (1998). Bulkhead location blamed for DLR blast. New Civil Engineer, Institute of Civil Engineers, February Issue, ICE (24). Docklands tunnel blowout down to elementary error, says judge. New Civil Engineer, Institute of Civil Engineers, January Issue, Ng, C.W.W., Rigby, D., Lei, G. & Ng, S. W.L. (1999). Observed performance of a short diaphragm wall panel. Géotechnique. Vol. 49, No.5, Ng, C.W.W., Rigby, D., Ng, S. W.L & Lei, G. (2). Field studies of well-instrumented barrette in Hong Kong. Journal of Geotechnical and Geo-environmental Engineering, ASCE. Vol. 126, No. 1, Ng, C.W.W. (29). Independent Expert Review Report: Review of Construction Method for the Central Kowloon Route. Highways Department. 6. Ng, C. W. W. & Wong, K.S. (212). Investigation of passive failure and deformation mechanisms due to tunnelling in clay in centrifuge. Canadian Geotechnical Journal. Accepted. 7. Ng, C.W.W., Lu, H. & Peng, S.Y. (212). Three-dimensional centrifuge modelling of twin tunnelling effects on an existing pile. Tunnelling and Underground Space Technology. In Press Wong, K. S., Ng, C. W. W., Chen, Y.M. & Bian, X.C. (212). Centrifuge and numerical investigation of passive failure of tunnel face in sand. Tunnelling and Underground Space Technology. Vol. 28,

46 EPB is generally applicable for clay to clayey sand where permeability is relatively low Fig. 1 Independent study -Ground conditions at Kowloon Bay (Ng et al, 2) ( ) Fig. 3 Grading curve related to the choice of EPB shield Fig. 2 Poor ground conditions at Kowloon Bay (Ng et al, 2) ( : -, ) Fig. 4 Tunnel Boring Machine ( )

47 17m < 1.5 times Tunnel Dia. 楔柕奺 Ma Tau Kok 楔柕奺㴟 Seawall at Ma Tau Kok ၡय़Нѳ Road level ᒾၰᢕ ᐒ ޔ ৩ሡ 2.5 TBM Diameter shall reach 2.5m ḅ漵䀋 Kowloon Bay 㴟㯜㯜ỵ Sea Level 2m < 1.5 times Tunnel Dia. Kai Tak 㴟 Seawall at Kai Tak ύΐᓪ ጕᒾၰ CKR Tunnel 㲍Ⰼ Soil 4% ࡋڵ Gradient of 4% 䞛Ⰼ Rock Bored tunnel method is not feasible 䪭᥆ ᐕᱟнਟ㹼Ⲵ Fig. 5 Proposed Bored Tunnel Construction Method for CKR 䪭᥆䳗䚃 ᐕ Ṹ The crater is 22 m wide and 7 m deep ൠ䶒 䲧ሜ 22 ˈ 7 ICE (1998) Ø Diameter = 5.2 m (䳗䚃ⴤᗁ 5.2 ) Ø Slurry shield boring machine using compressed air ( Ώ ᐕ ) Possible causes of failure ਟ㜭Ⲵ ഐ Ø Insufficient overburden above the tunnel (䳗䚃к㾶 㠚䟽н䏣) Ø High compressed air pressure (2.2 bar) within tunnel causing blowout failure (䳗䚃 ᆀ䶒 䙾བྷ) Fig. 6 Blowout Failure of Docklands Light Rail (extension) in UK on 23 Feb (1998 ᒤ 2 ᴸ 23 ᰕ 㤡 九 (Docklands)䕅䓼Ⲵ䳗䚃䮻᥆䶒⒗ ) Massive crater due to compressed air blast ഐ 㑞オ ᕅ䎧Ⲵབྷ䶒ぽൠ䶒 䲧 Fig. 7 Crater formed in the ground of George Green school (ICE, 1998 & 24) (George Green ᆖṑൠ䶒кᖒᡀⲴ 䲧)

48 z (mm) Pressure( ) Tunnel Face ( ) (a) (Wong et al., 212; Ng & Wong, 212) Proposed by Soubra (22) Medium dense sand C/D = 2.2 S x /D =.3 N gm = 73 a. Ng, C. W. W., Rigby, D., Lei, G. & Ng, S. W.L. (1999) b. Ng, C. W. W., Rigby, D., Ng, S. W.L & Lei, G. (2) c. Ng, C. W. W., Lu, H. & Peng, S.Y. (212)... d. Wong, K. S., Ng, C. W. W., Chen, Y.M. & Bian, X.C. (212). [J] x (mm) (b) Fig. 8(a) Passive Failure (blowout) of Tunnel Face ( ( ) ); (b) Measured displacement vectors leading to blowout 1. Ng, C. W. W. & Wong, K.S. (212). Investigation of passive failure and deformation mechanisms due to tunnelling in clay in centrifuge. Canadian Geotechnical Journal. Accepted. 2. Wong, K. S., Ng, C. W. W., Chen, Y.M. & Bian, X.C. (212). Centrifuge and numerical investigation of passive failure of tunnel face in sand. Tunnelling and Underground Space Technology. Vol. 28,

49 Ng, C. W. W., Rigby, D. B., Lei, G. H. & Ng, S. W. L. (1999). GeÂotechnique 49, No. 5, 681±694 Observed performance of a short diaphragm wall panel C. W. W. NG, D. B. RIGBY, G. H. LEI and S. W. L. NG The construction in Hong Kong of a 4 m deep excavated, large, rectangular-section barrette (i.e. a short diaphragm wall panel, 2 8 m long by 8 m wide) in sedimentary and weathered soils under bentonite has been heavily instrumented and closely monitored. During excavation, the maximum measured horizontal ground movements were only a few millimetres, with similar subsurface settlements around the panel. On the basis of three-dimensional numerical simulations of the excavation of the trench, an average mobilized shear strain greater than 1% around the excavated trench can be deduced. At the soil±wall interface, the initial lateral earth pressures decreased to hydrostatic bentonite pressures during excavation and increased above their initial K pressures after concreting. The measured lateral pressures just after concreting support a theoretical bilinear pressure envelope. KEYWORDS: deformation; diaphragm and in situ walls; earth pressure; numerical modelling; pore pressures. INTRODUCTION The use of slurry trenches to construct diaphragm walls for underground structures has become a well-known technique in civil engineering. The construction of the diaphragm walls will inevitably cause initial stress changes and deformations in the ground (Clough & O'Rourke, 199; Farmer & Attewell, 1973; Ng, 1992; Stroud & Sweeney, 1977; Symons & Carder, 1993). Various techniques, such as numerical modelling (Ng, 1992; Ng et al., 1995) and centrifuge modelling (Powrie & Kantartzi, 1996), have been attempted to investigate stress transfer mechanisms and ground deformations due to diaphragm walling. Field monitoring of different panel sizes in different ground conditions is vital to provide essential data for verifying numerical and centrifuge results. In the Far East, well-documented case histories of Manuscript received 24 September 1998; revised manuscript accepted 16 February Discussion on this paper closes 3 April 2; for further details see p. ii. Hong Kong University of Science and Technology. La construction d'une barrette aá grande section rectangulaire excaveâe aá 4 m de profondeur (c'est-aá-dire d'un court panneau de mur souterrain de 2,8 m de long par,8 m de large) dans des sols seâdimentaires et deâsagreâgeâs sous de la bentonite a fait l'objet de nombreux controãles aux instruments et a eâteâ observeâe de preás aá Hong Kong. Pendant l'excavation, les mouvements horizontaux maximum du sol qui ont eâteâ mesureâs n'ont eâteâ que de quelques millimeátres, avec des tassements similaires sous la surface autour du panneau. En nous basant sur des simulations numeâriques en trois dimensions de l'excavation de la trancheâe, nous avons pu deâduire une deâformation de cisaillement mobiliseâe moyenne supeârieure aá,1% autour de la trancheâe excaveâe. AÁ l'interface sol-mur, les pressions terrestres lateârales initiales baissent pour arriver au niveau des pressions de bentonite hydrostatiques pendant l'excavation et elles passent audessus de leurs pressions K o initiales apreás le beâtonnage. Les pressions lateârales mesureâes juste apreás le beâtonnage soutiennent la theâorie d'une enveloppe de pression bilineâaire. diaphragm walling are rarely reported in the literature, with one exception. Stroud & Sweeney (1977) carried out a detailed eld trial of a diaphragm wall panel, 6 1 m long by 1 2 m wide and about 36 m deep, constructed at Chater Road on Hong Kong Island. Maximum horizontal subsurface movements of 28 mm and 1 mm were recorded at 1 m and 2 m, respectively, away from the face of the trench at about 16 m below ground level. A settlement trough with a maximum value of 6 mm was observed 3 m away from the face of the trench. Recently the authors have had the opportunity to measure ground deformations and stress changes during the construction of a large, excavated, rectangular-section pile (barrette) for a University-led and industry-supported research project (Shen et al., 1998). The excavation and concreting procedures used for the barrette were identical to those used in the construction of a diaphragm wall panel. The barrette was tested for its ultimate vertical load capacity three weeks after construction. However, in this paper, only the performance of the short diaphragm wall panel during construction is 682 NG, RIGBY, LEI AND NG documented as a case history. The observed ground deformations during the excavation of the trench of the diaphragm wall panel are compared with a three-dimensional numerical analysis. Moreover, stress and pore water changes at the soil/wall interface during concreting are reported and discussed. SITE LOCATION AND GROUND CONDITIONS The site is located on the Kowloon peninsula of Hong Kong, to the east of a runway of the old Kai Tak International Airport, adjacent to the Public Works Central Laboratory at Kowloon Bay (Fig. 1). Figs 2 and 3 show a cross-section of the test barrette and the location of some relevant boreholes, together with uncorrected N values from standard penetration tests (SPTs) measured in each soil stratum. The site is on marine reclaimed land and the ground level is at approximately 4 5 m above sea level or Principal Datum (PD). The groundwater level is at about 3 m below the ground surface. The ground conditions consist of, in succession, approximately 6 m of ll material, 1 m of marine deposits, 12 m of alluvium of Quaternary age and 12 m of weathered granitic saprolites overlying granitic rocks of Upper Jurassic to Lower Cretaceous age (Strange, 199). The ground succession is similar to the site at Chater Road (Stroud & Sweeney, 1977). It can be seen from Fig. 2 that very scattered SPT N values were obtained in both the alluvium sand and the weathered granite. It should be noted that the idealized geological strata shown in the gure may only be applicable to a local area around the trench. From borehole information, the depth to the rock head has been found to increase quite signi cantly from the south-east to the north-west direction at the site. The initial horizontal stresses in the ground are not known for certain at Kowloon Bay. However, it is generally believed that the initial K values are less than 5 for soils in Hong Kong (Geotechnical Engineering Of ce, 1993). DETAILS OF CONSTRUCTION The test barrette or diaphragm wall trench was excavated using a traditional cable-operated grab. The size of the excavated trench was 2 8 m by 8 m in plan and 39 7 m deep (Figs 2 and 3). During construction, the trench was temporarily supported by bentonite slurry with a unit weight (ã b ) of 1:8 kn=m 3. Soil spoil, suspended in the bentonite slurry, was removed after pumping to a desanding unit at the ground surface. After desanding, the bentonite was recharged into the trench. Chiselling of the base took place when the excavated depth reached about 39 6 m below ground level. This caused a small overbreak at the base which was detected by a sonic pro ling system. When the excavation reached its nal level (39 7 m below ground level), three instrumented reinforcement cages were lowered and spliced together one by one into the trench, which was then concreted at an average rate of 1:32 m=h (or 23:12 m 3 =h) using a tremie pipe. The ordinary Portland cement (OPC) concrete used was grade C3/2 (design f cu ˆ 3 MPa) with a unit weight (ã c ) of 23:2 kn=m 3. It had a water-to-cement ratio of 47 and an average slump of 18 mm. The average uniaxial compressive strength measured from a core taken from the centre of the barrette was 37 5 MPa at 28 days. During concreting, the average temperature measured inside the trench was 27 68C. The excavation and concreting procedures of the barrette were in fact identical to those of the construction of a typical diaphragm wall panel. At the upper reinforcement cages, a sheathing zone was formed (see Fig. 2) with the intention of minimizing the skin friction developed between the pile and the upper layers of the surrounding soil during axial load testing. INSTRUMENTATION The test barrette at Kowloon Bay was heavily instrumented. The prime objectives of the instrumentation were to study ground deformations due to the construction of the barrette (the diaphragm wall panel) and, more importantly, to investigate the load transfer mechanism and load±settlement characteristics of the barrette during axial load testing. In this paper, only the instrumentation related to the construction of the barrette is reported. Details of other instrumentation and the research strategy are described by Shen et al. (1998). Figs 3 and 4 show the locations of various instruments both in and around the trench. Four Geokon vibrating-wire-total-earth-pressure cells, together with four vibrating-wire piezometers, were installed at the soil±wall interface at four elevations. Two total-earth-pressure cells were installed in both the alluvium and the weathered granite layers to measure total horizontal pressures at the soil±wall interface. The locations of the vibrating-wire piezometers were at about 8 mm above the corresponding pressure cells. In addition, one pneumatic piezometer was installed inside borehole BF-4 (see Fig. 4) at 35 m below ground level in the weathered granite to monitor pore water pressure changes during the construction and testing of the barrette. Magnetic extensometers were installed into three boreholes (see Fig. 4). The datum magnets were set into the rock, except in BF-3, owing to some construction dif culties encountered. The extensometers allowed subsurface soil movements to be measured by monitoring the location of each magnetic target with respect to the datum magnet. 681

50 Sea wall OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL Road Cheung Yip Street Bus depot Open storage 4.7 Bus depot Vehicle examination centre Kai Hing Road Seawall Kowloon Bay factory estate Tai Yip street Shun Yip St 4.3 Hoi Bun Road 684 NG, RIGBY, LEI AND NG Sheathing zone m 2. 8 m Soft base Soil profile Ground level Medium dense, fine to coarse sand with gravel and cobblesized quartz fragments (fill) 6. m bgl Soft, silty clay with traces of sand and shell fragments (marine deposits) m bgl Firm to stiff, silty clay with traces of sand and fine gravel (alluvium clay) 22. m bgl Medium dense, fine to coarse sand with fine gravel (alluvium sand) m bgl Extremely weak to strong, completely to slightly decomposed, medium-grained granite (weathered granite) m bgl m bgl (Rock) SPT N values BH-5 BT-1 BT-2 BF-3 BF-5 Fig. 2. Borehole logs and SPT N values at Kowloon Bay, Hong Kong (bgl, below ground level) Grass 4. 2 Fence Cargo working area Wall Public Works Central Laboratory The site Old Kai Tak Airport runway Fig. 1. Location plan of test pile site in Kowloon, Hong Kong Kwun Tong Typhoon Shelter Pier Three conventional inclinometer systems were installed in three boreholes, as shown in Figs 3 and 4. The bottoms of the inclinometers were xed in rock. The inclinometers were used to measure rotations and hence the lateral movement of the soil around the trial barrette. In addition, settlement markers were installed around the trench, as shown in Fig. 3. The settlement monitoring system consisted of a steel rod. The bottom of the rod was concreted in a hole of 1 5 m depth below ground level. The settlement markers were used to measure surface settlements of the ground due to the construction of the diaphragm wall panel. GROUND DEFORMATIONS DURING CONSTRUCTION OF THE BARRETTE The excavation of the trench started on the morning of 5 December 1997 and was completed at midnight 7 December During excavation, the level of bentonite was kept at about 1 5 m above the ground water table (i.e. at about 3 m PD). The trench was then concreted in the afternoon of 9 December The horizontal deformation pro les measured are illustrated in Fig. 5. The measurements from the conventional inclinometers are of poor quality. This was probably because the actual movements were too small, only about 2 5 mm maximum horizontal ground movement towards the trench was observed at the ground surface in BF-5 (4 m away from the trench). The resolution of the inclinometers ( 1 mm deviation per metre or a maximum deviation of 4 mm over the 4 m trench) was simply not good enough to differentiate any small movements induced on site. The measured horizontal displacements generally decrease with depth, except for the localized large displacements appearing at about 32 m below ground level. This localized large ground deformation could be due to overbreak and loss of ground during chiselling at the base of the trench. Generally, small lateral deformations were observed and these may be attributed to limited stress relief during excavation (i.e. the initial horizontal stresses in the ground were not very different from the bentonite pressure) and the signi cant effect of soil arching (Ng, 1992; Ng et al., 1995) around the relatively short panel length. Owing to the limited accuracy of the conventional inclinometers, no clear trend can be identi ed among the measured values from the three boreholes. In the measured pro les at Chater Road (Stroud & Sweeney, 1977) and at Kowloon Bay, more signi cant deep-seated deformation was observed at the former than at the latter site. This was probably due to substantial soil yielding at deeper levels (large deviatoric shear stress) for the longer panel (6 1 m) at Charter Road. Deviatoric shear stress due to the difference between the vertical and horizontal effective stresses would be induced in the soil around trenches during excavation. On the other hand, soil arching around the shorter panel (2 8 m) seems to minimize the ground deformation at Kowloon Bay. Details of soil arching and stress transfer mechanisms due to the construction of a diaphragm wall panel are discussed by Ng et al. (1995).

51 OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL NG, RIGBY, LEI AND NG N Working area 2. 8 m A 2. m 1. 7 m. 8 m 1. 5 m. 8 m BF-6 BF-5 BF-7 BF-3 BF-4 The trench M M ME6-1 M M ME5-1 ME3-1 M M M M Ground level Fill 6. m bgl SM13 SM14 SM15 SM16 SM17. 8 m SM9 SM1 SM11 BF-3 BF-4 SM5 BF-7 SM6 BF-5 SM7 BF-6 SM12 SM8. 8 m. 8m 1. 2 m 3. m 1. 4 m 1. 4 m 5. 8 m A SM1 SM2 SM3 SM4 3. m m m m. 8 m 1. 5 m m 1. 1 m. 4 m 1. 5 m 9. m 2. m M M M M M M ME6-8 DM6 P M M M M M M M ME5-8 (Rock) DM5 (Rock) ME3-2 Earth pressure cell and piezometer Pneumatic piezometer Spider magnet Datum magnet M M M M M M M M M M M M M M M M P m bgl PC2 P2 35. m bgl (Rock) PC4 P4 PC1 P1 PC3 P3 PC1, P1 PC2, P2 PC3, P3 PC4, P4 Fig. 4. Typical schematic cross-section A±A showing layout of instrumentation (not to scale) Marine deposits m bgl Alluvium clay 22. m bgl Alluvium sand m bgl Weathered granite m bgl m bgl (Rock) Levels m bgl m bgl m bgl m bgl Pressure cell and piezometer Magnetic extensometer Inclinometer Pneumatic piezometer Magnetic extensometer and inclinometer Settlement marker Lateral displacement: mm Lateral displacement: mm Lateral displacement: mm Fig. 3. Plan view of locations for instrumentation During concreting, the ground surface was pushed, on average, about 1 mm outward laterally away at 4 m and 6 m from the trench by the wetconcrete pressure (see Fig. 5). It should be noted that the theoretical concrete pressure acting on the soil face was higher than the initial horizontal stresses in the ground. At 2 3 m from the trench (BF-7), the observed ground deformation is somewhat unexpected as the measurements indicate that there was a signi cant lateral movement (about 11 mm) away from the trench at 11 m below ground level ( 6 5 m PD). Con rmation of the reliability of this substantial lateral movement can be obtained by considering the subsurface vertical movements measured by the extensometer installed in BF-3 (1 5 m away from the trench). Figure 6 shows the subsurface vertical movements recorded by the extensometer in BF-3. During the excavation, the measurements showed an increase in settlement at all levels as excavation continued, with a maximum settlement of 2 mm at ME3-5. After concreting, a small recovery of ground settlements was recorded by the uppermost two spider magnets, except at ME3-5, which recorded a substantial upward movement (heave) of 7 5 mm in the soft marine deposit. ME3-5 is 9 m below ground level, just above the spike of ground movement measured by the inclinometer in BF-7. The consistency between the inclinometer and magnetic extensometer readings suggests that these large local deformations in the two orthogonal directions adjacent to the trench might be due to the presence of a weak soil layer or an overbreak and loss of ground during excavation. All other spider magnets in the same borehole show an increase in settlement during concreting and that the settlement ceased after concreting. Depth below ground level: m BF-6 3 (6. m away from trench) 33 End of excavation 36 2 days after concreting BF-5 3 (4. m away from trench) 33 End of excavation 36 2 days after concreting Fig. 5. Ground deformation pro les at different stages of barrette construction (positive lateral displacement: towards the trench; best resolution 1 mm deviation=1 m depth) BF-7 3 (2. 3 m away from trench) 33 End of excavation 36 2 days after concreting

52 Vertical movement: mm Vertical movement: mm Vertical movement: mm Vertical movement: mm OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL 687 Date Date 1 Dec 5 Dec 9 Dec 13 Dec 17 Dec 3 Dec 7 Dec 11 Dec 15 Dec 1 Dec 5 Dec 9 Dec 13 Dec 17 Dec 3 Dec 7 Dec 11 Dec 15 Dec ME3-1, 1 m 4 ME3-1, 19 m 688 NG, RIGBY, LEI AND NG Date Date 1 Dec 5 Dec 9 Dec 13 Dec 17 Dec 3 Dec 7 Dec 11 Dec 15 Dec 1 Dec 5 Dec 9 Dec 13 Dec 17 Dec 3 Dec 7 Dec 11 Dec 15 Dec ME5-1, 2 m 4 ME6-1, 1 m 2 4 ME3-2, 3 m 2 4 ME3-11, 21 m 2 4 ME5-2, 6 m 2 4 ME6-2, 4 m 2 4 ME3-3, 5 m 2 4 ME3-12, 23 m 2 4 ME5-3, 1 m 2 4 ME6-3, 9 m ME3-4, 7 m ME3-5, 9 m ME3-6, 11 m ME3-7, 13 m ME3-8, 15 m ME3-9, 17 m Figure 7 summarizes the subsurface vertical movements measured by the upper magnets installed in both BF-5 and BF-6. The excavation of the trench caused maximum downward movements ME3-13, 25 m ME3-14, 27 m ME3-15, 29 m ME3-16, 31 m ME3-17, 33 m ME3-18, 35 m ME3-19, 37 m Excavation Concreting Excavation Concreting Fig. 6. Variations of subsurface vertical movement with time at BF-3 (1 5 m away from trench) (measured by magnetic extensometer: positive, settlement; negative, heave) of 1 5 mm and 2 5 mm, at 2 m depth in BF-5 and at 1 m depth BF-6, respectively. Generally, the deeper the magnet, the smaller the settlement recorded. Recovery of ground loss due to concret- 2 4 ME5-4, 14 m ing resulted in a substantial reversal of settlement in each borehole. This recovery of ground loss indicates that the lateral stress in the ground could not possibly be larger than the concrete pressure during construction. Low initial lateral stress in the ground can thus be deduced. Surface ground settlements were monitored using ordinary levelling techniques. Even allowing for the typical accuracy of 1 mm of the levelling instrument, the observed settlements at various distances from the trench are unexpectedly very small. The maximum settlement measured was 1 mm during the excavation of the trench. During concreting, no recovery of ground loss can be identi ed. CONTACT PRESSURES AT THE SOIL±WALL INTERFACE Four vibrating-wire-total-earth-pressure cells and piezometers were attached to appropriate locations of the reinforcement cage during the construction of the barrette. Once in position, the instruments were jacked horizontally from the ground surface to ensure contact with the surrounding soil. The prime objective of installing the piezometers was to record changes of pore water pressures during the proposed vertical-load test of the barrette a few weeks after construction. In order to prevent the ceramic tips of the piezometers from clogging with bentonite, the ceramic tips were plugged with solid soap. This was expected to take a few days to 2 4 ME6-4, 11 m Excavation Concreting Excavation Concreting Fig. 7. Variations of subsurface vertical movement with time at BF-5 and BF-6 (measured by magnetic extensometer) dissolve before allowing the piezometers to function properly after concreting. Initial readings of the piezometers and earth pressure cells were taken before the instruments were jacked into position. This allowed a comparison to be made with the calculated bentonite pressures, using a measured unit weight (ã b ) of 1:8 kn=m 3. There was very good agreement between the calculated bentonite pressures and the readings recorded by the vibrating-wire earth pressure cells (Fig. 8). The maximum difference between the calculated and measured pressures was less than 1 kpa. Some minor adjustments were made to the calibration factors. Some piezometers showed higher pressures than the calculated values (Fig. 9). This was possible because the solid soap used to plug the ceramic tips of the piezometers had little time to dissolve. After taking initial readings, the earth pressure cells and piezometers were bedded in by jacking them out horizontally against the excavated soil surface until cell readings equal to the assumed initial K (ˆ 5) earth pressure were achieved. The jacking pressure was reduced once the concrete level went a few metres above the location of each pressure cell. Since the initial horizontal stresses in the ground are not known for certain at Kowloon Bay, an initial K value, equal to (1 sin ö9), was assumed for each soil layer at Kowloon Bay, and the calculated lateral earth pressures are shown in Fig. 8. Low initial K values in ll and decom-

53 OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL 689 Lateral pressure: kpa NG, RIGBY, LEI AND NG Pore water pressure: kpa h c Depth below ground level: m PC1 PC2 Depth below ground level: m P1 P2 35 PC3 PC4 35 P3 P Assumed initial lateral earth pressure Just after concreting 8th day after concreting Fluid concrete pressure Bentonite stage 1st day after concreting Hydrostatic bentonite pressure Predicted bilinear pressure envelope Bentonite stage 1st day after concreting Hydrostatic water pressure Fluid concrete pressure Just after concreting 8th day after concreting Hydrostatic bentonite pressure Fig. 8. Variations of lateral pressure distribution with depth Fig. 9. Variations of pore water pressure distribution with depth posed granite have recently been veri ed by backanalysing a deep excavation using the non-linear Simpson's brick model (Malone et al., 1997). It was thus expected that there could be a stress reduction during the excavation of the barrette. For equilibrium, the initial total earth pressure had to reduce to the bentonite pressure. The reduction of the total initial stress in the ground due to excavation resulted in mainly horizontal ground movements. Swelling and softening of the soils followed and hence led to a decrease in shear strength and stiffness. On the basis of eld observations and theoretical considerations, Ng (1992) and Lings et al. (1994) proposed a theoretical bilinear pressure envelope for predicting lateral pressures developed at the soil±wall interface during concreting in a diaphragm wall panel. The theoretical bilinear equation derived is as follows: ó h ˆ ãcz z < h c (1) (ã c ã b )h c ã b z z. h c where ó h, z and h c are the total lateral pressure, the depth below the top of the panel and the critical depth, respectively. According to the guidelines given in CIRIA Report 18 (Clear & Harrison, 1985), the value of the critical depth is mainly governed by the type of cement used, the rate of concreting, the temperature and the size and shape of the trench. The critical depth calculated for the panel at Kowloon Bay is 6 3 m and the predicted bilinear envelope is shown in Fig. 8. The measured values just after concreting agree well with the theoretical line predicted by the equation, except at PC1. It is clear that the full uid concrete pressure did not develop over the full depth of the wall during concreting. The horizontal total pressures increased to some values higher than the assumed initial K pressures, except at PC1, and they pushed the surrounding soil away from the trench, as shown in Fig. 5. Measurements of lateral earth pressures 1 and 8 days after concreting are also shown in Fig. 8. There were substantial increases in lateral pressure at PC3 and PC4 in the weathered granite, but slight decreases at PC1 and PC2 in the alluvium sand. The increase of pressure in the weathered granite is somewhat unusual and might be due to swelling of the soil and stress redistribution. Figure 9 shows the observed pore water pressures at the soil±wall interface. As discussed previously, the measured values at the bentonite stage do not fall on the hydrostatic bentonite pressure line, probably because the ceramic tips were plugged with soap initially. Readings taken during and one day after concreting show that excess pore water pressures were generated because of concreting. Dissipation of the excess pore water pressure seemed to be complete 8 days after concreting. The pore water pressures at the soil±wall interface resemble hydrostatic conditions. Figure 1 shows the measured pore water pressures 1 5 m away from the face of the trench in BF-4 (at 35 m below ground level). The piezometer recorded an increase of 7 kpa above an average pore water pressure of 311 kpa during concreting. The excess pore water pressure largely dissipated

54 Pore water pressure: kpa OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL Oct 1 Nov 6 Nov 11 Nov 16 Nov 21 Nov 26 Nov 1 Dec 6 Dec 11 Dec 16 Dec 21 Oct within 2 days after concreting, indicating a high permeability of the weathered granite. The rate of pore pressure dissipation is consistent with the case history reported by Stroud & Sweeney (1977). THREE-DIMENSIONAL MODELLING OF THE TRENCH EXCAVATION Soil model, parameters and modelling procedures To model the three-dimensional installation processes of the diaphragm wall panel construction, the nite-difference program FLAC3D (Itasca, 1996) was adopted. Since the three-dimensional soil±structure interaction of diaphragm walling is rather complex, it was decided to choose a relatively simple soil model for ease of interpretation of the computed results. All the soils at Kowloon Bay have been modelled as simple linear elastic and perfectly plastic isotropic materials with a Mohr±Coulomb yield surface. The elastic shear modulus adopted for each soil layer was based on the measured velocity of shear waves (geophysical method) at the same site, from which the maximum shear stiffness at very small strains was determined (Ng et al., 1999). The model parameters are summarized in Fig. 11. To estimate the shear stiffness at moderate shear strains ( 1%) for each soil layer, eld-measured shear moduli of the weathered granite obtained from a self-boring pressuremeter (SBPM) were compared with the measured values obtained by the geophysical method. It was found that the elastic moduli of the weathered granite at very small strains obtained by the geophysical method were approximately three times higher than the stiffness at moderate strains measured by the Date Excavation Average Concreting Fig. 1. Variations of pore water pressure with time at BF-4 (1 5 m away from trench) (measured by pneumatic piezometer at 35 mbgl) SBPM (Ng et al., 1999). An assumption was thus made that the shear moduli at moderate shear strains in other soil strata would also be three times smaller than the elastic moduli obtained using the geophysical method. In this paper, two threedimensional analyses are described: one uses the small-strain elastic moduli obtained by the geophysical method; the other adopts the deduced shear moduli at moderate strains. The shear strength parameters are taken from published data available in the literature (Cowland & Thorley, 1984). For comparing the maximum observed ground deformations, only the excavation of the diaphragm wall trench was simulated. This was done by removing model soil elements inside the trench and applying a normal hydrostatic bentonite pressure on the trench faces simultaneously. Comparisons between measurements and computed results Figures 12 and 13 compare the measured and computed horizontal deformation pro les and vertical subsurface movements, respectively at various distances from the trench. It can be seen that the accuracy of the magnetic extensometers was generally better than that of the inclinometers. If the elastic moduli obtained from the geophysical method are used, the numerical simulation signi cantly underpredicts the actual ground deformations, especially for the vertical subsurface movements. This implies that the actual ground response substantially deviated from the elastic behaviour assumed in the three-dimensional analysis. Substantial plastic strains seem to have been induced during the excavation of the trench. 692 NG, RIGBY, LEI AND NG Depth: m m Fill 6 m 15 m 22 m m Varies Water table Marine deposits Alluvium I Alluvium II Weathered granite γ 18 kn/m 3 c kpa φ 28 K o. 53 γ 18 kn/m 3 c kpa φ 28 K o. 53 γ 2 kn/m 3 c kpa φ 33 K o. 46 γ 2 kn/m 3 c kpa φ 33 K o. 46 γ 2 kn/m 3 c kpa φ 39 K o. 35 Bedrock (Assumed rigid boundary) Fig. 11. Summary of soil parameters Lateral displacement: mm BF-7 (2. 3 m from face) Depth: m Depth Lateral displacement: mm BF-5 (4. m from face) Shear modulus at. 1% strain: MPa Lateral displacement: mm Fig. 12. Comparison between measured and computed horizontal deformation pro les after excavation (positive lateral displacements towards the trench) Depth: m BF-6 (6. m from face) Computed (full stiffness) Computed (1/3 stiffness) Measured

55 Depth below ground level: m Vertical movement: mm OBSERVED PERFORMANCE OF A SHORT DIAPHRAGM WALL PANEL 693 ME3-1 ME3-2 ME3-3 ME3-4 Computed (full stiffness) Computed (1/3 stiffness) Measured BF-7 (2. 3 m from face) Fig. 13. Comparison between measured upper four magnets and computed subsurface vertical deformations after excavation On the other hand, if all the soils are assumed to have operated at moderate shear strains ( 1%) during the excavation, the discrepancies between the computed and measured horizontal deformations generally become smaller. However, the discrepancy between the measured and computed vertical movements is still substantial. This discrepancy may be attributed to inappropriate soil parameters or the model used, or both. As an isotropic model was adopted in the analysis, the different responses in the two orthogonal directions seem to suggest that the soils exhibited a certain degree of anisotropy. The large discrepancies between the measured and computed deformations using the shear moduli at moderate shear strains ( 1%) may be attributed to the presence of the 3 m thick soft, layered strata (i.e. ll, marine deposit, alluvium clay and sand) above the decomposed granite. The stiffness of these soft strata is likely to be considerably less than the assumed moduli adopted in the analysis, possibly owing to the onset of substantial plastic yielding. This would result in a signi cant reduction of soil stiffness and lead to larger ground movements during excavation. Of course, larger ground deformations would have been computed if a degradation of soil stiffness with increasing strain was allowed in the analysis. According to the comparisons between the measured and computed results, an average shear strain well in excess of Vertical movement: mm ME5-1 ME5-2 ME5-3 ME5-4 BF-5 (4. m from face) Vertical movement: mm ME6-1 ME6-2 ME6-3 ME6-4 BF-6 (6. m from face) 1% was mobilized around the trench during the excavation. No further analyses have been attempted, owing to the limited accuracy of the inclinometers and the unavailability of good-quality soil stiffness parameters (small-strain shear moduli) for any of the soils except the weathered granites. CONCLUSIONS The construction in Hong Kong of an excavated large, rectangular-section barrette (i.e. a short diaphragm wall panel) in weathered and sedimentary soils under bentonite has been heavily instrumented and closely monitored. The size of the panel was 2 8 m long by 8 m wide by 4 m deep. During excavation, the maximum horizontal ground movements recorded were of the order of a few millimetres, with negligible surface settlements around the panel. A small amount of recovery of the horizontal ground movement was measured during concreting. The observed ground deformations are substantially smaller than the measured values obtained during the construction of a longer panel (6 1 m long by 1 23 m wide in plan) excavated on a site with similar ground conditions. On the basis of three-dimensional numerical simulations of the excavation of the trench, an average mobilized shear strain greater than 1% around the excavated trench can be deduced. 694 NG, RIGBY, LEI AND NG At the soil±wall interface, the initial lateral earth pressures decreased to the hydrostatic bentonite pressures during excavation and then increased above the assumed initial K pressures after concreting. The measured lateral pressures just after concreting show some agreement with a pressure distribution based on a theoretical bilinear pressure envelope. Excess pore water pressures at the soil±wall interface and around the panel were recorded during construction. Dissipation of the excess pore water pressures seems to have been rapid and to have been completed within a few days. ACKNOWLEDGEMENTS This research project is supported by a research grant (HIA96/97.EG3) from the Hong Kong University of Science and Technology and a research grant (CRC96/99.EG4) from the Research Grant Council of Hong Kong. The authors would like to acknowledge the contribution provided by Paul Y. Foundation Ltd, who constructed and tested the heavily instrumented barrette. Other sponsors of this test barrette include the Geotechnical Engineering Of ce of the Hong Kong Government, Fong On Construction Ltd, Mass Transit Railway Corporation and Geotechnical Instruments Ltd. Technical input and support from Professors C. K. Shen and Wilson Tang of the Hong Kong University of Science and Technology and Messrs Martin Pratt and David Ng of Bachy Soletanche Group are highly appreciated. REFERENCES Clear, C. A. & Harrison, T. A. (1985). Concrete pressure on formwork. Report 18. London: CIRIA. Clough, G. W. & O'Rourke, T. D. (199). Construction induced movements of in-situ walls. In Design and performance of an earth retaining structure, Geotechnical Special Publication 25, pp. 439±47. New York: American Society of Civil Engineers. Cowland, J. W. & Thorley, C. B. B. (1984). Ground and building settlement associated with adjacent slurry trench excavation. Proceedings of the international conference on ground movements and structures. pp. 723±738. Farmer, I. W. & Attewell, P. B. (1973). Ground movements caused by a bentonite-supported excavation in London Clay. GeÂotechnique 23, No. 4, 577±581. Geotechnical Engineering Of ce (1993). Guide to retaining wall design. 2nd edn. Hong Kong: Geotechnical Engineering Of ce, Civil Engineering Department. Itasca (1996). Fast Lagrangian Analysis of Continua (FLAC-3D), version 1.1, user's manuals. Itasca Consulting Group, Inc., Minnesota. Lings, M. L., Ng, C. W. W. & Nash, D. F. T. (1994). The pressure of wet concrete in diaphragm wall panels cast under bentonite. Proc. Instn Civ. Engrs, Geotech. Engng, 17, 163±172. Malone, A., Ng, C. W. W. & Pappin, J. (1997). Country report (invited): collapses and displacements of deep excavations in Hong Kong. 3th year Anniversary Symposium of the Southeast Asian Geotechnical Society, Bangkok, pp, 5-124± Ng, C. W. W. (1992). An evaluation of soil±structure interaction associated with a multi-propped excavation. PhD thesis, University of Bristol. Ng, C. W. W., Ling, M. L., Simpson, B. & Nash, D. F. T. (1995). An approximate analysis of the three-dimensional effects of diaphragm wall installation. GeÂotechnique, 45, No. 3, 497±57. Ng, C. W. W., Pun, W. K. & Pang, P. L. R. Small strain stiffness of natural granitic saprolites in Hong Kong. J. Geotech. Geoenv. Engng. ASCE (in press). Powrie, W. & Kantartzi, C. (1996). Ground response during diaphragm wall installation in clay: centrifuge model tests. GeÂotechnique, 46, No. 4, 725±739. Shen, C. K., Ng, C. W. W., Tang, W. H. & Rigby, D. (1998). Invited discussion paper: testing a friction barrette in decomposed granite in Hong Kong. Proc. 14th Int. Conf. Soil Mech. Found. Engng, Hamburg, vol. 4, 2325±2328. Strange, P. J. (199). The classi cation of granitic rocks in Hong Kong and their sequence of emplacement in Sha Tin, Kowloon and Hong Kong Island. Geo. Soc. Hong Kong Newslett., 8, Part 1, 18±27. Stroud, M. A. & Sweeney, D. J. (1977). Discussion appendix. In A review of diaphragm walls, pp. 142±148. London: Institution of Civil Engineers. Symons, I. F. & Carder, D. R. (1993). Stress changes in stiff clay caused by the installation of embedded retaining walls. In Retaining structures, (ed. C. Clayton), pp. 227±236. London: Thomas Telford.

56 FIELD STUDIES OF WELL-INSTRUMENTED BARRETTE IN HONG KONG INTRODUCTION By Charles W. W. Ng, 1 Member, ASCE, Douglas B. Rigby, 2 Member, ASCE, Sean W. L. Ng, 3 and G. H. Lei 4 ABSTRACT: A large excavated rectangular pile (barrette) with lateral earth pressure and pore-water pressure cells was successfully constructed and tested in a sequence of marine, alluvial, and weathered granite soils. A soft base formed beneath the bottom of the barrette permitted over 1 mm of vertical settlement, completely mobilizing the shaft friction at the barrette-soil interface. During the vertical load tests, an unusual and complex response of pore-water pressures and earth pressures at the barrette-soil interface was measured. During each vertical loading cycle (except the last one) and before interface slippage of the barrette occurred, excess positive pore-water pressures were recorded in all soil layers. Upon the initiation of slip at the barrette-soil interface, a sudden drop in the measured pore pressures as well as a substantial drop in lateral earth pressures generally resulted. Subsequent loading or unloading slippage events did not show the same dramatic behavior unless a period of consolidation/recovery was allowed first. This implies that caution must be used in design of barrettes relying heavily on skin friction when shearing induces contractive soil behavior. The current test results indicated that the empirical uncorrected SPT-N value approach and the effective stress -method were inconsistent. Limited space and high demand have made land in Hong Kong extremely expensive. Tall buildings are built to optimize the floor area to land area ratio. Many of the tall buildings located along the Victoria Harbor on the Hong Kong Island and the Kowloon peninsula are commonly founded on reclaimed land. Thus deep foundations are required to resist both vertical and horizontal loads due to the weight of the building and wind, respectively. The prevailing deep foundation types for tall buildings on these reclaimed lands are large bored and excavated piles, which are very long, normally in excess of 5 m. These piles can be circular(bored piles/drilled shafts) or rectangular(barrette) in shape and must extend through the fill, underlying soft marine clay, sandy clay, and alluvial sand deposit down into the deep weathered granite soil(saprolite), which is typically less weathered with an increase in depth. Thethicknessoftheweatheredgranitecanbeupto8min some places, and its depth can extend to more than 1 m from the ground surface. Over the last 15 years, barrette foundations have become increasinglypopularinpartsofasiasuchashongkongand Malaysia for many civil engineering structures and tall buildings. The construction method for barrettes is very similar to that adopted for diaphragm walls, where a rectangular trench is excavated under bentonite by heavy grabs or hydrofraise and filled with tremie concrete. In Hong Kong, single barrettes uptoasizeof1.5mwide 6.6mlong(onplan)havebeen constructed (Pratt and Sims 199). Due to their rectangular shape, barrette foundations are particularly suitable to resist large vertical and significant horizontal loads in a chosen direction. For deep rectangular piles, the current design procedures adopted in Hong Kong are relatively conservative, and they 1 Assoc. Prof., Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and Technol., Clear Water Bay, Kowloon, Hong Kong. 2 Asst.Prof.,Dept.ofCiv.Engrg.,HongKongUniv.ofSci.andTechnol., Clear Water Bay, Kowloon, Hong Kong. 3 Res. Student, Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and Technol., Clear Water Bay, Kowloon, Hong Kong. 4 Res. Student, Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and Technol., Clear Water Bay, Kowloon, Hong Kong. Note. Discussion open until June 1, 2. To extend the closing date one month, a written request mustbe filed withtheascemanagerof Journals. The manuscript for this paper was submitted for review and possiblepublicationonapril1,1999.thispaperispartofthejournal of Geotechnical and Geoenvironmental Engineering, Vol. 126, No. 1, January, 2. ASCE, ISSN //1-6 73/$8. $.5 per page. Paper No / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 assume a heavy reliance on end-bearing resistance of bedrock in many instances. Without performing at least one full-scale pileloadtestonsite,skinfrictioninexcessof1kpaisnot permitted normally by the regulations. In some areas, however, bedrockisfoundatdepthsofover1m.undersuchcircumstances, excavation of deep foundations to bedrock becomes difficult, time consuming, and expensive. Exceeding the nominal permitted skin friction requires costly and time-consuming full-scale pile tests to verify design values of skin friction. Many Hong Kong engineers would welcome improved design guidelines based on more rational design approaches that would allow for higher default values of skin friction along a pile to be used, or a reduction in the number of verification piles in similar site conditions. The problem is that estimation of skin friction development alongalongbarrette(between4and1m)isaverydifficult task. The method of construction, workmanship, rheological properties of the slurry, and concrete placement affect its behavior. Any attempt to increase the design skin friction value must be done with caution. A task force has recently been formed, with participants from the government, some contractors and consultants, and the Hong Kong University of Science and Technology, to carefully study this problem aiming at the development of a more reasonable design guideline for deep pile foundations in Hong Kong. Currently a university-led and industry-supported three-year research project is under way to study skin friction on barrettes founded in weathered granites in Hong Kong by full-scale pile testing, laboratory tests, numerical and centrifuge modeling, and reliability analysis(shen et al. 1997). Initially, two piling test sites are investigated: one atkowloonbayandtheotherinthecentraldistrict.inthis paper, the construction of a 2.8 m long by.8 m wide and 39.7 m deep barrette at Kowloon Bay, its vertical load-deflection characteristics, and its pore-water pressure and lateral stress changes at the soil/barrette interface are reported and discussed. In addition, the measured skin friction is compared with other test results in Hong Kong. SITE LOCATION AND GROUND CONDITIONS ThetestsiteislocatedontheKowloonpeninsulaofHong Kong,totheeastofarunwayoftheoldKaiTakinternational airport, at the Kowloon Bay area (Fig. 1). Fig. 2 shows the geology and some relevant borehole information obtained at Kowloon Bay. The site is on marine reclaimed land andthe ground level is at approximately 4.48 m above Principal Datum(PD).Theground-waterlevelisabout3mbelowground surface. The ground conditions consist of about 6. m fill ma- FIG. 1. Location of Test Barrette at Kowloon Bay, Hong Kong FIG. 2. terial overlying a succession of approximately 9.5 m marine clay deposits, 7.5 m of sandy clay(probably alluvial), 4.8 m alluvialsandofquaternaryage,andabout12mofweathered granitic saprolites that overlie granitic rocks of Upper Jurassic to Lower Cretaceous age(strange 199). Detailed descriptions and measured N-values by Standard Penetration Tests(SPT) foreachtypeofmaterialsaregiveninfig.2.itcanbeseen that scattered SPT-N values were obtained in both alluvial sand and weathered granite. The former shows a decreasing N-values with depth, whereas the latter illustrates an opposite trend. Based on results of drained triaxial compression tests on weathered granites, effective cohesion and angle of friction werefoundtobekpaand39,respectively.typicalatterberg limits for the sandy clay are4, 2, and 2 for the LL, PL, and PI, respectively(geo 1996a). Strictly speaking, the site is not ideal for studying skin friction of excavated piles in weathered granite as the thickness of the granite is relatively thin and the measured SPT-N values are relatively low. However, due to the limited availability of landinhongkongforpurelyresearchpurposesandthetime restraints, this existing government test site was chosen. A distinct advantage is that the site investigation records are very comprehensive as this site has been a test site for the Geotechnical Engineering Office of the Hong Kong Special Administrative Region over the years. Various in situ and laboratory tests (Ng et al., unpublished paper, 1999) have been carried out on this site, resulting in ground conditions that are well known. DETAILS OF CONSTRUCTION The test barrette or the diaphragm wall trench was excavated using a traditional cable-operated grab. The size of the excavatedtrenchwas2.8by.8monplaneand39.7mdeep (Figs. 2 and 3). During construction, a concrete guide wall was first placed, and at deeper levels the trench was supported Borehole Logs and SPT-N Values at Kowloon Bay, Hong Kong JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 61

57 FIG. 3. Layout of Instrumentation (Plan View) bybentonitewithunitweight( b)ofabout1.8kn/m 3.Soil spoil, suspended in the bentonite slurry, was pumped to a desanding unit at the ground surface. After desanding, the bentonite was recharged into the trench. Chiseling of the base was carried out when rock was encountered at a more shallow depth than expected, about 39.6 m below ground. This caused asmalloverbreakatthebase,whichwasdetectedbyasonic profiling system. The entire excavation took 62 h to complete andthefinalexcavateddepthwas39.7mbelowground.the rate of excavation could have been improved if a 24-h nonstop excavation schedule were followed, as is the normal practice for commercial test barrettes in Hong Kong. After completion of the excavation, three instrumented reinforcement cages were lowered one-by-one into the trench. Concreting was carried out 43 h after completion of theexcavation. The average rate of concreting was 1.32 m/h by using a tremie pipe. The whole barrette trench was filled with concrete in 4.5 h. The Ordinary Portland Cement(OPC) concreteusedwasgradec3/2withunitweight( c)of23.2kn/ m 3. It had a water-to-cement ratio of.45 and an average slump of 18 mm. During concreting, the average temperature measuredinsidethetrenchwas27.6 Cinsidetheslurry.The excavation and concreting procedures of the barrette, in fact, were identical to the construction of a typical diaphragm wall panel. Thetop2mofthebarretteconsistedofareduced-section sheathing zone(fig. 2) built with the intention of minimizing the interface skin friction developed between the barrette and the upper soil layers. This sheathing layer consisted of four layers: a 3 mm steel plate welded onto the reinforcement cage, a coating of bitumen, a flexible and weak voltex layer(geotextile infilled with sodium bentonite), and a thin sheet of plywood. However, the final result was that the plywood was unfortunately attached to the steel plate with a dense matrix of high-strength screws, precluding the possibility of shear between the intermediate soft layers. As a consequence, the theoreticalgapofabout8mmbetweentheplywoodandthe surrounding soil was not back-filled with gravel as planned, so that a weak friction zone would hopefully exist. However, steel rods inserted into this suspected bentonite-filled gap about 2 weeks after concreting were unable to probe beyond a meter or two all round the barrette. Either concrete overflow had partially filled the gap, construction activities failed the soil infilling the gap, or surface materials mixed with solidified bentonite. Thus, in the end, the sheathing zone was not expected to function to effectively reduce skin friction over the top half of the barrette. Atthebottomofthebarrette,a soft basewasformedto minimize the effects of end-bearing for mobilizing full skin friction at the soil-wall interface. This was done by placing a m in height steel box to the bottom of the trench, before the lowering of the main reinforcement cages. Theboxwasmadeof3mmthicksteelplate,anditwasinitially filled with fine round sand. Seven days after concreting, the sand-filled steel box was drilled through and flushed with pressurized water via two cast-in flushing pipes and one concrete core hole in the middle of the barrette (Fig. 3). Great care was taken to ensure that most of the sand was flushed out to form a soft base (i.e., void) underneath the barrette. INSTRUMENTATION To study the load transfer mechanism and load-settlement characteristics of the barrette constructed at Kowloon Bay, a substantial amount of instrumentation was installed. A summary of the instruments installed inside the barrette is given intable 1. In addition, four sets of standard dial gaugestogether with surveying were used to monitor the vertical settlement of the top of the barrette and reference beams during testing. Strain gauges were placed at 27 levels on the reinforcement cages(figs.3and4).foursurfacemountedandfourembedded strain gauges were placed alternatively at different levels TABLE 1. Summary of Instrumentation at Kowloon Bay Instrument (1) Quantity (2) Strain gauges 132 Rod extensometers 1 In-place inclinometers 2 Vibrating wire piezometers 4 Earth pressure cells 4 FIG. 4. to measure vertical strains induced in the reinforcing bars and in concrete, respectively. Moreover, four levels of horizontally embedded strain gauges, with four gauges in each level, and four levels of dummy gauges(eight in total) were installed in the cage. The horizontally embedded strain gauges were used to determine any Poisson s ratio effects(results are not relevanttothispaper).atotalof132gauges,typicallyat3mand 1 m intervals in the sheathed and unsheathed zones, respectively, were installed to determine the strain distributions along the entire depth. It was found that similar results were recorded by both the surface mounted and embedded strain gauges. Thus, no further distinctions between the two types of gauges aremadeinthispaper. Ten aluminum rod extensometers were sleeved individually inpvctubesandinstalledtofivedepthsattwodifferentlocations inside the barrette to monitor displacements between eachdepthandreferencesteelplateatthetopofthebarrette (Figs.3and4).Afterthepiletestitwasfoundthatalthough the extensometers reflected a reasonable pattern of elastic pile shortening and rebound during loading and unloading cycles, some of the relative magnitudes measured were clearly unreliable. This was likely caused by friction developed between the metal rods and the PVC tubes having a 12.5 mm outer Layout of Instrumentation (Elevation) diameter and 12.6 mm inner diameter, respectively. Recently, extensometers consisting of 15 mm steel rods placed in 5 mm steel tubes filled with oil have given reliable results in Hong Kong. A total of 38 biaxial servoaccelerometer sensors were installedat19levels(mostofthemat2minterval)intwocastin pipes inside the barrette. The locations of the in-place inclinometers and the levels of sensors were indicated in Figs. 3 and 4, respectively. The bottoms of the inclinometers were socketed in rock. These two in-place inclinometers were designed to measure rotations and hence horizontal movements of the barrette during loading. From the measurements, it was concluded that no significant bending deflection of the barrette was induced during the vertical load tests. A total of four vibrating wire total earth pressure cells, together with four vibrating wire piezometers, were installed at the barrette-soil interface at four levels within the layers of sandy clay, alluvial sand, and weathered granite. The depths of the earth pressure cells and piezometers are shown in Fig. 4. These total earth pressure cells and piezometers were attached to appropriate locations of the reinforcement cage during the construction of the barrette. Once in position, the instruments were jacked out horizontally to ensure contact at 62 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 63

58 reasonable pressures with the surrounding soil. Each vibrating wire piezometer was located about 8 mm above the corresponding pressure cell. LOAD AND DISPLACEMENT BEHAVIOR OF PILE The loading system for the test barrette consisted of two 1, ton hydraulic jacks pushing against kentledge formed from steel billets. The kentledge rested on two parallel sets of concreteblocksspaced6mapart,centertocenter.apilecap atthetopofthebarretteandasteelspreaderbeamabovethe jacks were used to transfer the manually applied loads. The pile head displacement was measured by using four dial gauges symmetrically resting on two reference beams. Settlement of these reference beams was monitored by the conventional surveying technique. The test program originally comprised four loading and unloading cycles (Fig. 5). However, after the applied load reached 7,455 kn at the second cycle, substantial settlement wasrecordedandtheappliedloadcouldnotbeheldconstant within the prescribed maximum settlement tolerance of.5 mm/1 min. It was therefore decided to unload the barrette to 4,555 kn and to hold it for 8 h (about 3 days). After the holding period, the testing program resumed and two more loading cycles were performed. The barrette-soil interface appeared to gain strength as a result of consolidation, which is discussed later. Due to the presence of the soft base (i.e., void) underneath the barrette, the barrette ultimately settled about 1 mm, enabling the skin friction to be fully mobilized along the shaft. Prior to calculating the distribution of axial load and shear stressalongthelengthofthebarrette,itisnecessarytoadopt appropriate values of Young s modulus for the barrette. Since the conditions of concrete curing inside the trench are very different from those in a standard curing tank in the laboratory, continuous concrete cores were taken from at the center along the depth of the barrette to determine the Young s modulus of the in situ concrete. The measured secant Young s modulus varyingwithdepth(atabout1mspacing)isshowninfig.6. Although there is a general increase of Young s modulus with depth due to the natural compaction process of the concrete under its own weight, the measured data are fairly scattered. For accurately converting the measured strains in the barrette to stresses, the actual measured modulus at each tested depth and corrected for steel present was adopted in the calculations. Fig. 7 shows the deduced axial load versus depth for the FIG. 6. Variation of Measured Young s Modulus of Concrete with Depth last loading cycle. An axial stress was calculated from the average measured strains at each level and the corresponding measured Young s moduli (Fig. 6) from concrete cores taken from the center of the barrette. From the calculated axial stress, the axial load with depth was then determined considering the local barrette cross-sectional area. It can be seen that the deduced axial load at the top of the barrette is consistent with the applied load recorded by the hydraulic jacks (shown as dotted lines). The load distribution along the depth shows features common to a typical friction barrette. Due to the presence of the soft base, minimal base resistance was mobilized, except at the maximum load of 8,95 kn, where the base resistance increased substantially from 56 to 1,6 kn. This substantial increase was likely caused by the mobilization of some end bearing resistance due to the crushing of the 3 mm steel box underneath the barrette (possibly not all of the sandwasflushedoutfromthebox). BARRETTE SKIN FRICTION From the gradient of the barrette normal stresses with depth, mobilizedskinfriction(interfaceshearstress, )iscalculated and plotted against deduced local displacement of the barrette ineachsoilstratumasshowninfig.8.shearstresswasfully mobilized in all soil strata when the displacement reached approximately 4 mm (or at 2.4% of the 1.7 m equivalent diameter of the barrette), after which the magnitude of shear stress remained essentially constant. Some peak strength behavior with softening at the interface was observed in the fill, sandy clay, and alluvial sand. The large skin friction mobilized in the fill material may possibly be attributed to the surcharge effects resulting from the dead weight of the concrete blocks supporting the kentledge load. On the other hand, no peak strength behavior could be identified at the barrette-weathered granite interface. Fig. 9 shows the distribution of mobilized skin friction with depth. Within the sheathed zone, a constant and substantial amount of shear stress was mobilized, apparently indicating that no weak zone existed around the sheathed barrette (except at depth below 15 m; see Fig. 7). The mobilized shear stress in the sandy clay was smallest, whereas the mobilized shear stresses in the fill, marine deposits, and weathered granite are in the same order of magnitude at the maximum applied load. Below the sheathed zone, the distribution of maximum mobilized shear stress follows the trendofthemeasuredspt-nvalues(fig.2).atthemaximum applied load(8,95 kn), the mobilized skin friction of about 3kPawasfoundforthefill,marineclay,alluvialsand,and weathered granite, and about 15 kpa for the sandy clay. For comparing shear stress mobilized in similar soils at different construction sites, it is a common practice to normalize the measured shear stress with an average uncorrected SPT-N value, i.e., /N, (beforeconstruction)inhongkong.fig.1 shows the comparison between the normalized maximum shearstressmeasuredatthecurrenttestsiteandattheinternational Trademart, which is about 1.4 km away in Kowloon Bay(Ho1994).Atthelattersite,a2.2mby.8mby56.8 m deep barrette was also excavated by cable-operated grabs andfoundedinabout2mthicklayerofweatheredgranite, which has an average higher value of SPT-N value(typically rangingfrom15to11)thantheformersite.forspt-nvalues smaller than 35, the magnitude of measured skin friction is consistent between two sites. No comparison can be made between the measurements from the two sites for higher SPT-N values.however,therangesofthevalues /N obtainedfrom the former and the latter sites are.9 to 2.9 and 1.3 to 2.3, respectively. PORE PRESSURE RESPONSE AT SOIL/BARRETTE INTERFACE After installation of the piezometers, readings were taken continuously to compare them with the initial hydrostatic porewater pressures in the ground (the initial ground-water table waslocatedatabout1.3mpd).itwasfoundthatthemeasured pore-water pressure at gauge P1 at the sandy clay layer(fig. 11) was slightly higher than the corresponding hydrostatic value (1.3 mpd) before the loading test. Piezometric level (head)isdefinedasthesumofthepore-waterpressurehead and the elevation head at each location. The measured piezometric heads recorded at P3 and P4 were a little lowerthan the hydrostatic conditions in the weathered granite before the commencement of the load test. Labels LC1 LC4 denote the commencement of the first to the fourth load cycles, respectively. Similarly, labels UC1 UC4 represent the start of the first to the fourth unloading cycles. Whenthefirstloadingcycle(LC1)wascarriedout,allpiezometers responded positively to each increment of applied load [Fig. 11(b)], recording an ultimate increase of head of almost3minthesandyclay(p1)andabout1minboththe alluvial sand(p2) and weathered granite(p3 and P4). During the first unloading cycle(uc1), the pore pressure in the sandy clay and alluvial sand remained steady while a slight and gradualreductionofporepressurewasrecordedatp3andp4in the weathered granite. Each loading and unloading increment is clearly confirmed by the pile shortening and rebound measurements from the rod extensometer data. During loading cycle 2(LC2), again increments of applied load resulted in increments of positive pore pressure until failure occurred. A further increase of approximately 6 m head of waterwasmeasuredinthesandyclay,2minthealluvialsand FIG. 5. Load-Settlement Response of Test Barrette FIG. 7. Deduced Axial Load Distribution at Load Test Cycle 4 FIG. 8. Mobilization of Skin Friction with Local Displacement 64 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 65

59 FIG. 9. Distribution of Mobilized Skin Friction (LC4) and4 5mintheweatheredgranite,respectively.Theminor drops in water pressure occur together with reduction in applied loading due to continued settlement of the pile. At failure the settlement rate of the barrette increased significantly and thesettlementincreasedtoabout25mm(fig.5)andthepore pressuresbegantosuddenlydropatthesametimeinallsoil layers[aftertheloadingslipshowninfig.11(b)].slipatthe barrette-soil interface is inferred at failure since all rod extensometer data held constant, indicating that the entire barrette wasmovingdownwardasarigidbody(about2mm).after about5minofslippage,theloadwasreducedandthensubsequently maintained for a holding period of 8 h. During this time the excess pore pressures dissipated almost completely in the weathered granite but only slightly in the sandy clay and, surprisingly, in the alluvial sand [Fig. 11(a)]. Also visible in thisfigureatp2inthealluvialsandaremeasurementsoftidal action. The shifted frequency and the attenuated relative magnitudes of the high and low tides correspond to actual tidal values from 12/3/97 to 1/2/98( Quarry 1997, 1998) almost perfectly, indicating proper functioning of the P2 pore-water pressure gauge. Pore pressure behavior during the third loading cycle(lc3) was very similar to cycle 2. Loading increments resulted in corresponding increases in all four piezometers; overall, the increaseofexcessporewaterpressureamountedtoabout4m head in the sandy clay, 3 m in the alluvial sand, and 2.5 m head in the weathered granite[fig. 11(c)]. Failure and slippage of about 4 mm occurred at approximately the same load (7,455 kn) as cycle 2, again accompanied by significant and sudden drops of pore pressure at all thepiezometers(dueto FIG. 11(b). Variations of Piezometric Level during First Two Load Test Cycles FIG. 11(c). Variations of Piezometric Level during Last Two Load Test Cycles FIG. 1. Relationship between Normalized Maximum Shear Stress in Weathered Granite and Average SPT Value, N FIG. 11(a). Variations of Piezometric Level During Load Testing loading slip). During unloading(uc3) the measured pore pressures continued to drop slightly until another sudden drop of pore pressures was observed atthe lastunloading stageto kn(exceptp2)whichwasmostsignificantinthesandyclay. Loading cycle 4(LC4) varied from previous loading cycles since it occurred shortly after a barrette-soil interface failure. Increments of applied load in LC4 caused only slight increases of pore-water pressure in the alluvial sand and weathered granite while pore pressures remained essentially steady in the sandy clay [Fig. 11(c)]. As the load reached its maximum valueof8,95kn(fig.5),failureoccurred(about3mmof downward slippage), again resulting in substantial drops of pore-water pressures in all soil layers but the alluvial sand whereanincreaseofwaterpressurefollowedbyadropwas measured[fig. 11(c)]. One piezometer in the weathered granite recorded a significant drop in pore pressure to a negative value, lower than the original in situ water pressure. Like UC3, water pressure values were steady until the last unloading step, when sudden changes in pore-water pressure resulted in all layers except the alluvial sand. Two of the piezometers recorded the standard drop in pore pressure, while the third negative reading piezometer showed a sharp increase back to a more reasonable value. After the load test, dissipation of excess pore-water pressures continued[fig. 11(a)]. As expected, the rate of dissipationwasmuchslowerinthesandyclaysoil(p1)thaninthe weathered granite(p3 and P4). Dissipation at the alluvial sand interface was slow during loading, but generally the excess pore pressures generated returned near their equilibrium value (around2mofhead)fairlyquicklyafterslip.nearlyallthe excess pore-water pressure was dissipated in the weathered granite by January 6, 1998, i.e., 8 h after the load test.on the other hand, about 3 m and 1 m excess pore-water head still remained in the sandy clay and alluvial sand layers, respectively, at this same time. What appeared to be tidal behavior was picked up at P2 and P4, but it is difficult to see since actual peak tide magnitudes dropped by 5% and the scan frequency of instruments was reduced substantially after the pile test. CHANGES OF LATERAL STRESS AT SOIL/ BARRETTE INTERFACE After the vibrating wire earth pressure cells had been lowered into the excavated trench filled with bentonite, initial readings were taken before the instruments were jacked into position. This allowed an in situ calibration of the pressure cellsto bemadebetween themeasuredvaluesandthetheoretical bentonite pressures, using a measured unit weight of bentonite( b)=1.8kn/m 3.Itwasfoundthattherewasgood agreement between the calculated bentonite pressures and the readings recorded by the earth pressure cells; the maximum difference between the calculated and measured pressure was lessthan1kpa.atthispointthecellswerejackedoutlaterally to engage the soil. Just after concreting, the lateral earth pressures increased to values close to the assumed in situ earth pressures before excavation. The measured total earth pressures at the soil/concrete interface also corresponded well with the theoretical bilinear concrete pressure envelope proposed by Ng(1993) and Lings et al. (1994) for concrete case under bentonite during diaphragm wall construction. Over the next two days lateral 66 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 67

60 pressures remained essentially unchanged in the sandy clay and alluvial sand soils, but increased about 1 kpa in the weathered granite. Further details of the earth pressure measurements and interpretations during the construction of the barrettearereportedbyngetal.(1999). During the subsequent three-week curing period before the vertical load test, a gradual continuous decrease of lateral earth pressure was measured at all lateral earth pressure cells(about 3kPainthesandyclayandalluvialsandandabout6kPa in the weathered granite). The observed reduction in the contact earth pressure may in large part be due to soil consolidation as indicated by the dissipation of excess positive porewater pressures generated during the construction of the barrette. In addition, the reduction in earth pressure might be causedbysmallshrinkageofthepressurecellunitsasaresult ofafallintemperatureafterhydrationofcementintheconcrete. Lings et al.(1994) also reported reductions in earth pressure at the soil/diaphragm wall interface of a heavily overconsolidated stiff clay. After the three-week curing period, the final result was an overall decrease of about 2 3 kpa below the assumed initial lateral earth pressures in the sandy clay and alluvial sand, and an increase of about 6 kpa above the original pressures in the weathered granite. This observed net increase in lateral earth pressure could be attributed to the swelling of weathered granite as result of stress relief during the excavation of the trench for constructing the barrette. Davies and Henkel(1981) have reported swelling behavior in weathered granite during the construction of diaphragm wall panels. The measured earth pressures during the four cycles of load testing are shown in Fig. 12.The earth pressurecellsallremained virtually constant during the first load and unload cycles(lc1 and UC1) when the barrette displacement was small [Fig.5;Figs.12(aandb)].Withthebeginningofloadcycle 2,somesmalldropsoflateralearthpressurewereseeninthe weathered granite soil layer until the onset of pile failure(interface slip). Upon application of the maximum vertical load of7,455kn,asuddenandlargereductionofabout1kpa wasmeasuredatcellspc3andpc4intheweatheredgranite. Atthesametimetherewerenoticeablefallsintheotherearth pressurecellsalso(about2kpaatpc1insandyclayand45 kpa at PC2 in the alluvial sand). During the partial unloading and early holding period, there was a small recovery in earth pressureatallcells.overtherestoftheholdingperiod,aslow but steady 25 kpa increase of pressure was measured in the weathered granite, while pressures in the sandy clay and alluvial sand remained almost constant [Figs. 12(a and b)]. Twice-a-day uniform fluctuations in lateral pressure were visibleatallcellsduetotidalpressures. After the period of maintained load, further loading(lc3) did not cause a significant variation in earth pressures until a verticalloadof7,455knwasappliedandslipatthebarrettesoil interface occurred. At this time a significant reduction in earth pressures (75 to 1 kpa) was again recorded by PC3 and PC4[Fig. 12(c)] in the weathered granite. Small drops in the other earth pressure cells were also recorded at this time. Duringunloadingofcycle3(UC3),pressureatPC1andPC2 basically held constant while a slight recovery was again observed for the cells in the weathered granite. Application of loading and unloading cycle 4 resulted in an FIG. 12(c). Variations of Lateral Earth Pressure during Last Two Load Test Cycles FIG. 12(a). Variations of Lateral Earth Pressure during Load Testing FIG. 12(b). Variations of Lateral Earth Pressure during First Two Load Test Cycles FIG. 13. Details of Pressure Changes at Interface during Load Cycle 2 68 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 69

61 overall constant lateral stress readings at all cells except for minordropsinpressureinthreeofthefourcellsattheonset ofslipatabout7,kn.thelateralearthpressureoverthe next week remained constant in all soil layers. SUMMARY AND POSSIBLE EXPLANATION OF BARRETTE-SOIL INTERFACE BEHAVIOR The barrette-soil interface behavior consisting of shear strains, relative movements, pore pressure changes, and lateral earth pressure changes are clearly related. To more easily examine the relationships involved, pore pressure and lateral earth pressure measurements are combined on the same figure for two load cycles. Fig. 13 presents cycle 2 data illustrating virgin loading, slip, and partial unloading behavior(see also cycle 1 and 3 data). Fig. 14 gives details of cycle 4, which shows nonvirgin loading, slip, and unloading behavior soon after pile failure (cycle 3). In general, the behavior can be summarized as follows: Vertical pile loading produces shear strains in the soil near the barrette, inducing increased pore-water pressure and no significant changes in lateral earth pressures(highest inthesandyclay,lowestinthealluvialsand). Vertical pile unloading leaves pore pressures and lateral earth pressures essentially unchanged. At relatively large loading increments near pile failure, virgin slippage (shear failure) along the barrette-soil interface results in a reduction of pore pressures (usually highest in the sandy clay) and significant drops in lateral earth pressures(highest in the weathered granite). Finally, subsequent reloading of the pile soon after a failure does not result in pore pressure increases(except in alluvial sand), but when slip initiates, pore pressures again drop while lateral earth pressures maintain their values. The mainly elastic simple shear of the interface soils apparently produces a loose or contractive volumetric response at the soil layers. Since the sandy clay and weathered granite soils have significant clay content, their shearing oc- curs under undrained conditions, producing increases in the pore-water pressure. Induced excess pore pressures may be higher in the normally consolidated sandy clay layer because of the higher clay content and since more simple shear occurs higher up the barrette(due to increased shortening of the pile). The increases in pore pressure in the alluvial sand layer are somewhat difficult to understand, but the excavation under bentonite would produce a normally consolidated bentonite cake layer at the barrette-sand boundary. With its high clay content, the sandy clay may act in an undrained manner with no reductions in lateral earth pressure, while the small drops of earth pressure in the weathered granite are likely the result of higher permeability, allowing some consolidation to occur. The very slow dissipation of pore pressures in the sandy clay and relatively rapid dissipation in the weathered granite soil provide further confirmation[figs. 11(a andb)]. Barrette-soil interface slip triggers consequential new behavior with the general sudden drop in both pore pressure and lateral earth pressure. The mechanisms involved in this basically consistent and repeated behavior are complex. It is possiblethatthesoilbehavesortendstobehaveinacontractive manner when it is subjected to shearing. The pore pressure may drop because slip occurs during stress reversal, and this allows relaxation of the shear strains leading to a reduction of previous undrained pore-water pressure buildup. On the contrary, this slip also apparently permits some localized drainage andhencecontractionofthesoiltooccur(possiblyduetothe sudden stress/strain reversal or reorganization of stresses in the localsoilmatrix).thiswouldcauseareleaseofthelargeconfining stresses induced during previous shearing because the concrete wall of the barrette is rigid and does not act as a constant pressure interface. Any contraction of the soil away fromthewallwillleadtoareductionintotalcontactpressure. This might explain the sudden and large reductions of lateral earth pressure compared to the much smaller reductions in pore-water pressure. It should be noted that the concrete wall has a much higher stiffness relative to that of the adjacent soils. The observed complex pattern of interface pore pressure behavior described in the present barrette field test has been observedbothinthefield(earth1986;matlock1992)andthe laboratory where pile/normally consolidated clay interface tests have been carried out measuring excess pore pressures during two-way cyclic loading as reported by Rigby and Desai (1996) and Rigby(1997). One observation of these authors is that the two-way cyclic shear of a pile-clay interface results in a continuous increase in excess pore pressure unless interface slip failure occurs, causing a sudden drop in excess porewater pressure. Aftercycle2,attheKowloonBaytest,whenthesoilwas allowed to consolidate and strengthen, subsequent shearing of thesoilduringloadcyclethreeresultedinthetypeofvirgin loading behavior described previously for cycle 2[Figs. 11(b and c)]. In contrast, load cycle 4 immediately followed cycle 3 so the behavior was more consistent with continued slip behavior since the interface was still weak. A further interesting point is the occurrence of the same slip pattern of behavior at a small scale [Figs. 11(c) and 12(c)],duringthelastpileunloadingstepforcycles3and4. In cycle 1 failure had not occurred yet, and in cycle 2 the barrettewasnotfullyunloaded,sothesuddendropsofpore pressureobservedinbothcycles3and4seemtobeindicative of possible additional slip occurring. STANDARD PRACTICE FOR DESIGN OF BARRETTES IN HONG KONG Barrette design in weathered granite in Hong Kong has been essentially based on empirical approaches (Pratt and Sims 199). An empirical relationship between an uncorrected SPT- N value before construction and allowable skin friction of.5nkpawithalimitof1kpaandanallowableendbearing pressureof5nwithalimitof1,2kpaarecommonlyused in Hong Kong. With this allowable load approach, the overall factor of safety (FOS) is not explicitly defined. However, a FOS greater than 2 is anticipated according to experience gained in Hong Kong. To compare the current measured skin friction at Kowloon Bay with other tested barrettes in weathered granite in Hong Kong, the two most common approaches are adopted. Figs. 15 and 16 show the interpreted field test results from seven relatively well-documented case histories for barrette constructioninweatheredgranite.detailsofb1 B5testsandtestB6 are given by GEO (1996b) and Lo (1997), respectively. The test at the International Trademart in Kowloon Bay(Ho 1994) isidentifiedasb5,andthecurrenttestismarkedasb7inthe figures.theshaftfrictioncoefficient (GEO1996b)isdefined as the ratio between the ultimate skin friction and the mean vertical effective stress, assuming that the effective cohesion iszero. It can beseen thatthereisalargescatterinthededuced skin frictions; varies from.1 to.46 whereas /N fallsbetween.77and2.3.thevariabilitymaybearesultof different methods of construction, quality of workmanship, quality and consistency of SPT testing, natural variations of ground conditions, and methods of interpretation. Byplottingthecurrenttestresult(B7)inFig.15,themobilizedshearstressoverSPT-Nratioiscloseto1.5.ThissuggestsanimplicitFOSgreaterthan2.Ontheotherhand,the deduced valuebasedontheeffectivestressprincipleisonly.1(fig.16),whichisonlyhalfofthevalueobtainedfroma FIG. 14. Details of Pressure Changes at Interface during Load Cycle 4 7 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 FIG. 15. Relationship between Skin Friction and Mean SPT Value, N for Barrettes JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 71

62 Matlock, H.(1992). Field and laboratory testing. NSF Workshop on RecentAccomplishments and Future Proc., TrendsinGeomech. in the 21st Century,M.Zaman,C.S.Desai,andA.P.S.Sel- U.S.-Canada vadurai, eds., Ng, C. W. W. (1993). Lateral stress development in diaphragm walls during concreting. Proc. Int. Conf. on Retaining Struct., Thomas Telford, London, Ng,C.W.W.,Rigby,D.,Lei,G.H.,andNg,S.W.L.(1999). Observed performance of a short diaphragm wall panel. Géotechnique, London, 49(5), Pratt,M.andSims,M.J.(199). Theapplicationofnewtechniquesto solve deep basement and foundation problems. Deep Foundation Proc. Int. Conf. on Practice, Quarry bay times and heights of high and low waters (1997). Tide Tables for Hong Kong 1997, Royal Observatory, Hong Kong, 25. Quarry bay times and heights of high and low waters (1998). Tide Tables for Hong Kong 1998, Royal Observatory, Hong Kong, 2. Rigby, D. B. (1997). Pore pressure behavior of clay-steel interfaces. Proc., 2nd Int. Symp. on Struct. and Found. in Civ. Engrg.,HKUST, Hong Kong, Rigby, D. B., and Desai, C. S.(1996). Testing and constitutive modeling of saturated interfaces in dynamic soil-structure interaction. Nat. Sci. Rep. to Found., Dept. of Civ. Engrg. and Engrg. Mech., University of Arizona, Tucson, Ariz. Shen,C.K.,Ng,C.W.W.,Tang,W.H.,andRigby,D.(1997). Testing a friction barrette in decomposed granite in Hong Kong. Int. Conf. Soil Mech. and Found. Proc., 14th Engrg., Vol. 4, Strange, P. J.(199). The classification of granite rocks in Hong Kong and their sequence of emplacement in Sha Tin, Kowloon and Hong Kong Island. Geotech. Soc. of Hong Kong Newsletter, Hong Kong, 8(1), FIG. 16. Relationship between Skin Friction and Mean Vertical Effective Stress for Barrettes similarsite(i.e.,b5).thislow valuemaybeattributedto the unintended long delay in concreting after completion of the excavation (43 h). A thicker bentonite cake may have formed at the granite/barrette interface leading to a lower skin friction mobilized at the granite/barrette interface. The test resultsshowninfigs.15and16clearlyindicatedthatthecurrent empirical uncorrected SPT-N value approach and the effectivestress -methodwereinconsistent. CONCLUSIONS 1. A well-instrumented 2.8 m by.8 m by 39.7 m deep excavated concrete pile(barrette) was successfully constructed and tested purely for research purposes in Hong Kong. This test pile was a joint effort and collaboration between the university, government agencies, and the industry. Due to the formation of a soft base at the bottom of the barrette, vertical displacement of more than 1 mm(6% of equivalent diameter) was permitted. This large vertical displacement of the barrette enabled skin friction at the barrette/soil interface to be fully mobilized. 2. Theshearstressinallsoillayerswasfullymobilizedat about 2.4% of the equivalent pile diameter with no or little softening behavior observed. At the maximum applied load (8,95 kn), the mobilized skin friction of about3kpawasfoundforthefill,marineclay,alluvial sand, and weathered granite, and about 15 kpa for the sandyclay.normalizedshearstress( /N )oftheweathered granite ranged from.9 to 2.9 were obtained. 3. During the vertical load tests, specially installed instrumentation allowed the measurement of an unusual and complex response of pore-water pressures and earth pressures at several points along the barrette/soil interface. During each load cycle(except LC4), a buildup of excess positive pore-water pressure was recorded in the sandy clay and weathered granite as the vertical applied load on the barrette increased. The increase in pore-water pressure was likely caused by the undrained contractive behavior of the soils. There was no significant change in lateral earth pressure during each load and unload cycle, except at the occurrence of a large vertical displacement of the barrette. When loading caused significant slippage of the barrette within the soil, a consistent and substantial reduction in lateral total earth pressures resulted together with a drop of excess pore-water pressures. The earth pressure drop was most significant in the weathered granite soils, the changes in pore-water pressure more significant in the sandy clay. 4. The increases of pore-water pressure during pile loading 72 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 and the sudden significant drop of lateral earth pressure attheonsetofpilefailureforsomesoillayersillustrate that the complex barrette-soil interaction will likely be best understood using a soil mechanics perspective. A possible mechanism to explain the behavior is that slip occurring during stress reversal permits some localized drainage and contraction of the soils. Since the concrete wallofthebarretteisrigidanddoesnotactasaconstant pressure interface, any contraction of the soils away from thewallwillleadtoareductionintotalcontactpressure. 5. The current test results indicated that the empirical SPT- Nvalueapproachandthe -methodwereinconsistent. ACKNOWLEDGMENTS The writers are grateful for the contribution provided by Paul Y. Foundation Ltd., who constructed the heavily instrumented research pile. Other sponsors of this test barrette include the Geotechnical Engineering Office of the Hong Kong Government, Fong On Construction Ltd., MTRC, Geotechnical Instruments Ltd., and the Hong Kong Institution of Engineers. In addition, the writers would like to acknowledge the financial support fromthetworesearchgrants,hiaandcrc,awardedbythehongkong University of Science and Technology and the Research Grant Council of Hong Kong, respectively. Technical input and support from Professors C. K. Shen andwilsontang of the Hong Kong University of Science and Technology, and Martin Pratt and David Ng of Bachy Soletanche Group, are highly appreciated. APPENDIX. REFERENCES Davies, R. V., and Henkel, D. J.(1981). Geotechnical problems associated with the construction of Chater Station, Hong Kong. Geotech. Seminar, Session No. 5, HKIE. Earth Technology Corporation.(1986). Pile segment tests Sabine Pass: Some aspects of the fundamental behavior of axially loaded piles in clay soils. ETC Rep. No. 85-7, Houston/Long Beach. Geotechnical Engineering Office(GEO).(1996a). R&D project PIL1a research on large-section excavated piles trial piles, TNK959, Kowloon Bay. SoilTestingRep.No.631, Department of Civil Engineering of the Hong Kong Government of the Special Administrative Region, China. GEO. (1996b). Pile design and construction. GEO Publ. No. 1/96, Department of Civil Engineering of the Hong Kong Government of the Special Administrative Region, China. Ho, K. K. S.(1994). Behaviour of the instrumented trial barrette for the Trademart Development at N.K.I.L. 632, Kowloon Bay. Rep. Spec. Proj. SPR8/94, Geotechnical Engineering Office, Department of Civil Engineering of the Hong Kong Government of the Special Administrative Region, China. Lings,M.L.,Ng,C.W.W.,andNash,D.F.T.(1994). Thepressureof wet concrete in diaphragm wall panels cast under bentonite. Engrg. Proc. Instn. of Civ. Engrs., Geotech. U.K., London, 17(July), Lo, D. O. K.(1997). Behavior of the instrumented test barrette for the west Kowloon corridor stage IV project. Spec. Proj. Rep. SPR 2/97, Geotechnical Engineering Office, Civil Engineering Department of the Hong Kong Special Administrative Region, China. JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2 / 73

63 Tunnelling and Underground Space Technology xxx (212) xxx xxx 2 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile C.W.W. Ng, H. Lu, S.Y. Peng Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong a r t i c l e i n f o Article history: Received 6 March 212 Received in revised form 11 July 212 Accepted 23 July 212 Available online xxxx Keywords: Twin tunnelling Piles Three-dimensional Soil structure interaction Centrifuge modelling Dry sand 1. Introduction a b s t r a c t Tunnel construction inevitably causes soil stress changes in the ground and hence induces ground movements. Uncontrolled ground movements induced by tunnelling may cause cracking in buildings and gas mains, or induce additional loads on piles of nearby structures. In urban cities, it is not uncommon to encounter existing piles during tunnel constructions. Estimation of the effects of tunnelling on existing pile foundations of buildings poses a major challenge to designers. It is particularly vital to estimate the tunnelling effects when two new tunnels are to be built near an existing pile. Bezuijen and Schrier (1994) studied the influence of bored tunnels on pile foundations. They pointed out that the pile settlement can be quite significant if the distance between pile and tunnel is less than the tunnel diameter. Loganathan et al. (2) assessed tunnelling-induced ground deformations and their adverse effects on pile foundations in clay. Tunnelling-induced bending moment and axial force in the piles of a pile group were investigated by modelling volume loss of a tunnel in a single stage. They concluded that the tunnelling-induced bending moment may be critical when Corresponding author. Tel.: ; fax: addresses: [email protected] (C.W.W. Ng), [email protected] (H. Lu), muffinpp@ ust.hk (S.Y. Peng) /$ - see front matter Ó 212 Elsevier Ltd. All rights reserved. Contents lists available at SciVerse ScienceDirect Tunnelling and Underground Space Technology journal homepage: www. elsevier. com/ locate/ tust Tunnelling activity inevitably induces soil stress changes and ground deformation, which may affect nearby existing pile foundations. Although a number of studies have been carried out to investigate the effects of tunnelling on existing piles, the excavation of only one tunnel is often considered. The fundamental interaction between twin tunnel construction and an existing pile foundation has not been thoroughly studied. In this study, a series of three-dimensional centrifuge model tests investigating the effects of twin tunnel construction on an existing single pile in dry sand were conducted. The influence of the depth of each tunnel relative to the pile was investigated by constructing the twin tunnels either close to the mid-depth of the pile shaft or near the pile toe. The pile settlement induced by the excavation of the twin tunnels is found to be closely related to the depth of each tunnel relative to the pile. The measured cumulative pile settlement due to tunnelling near the toe is about 2.2 times of that due to tunnelling near the mid-depth of the pile shaft. Apparent losses of pile capacity of 36% and 2% are identified due to the construction of twin tunnels near the pile toe and at the mid-depth of the pile, respectively. Although there is an increase in the axial force induced in the pile when a tunnel is constructed at the mid-depth of the pile, significant increases in bending moment is not observed in any of the tests. Ó 212 Elsevier Ltd. All rights reserved. the centerline of the tunnel is located near the pile toe. Jacobsz et al. (24) investigated the adverse effects of tunnelling beneath a pile in dry sand. An influence zone was identified above and around the tunnel in which the pile could suffer significant settlement, depending on the volume loss induced by the tunnelling. Lee and Chiang (27) studied the tunnelling-induced bending moment and axial force in a single pile in saturated sand. Tunnels were embedded at various cover-to-diameter ratios. The authors concluded that the depth of the tunnel relative to the pile has a significant influence on the distribution of the bending moment along the pile. As far as the authors are aware, the three-dimensional centrifuge modelling of the effects of twin tunnelling on a single pile has not been reported. Mroueh and Shahrour (22) carried out three-dimensional elastoplastic finite element analyses to study the influence of a tunnel construction on a single pile as well as on pile groups. The numerical results showed that there is a significant reduction in the tunnellinginduced axial force and bending moment in the piles furthest away from the tunnel due to the group effect. Lee and Ng (25) carried out a three-dimensional, elasto-plastic, coupled-consolidation numerical analysis to investigate the effects of an open face tunnel excavation on an existing loaded pile. It is shown that the factor of safety (FOS) of a pile can be reduced from 3. to 1.5 due to the additional settlement of a pile induced by tunnelling when a settlement-based failure criterion (Ng et al., 21a) is used. The effects of tunnel construction on the nearby pile foundations are obviously three dimensional. A fundamental understanding of the three-dimensional tunnel-soil-pile interaction is needed. In addition, few researchers have investigated the effects of twin tunnelling on piles, except Pang (27), who reported the field monitoring and numerical study of the effects of twin tunnelling on an adjacent pile foundation in Singapore. A northbound tunnel and a southbound tunnel were constructed near piles one after the other. The smallest clear distance between the tunnels and piles was 1.6 m. Results of the field study showed that the piles were subjected to a large dragload due to an induced soil settlement in residual soil. However, strain gauges are only instrumented along pile portion near tunnels. The pile settlement due to tunnelling is not reported. In this study, a series of three-dimensional centrifuge model tests were performed to investigate the behavior of a single pile due to the construction of twin tunnels one after the other. The effects of the three-dimensional tunnel excavation process were simulated in-flight by controlling the volume loss at 1.% in each stage of the three-dimensional excavation of each tunnel. The twin tunnels in each test are located at either mid-depth of the pile or the pile toe. In addition to measurements of ground surface settlement and pile settlement, the bending moment and axial force induced in the pile by the tunnelling in different stages of construction were captured. The objective of this study is to investigate the response of an existing pile when a new tunnel excavation is to be carried out nearby. It is intended that results from this study can assist engineers and designers to choose and design the location (i.e., the depth) of the new tunnel. 2. Centrifuge modelling 2.1. Experimental program and setup The fundamental principle of centrifuge modelling is to recreate the stress conditions, which would exist in a full-scale problem, in a model of a greatly reduced scale. This is done by subjecting model components to an enhanced body force, which is provided by centripetal acceleration (rx 2 ) when a centrifuge rotates at a constant angular velocity (x) about the center of the centrifuge arm with radius, r. For instance, an 1 m prototype stress conditions can be replicated in a centrifuge by an 1 m height model when the Earth s gravity (g) is enhanced by 1 times (i.e., rx 2 = Ng = 1g). Thus, centrifuge is suitable for simulating stress-dependent materials such as soils. More details, scaling laws and centrifuge applications are given by Schofield (198), Taylor (1995) and Ng et al. (26). Table 1 summarizes all the relevant scale laws in this study. In total, four centrifuge model tests were carried out at the Geotechnical Centrifuge Facility of the Hong Kong University of Science Table 1 Centrifuge scaling factor. Physical quantity Gravitational acceleration n Length 1/n Area 1/n 2 Volume 1/n 3 Settlement n Stress 1 Strain 1 Force 1/n 2 Density 1 Mass 1/n 3 Flexural stiffness 1/n 4 Bending moment 1/n 3 Scaling factor (model/prototype) and Technology (Ng et al., 21b, 22). The 4g ton centrifuge has an arm radius of 4.2 m and is equipped with a two-dimensional hydraulic shaking table and a four-axis robotic manipulator. All of the centrifuge tests were carried out at an acceleration of 4g. Fig. 1a shows the schematic elevation view of Test T. A single pile is located at the center of each model container. Test T is designed to investigate the behavior of a pile due to a single tunnel constructed near pile toe in dry sand. The model pile had a diameter of 2 mm (.8 m in prototype) and was 6 mm long (24. m in prototype). The pile cap was elevated by 11 mm and therefore the embedded depth of each pile was 49 mm (19.6 m in prototype). The tunnel diameter (D) was 152 mm (6.8 m in prototype). The C/D ratio (cover-to-diameter ratio) of the tunnel is 2.7. The horizontal distance from the centerline of the tunnel to the pile was.75d. In addition, a separate test (Test L) is carried out to obtain the load settlement curve of the single pile without tunnelling effects. This test has the same configuration of Test T but only without the model tunnel. As shown in Fig. 1b, Test TT was designed to study the effects on the pile induced by the construction of twin tunnels, one after the other, near the pile toe. The C/D ratio of each tunnel is 2.7, same as that in Test T. Fig. 1c shows the schematic elevation view of Test SS. This test was designed to investigate pile responses induced by the construction of twin tunnels near the mid-depth of pile shaft. The C/D ratio of each tunnel is 1.5. A summary of the test program is given in Table 2. Fig. 2a and b shows the plan views for Test T and Tests TT and SS, respectively. In Test T, the model tunnel had a length of 228 mm, which was equal to 1.5D. The three-dimensional tunnel construction was simulated in three stages, with the tunnel face advancing by a distance of.5d in each stage. In Tests TT and SS, the longitudinal length of each tunnel was 38 mm, which was equivalent to 2.5D. The tunnel excavation was simulated in five stages, again with the tunnel face advancing by a distance of.5d in each stage. A photograph of the model package is shown in Fig. 3a Simulation of tunnel construction In simulating of tunnel advancement, it is common to model overall volume loss resulting from tunnelling effects in practice (Taylor, 1995; Mair, 28), rather than trying to simulate different construction steps in centrifuge. Obviously, this implies that some construction details like erection and deformation of tunnel liners, stiffness of liners and workmanship are not simulated separately. Only the overall result like volume loss due to actual tunnelling is simulated. Obviously, this type of modelling is not ideal. However, it does capture the essential effects (i.e., volume loss) of tunnelling and can meet the comparative objective of different simulated cases in this paper. In the single tunnel test, the model tunnel consisted of three cylindrical rubber bags. In the twin tunnel tests, each model tunnel consisted of five cylindrical rubber bags (see Fig. 3b). Between two rubber bags was a rigid aluminum divider to control and separate the volumes of water inside so that volume change in each rubber (i.e., the tunnel volume loss) can be controlled independently. Each rubber bag was filled with de-aired water. Three-dimensional tunnel construction was simulated in-flight by draining away a controlled amount of water from each rubber bag one by one. The amount of water drained away was controlled as 1.% of the total volume of the cylindrical rubber bag. This is to simulate an equivalent volume loss of 1.% of excavated cross section area of the tunnel face during each stage of tunnel construction. Since the effects of tunnel excavation were modelled by inducing an equivalent volume loss resulting from various construction factors and tunnel Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212), Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

64 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx 3 4 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx Working load 11 Strain gauges (a) Pile Tunnel 1 Toyoura sand 85 Working load Working load Pile 2nd tunnel 1246 (b) 1st tunnel Toyoura sand 85 Toyoura sand Monitoring section Hydraulic chamber Note: L=LVDT Tunnel L1 Pile L3 L5 1 x 2 L9 3 L2 L4 y Video Load cell Single pile Strongbox Sand (a) Hydraulic jack LVDTs Servo-valve Table 2 Test program. liner in the field, model tunnel liner is not needed to be simulated in the centrifuge tests Model piles and instrumentation The instrumented model pile was fabricated from an aluminum tube (see Fig. 1a). Nine levels of full Wheatstone bridge of strain gauges were installed to record bending moment and axial force along the entire pile length. The strain gauges were protected by a thin layer of epoxy. Prior to a centrifuge test, calibration was carried out to obtain a relationship between an applied bending moment to the pile and the corresponding reading of each full Wheatstone bridge. For structural elements, the bending stiffness (EI) of pile was chosen as the vital governing property to be satisfied. It can be derived that the scaling requirement for a model pile Pile 2nd tunnel 1246 (c) 114 1st tunnel Toyoura sand Fig. 1. Elevation view of centrifuge model: (a) Test T; (b) Test TT; (c) Test SS. All dimensions are in mm in model scale. Test ID C/D Remark Relative density of sand (%) L N/ Pile load test 6 A T 2.7 Single tunnelling near pile toe 6 TT 2.7 Twin tunnelling near pile toe 65 SS 1.5 Twin tunnelling near pile shaft 62 N/A denotes not applicable. 85 (EI) m is equal to N 4 (EI) p (refer to Table 1), where N is the number of times of gravity enhanced in a centrifuge test, E is Young s modulus, I is the second moment of area for a cross-section, and subscripts m and p refer to the model and prototype scale, respectively. The model pile had an axial rigidity (E m A m ) of 7473 kn and a bending rigidity (E m I m ) of 273 N m 2, which are corresponding to prototype E p A p of 11,957 MN and E p I p of 71 MN m 2 of a real concrete pile, respectively. The pile is wished-in-place in the sand bed. Thus, pile installation effects are not simulated. In each test, the pile was loaded in-flight (i.e., while the centrifuge is spinning) using a hydraulic jack. A load cell was installed at the piston of the jack to control applied load. Settlement of the pile was measured by a linear variable differential transformer (LVDT) located at the pile head Model preparation Dry Toyoura sand (G s = 2.65, e max =.977, e min =.597, u cv ¼ 31 ) (Ishihara, 1993) was used in each test. Each centrifuge model was prepared by the pluvial deposition method. Sand was rained from a hopper, which was kept 5 mm above the surface. The measured average relative density of each test is also given in Table 1. Since the scaling factor for soil density is one (see Table 1), soil with a relative density of 6 65% (refer to Table 2) in prototype was modelled in each centrifuge test. It is well understood that the stiffness of a soil (i.e., shear modulus, G or Young s modulus, E s ) is L8 Toyoura sand 2nd tunnel L7 Monitoring section dependent on its density and stress level. The stress level is simulated correctly and prototype soil density is prepared in a centrifuge. Therefore, the stiffness of the ground should be comparable to that of full scale in prototype Test procedure L Pile (a) L (b) Hydraulic chamber Note: L=LVDT 1st tunnel After model preparation, the acceleration of the centrifuge was increased to 4g. The model pile was loaded in-flight at 4g in a number of steps. In each step, an incremental vertical load of 1 N (16 kn in prototype) was applied. Each load increment was maintained for 3 min. Once the load had reached the working load (12 N), tunnel construction with the designed volume loss of 1.% was carried out. Three construction stages were simulated y L1 L3 x Fig. 2. Plan view of centrifuge model: (a) Test T; (b) Test TT and Test SS. All dimensions are in mm in model scale. L2 L4 59 L in-flight by draining away water from each of the rubber bags one after the other in Test T, as shown in Fig. 2a. A similar procedure was adopted in Tests TT and SS, except five tunnel construction stages were simulated. Throughout each test, ground surface settlement, settlement of the pile, induced bending moment and axial force along the pile were recorded. 3. Test results Model tunnels Strain gauges (b) Model pile Sand Fig. 3. (a) A typical model setup; (b) model tunnels and model pile. All test results presented in this paper are in prototype scale, unless stated otherwise Determination of the axial load carrying capacity of the pile Prior to tunnelling, it is necessary to obtain the capacity of the pile so that the working load can be deduced. A pile load test Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212), Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

65 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx 5 6 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx Settlement (mm) (i.e., Test L) was carried out. Fig. 4 shows the measured load settlement relationship. The load applied to the pile cap was gradually increased to 4 MN at increments of 1 kn in each step. The ultimate axial load capacity was determined based on a displacement-based failure criterion proposed by Ng et al. (21a). This failure criterion is expressed as follows: d ph;max ffi :45d p þ 1 2 After tunnelling Failure criterion proposed by Ng et al. (21a) P h L p A p E p where d ph,max is the maximum pile head movement which defines the ultimate load, P h is the pile head load, L p is the pile length, E p is the elastic modulus of pile shaft, A p is the cross-sectional area of the pile, and d p is the pile diameter. The failure criterion proposed by Ng et al. (21a) is a semi-empirical method for estimating an approximate interpreted failure load for pile. The method is based on a moderately conservative estimation of the movement required to mobilize toe resistance and incorporates observations of shaft shortening from actual pile loading tests. Both the 5%D criterion proposed by O Neill and Reese (1999) and 1%D criterion proposed by Weltman (198) do not include shortening of pile shaft and thus they may not be appropriate for long piles. Both criteria are displacement-based failure criteria. The 5%D and 1%D criterion defines the failure load of pile as the load causing a settlement of 5% and 1% of the pile diameter, respectively. As shown in the figure, the ultimate load capacity of the pile was 2.88 MN. A working load of 1.92 MN was adopted with a factor of safety of 1.5. A pile settlement of 1.6%d p was observed due to the applied working load Ground surface settlement Load (kn) Ultimate load capacity of pile Equivalent pile load after tunnelling Working load Fig. 4. Load settlement relationship obtained from load test without tunnelling (Test L). Fig. 5a shows the extents of ground surface settlement (S) measured by LVDTs 1 8 (as shown in Fig. 2) in Tests T and TT. Both the measured surface settlement and the transverse distance from the centerline of the first tunnel in Test TT and the only tunnel in Test T (x) were normalized by the diameter of tunnel (D). In Test T, only half of the ground surface settlement trough was measured. At the end of the three stages of tunnel excavation, the measured maximum settlement was.27%d above the centerline of the tunnel. At a distance of 2D from the centerline of the tunnel, a surface settlement of.3%d was also measured. In Test TT, a maximum settlement of.24%d occurred at the centerline of the first tunnel after its completion. The consistency between the ð1þ x/d two measured maximum settlement values illustrates the repeatability of the tests. After the excavation of the second tunnel in Test TT, the measured maximum surface settlement above the second tunnel was almost the same as that due to the construction of the first tunnel. The measured maximum accumulated settlement was.35%d after the excavation of the twin tunnels. Since no LVDT was installed between the centerlines of the two tunnels, the location where the maximum settlement occurred could not be identified. However, the maximum settlement was unlikely to have occurred above the centerline of the second tunnel. The ground surface settlement profile due to tunnel construction may be represented by a Gaussian distribution (Peck, 1969). Ground surface settlement S is defined as S ¼ S max expð x 2 =2i 2 Þ V s S/D (%) ndtunnel 1st tunnel 1. Measured, after 1st tunnelling (T) Measured, after 1st tunnelling (TT) Fitted, after 1st tunnelling (TT) Measured, after 2nd tunnelling (TT, cumulative) Predicted, after 2nd tunnelling (TT) S/D (%) 2nd tunnel (a) x/d 1st tunnel (b) Measured, after 1st tunnelling (SS) Fitted, after 1st tunnelling (SS) Measured, after 2nd tunnelling (SS, cumulative) Predicted, after 2nd tunnelling (SS) Fig. 5. Ground surface settlement induced by tunnel excavation: (a) Test T and Test TT; (b) Test SS. S max ¼ pffiffiffiffiffiffi ð3þ 2p i where S max is the maximum settlement at the tunnel centerline, i and x are the lateral distances from the tunnel centerline to the point of inflection and any other points on the settlement trough, respectively. V s is the volume of settlement trough. Eq. (2) was adopted to fit the measured values of ground surface settlement induced by excavation of the first tunnel in Test TT. As expected, the measured S max was located above the centerline of the first tunnel. The best-fitted curve using an i of 7.5 m is shown in Fig. 5a. As proposed by O reilly and New (1982), i can be represented by KZ, where Z is the vertical distance from ground surface to the center of tunnel. Following the above equation, the deduced ð2þ K value for the best-fitted curve is.39, which lies between the values.25 and.45 suggested by Mair and Taylor (1997) for tunnelling in sand. Based on the design chart proposed by Peck (1969), the maximum settlement calculated using Eq. (3) is 38 mm (.63%D), which is more than twice the measured data. This is consistent with the findings by Attewell and Farmer (1974), who also observed that the method proposed by Peck (1969) overestimates the maximum settlement for tunnelling in sand. To predict the surface settlement above twin tunnels, Attewell et al. (1986) suggested summing the Gaussian curves induced by two tunnels. In this study, the best-fitted Gaussian curve is deduced from excavation of the first tunnel, as shown in Fig. 5a. Incremental settlement due to the second tunnelling is assumed the same as that during the first tunnel excavation. Superimposed curve based on the two Gaussian curves are also shown in the figure. It can be seen that the combined settlement curve fits quite well with the measured values at the end of the second excavation, except that settlement above the shoulder of the first tunnel. Fig. 5b shows the measured surface settlement profiles in Test SS. After excavation of the first tunnel, a maximum settlement of.5%d occurred above the centerline of the first tunnel. After excavation of the second tunnel, the location of maximum surface settlement shifted to above the centerline of the second tunnel and the measured maximum value was.67%d. By fitting a Gaussian curve to the measured surface settlement profile after the construction of the first tunnel in Test SS, the fitted K value was found to be.39, consistent with that in Test TT. Similarly, by summing the two best-fitted Gaussian curves, the resulting settlement distribution is also included for comparison in Fig. 5b. The maximum settlement obtained from the summation of the two fitted curves was located above the centerline of the twin tunnels. Moreover, the measured settlement was 16% larger than the combined maximum value. It is evident that the incremental maximum settlement induced by the excavation of the second tunnel was larger than that induced by the first tunnel. Addenbrooke (1996) and Chapman et al. (27) also found that the incremental surface settlement induced by the second tunnel is larger than that induced by the first tunnel when C/D of twin tunnels is larger than 3., which is higher than Test TT in this study. However, it should be noted that volume loss in each individual tunnel was not controlled in their studies. On the contrary, this study simulates tunnelling by controlling the volume loss to be 1% in all tests. This might be the reason why the magnitudes of incremental settlements induced by the twin tunnelling are close in Test TT. Comparing the measured surface settlements in Tests TT and SS, it can be observed that the maximum surface settlement in Test TT (i.e.,.35%d) was substantially smaller than that in Test SS (i.e.,.67%d). This was because C/D (=1.5) the twin tunnels in Test SS was smaller than that in Test TT (C/D = 2.7). Mair et al. (1993) reported that the maximum surface settlement above one tunnel is inversely proportion to the depth of the tunnel for a given volume loss, tunnel diameter and constant value of K. Based on this study, it is evident that the surface settlement induced by twin tunnelling also decreases with increasing C/D ratio Pile settlement and apparent loss of pile capacity due to tunnelling Fig. 6 shows the development of the normalized pile settlement (S p ) during each tunnel construction stage. Location of the tunnel at any stage is indicated by the distance between tunnel face to the centerline of the pile (y). Both the measured S p and the distance from the tunnel face to the centerline of the pile (y) were normalized by the tunnel diameter (D). In Test T, as the tunnel face advanced at a depth of C/D = 2.7 from y/d =.75 to.25, a pile settlement of.4%d was T TT, first tunnel Sp/D (%) TT, second tunnel only SS, first tunnel.25 SS, second tunnel only.3 y/d Fig. 6. Pile settlement induced by excavations of twin tunnels. induced. A significant increase in pile settlement (.12%D) occurred when the tunnel face advanced from y/d =.25 to.25. When the tunnel face reached y/d =.75, the pile settlement increased to.22%d (1.7% of the pile diameter). About 55% of the pile settlement was induced when the tunnel face was between y/ D =.25 and.25. In Test TT, it can be seen that the development of the pile settlement during each tunnel construction stage was very similar, in terms of the pile settlement profile and magnitude, to that observed in Test T. Significant increases in pile settlement also occurred when the tunnel face was between y/d =.25 and.25. When the face of either tunnel advanced from y/d = , the pile settlement only increased from.22%d to.25%d (1.9% of the pile diameter). About 9% of the pile settlement induced by each tunnel occurred when the face of each tunnel was between y/d =.75 and.75. Thus, the tunnelling influence zone of pile settlement was identified to be between y/d =.75 and.75. At C/D = 2.7, surprisingly the excavation of the first tunnel had almost the same effects on the pile settlement as the excavation of the second tunnel. It may suggest that each individual tunnel construction induced limited plastic strains around the pile toe. This implies that soil around the pile toe remains almost elastic. Hence, the influence of the second tunnelling on pile settlement is almost the same as that resulted from the first tunnel. In Test SS, the induced settlement increased almost linearly as the excavation of the first tunnel progressed at C/D = 1.5. After the excavation of the first tunnel, a pile settlement of.15%d (1.1% of the pile diameter) was measured. The induced pile settlement due to the first tunnel in this test was 4% less than that due to the first tunnel in Test TT. Based on their centrifuge tests, Lee and Chiang (27) also reported that pile settlement induced by tunnelling near the mid-depth of pile shaft is smaller than that induced by tunnelling near the pile toe. This is because the load transfer mechanisms of the two cases are different (to be discussed in detail later). The pile settlement due to the excavation of the second tunnel in this study was.7%d (.5% of the pile diameter), which was only about 47% of that due to the first tunnel. This may be because a significant amount of the vertical load applied to the pile was transferred downwards to the pile toe after the excavation of the first tunnel. Therefore, the effects of the construction of the second tunnel on pile settlement were smaller than those of the construction of the first tunnel. By comparing the measured results between Tests TT and SS, it is evident that the pile settlement induced by twin tunnelling is closely related to the relative location of a tunnel to the pile. Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212), Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

66 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx 7 8 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx Sp /D (%) C/D T, after tunnelling TT, after 1st tunnelling SS, after 1st tunnelling Fig. 7 shows the normalized pile settlement induced by tunnelling at different C/D ratios. Open symbols denote pile settlements induced by the first tunnel, whereas solid symbols are cumulative pile settlements induced by both tunnels. After excavating the first tunnel, the pile settlement was about.15%d for C/D = 1.5 in Test SS, whereas the pile settlements were about.23%d for C/D = 2.7 in Test T and Test TT. It is clear that tunnelling near the pile toe induced 1.5 times of the pile settlement than tunnelling at the mid-depth of the pile shaft. After the excavation of the twin tunnels in Tests TT and SS, the cumulative pile settlement increased to.23%d for C/D = 1.5 in Test SS and to.5%d for C/D = 2.7 in Test TT. It is evident that twin tunnelling near the pile toe induced about 2.2 times of pile settlement than tunnelling at the mid-depth of pile shaft. Since pile capacity is often interpreted using settlement criteria, the induced pile settlement due to tunnelling can be considered as an apparent loss of pile capacity (ALPC). Before tunnel excavation, pile settles 12 mm due to the initial applied working load (1.92 MN). An additional pile head settlement of 15 mm (.24%D) is induced due to tunnelling in Test T (see Fig. 6). By using the load-settlement relationship shown in Fig. 4, a total pile settlement of 27 mm may be regarded as an equivalent load of 2.49 MN applied to the pile. Thus, the increase in equivalent pile load can be calculated to be.57 MN (i.e., MN) due to the tunnel excavation. Since the ultimate load capacity of the pile was 2.88 MN as obtained from the load settlement curve using the displacement-based failure criterion proposed by Ng et al. (21a), it can be considered that an ALPC of 2% occurred due to the tunnel excavation. The ALPC was about 21% after excavating the first tunnel in Test TT, but it increased to about 36% after the second tunnel was constructed. In Test SS, the ALPCs were about 14% and 2% after the first and second tunnels were constructed, respectively. The ALPCs suggest that the serviceability limit state of the pile after tunnelling should be considered Axial forces along the pile TT, after 2nd tunnelling (cumulative) SS, after 2nd tunnelling (cumulative) Fig. 7. Pile settlement induced by tunnelling at different C/D ratios. Fig. 8a shows the measured axial force along the pile in Test T. The depth (z) was normalized by the tunnel diameter. Before the tunnel face reached the pile (i.e., y/d =.25), no significant change in axial force was recorded. As the tunnel face reached the pile toe (i.e., y/d =.25), the axial force along the pile decreased. This may imply that the shaft resistance of the pile was further mobilized when the pile settled more than the surrounding soil. This is consistent with the significant pile settlement in Fig. 6. The maximum reduction in axial force occurred at z/d = 2. and the magnitude of maximum reduction was kn (i.e., 6.2% of the working load). Although the end bearing resistance was not measured at the pile toe directly in this study, the measured axial force at z/d = 3.1 (.6 m above the toe of the 19.6 m pile, as shown in Fig. 1) may be used to deduce the variation of end bearing resistance due to tunnelling. The measured axial force at z/d = 3.1 decreased as the tunnel face reached the pile toe. This revealed that there was a reduction in toe resistance due to stress relief which resulted from the 1% volume loss during tunnelling. As the tunnel face passed the pile (i.e., y/d =.75), the axial force along the pile increased. It can be observed that the maximum reduction in axial force occurred when tunnel face finally reached the piles. The magnitude of maximum reduction in axial force cannot be captured in plane strain model tests. This illustrates the importance of carrying out three-dimensional model tests. Fig. 8b shows the measured axial forces along the pile in Test TT. The development of axial force due to the first tunnelling was consistent with that in Test T. As expected, the maximum reduction in axial force also occurred when the first tunnel finally reached the pile (i.e., y/d =.25). As the excavation of the first tunnel continued to a distance of y/d = 1.25 away from the pile, the axial force almost reduced to the value before tunnel excavation. When the second tunnel was being excavated, the maximum reduction in axial force was kn at z/d = 1.9. The reduction was more than 197% larger than that caused by the excavation of the first tunnel (118.6 kn). After the excavation of the second tunnel, the final reduction in axial force decreased to 75% of its maximum value. This illustrates the complex load transfer among the soil, the pile and the tunnels during tunnelling which could only be captured in three-dimensional simulations. Fig. 8c shows the axial forces along the pile in Test SS. In contrast to the reduction in axial forces measured in Tests T and TT, the axial force in the pile increased during the first tunnelling at the mid-depth of the pile (i.e., z/d = 2.). The maximum increase in axial force recorded occurred at z/d = 1.55 above the first tunnel when its face reached y/d = When the second tunnel was being constructed, the measured axial force continued to increase but at a reduced rate until the end of tunnel construction. The final maximum recorded increase in axial force was kn (22% of the working load), which was 32% larger than that measured at the end of the first tunnel (32. kn or 16.7% of the working load). By inspecting the distributions of axial force closely, it can be seen that the axial force increased at depths above the tunnels. This suggests that there was a decrease in shaft resistance caused by downward soil movement and also by the reduction in confining stress due to the 1% volume loss from each tunnel excavation. In contrast to the induced axial force above z/d = 1.5, the induced axial force below z/d = 1.5 decreased with depth. This implies that there was an increase in shaft resistance below z/d = 1.5 when the pile settled during tunnel construction as shown in Fig. 6. The measured axial force at z/d = 3.1 (near the toe) increased by kn (i.e., 6.5% of the working load) at the end of the construction of the first tunnel Shaft resistance and load transfer mechanisms Fig. 9a shows the average shaft resistance along the pile in Test T and Test TT. Each pile shaft is divided into two parts: the upper portion of pile shaft above the depth of tunnel crown and the lower portion below tunnel crown. Based on the axial force measured using strain gauges, the average unit skin friction (f) may be calculated as follows: f ¼ DQ pl where DQ is the difference between measured axial loads from any two consecutive strain gauges, l is the length of each portion, and p ð4þ z/d Axial force (kn) Before tunnelling y/d=-.25 y/d=.25 y/d=.75 Tunnel springline (a) is the perimeter of the pile. In this way, load transfer mechanism of the pile can be clearly illustrated. In Test T, as the tunnel face advanced from y/d =.75 to.25, no significant change in shaft resistance was observed. However, as the tunnel face further z/d -.5 z/d Before tunnelling Axial froce (kn) 1st tunnel_y/d=-.75 1st tunnel_y/d=.25 1st tunnel_y/d=1.25 2nd tunnel_y/d=-.75 2nd tunnel_y/d=.25 2nd tunnel_y/d=1.25 Tunnel springline (b) Before tunnelling Axial force (kn) 1st tunnel_y/d=-.75 1st tunnel_y/d=.25 1st tunnel_y/d=1.25 2nd tunnel_y/d=-.75 2nd tunnel_y/d=.25 2nd tunnel_y/d=1.25 Tunnel springline (c) Fig. 8. Axial forces of pile at various tunnelling stages: (a) Test T; (b) Test TT; (c) Test SS. advanced from y/d =.25 to.25, the average shaft resistance of the pile below the tunnel crown (i.e., z/d > 2.7) decreased significantly and that above the tunnel crown (i.e., z/d < 2.7) increased significantly. Due to reduction in confining stress induced by tunnel Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212), Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

67 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx 9 1 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx Average shaft resistance (kpa) Average shaft resistance (kpa) y/d (a) T, above tunnel crown T, below tunnel crown TT, 1st tunnel, above tunnel crown TT, 1st tunnel, below tunnel crown TT, 2nd tunnel, above tunnel crown TT, 2nd tunnel, below tunnel crown SS, 1st tunnel, above tunnel crown SS, 1st tunnel, below tunnel crown SS, 2nd tunnel, above tunnel crown SS, 2nd tunnel, below tunnel crown y/d (b) Fig. 9. Average shaft resistance along pile: (a) Test T and Test TT; (b) Test SS. excavation, the end bearing resistance and shaft resistance of the pile at depths below the tunnel crown (i.e., z/d > 2.7) was reduced. In order to maintain vertical equilibrium, the pile had to settle more than the surrounding soil to mobilize shaft resistance at the upper portion of the pile shaft. This caused an increase in load taken by the upper portion and is called upward load transfer. As the tunnel face passed the pile (i.e., as it advanced from y/ D =.25.75), the shaft resistance of the pile above the tunnel crown (i.e., z/d < 2.7) decreased whereas that below the tunnel crown (i.e., z/d > 2.7) increased. Due to the downward soil movement and the decrease in normal stress acting on the pile induced by tunnel excavation, the shaft resistance decreased along the pile above the tunnel crown (i.e., z/d < 2.7). To maintain the vertical equilibrium, the shaft resistance along the pile below the tunnel crown (i.e., z/d > 2.7) and the end bearing resistance increased as shown in Fig. 8c. The decrease of pile load from the upward portion resulted in an increase of load in the lower portion of the pile for maintaining the vertical equilibrium. This is called downward load transfer. This is in contrast to the load transfer mechanism when tunnel face was close to the pile (y/d =.25 and.25). As a result, the axial force increased when the tunnel face passed the pile and advanced from y/d = Therefore, the maximum reduction in axial force occurred when the tunnel face reached the pile as shown in Fig. 8. This is due to the two distinct load transfer mechanisms. In Test TT, during the excavation of the first tunnel, the development of average shaft resistance above and below the tunnel crown was consistent with that in Test T. Two types of load transfer can again be observed. When the second tunnel was being excavated, the reduction in average resistance below the tunnel crown was larger than that in the excavation of the first tunnel. This is because part of the applied working load had been transferred to the lower part of the pile after the excavation of the first tunnel, as illustrated in Fig. 8b. Fig. 9b shows the average shaft resistance along the pile in Test SS. Unlike in Test TT, only downward load transfer was observed in Test SS. During the excavation of the first tunnel, the average shaft resistance decreased significantly above the tunnel crown (i.e., z/ D < 1.5) and increased below the tunnel crown (i.e., z/d < 1.5). Because vertical equilibrium of pile can be achieved by transferring the load taken by the upper part of the pile to the lower part of the pile, settlement of pile was relatively small compared with that when the tunnel face was near the pile toe as illustrated in Fig. 6. This is consistent with the centrifuge results reported by Lee and Chiang (27). When the second tunnel was being excavated, the average shaft resistance decreased above the tunnel crown (i.e., z/d < 1.5) and increased below the tunnel crown (i.e., z/d < 1.5), similar to that when the first tunnel was being excavated. At the end of the construction of the two tunnels, the mobilized shaft resistance of the pile was significantly reduced to almost zero above the crown of each tunnel. The change in average shaft resistance due to the excavation of the second tunnel was much smaller than that due to the excavation of the first tunnel. This is consistent with the measured pile settlement results shown in Fig. 6. The ALPC induced by the excavation of the second tunnel (6%) was only 43% of that induced by the first tunnelling (14%). After the excavation of the first tunnel, a significant amount of applied working load was transferred to the pile shaft below the tunnel crown and the pile toe. Thus, the effect of the second tunnel at the same depth was smaller. Clearly, the three-dimensional load transfer mechanism for twin tunnelling near the pile shaft is different from that for twin tunnelling near the pile toe. A pile suffers significant settlement when an upward load transfer is observed, but it experiences smaller settlement when the load transfer is downward. In this study, the ALPC for twin tunnelling near the pile toe (36%) was 1.8 times of that for tunnelling near the mid-depth of pile shaft (2%), as illustrated in Fig Induced bending moments in the pile Fig. 1a shows the measured tunnelling induced bending moment along the pile in Test T and Test TT. Bending moments were taken to be positive if tensile stress was induced at the side facing the first tunnel. In Test T, the maximum tunnelling-induced bending moment occurred near the tunnel crown (i.e., z/d = 2.7) with a magnitude of 61.5 kn m, which was 7.7% of the bending moment capacity of the pile (M yield = 8 kn m). In Test TT, the maximum bending moment occurred near the crown of the first tunnel (i.e., z/d = 2.7) after its excavation. The maximum induced bending moment was 84.3 kn m (1.5% of M yield ). After the excavation of the second tunnel, the maximum bending moment decreased to 7.8 kn m (8.9% of M yield ). Centrifuge test results reported by Loganathan et al. (2) are also included in the same figure for comparison. The maximum bending moment also occurred near the tunnel crown based on their test results. The bending moment at z/d < 2. in this study is different from that reported by Loganathan et al. This may be because the pile head in this study was constrained by the hydraulic jack. The general profiles of bending moment are similar though. Fig. 1b shows measured tunnelling induced bending moment along the pile in Test SS. After the first tunnel was excavated, the maximum bending moment occurred above the tunnel crown (i.e., z/d =.8). The magnitude is kn m (14.2% of M yield ). A bending moment of 38.7 kn m occurred near the ground surface. This may be also due to constrain of the hydraulic jack. After the second tunnelling, the maximum bending moment was 136. kn m (17.% of M yield ). It is the maximum value in all the three tests. Therefore, it can be concluded that the induced bending moment to the pile by twin tunnelling is relatively insignificant. 4. Summary and conclusions A series of centrifuge model tests were carried out to investigate the effects of twin tunnel construction on an existing single pile in dry sand. In each centrifuge model test, two tunnels were simulated three-dimensionally one after the other in-flight. Based on the test results, the following conclusions may be drawn: (a) The settlement of a pile induced by twin tunnelling is closely related to the depth of each tunnel relative to the pile. Near the pile toe (i.e., Test TT), the excavation of the first tunnel results in a pile settlement of about 1.9% of the pile diameter. The magnitude of incremental pile settlement due to the construction of the second tunnel only is about the same. Based on the displacement-failure load criterion proposed by Ng et al. (21a), the apparent loss of pile capacity (ALPC) is about 21% after the construction of the first tunnel construction, and increases to about 36% (cumulative) after the construction of the second tunnel. (b) For twin tunnelling near the mid-depth of the pile shaft (i.e., Test SS), the pile settlement induced by excavating the first tunnel is 1.1% of the pile diameter, which is only about 6% of that induced by the first tunnel constructed near the toe in Test TT. Due to the load transfer from the upper to lower part of the pile after the construction of the first tunnel, the construction of the second tunnel (also near T, after tunnelling z/d TT, after 1st tunnelling TT, after 2nd tunnelling Centrifuge test in clay (Loganathan et al., 2) 3.5 Induced bending moment (knm) (a) Tunnel springline Induced bending moment (knm) z/d Tunnel springline After 1st tunnelling After 2nd tunnelling (b) Fig. 1. Tunnelling-induced bending moment on pile: (a) Test T and Test TT; (b) Test SS. the pile shaft) induces a much smaller pile settlement of only.5% of the pile diameter. The ALPCs are about 14% and 2% after the excavation of the first and second tunnels, respectively. (c) The pile settlement induced by twin tunnelling is closely related to C/D ratios (cover-to-diameter ratio) of tunnels and the relative location of a tunnel to the pile. The cumulative pile settlement due to tunnelling near the toe is about 2.2 times of that due to tunnelling near the mid-depth of the pile shaft. (d) For construction of each tunnel near the pile toe, two distinct load transfer mechanisms, i.e., an upward first and then a downward load transfer can be identified in Tests T and TT. An upward load transfer is identified when a tunnel face approaches to within.25d of the pile. Due to the reduction in confining stress as a result of the approaching tunnel near the toe, both the shaft resistance near the toe and the end bearing resistance of the pile decrease. As a result, a larger shaft resistance is mobilized in the upper half of the pile (or upper load transfer) while the pile settles to maintain the vertical equilibrium under the applied pile load. On the contrary, a downward load transfer is observed after the tunnel face has passed the pile by a distance of.25d. The shaft resistance along the upper portion of the pile decreases due to downward soil movement and stress relief induced by tunnelling, resulting in an increase in the shaft resistance along the lower portion of the pile and in the end bearing resistance so that the vertical equilibrium of the pile can be maintained. Due to the two distinct types of load transfer, the maximum reduction in axial force occurs when a tunnel face reaches the pile (in Test TT). (e) When each tunnel is excavated near the mid-depth of the pile shaft (in Test SS), only a downward load transfer can be identified. The shaft resistance along the upper half of the pile decreases whereas the shaft resistance along the lower half of the pile increases as tunnel advances. However, there is an increase of 32 kn and 423 kn (i.e., equivalent to Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212), Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

68 C.W.W. Ng et al. / Tunnelling and Underground Space Technology xxx (212) xxx xxx 11 Tunnelling and Underground Space Technology 28 (212) % and 22% of the working load) in the pile axial force above the crown after the construction of the first and second tunnels, respectively. At the end of the construction of the two tunnels, shaft resistance of the pile is significantly reduced to almost zero above the crowns of the tunnels. (f) The induced bending moment due to twin tunnelling is insignificant. The maximum bending moment induced in the pile by excavation of single or twin tunnels is only about 17% of the ultimate bending moment capacity. (g) When C/D is 1.5, the surface settlement induced by the excavation of the second tunnel is larger than that induced by the excavation of the first tunnel. Simply predicting ground surface settlement by summing the two Gaussian curves can underestimate the actual value by up to 16%. (h) In this study only volume loss induced due to tunnelling is modelled. The effect of removal of soil inside of tunnel is not simulated. Since all the tests were carried out using the same method (by simulating volume loss), measured results obtained from different cases investigated are comparable and conclusions drawn from this study should not be affected. Acknowledgements The authors would like to acknowledge the financial support provided by the Research Grants Council of the HKSAR (General Research Fund Project No ). References Addenbrooke, T.I., Numerical Modeling in Stiff Clay. Ph.D. Thesis, Imperial College, London. Attewell, P.B., Farmer, I.W., Ground deformations resulting from shield tunnelling in London clay. Can. Geotech. J. 11, Attewell, P.B., Yeates, J., Selby, A.R., Soil Movements Induced by Tunnelling and Their Effects on Pipelines and Structures. Chapman and Hall, New York. Bezuijen, A., Schrier, J.S., The influence of a bored tunnel on pile foundations. Centrifuge 94, Singapore, pp Chapman, D.N., Ahn, S.K., Hunt, D.V.L., 27. Investigating ground movements caused by the construction of multiple tunnels in soft ground using laboratory model test. Can. Geotech. J. 44 (1), Ishihara, K., Liquefaction and flow failure during earthquakes. Géotechnique 43 (3), Jacobsz, S.W., Standing, J.R., Mair, R.J., Hahiwara, T., Suiyama, T., 24. Centrifuge modeling of tunnelling near driven piles. Soil Found 44 (1), Lee, C.J., Chiang, K.H., 27. Responses of single piles to tunnelling-induced soil movements in sandy ground. Can. Geotech. J. 44 (1), Lee, T.K., Ng, C.W.W., 25. Effects of advancing open face tunnelling on an existing loaded pile. J. Geotech. Geoenviron. Eng., ASCE 131 (2), Loganathan, N., Poulos, H.G., Stewart, D.P., 2. Centrifuge model testing of tunnelling-induced ground and pile deformations. Géotechnique 5 (3), Mair, R.J., Taylor, R.N., Bracegirdle, A., Subsurface settlement profiles above tunnels in clay. Géotechnique 43 (3), Mair, R.J., Taylor, R.N., Bored tunnelling in the urban environment. State-ofthe-art report and theme lecture. In: Proceedings of 14th International Conference on Soil Mechanics and Foundation Engineering, vol. 4, Hamburg, Balkema, pp Mair, R.J., 28. Tunnelling and geotechnics: new horizons. Geotechnique 58 (9), Mroueh, H., Shahrour, I., 22. Three-dimensional finite element analysis of the interaction between tunnelling and pile foundations. Int. J. Numer. Anal. Methods Geomech. 26, Ng, C.W.W., Yau, T.L.Y., Li, J.H.M., Tang, W.H., 21a. New failure load criterion for large diameter bored piles in weathered geomaterials. J. Geotech. Geoenviron. Eng., ASCE 127 (6), Ng, C.W.W., Van Laak, P.A., Tang, W.H., Li, X.S., Zhang, L.M., 21b. The Hong Kong geotechnical centrifuge. In: Proc. 3rd Int. Conf. Soft Soil Engineering, pp Ng, C.W.W., Van Laak, P.A., Zhang, L.M., Tang, W.H., Zong, G.H., Wang, Z.L., Xu, G.M., Liu, S.H., 22. Development of a four-axis robotic manipulator for centrifuge modeling at HKUST. In: Proc. Int. Conf. Physical Modeling in Geotechnics, St. John s Newfoundland, Canada, pp Ng, C.W.W., Zhang, L.M., Wang, Y.H., 26. In: Proceedings of 6th Int. Conf. on Physical Modelling in Geotechnics (TC2), vols. 1 and 2. Taylor & Francis. O Reilly, M.P., New, B.M., Settlements above tunnels in the United Kingdomtheir magnitude and prediction. In: Proceedings of the Conference Tunnelling 82, IMM London, pp O Neill, M.W., Reese, R.C., Drilled Shaft: Construction Procedures and Design Methods. Federal Highway Administration, Washington, DC. Pang, C.H., 27. The Effect of Tunnel Construction on Nearby Pile Foundation. Ph.D. Thesis, National University of Singapore. Peck, R.B., Deep excavation and tunnelling in soft ground. In: Proc. 7th Int. Conf. Soil Mech. Found. Engng., pp Schofield, A.N., 198. Cambridge geotechnical centrifuge operations. Géotechnique 3 (3), Taylor, R.N., Geotechnical Centrifuge Technology. Blackie Academic and Professional, London. Weltman, A.J., 198. Pile Load Testing Procedures. DOE and CIRIA Piling Devel. Group Rep. PG7. Construction Industry Research and Information Association, London. Centrifuge and numerical investigation of passive failure of tunnel face in sand K.S. Wong a,, C.W.W. Ng a, Y.M. Chen b, X.C. Bian b a Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong b Department of Civil Engineering, Zhejiang University, 388 Yuhangtang Road, Hangzhou 3158, China a r t i c l e i n f o Article history: Received 19 July 211 Received in revised form 2 October 211 Accepted 13 December 211 Available online 9 January 212 Keywords: Tunnel face Failure mechanism Passive failure pressure Surface displacement Sand Contents lists available at SciVerse ScienceDirect Tunnelling and Underground Space Technology journal homepage: www. elsevier. com/ locate/ tust a b s t r a c t Tunnelling is one of the major construction methods to sustain the increasing demand on development in cities. Although many studies have been carried out to investigate the active failure mechanism of tunnel face in sand, the study of passive failure of tunnel face is relatively rare and most of studies are analytical solutions based on the upper bound theorem. In this paper, centrifuge model tests and three-dimensional finite element analyses have been conducted to study the passive failure mechanisms of tunnel face in sand for tunnels located at cover over diameter (C/D) ratios equal to 2.2 and 4.3. Passive failure pressures of tunnel face as well as ground surface displacements were investigated in centrifuge tests. From both centrifuge and numerical investigations, it is found that for a tunnel located at C/D ratio equals to 2.2, the soil in front of the tunnel face is displaced by the advancing tunnel face while the soil further away from the tunnel face is forced upwards to the ground surface. A funnel-type failure mechanism is observed and the extent of the failure mechanism is narrower than a five-block failure mechanism commonly assumed in an existing upper bound solution. However, the calculated passive failure pressure by the upper bound solution is fairly consistent with the measured face pressure. It is observed that the funnel-type failure mechanism induces surface heaves. Both observed longitudinal and transverse heaves are well described by Gaussian distributions. For a tunnel located at C/D ratio equals to 4.3, the displacements of soil are confined around the vicinity of an advancing tunnel face and a localised failure mechanism associated with ground settlement is observed and computed. There is a large discrepancy between the localised failure mechanism and the five-block failure mechanism. The calculated failure face pressure is higher than the corresponding measured value by 5%. However, reasonable consistency can be found between measured and computed passive face pressures. Ó 211 Elsevier Ltd. All rights reserved. 1. Introduction In cities, tunnelling has become one of the major construction methods to sustain the increasing demand on development. When a tunnel is excavated, the tunnel face pressure shall be maintained within the minimum and maximum support pressures to prevent active or passive failure at the tunnel face. The failure may endanger human life and cause catastrophic damage to the structures within the influence zone. Over the decades, numerous theoretical and experimental studies have been performed to investigate the active failure of tunnel face (Anagnostou and Kovà ri, 1994, 1996; Chambon and Corte, 1994; Dias et al., 28; Leca and Dormieux, 199; Leca and Panet, 1988; Mollon et al., 21; Soubra, 22). But the study of passive failure of tunnel face is relatively rare and most of them are analytical solutions and numerical simulations (Dias et al., 28; Leca and Dormieux, 199; Mollon et al., 21; Soubra, 22; Soubra Corresponding author. Tel.: / ; fax: addresses: [email protected] (K.S. Wong), [email protected] (C.W.W. Ng), [email protected] (Y.M. Chen), [email protected] (X.C. Bian). et al., 28). Passive failure during drilling of the 2nd Heinenroord Tunnel (Bezuijen and Brassinga, 25) imply that passive failure at tunnel face is not a theoretical risk. It is of important to know the passive failure pressure of tunnel face to prevent the passive failure. Leca and Panet (1988) and Leca and Dormieux (199) used lower and upper bound theorems to predict the passive failure pressure of tunnel face in sand. For upper bound solutions, they adopted a truncated cone as the failure mechanism. It is a planar on vertical plane of symmetry running longitudinally along centreline of tunnel. Soubra (22) improved the upper bound solutions by introducing a log spiral instead of planar on the vertical plane of symmetry. The log spiral was idealised using multiple truncated rigid cones. But only an inscribed elliptical area to the entire circular tunnel face is involved in deriving the upper bound solutions, due to the conical shape of the rigid cones used in the derivation (Mollon et al., 21). While existing analytical and numerical analyses provide valuable information and knowledge on passive failure of tunnel face, the problem has not been studied systematically and fully understood. In view of this, centrifuge model tests were carried out to /$ - see front matter Ó 211 Elsevier Ltd. All rights reserved. doi:1.116/j.tust Please cite this article in press as: Ng, C.W.W., et al. Three-dimensional centrifuge modelling of the effects of twin tunnelling on an existing pile. Tunnel. Underg. Space Technol. (212),

69 298 K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) investigate passive failure of tunnel face in sand for tunnels located at cover over diameter, C/D ratios equal to 2.2 and 4.3. In addition, finite element analyses were conducted to back-analyse the centrifuge model tests and provide further information to understand the problem. This paper presents: (a) details of centrifuge model setup and test procedures; (b) details of the finite element analyses; (c) failure mechanisms and passive failure pressure of tunnel face obtained from the centrifuge model tests and finite element analyses as well as comparisons with the existing analytical solutions; and (d) the induced ground surface displacement due to tunnel face displacement obtained from centrifuge tests. 2. Centrifuge modelling 2.1. Experimental program The centrifuge model tests were performed in the Geotechnical Centrifuge Facility (GCF) at the Hong Kong University of Science and Technology (HKUST). The geotechnical centrifuge at HKUST (Ng et al., 21, 24) has a rotating arm of approximately 8.4 m in diameter. The maximum modelling capacity of the centrifuge is 4g ton. It is capable of simulating an elevating gravity field over 15 times that of the earth s gravity (g) for static model. Three tunnelling cases with different C/D ratios were carried out in saturated sand to study the passive failure of tunnel face. Tests S2 and S4 were designed to investigate the passive failure of tunnel face for tunnels located at C/D ratios equal to 2.2 and 4.3 respectively. The performance and reliability of load cells submerged in the water deteriorated over time due to water ingress. Tunnel located at C/D ratio equals to 2.2 was repeated in Test S2R using an internal load cell as described in Section 2.2 to obtain the variation of tunnel face pressure with tunnel face displacement. Some details of the centrifuge tests are given in Table Model setup Fig. 1 shows the plan and elevation views of the centrifuge model setup. A rectangular model container of plan dimensions mm and depth 85 mm with a perspex viewing window in the front face was used in the centrifuge model tests. A 12.7 mm thick glass, measuring 85 mm by 714 mm, was bolted to 25.4 mm thick perspex with similar dimensions to form a composite panel. The composite panel was attached to the front face of the model container with the glass side in contact with sand. The glass served the purpose for ease of PIV control marker placement while reducing the friction between the front face and the sand. The face of the glass which was in contact with the sand formed the vertical plane of symmetry. A 25.4 mm thick aluminium plate was used to separate the sand from the loading system. The aluminium plate was braced by six aluminium struts with diameter of 41.3 mm attached to the side wall of the model container. Fig. 2 shows the model tunnel used in the Test S2R. A tunnel lining consisted of an aluminium hollow tube, 5 mm in diameter and 2 mm long with wall thickness of 2.7 mm, which was split longitudinally along the centre plane. A hollow loading piston, 2 mm in diameter and 14 mm long was screwed to the a 6 mm long tunnel face block. A 25 mm long end block was placed Table 1 Test program. Test C/D Relative density (%) Unit weight (kn/m 3 ) Remarks at another end of the loading piston. The tunnel lining was placed on the tunnel face block and bolted to the end block. The tunnel face block consisted of a tunnel face, an internal load cell and a sleeve. The internal load cell was made of hollow aluminium tube with semiconductor strain gauges bonded on the external surface. Epoxy coating was used to protect the strain gauges from abrasion. Full Wheatstone bridge strain gauges was arranged to compensate for temperature effects. The load cell was attached to the tunnel face and protected by the sleeve. There were two O-rings at both ends of the load cell in contact with the inner face of the sleeve. This minimised the friction between sleeve with lining/soil and glass to be transferred to the load cell. The O-rings S Performance of load cells submerged in water deteriorated over time S Performance of load cells submerged in water deteriorated over time S2R (a) (b) Fig. 1. Basic configuration of a centrifuge package (dimensions in mm): (a) plan view and (b) elevation view. End block Lining Loading piston Tunnel face block Internal load cell Fig. 2. Model tunnel. Crown Invert also served the purpose to isolate the load cell from water ingress. There was a ring of filler to separate the sleeve from the tunnel face to ensure load from the tunnel face was measured by the load cell only. The silica gel was squeezed into the hollow section of the load cell and isolated the strain gauges from water. Model tunnel used in Tests S2 and S4 was similar to that used in Test S2R except the loading piston and tunnel face block also served as load cells with full Wheatstone bridge strain gauges bonded to its external surface and protected by epoxy coating. The performance and reliability of the load cells deteriorated over time due to water ingress. Fig. 3 shows the loading system used in the centrifuge tests. The loading system consisted of a hydraulic actuator, mounted on an actuator support by using two L-shape brackets and connected to an oil supply system through two oil supply tubes. A linear variable differential transformer (LVDT) was fastened to the actuator support and its core was bolted to a fitting cap. The fitting cap was attached to the piston of the actuator. The actuator support was made of four 12.7 mm thick aluminium plates and bolted to a 25.4 mm thick base plate, which was mounted to the bottom of the model container. A connecting piston used to connect the loading system and model tunnel was slot to the fitting cap and secured by a L-shape fitting. The connecting piston also served as an external load cell with full Wheatstone bridge strain gauges bonded to its external surface and protected by epoxy coating. The model tunnel was supported by an aluminium column at one end and bolted to the aluminium plate via two screws at another end as shown in Fig. 1. The connecting piston of the loading system passing through the aluminium plate was connected to the loading piston of the model tunnel. The hole on aluminium plate had a groove, fit with an elastomeric O-ring, which encircled the connecting piston to ensure watertight during the movement of the piston. Perforated drainage pipes wrapped with geotextiles were used to form a drainage system. The drainage system was placed on the bottom of the model container Instrumentation In Tests S2 and S4, tunnel face pressure was given by an external load cell as shown in Fig. 3. In Test S2R, tunnel face pressure was given by an internal load cell in the tunnel face block as described in the previous section. Horizontal displacement of tunnel face was measured by LVDT attached to the actuator support. Particle image velocimetry (PIV) and close-range photogrammetry originally developed by White et al. (23) was used to monitor the subsurface soil displacement on the vertical plane of symmetry. The precision of the measurement is.1 mm. Digital images were captured using in flight camera mounting on the swinging platform. Each image was divided into soil patches and each patch represented Oil supply tubes LVDT Connecting piston Actuator support Fig. 3. Loading system. L shape fitting Externall loadcell L-shape brackets Hydraulic actuator Fitting cap a measurement point. The movement of soil patches between two successive images was traced based on cross-correlation. PIV can only measure the displacement in the image-space coordinates. Close range photogrammetry was used to convert the image-space coordinates to object-space coordinates. Details of PIV and closerange photogrammetry can be obtained from White et al. (23). Longitudinal surface displacements were deduced from the PIV results. The transverse surface displacements were monitored using LVDTs at Sections S1 and S2 located at 5.6D and 7.9D respectively along the x direction as indicated in Fig. 1. For Test S2, four LVDTs were placed at Section S1. Four LVDTs were placed at Section S2 for Test S4. Four LVDTs were placed at Section S1 and three LVDTs were placed at Section S2 for Test S2R Model preparation Toyoura sand used in the tests has the maximum and minimum void ratios of.977 and.597 respectively. The sand has a specific gravity of 2.65 (Verdugo and Ishihara, 1996). The sand was pluviated to the strong box through a hopper and the drop height of the sand was set as 5 mm. The relative density, D r, of the sand sample at 1g was 67%, 56% and 63% for Tests S2, S4, and S2R respectively. The relative density and corresponding saturated unit weight of the sand sample for Tests S2, S4 and S2R are listed in Table 1. In order to saturate the sand, aluminium cover with elastomeric O-ring was placed on top of the model container. Vacuum was applied to the sand for 2 h. Then, carbon dioxide was injected through drainage system to replace the residual air. After that, the vacuum was reapplied for another 3 h. While maintaining the vacuum, deaired water from water tank was supplied through the drainage system to the soil mass. When the water level reached required level above the ground surface of the sand sample, the saturation process was terminated. The whole saturation process required around 4 h Test procedures After completion of model preparation and final check, the model container was transferred to the swinging platform. The water level was maintained at around 15 mm above the ground surface after connecting the strong box to the stand pipe on the platform. Four cameras were set up to capture photographs during centrifuge testing. Three videos were installed to monitor the test. The data logger was then set to record data at 1 Hz, and upon centrifuge spinup, photographs were taken at 15 s intervals and saved to the computer. When the acceleration of the centrifuge reached 1g and equilibrium condition was achieved, the camera settings were changed to take photographs every 3 s. The tunnel face block was pushed toward the sand in.2 mm per second. The speed was chosen to ensure enough time for the water to flow into the tunnel lining so that suction was not created behind tunnel face block. The tunnel face displacement was prescribed and the corresponding face pressure was measured. After pushing the tunnel face block for a maximum displacement of 35 mm, the centrifuge was spun down. It should be noted that in actual shield tunnelling, pressure controlled is normally adopted. The measured results from centrifuge tests may not be directly applicable to the actual tunnel construction. On the other hand, pressure controlled was used in the numerical simulations described later in this paper. 3. Three-dimensional numerical modelling 3.1. Finite element mesh and boundary conditions Three-dimensional numerical modelling were performed using Plaxis finite element code (Brinkgreve and Broere, 24). Fig. 4

70 3 K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) shows the finite element mesh used in the numerical modelling for tunnel located at C/D ratio equals to 2.2. The finite element mesh was 7 mm long, 3 mm high and 3 mm wide. For tunnel located at C/D ratio equals to 4.3, the finite element mesh was 7 mm long, 4 mm high and 3 mm wide. This is similar to the dimensions of the soil samples in the model container. Only half of the tunnel was modelled, taking advantage of symmetry about y = mm. 15-noded wedge elements and 8-noded plate elements were used to model the sand and the tunnel lining respectively. On the left and right boundaries of the mesh, the movement in the y direction was restrained. The movement in the x direction on the front and rear boundaries was restrained. Pin supports were applied to the base boundary to restrict movements in the x, y and z directions. The water table was located at the ground surface, with a hydrostatic initial pore-water pressure profiles. The tunnel lining and shield was submerged under water Constitutive models and model parameters The response of the sand is modelled using a non-linear hardening-soil (H-S) constitutive model employed the Mohr Coulomb failure criterion with a non-associated flow rule. The H-S model is a non-linear elastic plastic formulation using multiple yield loci as a function of plastic shear strain and a cap to capture volumetric hardening as described by Schanz et al. (1999). Loading and unloading within the current yield surface, which is defined by a unloading and reloading modulus, E ur, are assumed to be elastic. Critical state angle of friction, / cs is defined by maximum angle of dilation, w and peak angle of friction, / according to Eq. (1a). The mobilised dilation angle, w m is related to mobilised angle of friction, / m and / cs as shown in Eq. (1b). sin / cs ¼ ðsin / sin wþ=ð1 sin / sin wþ sin w m ¼ ðsin / m sin / csþ=ð1 sin / m sin / csþ ð1aþ ð1bþ Soil dilatancy, defined by the ratio of the plastic volumetric strain rate to the plastic shear strain rate is equal to sinw m. A cut-off for dilation is allowed when void ratio reaches the nominated maximum void ratio. Under triaxial condition, the model predicts a hyperbolic relationship for the drained secant Young s modulus, E, as given in Eq. (2). E ¼ 2E 5 ð1 R f q=q f Þ; E 5 ¼ E ref 5 ðr 3 =p ref Þ m ð2þ where q is the deviatoric stress; E ref 5 is the E value when q is 5% of the maximum deviatoric stress, q f at the reference confining stress, 3 Tunnel axis Tunnel Centreline Left boundary 7 Front boundary 3 y z Tunnel face x Lining Rear boundary Direction of tunnel advancement Right boundary Fig. 4. Finite element mesh adopted for three-dimensional numerical modelling (dimensions in mm). Table 2 Soil parameters used in the finite element analyses. Parameter p ref ; R f is the failure ratio and m controls the stress level dependence of the stiffness; r 3 is the confining stress. In this study, p ref was taken as 1 kpa. Beside, effective oedometer modulus, E oed which control the cap of the model is also stress level dependence. The relative density of the sand was taken as 6%, which corresponds to a saturated unit weight, c sat of 19. kn/m 3. According to Verdugo and Ishihara (1996), the critical state angle of friction, / cs for Toyoura sand is 31 under triaxial condition. In order to obtain shear strength and stiffness parameters for numerical simulations in this study, triaxial compression tests conducted by Maeda and Miura (1999) were simulated numerically. The fitted / and w are 37 and 7 respectively. Small cohesion, c of 1 kpa was assigned to the soil. In the numerical simulations, E 5 at confining stress of 1 kpa was taken as 27 MPa. E oed was set as 27 MPa (Schanz and Vermeer, 1998). The unloading and reloading modulus, E ur was taken as 81 MPa, which is three times of E 5 (Brinkgreve and Broere, 24). The Poisson s ratio of the sand was taken as.2. The coefficient of earth pressure at-rest was set equal to.5. Soil parameters used for the numerical modelling are summarised in Table 2. The tunnel lining was modelled as linear elastic material with a Young s modulus of kpa, a Poisson s ratio of.15, and a thickness of 2.7 mm Numerical modelling procedures The analysis was started by applying an acceleration of 1g to increase the gravity of the numerical model to simulate the stress state in the centrifuge test. As tunnel was assumed to be wishedin-place and submerged under water, tunnel excavation was simulated by deactivating the soil elements within tunnel excavation zone and activated the shell elements of the lining in a single step. Such a simplified modelling approach had been adopted successfully in previous studies (Li et al., 29). In order to ensure the sand around the vicinity of tunnel face remained at at-rest condition, a constant pressure which equal to the at-rest earth pressure at the centre of tunnel face was applied to the tunnel face. Subsequently, pressure controlled boundary was adopted at the tunnel face to investigate the passive failure of tunnel face. 4. Failure mechanism Value Saturated unit weight, c sat (kn/m 3 ) 19. Effective cohesion, c (kpa) 1. Effective angle of friction, / ( ) 37 Angle of dilation, w ( ) 7 Effective secant modulus, E 5 (MPa) 27 Effective unloading/reloading modulus, E ur (MPa) 81 Effective oedometer modulus, E oed (MPa) 27 Poisson s ratio, t.2 m.5 Failure ratio, R f.9 At-rest earth pressure coefficient, K o.5 Fig. 5a shows the measured normalised displacement vectors on the vertical plane of symmetry at normalised tunnel face displacement, S x /D of.8, for tunnel located at C/D ratio equals to 2.2. The corresponding normalised tunnel face pressure, N cm is 91. The normalised tunnel face pressure following Leca and Dormieux (199) is given by N cm = r t /cd, where r t = tunnel face pressure and c = effective unit weight of sand. The displacement vectors were obtained from the centrifuge tests using PIV analyses and normalised by tunnel face displacement. The displacement vectors illustrate that the soil in front of the tunnel face is displaced by the (a) z(mm) (b) z(mm) Proposed by Soubra (22) C/D = S x /D =.8 (Prescribed) N γ m = 91 (Measured) x (mm) Failure zone Proposed by Soubra (22) C/D = 2.2 S x /D =.4 (Computed) N γ m = 67 (Presribed) x (mm) Fig. 5. Comparison of (a) measured and (b) computed normalised displacement vectors for tunnel located at C/D ratio equals to 2.2. advancing tunnel face, while the soil further away from the tunnel face is forced upwards to the ground surface and hence the soil surface heaves. The observed failure mechanism is compared to a five-block failure mechanism (dashed lines in the figure) proposed by Soubra (22) in obtaining upper bound solutions. The fiveblock failure mechanism is obtained by assuming the sand obeys normality where w = / cs, which is rarely observed during drained failure of sand. The normality assumption may be one of the reasons that the five-block failure mechanism is wider than the observed failure mechanism. Previous studies (De Borst and Vermeer, 1984; Loukidis and Salgado, 29; White et al., 28) also revealed that a wider failure mechanism is obtained when the soil is more dilative. When the observed failure mechanism is idealised by solid lines, a funnel-type failure mechanism may be postulated. Fig. 5b shows the computed normalised displacement vectors at S x /D of.4 and N cm of 67. The computed displacement vectors are obtained from numerical back-analysis and normalised by tunnel face displacement. The computed displacements vectors also show a funnel-type failure mechanism. The finite element results reveal that the soil elements in front of the tunnel face and those further away from the tunnel face at which soil is forced upwards to the ground surface. The elements reached failure are indicated by the failure zone illustrated in the figure. Fig. 6a shows the measured normalised displacement vectors are localised around the tunnel face at S x /D of.8 for tunnel located at C/D ratio equals to 4.3. The corresponding N cm is 185. The soil in front of the tunnel face is displaced forwards, whereas the soil in regions located further away from the tunnel axis is forced outwards. Obviously, the observed failure mechanism illustrated by the displacement vectors is differed from a five-block failure mechanism (dashed lines in the figure) proposed by Soubra (22). (a) z (mm) (b) z (mm) x (mm) Fig. 6b shows the computed displacement vectors at S x /D of.8 and N cm of 249. The failure mechanism illustrated by the computed displacement vectors shows fairly close agreement with the localised failure mechanism. The finite element results reveal that the soil elements adjacent to the tunnel face reached failure. Computed failure zone is reasonably consistent with the observed displacement vectors shown in Fig. 6a. The displacement vectors at a region close to the ground surface are scaled up to 2 times for clarity. The displacement vectors illustrate surface settlements. The mobilised shear above the crown of tunnel face might drag the soil mass forwards and induce the surface settlements. 5. Passive failure pressure of tunnel face Proposed by Soubra (22) C/D = 4.3 S x /D =.8 (Presribed) N γ m = 185 (Measured) Failure zone 6 65 Proposed by Soubra (22) C/D = 4.3 S x /D =.8 (Computed) N γ m = 249 (Presribed) x (mm) Fig. 6. Comparison of (a) measured and (b) computed normalised displacement vectors for tunnel located at C/D ratio equals to 4.3. Fig. 7 shows the variations of N cm with S x /D for tunnels located at C/D ratios equal to 2.2 and 4.3. For tunnels located at C/D ratio equals to 2.2, N cm increases with S x /D but at a reducing rate and approaches a steady state. The variation of N cm with S x /D for Test S2 is comparable to that obtained from Test S2R. Some of data was missing due to improper electrical gain was used in Test S2. It is observed that N cm in Test S2 is larger than that in Test S2R. This may be reasonable as the N cm obtained from external load cell in Test S2 was affected by the friction between tunnel face block and the lining, glass or sand. It is fortuitous that the computed pressure-displacement curve is in good agreement with the measured response. It should be noted that a displacement controlled

71 32 K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) K.S. Wong et al. / Tunnelling and Underground Space Technology 28 (212) Normalised tunnel face pressure, N γm boundary was used in the centrifuge test while pressure controlled was adopted in the numerical simulation. Calculated passive failure pressure by using the upper bound solution, which were derived by Soubra (22) are included for comparison. The calculated passive failure pressure is fairly consistent with the measured face pressure. For tunnel located at C/D ratio equals to 4.3, the measured N cm increases slowly when S x /D is less than.1. Subsequently, N cm increases rapidly but in a reducing rate. The computed N cm increases in a reducing rate and a steady state value was not reached at S x /D ratio equals to.8. The computed N cm is larger than the measured one. This may be due to the presence of localised loose deposit in front of tunnel face created at 1g condition. The calculated passive failure face pressure is higher than the corresponding measured value by 5%. The discrepancy may be because the measured pressure did not reach the failure one as the pressure was still increasing. This might be also attributed to the large discrepancy between the localised failure mechanism and the five-block failure mechanism. 6. Surface displacement It is well known that measured transverse surface settlements due to tunnelling may be represented by a Gaussian distribution as suggested by Peck (1969). The Gaussian distribution can be described by: D ¼ D max expð s 2 =2i 2 Þ C/D = S2 (Measured) C/D = S2R (Measured) C/D = S4 (Measured) C/D = 2.2 (Computed) C/D = 4.3 (Computed) Soubra (22) Normalised tunnel face displacement, S x /D C/D = 4.3 C/D = 2.2 Fig. 7. Comparison of measured and computed variations of normalised tunnel face pressure with tunnel face displacement for tunnels located at C/D ratios equal to 2.2 and 4.3. where D is the transverse surface settlement; D max is the maximum transverse surface settlement on the tunnel centreline; s is the horizontal distance from the tunnel centreline; and i is the point of inflection of the settlement trough. The point of inflection is equal to Kz o, whereas z o is depth of tunnel. The relation was proposed by O Reilly and New (1982) and validated using field data by (Lake et al., 1992; Mair and Taylor, 1997). Mair and Taylor (1997) found that the K values vary from.25 to.45 for sand and gravel. Instead of fitting settlement profiles, Gaussian distribution is attempted to fit measured heaves in this study. Fig. 8a shows the measured normalised soil displacements, D/D at.6d below the ground surface on the vertical plane of symmetry for tunnels located at C/D ratios equal to 2.2 and 4.3. The soil displacements along the longitudinal direction were obtained from the PIV analyses. Gaussian distributions are obtained by setting K equal to.27 and the D max are deduced from the measured heaves. For tunnel located at C/D ratio equals to 2.2, heaves increase with S x /D. This is consistent with the funnel-type failure mechanism, which extends to the ground surface. The measured heaves are ð3þ (a) Heave, /D (%) (b) Surface heave, /D (%) well described by the Gaussian distributions. The extent of the heaves and location of the maximum heave remain around 4D and 1.7D respectively from the initial position of tunnel face for different S x /D. For tunnel located at C/D ratio equals to 4.3, instead of heaves, settlements are induced as S x /D is increased. This is consistent with the localised failure mechanism, which induces surface settlement as shown in Fig. 6b. The maximum settlement are located near to the initial position of tunnel face. Fig. 8b illustrates the measured normalised transverse surface heaves induced by tunnel face displacements at a section located at 3.4D in front of the initial position of the tunnel face. The surface heaves were measured by using LVDTs. For tunnel located at C/D ratio equals to 2.2, heaves increase with S x /D and the extent of heave remains at 3D from the longitudinal tunnel axis. Gaussian distributions are fitted to the transverse surface heaves. The fitted values of Gaussian distributions are obtained by using K equal to.4. The transverse surface heaves are well described by the Gaussian distributions. For tunnel located at C/D ratio equals to 4.3, settlements are induced and the extent of the settlements is around 3D from the tunnel axis. 7. Conclusions Initial position of tunnel face 1.7D -2. C/D = 2.2, Sx/D =.3 (Measured) C/D = 2.2, Sx/D =.5 (Measured) C/D = 2.2, Sx/D =.6 (Measured) -4. C/D = 4.3, Sx/D =.3 (Measured) C/D = 4.3, Sx/D =.5 (Measured) -6. C/D = 4.3, Sx/D =.6 (Measured) C/D = 2.2, Sx/D =.3 (Fitted) -8. C/D = 2.2, Sx/D =.5 (Fitted) C/D = 2.2, Sx/D =.6 (Fitted) x/d C/D = 2.2, Sx/D =.3 (Measured) -1. C/D = 2.2, Sx/D =.5 (Measured) C/D = 2.2, Sx/D =.6 (Measured) -2. C/D = 4.3, Sx/D =.3 (Measured) C/D = 4.3, Sx/D =.5 (Measured) -3. C/D = 4.3, Sx/D =.6 (Measured) C/D = 2.2, Sx/D =.3 (Fitted) -4. C/D = 2.2, Sx/D =.5 (Fitted) C/D = 2.2, Sx/D =.6 (Fitted) y/d Fig. 8. Measured heaves at different tunnel face displacement for tunnels located at C/D ratios equal to 2.2 and 4.3: (a) longitudinal direction and (b) transverse direction. The results of centrifuge model tests and finite element backanalyses investigating the passive failure of tunnel face in sand are reported. The soil failure mechanisms and passive failure pressures of tunnel face are described and discussed. In addition, the induced ground surface displacements due to the increase in tunnel face displacement are also reported. For both centrifuge and numerical investigations, it is found that for tunnel located at C/D ratio equals to 2.2, the soil in front of the tunnel face is displaced by the advancing tunnel face. The soil further away from the tunnel face is forced upwards to the ground surface and hence the soil surface heaves. A funnel-type failure mechanism is observed and the extent of the funnel-type failure mechanism is narrower than a five-block failure mechanism commonly assumed in the existing upper bound solution. However, the calculated passive failure pressure by the upper bound solution is fairly consistent with the measured face pressure. Besides, the computed pressure-displacement curve is in good agreement with the measured response. It is found that the extent of longitudinal surface heaves and the location of maximum heave remain around 4D and 1.7D respectively in front of the initial position of tunnel face at different S x /D ratio. The extent of transverse surface heaves remains at 3D from longitudinal tunnel axis for different S x /D ratio at a section, which is 3.4D in front of the initial position of tunnel face. Both observed longitudinal and transverse heaves are well described by Gaussian distributions. For a tunnel located at C/D ratio equals to 4.3, the displacements of soil are confined around the vicinity of an advancing tunnel face and a localised failure mechanism associated with ground settlements is observed and computed. There is a large discrepancy between the localised failure mechanism and the five-block failure mechanism. The calculated failure face pressure is higher than the corresponding measured value by 5%. However, reasonable consistency can be found between measured and computed passive face pressures. Acknowledgements The authors would like to acknowledge the Research Grant from the Research Grants Council of HKSAR, the National Science Foundation of China for awarding the second author the Overseas and Hong Kong, Macau Young Scholars Collaborative Research Fund (No ) and financial support from the National Basic Research Program of China (Research Grant: 27CB7141). References Anagnostou, G., Kovà ri, K., The face stability of slurry-shield-driven tunnels. Tunnelling and Underground Space Technology incorporating Trenchless 9 (2), Anagnostou, G., Kovà ri, K., Face stability conditions with earth-pressurebalanced shields. Tunnelling and Underground Space Technology 11 (2), Bezuijen, A., Brassinga, H.M., 25. Blow-out pressures measured in a centrifuge model and in the field. Tunnelling: In: Bezuijen, A., Lottum, H.v. (Eds.), A Decade of Progress GeoDelft Taylor & Francis, London, pp Brinkgreve, R.B.J., Broere, W., 24. PLAXIS 3D Tunnel Version 2, PLAXIS bv, Netherlands. Chambon, P., Corte, J.F., Shallow tunnels in cohesionless soil: stability of tunnel face. Journal of Geotechnical Engineering ASCE 12 (7), De Borst, R., Vermeer, P.A., Possibilities and limitations of finite elements for limit analysis. Geotechnique 34 (2), Dias, D., Janin, J. P., Soubra, A. H., & Kastner, R., 28. Three-dimensional face stability analysis of circular tunnels by numerical simulations. In: Proc., Geocongress 28, ASCE, Geotechnical Special Publication vol. 179, pp Lake, L.M., Rankin, W.J., Hawley, J., Prediction and Effects of Ground Movements Caused by Tunneling in Soft Ground Beneath Urban Areas. CIRIA Report 3, Construction Industry Research and Information Association, London. Leca, E., Dormieux, L., 199. Upper and lower bound solutions for the face stability of shallow circular tunnels in frictional material. Geotechnique 4 (4), Leca, E., Panet, M., Application du Calcul a la Rupture a la stabilitb du front de taille dun tunnel. Revue FranGaise de Giotechnique 43, Li, Y., Emeriault, F., Kastner, R., Zhang, Z.X., 29. Stability analysis of large slurry shield-driven tunnel in soft clay. Tunnelling and Underground Space Technology 24 (4), Loukidis, D., Salgado, R., 29. Bearing capacity of strip and circular footings in sand using finite elements. Computers and Geotechnics 36 (5), Maeda, K., Miura, K., Relative density dependency of mechanical properties of sands. Soils and Foundations 39 (1), Mair, R.J., Taylor, R.N., Bored tunnelling in the urban environment. In: Proc. 14th Int. Conf. Soil Mech. Found. Engng, vol. 4, pp Mollon, G., Dias, D., Soubra, A.H., 21. Face stability analysis of circular tunnels driven by a pressurized shield. Journal of Geotechnical and Geoenvironmental Engineering 136 (1), Ng, C.W.W., Van Laak, P., Tang, W.H., Li, X.S., Zhang, L.M., 21. The Hong Kong geotechnical centrifuge. In: Proc. 3rd Int. Conf. Soft Soil Engineering, pp Ng, C.W.W., Li, X.S., Van Laak, P.A., Hou, Y.J., 24. Centrifuge modeling of loose fill embankment subjected to uni-axial and bi-axial earthquakes. Journal of Soil Dynamics and Earthquake Engineering 24 (4), O Reilly, M.P., New, B.M., Settlements above tunnels in the United Kingdom their magnitude and prediction. In: Tunnelling 82. Papers Presented at the 3rd International Symposium, pp Peck, R.B., Deep excavations and tunneling in soft ground. In: Proceedings of the 7th International Conference on Soil Mechanics and Foundation Engineering, State of the Art, pp Schanz, T., Vermeer, P.A., On the stiffness of sands. Pre-failure Deformation Behaviour of Geomaterials Schanz, T., Vermeer, P.A., Bonnier, P.G., The hardening soil model: formulation and verification. Beyond 2 in computational geotechnics. In: Ten Years of PLAXIS International. Proceedings of the international symposium, Amsterdam, March 1999, pp Soubra, A.H., 22. Kinematical approach to the face stability analysis of shallow circular tunnels. Proceedings of the Eight International Symposium on Plasticity 443, 445. Soubra, A.H., Dias, D., Emeriault, F., Kastner, R., 28. Three-dimensional face stability analysis of circular tunnels by a kinematical approach. Geotechnical Special Publication 894, 91. Verdugo, R., Ishihara, K., The steady state of sandy soils. Soils and Foundations 36 (2), White, D.J., Take, W.A., Bolton, M.D., 23. Soil deformation measurement using particle image velocimetry (PIV) and photogrammetry. Geotechnique 53 (7), White, D.J., Cheuk, C.Y., Bolton, M.D., 28. The uplift resistance of pipes and plate anchors buried in sand. Geotechnique 58 (1),

72

73

74