請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/17066
標題: | 邁向高時空分辨率環境影響監測 Towards High Spatio-Temporal Resolution Monitoring of Environmental Impact |
作者: | Maikel Issermann 伊瑟曼邁克 |
指導教授: | 張斐章(Fi-John Chang) |
關鍵字: | 能源模擬,城市洪水,不確定性量化,地理空間建模,功能模型介面,3D 數據可視化, Energy simulation,Urban inundation,Uncertainty quantification,Geospatial modelling,Functional Mockup Interface,3D data visualization, |
出版年 : | 2020 |
學位: | 博士 |
摘要: | 市政當局及其公共事業在全球面臨越來越大的壓力,需要製定可持續的資源戰略並實現碳中和,只能依靠數據驅動的規劃方法,來解決耦合基礎結構系統和創建循環供應路徑的複雜性。但是,由於所需的能力和成本過高,很少有市政當局進行全面的數據收集。因此,本研究試圖透過以高時空分辨率,進行對城市環境影響的模型建造和模擬來減輕負擔。能源需求和城市徑流的影響是起點,住宅能源需求估算由城市建築能源模型(UBEM)執行,該模型通過 Functional Mock-up Interface(FMI)實施。以 FMI 為基礎並與 UBEM 的交互作用,無需重新初始化即可模擬各種環境條件。它能讓建築能源模擬 EnergyPlus 與外部模型耦合, 根據每棟建築物的鄰近環境條件,估算具有高時間分辨率的能源需求。為改善適用性,以 FMI 為基礎的 UBEM 通過合併自動程序,通過篩選的 Poisson 曲面重建算法,從城市範圍的點雲派生 3D 建築模型,顯示高幾何保真度,進而豐富該模型。建築和建築系統是根據歐盟 TABULA 項目編制的住宅建築類型定義的。為了演示擬議的 UBEM 的功能,使用城市小氣候數據和熱泵代替其他供熱和熱水系統構建了一個方案。在變化的佔用率和天氣條件下,對能源需求的時間密集估計,可以預測對下降的峰值負荷做出需求響應的需求。並且將模擬年度能源使用強度,與參考研究的結果進行比較,得出了令人滿意的結果,因此,該建模概念被認為是適當的。關於城市徑流,簡化的數學模型“元胞自動機”與“運動成本”字段相結合,將穿越區域的難度記入城市洪水模型 CAMC。它通過簡單的矩陣和邏輯運算,達到高計算效率。該開發,集中應用在具有眾多建築物的密集城市建成區中。CAMC 足夠高效和靈活,可用於具有當前天氣狀況的“實況”城市洪水預警系統。將 CAMC 與基於淺水方程的 ANUGA 模型進行了比較。在高空間分辨率下,CAMC 大約比 ANUGA 快 5 倍,並且能夠保持數值穩定性。案例研究在德國伍珀塔爾市的市區中進行,該市區有大約 12,000 座建築物,據估能源需求約為5736 座。根據翻新及供暖系統的狀態,歸納出 41 種類型和 8 種可能配置的多種建築材料。通過網絡界面可視化使用“實況”天氣的 3D 建築模型和模擬結果,該界面通過地理空間 3D 框架 NASA WorldWind 呈現。“實況”模擬模式可以估算出住宅能源和其碳濃度、住宅需水量以及降雨事件下的徑流深度。為了評估時空模型中的不確定性,基於點估計方法(PEM)開發了不確定性量化方法。相較對通用但計算要求很高的 Monte Carlo(MC)模擬,PEM 提供了的替代方法。PEM 使用不確定變量的概率分佈之代表值,重新運轉模型。結果可以估算輸出分佈的統計動差。Hong 的方法是針對案例研究實施的特定 PEM。該案例研究使用 ANUGA模型對英國格拉斯哥地區的水流進行了模擬,海拔高度為不確定性的來源。逐步高斯模擬法會產生隨機誤差範圍,可用在輸入任何空間模型,MC 模擬的輸出用於驗證。使用 Grams-Charlier 擴展來擬合分佈並生成概率洪水氾濫圖。PEM 不需要至少進行 500 次 MC 模擬,而只需進行 3 次模擬。因此,Hong 的方法對於近似時空模型的不確定性似乎很有吸引力。 Municipalities and their public utilities are worldwide under increasing pressure to develop sustainable resource strategies and achieve carbon neutrality. The complexity to increase efficiencies by coupling infrastructure systems and creating circular supply pathways can only be coped with by relying on data driven planning approaches. However, comprehensive data collection by municipalities is rare due to the required competency and costs. This research thus attempts to lower the burden by facilitating the modelling and simulation of urban environmental impacts with high spatio-temporal resolution. The impacts of energy demand and urban runoff are the starting points. The estimation of residential energy demand is performed by an Urban Building Energy Model (UBEM), which is implemented with Functional Mockup Interface (FMI). The FMI-based UBEM enables interactive capacities to simulate diverse environmental conditions without reinitialisation. It allows to couple the building energy simulation EnergyPlus with external models. The results are estimates of energy demand with high temporal resolution based on the adjacent environmental conditions of each building. In order to ameliorate the applicability, the FMI-based UBEM is further enriched by incorporating an automatic procedure to derive 3D building models, which display high geometrical fidelity, from city-wide point clouds through the screened Poisson surface reconstruction algorithm. Construction and building systems were defined according to the residential building typology compiled by the EU project TABULA. To demonstrate the functionality of the proposed UBEM, a scenario was constructed with urban microclimate data and heatpumps instead of other heating and hot water systems. The temporally dense estimation of energy demand under changing occupancy and weather conditions can predict the need for demand re-sponse to shed peak loads. The modelling concept is deemed appropriate because the comparison of simulated annual energy usage intensities with results from reference studies yielded a satisfying outcome. Regarding urban runoff, a new and efficient inundation model was developed, called CAMC. It is based on Cellular Automata (CA), which is a simplified mathematical model, and Motion Cost fields (MC), which indicate the energy effort for an agent to traverse an area. Simple matrix and boolean operations ensure that the simulation has a low runtime. The intended application focuses on areas with compact urban fabrics. Because of CAMC’s flexibility and efficiency, it can run inundation simulations fast enough to apply current weather conditions, and thus provide “live”urban flood warnings. The shallow water equations-based model ANUGA was the reference model and CAMC performed about five times more efficient at high spatial resolution, while still been numerically stable. A ”live” assessment system for urban environmental impact was established for an area in the German city of Wuppertal. The area contains about 12,000 buildings, where the energy demand of 5,736 residential buildings was estimated. The characterization generated a diverse building stock with 41 types and 8 possible configurations depending on the state of refurbishment and heating system. The 3D building models and simulation results using ”live” weather are visualized by a web-interface, which is implemented with the geospatial 3D framework NASA WorldWind. The “live”simulation mode produces estimates of residential energy and its carbon intensity, residential water demand as well as water depths in the case of a rainfall event. To evaluated the uncertainty in spatio-temporal models, an uncertainty quantification method was developed on the basis of Point Estimate Methods (PEMs). In contrast to the widely used Monte Carlo (MC) simulation, PEM offers a different and computationally less expensive approach. In order to predict the statistical moments of the probability distribution of the output variable, PEMs evaluated the considered model with representative values of the probability distribution of the uncertain input variable. Amongst the PEMs, the Hong’s method is spatially implemented and random error fields are generated by Sequential Gaussian Simulation to modify the input data. An urban flood simulation for an area in Glasgow, UK, was conducted with the inundation model ANUGA. Elevation was in this case considered the uncertain input variable. As a reference method in uncertainty quantification, MC simulation was used. The moment estimates of PEM can be applied to derive probability distributions by the means of the Grams-Charlier Expansion. Probabilistic flood-inundation maps are subsequently produced. Three simulations are sufficient for PEM, while 500 simulations were performed for MC. Model performance indicators showed an acceptable level of agreement in the approximation of the uncertainty in the spatio-temporal model by the Hong’s method. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/17066 |
DOI: | 10.6342/NTU202002619 |
全文授權: | 未授權 |
顯示於系所單位: | 生物環境系統工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-0708202013191300.pdf 目前未授權公開取用 | 40.1 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。