本文選題：重氣擴(kuò)散 + 數(shù)值模型�。� 參考：《北京交通大學(xué)》2017年碩士論文

【摘要】：近年來,隨著化工產(chǎn)業(yè)的快速發(fā)展,化學(xué)品的使用越來越多。其中,一些化學(xué)品由于儲存不正確,發(fā)生意外泄漏,產(chǎn)生比空氣密度大的氣云,稱為重氣,它沿著地表進(jìn)行擴(kuò)散,引發(fā)中毒、燃燒、爆炸等危險事故的概率急劇增加,嚴(yán)重威脅周圍人員人身安全和財產(chǎn)安全。重氣擴(kuò)散過程受很多外界因素影響,如風(fēng)速、障礙物等,其擴(kuò)散過程復(fù)雜。因此,開展重氣擴(kuò)散過程的研究,對于科學(xué)預(yù)防重氣泄漏引發(fā)事故的發(fā)生以及指導(dǎo)應(yīng)急救援具有重要理論價值和實踐意義。重氣擴(kuò)散過程相當(dāng)復(fù)雜,大型現(xiàn)場試驗雖然能夠收集到大量的真實數(shù)據(jù),但是花費巨大且不能重復(fù)試驗,經(jīng)驗?zāi)Ｐ瓦m用性較窄且精度不夠。針對以上問題,本文選用修正的湍流(k-ε)模型修飾的計算流體力學(xué)(CFD)數(shù)值模型,以氯氣為例,開展重氣在有風(fēng)條件下繞障礙物擴(kuò)散的研究。本文主要的研究工作如下:(1)綜述了重氣擴(kuò)散數(shù)值模型的國內(nèi)外研究現(xiàn)狀,在對比分析了唯象模型、箱模型、淺層模型和計算流體力學(xué)模型的優(yōu)缺點之后,選用計算流體力學(xué)模型模擬重氣擴(kuò)散過程,并且選用修正的k-ε模型更為真實地刻畫重氣擴(kuò)散過程中的湍流運動。選取實驗場景,建立三維坐標(biāo)系,結(jié)合坐標(biāo)系建立CFD模型的數(shù)學(xué)方程式,簡化公式待求解。(2)在求解CFD數(shù)值模型時,由于計算區(qū)域網(wǎng)格劃分?jǐn)?shù)目巨大,計算任務(wù)艱巨,耗時長。目前在高性能計算領(lǐng)域內(nèi)發(fā)展非常迅速的CPU-GPU異構(gòu)體系結(jié)構(gòu),其由于巨大的并行計算潛力被重多學(xué)者所關(guān)注。但是,CPU-GPU異構(gòu)體系的OpenCL編程模型缺乏擴(kuò)展到多個計算節(jié)點的能力,傳統(tǒng)的并行編程模型如共享存儲編程模型的OpenMP,又難以發(fā)掘異構(gòu)系統(tǒng)所有的計算能力,所以本文將OpenCL與OpenMp結(jié)合起來,以MPI作為進(jìn)程間的消息傳遞機制,提出基于MPI+OpenMP+OpenCL的三級混合并行編程模型,以此對CFD數(shù)值模型求解過程進(jìn)行并行優(yōu)化,大幅度提高計算速度和精度,保證計算的實時性。當(dāng)網(wǎng)格規(guī)模為200*200*80時,該模型在128個計算節(jié)點上獲得了 12.56的加速比,并行效率為62.1%。(3)將CFD模型的數(shù)值計算結(jié)果與三維仿真軟件Unity3D結(jié)合,實時模擬氯氣在有風(fēng)條件下繞障礙物擴(kuò)散的過程,仿真效果真實、美觀,動畫效果流暢自然。此外,根據(jù)氯氣濃度繪制出傷害區(qū)域,結(jié)合傷害模型和個人風(fēng)險,對氯氣中毒事故給出風(fēng)險評估。
[Abstract]:In recent years, with the rapid development of chemical industry, more and more chemicals are used. Among them, some chemicals are accidentally leaked due to incorrect storage, producing a gas cloud with a higher density than air, called heavy gas, which diffuses along the surface, causing a sharp increase in the probability of dangerous accidents such as poisoning, combustion, explosion, etc. It is a serious threat to the personal and property safety of the surrounding people. Heavy gas diffusion process is affected by many external factors, such as wind speed, obstacles and so on. Therefore, the research of heavy gas diffusion process is of great theoretical value and practical significance in preventing accidents caused by heavy gas leakage scientifically and guiding emergency rescue. The heavy gas diffusion process is quite complex. Although large scale field tests can collect a large amount of real data, it is expensive and cannot be repeated, and the applicability of the empirical model is narrow and the accuracy is not enough. To solve the above problems, the modified turbulence k- 蔚 model modified computational fluid dynamics (CFD) numerical model is used to study the diffusion of heavy gas around obstacles under the condition of wind, taking chlorine gas as an example. The main research work of this paper is as follows: (1) the present research situation of heavy gas diffusion numerical model at home and abroad is summarized. After comparing and analyzing the advantages and disadvantages of phenomenological model, box model, shallow model and computational fluid dynamics model, The computational fluid dynamics model is used to simulate the heavy gas diffusion process, and the modified k- 蔚 model is used to describe the turbulent motion in the heavy gas diffusion process more realistically. Selecting experimental scene, establishing three-dimensional coordinate system, establishing mathematical equation of CFD model combined with coordinate system, simplifying formula to be solved. (2) in solving the numerical model of CFD, because of the huge number of grid division in calculation area, the calculation task is arduous, and the time consuming is long. At present, CPU-GPU heterogeneous architecture, which has been developing rapidly in the field of high performance computing, has attracted much attention from many scholars because of its huge parallel computing potential. However, the OpenCL programming model of CPU-GPU heterogeneous system lacks the ability to extend to multiple computing nodes. The traditional parallel programming model, such as OpenMPof shared storage programming model, is difficult to exploit all the computing power of heterogeneous systems. Therefore, this paper combines OpenCL and OpenMp, takes MPI as the message passing mechanism between processes, proposes a three-level hybrid parallel programming model based on MPI OpenMP OpenCL, and optimizes the CFD numerical model solving process in parallel. The calculation speed and accuracy are greatly improved to ensure the real-time calculation. When the grid size is 200 / 200 / 80, the acceleration ratio of 12.56 is obtained on 128 computing nodes, and the parallel efficiency is 62.1. the parallel efficiency of this model is 62.1. the model combines the numerical results of the CFD model with the 3D simulation software Unity3D. The real time simulation of chlorine diffusion around obstacles under the condition of wind shows that the simulation effect is real, beautiful, and the animation effect is smooth and natural. In addition, according to the concentration of chlorine gas, the injury area is drawn, and the risk assessment of chlorine poisoning accident is given by combining the injury model and personal risk.
【學(xué)位授予單位】：北京交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：O241.82;O552.2

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 孫寶江;;重氣泄漏擴(kuò)散影響因素分析及模型研究[J];科技視界;2015年21期

2 何友龍;;氯氣泄漏擴(kuò)散影響因素的數(shù)值模擬研究[J];江西化工;2015年03期

3 高玉勵;周秀娟;張國凱;;基于CPU/GPU異構(gòu)體系結(jié)構(gòu)的混合編程模型[J];信息通信;2015年04期

4 王鵬;王全國;黨文義;韓中樞;;大型LNG儲罐翻滾事故放空氣擴(kuò)散后果模擬[J];中國安全生產(chǎn)科學(xué)技術(shù);2014年11期

5 王春華;;液氯泄漏事故后果風(fēng)險研究[J];江蘇氯堿;2013年06期

6 鄭遠(yuǎn)攀;錢慎一;蘇曉珂;;CH_4/CO_2混合氣體瞬時擴(kuò)散特性的理論與仿真研究[J];防災(zāi)減災(zāi)工程學(xué)報;2013年03期

7 王星捷;李春花;;基于Unity3D平臺的三維虛擬城市研究與應(yīng)用[J];計算機技術(shù)與發(fā)展;2013年04期

8 張慶科;楊波;王琳;朱福祥;;基于GPU的現(xiàn)代并行優(yōu)化算法[J];計算機科學(xué);2012年04期

9 陳曉花;李春光;;SIMPLEC算法在同位網(wǎng)格中的研究及應(yīng)用[J];甘肅聯(lián)合大學(xué)學(xué)報(自然科學(xué)版);2011年03期

10 倪樂波;戚鵬;遇麗娜;王婧;;Unity3d產(chǎn)品虛擬展示技術(shù)的研究與應(yīng)用[J];數(shù)字技術(shù)與應(yīng)用;2010年09期

相關(guān)博士學(xué)位論文 前2條

1 曹維;大規(guī)模CFD高效CPU/GPU異構(gòu)并行計算關(guān)鍵技術(shù)研究[D];國防科學(xué)技術(shù)大學(xué);2014年

2 魏利軍;重氣擴(kuò)散過程的數(shù)值模擬[D];北京化工大學(xué);2000年

相關(guān)碩士學(xué)位論文 前9條

1 齊佳;氣體爆炸模型研究及三維仿真實現(xiàn)[D];北京交通大學(xué);2016年

2 馬俊峰;基于OpenCL的多GPU并行計算的研究與應(yīng)用[D];哈爾濱理工大學(xué);2014年

3 崔月凱;基于FLUENT二次開發(fā)的罐式車輛運輸毒性氣體泄漏仿真系統(tǒng)研究[D];長安大學(xué);2013年

4 蔣忠強;基于GPU加速的中性氣體泄漏模擬與救援研究[D];大連理工大學(xué);2011年

5 游佐勇;OpenMP并行編程模型與性能優(yōu)化方法的研究及應(yīng)用[D];成都理工大學(xué);2011年

6 方旭東;面向大規(guī)�？茖W(xué)計算的CPU-GPU異構(gòu)并行技術(shù)研究[D];國防科學(xué)技術(shù)大學(xué);2009年

7 朱廣萍;液氯泄漏事故后果模擬分析與應(yīng)急管理[D];中南大學(xué);2009年

8 巴特;基于ArcGIS Engine的突發(fā)性大氣污染事故應(yīng)急響應(yīng)系統(tǒng)研究[D];大連理工大學(xué);2009年

9 胡世明;氣體釋放源的三維瞬態(tài)重氣擴(kuò)散模型及數(shù)值研究[D];北京化工大學(xué);2000年

，

本文編號：1941341