Ice分散式運算技術於都市雨水下水道水理模擬之應用

Ice Distributed Computing Technology for Urban Storm Sewer Simulation

程連育(Lien-Yu Cheng)

指導教授 ： 林旭信

中原大學/工學院/土木工程研究所/碩士(2011年)

https://doi.org/10.6840/cycu201101010

摘要

分散式運算技術利用現有個人電腦運算資源，透過網路形成一強大的高效能運算架構。然而，由於各種分散式運算環境所使用的程式語言與作業系統不同，造成資料交換與訊息溝通的問題。
本研究主要目的為應用分散式運算的概念進行雨水下水道水理模擬，在水理模式間複雜的演算中加速運算效能以提升演算之效率。研究中以SWMM(Storm Water Management Model)之地表逕流模式(SWMM-RUNOFF)與雨水下水道模式(SWMM-EXTRAN)為例，研發一分散式雨水下水道模擬系統（Ice Distributed Simulation Sewer System，簡稱IDSS），IDSS以ZeroC公司開發之Ice (Internet Communications Engine)中介軟體(Middleware)為計算平台，降低分散式系統開發資料交換之複雜性與作業系統、程式語言之相依性，提升SWMM模式之演算效率。
以台北市第四排水分區之保儀、實踐及中港抽水站系統為案例，5年24小時降雨設計雨型進行水理模擬，客戶端將排水分區水理資料透過IDSS系統分散至三台伺服端，經伺服端進行模擬後再將結果回傳至客戶端，將IDSS系統模擬結果與SWMM模擬結果做比較，其結果一致，測得網路延遲平均為2.93秒。IDSS系統利用分散式運算架構可節省20%的運算時間，能有效提升雨水下水道水理模擬演算效能。

關鍵字

IceTM中介軟體 ； SWMM ； 分散式運算技術 ； 訊息交換

並列摘要

Distributed computing technology uses the resources of personal computing over Internet forming a powerful and high-performance computing infrastructure. However, the diversity of distributed computing environment with various programming languages and operating systems will lead a lot of problems, such as information exchange and communication, which increase in difficulties of software developments.
The main objective of this study is to apply the concepts of distributed computing to simulating storm sewer flow. In the interests of efficiency of computational complexities of storm sewer simulation, an Ice (Internet Communications Engine) distributed simulation sewer system (IDSS) is developed by using RUNOFF and EXTRAN modules of the SWMM (Storm Water Management Model) model based on Ice middleware of ZeroC Company. IDSS can increase in the calculations efficiency of SWMM model and reduce the complexity of information exchange with the operating system dependencies of distributed system developments.
In Taipei city, the 4th drainage system, Bao-yi, Shi-Jian and Zhung-Gung drainage systems are employed as a case study. Clients simulate the storm sewer flow of the three drainage systems by the IDSS system to which the design storm with 5-year return period and 24-hour duration for Taipei city as inputs. Meanwhile, the IDSS system distributes the input data to three sets of servers running SWMM simultaneously and waits for the results from the servers. The simulated results from the servers show that the results of IDSS correspond to the results of original SWMM. The network traffic on the IDSS system network delay is 2.93 seconds. In summary, the IDSS system not only gives identical simulation results, but saves 20% computing time and speeds up the efficiency of storm sewer simulation.

並列關鍵字

IceTM middleware ； SWMM ； distributed computing technology ； information exchange

參考文獻

21. 蔡文雄, 2009,“整合雨水下水道即時水理模擬與Web GIS展示技術於都市內水預警之應用─以台北市第四排水分區為例,”中原大學土木工程學研究所碩士論文

19. 邱煌升, 2008,“以OpenMI架構整合都市降雨逕流模式與雨水下水道模式,”中原大學土木工程學研究所碩士論文。

22. 廖烜欣, 2009,“街道與雨水下水道淹排水模式之研究,”國立成功大學水利及海洋工程學系碩士論文

16. 李玉婷, 2002,“具錯誤移轉機制之分散式平行運算平台設計與建置,”元智大學資訊管理研究所碩士論文

18. 美商傑明工程顧問股份有限公司台灣分公司, 2002, “台北市排水系統調查檢討及資料建檔規劃(第四分區)－排水系統檢討分析期中報告書,”台北市政府工務局養護工程處。