原子能科学技术
原子能科學技術
원자능과학기술
ATOMIC ENERGY SCIENCE AND TECHNOLOGY
2014年
11期
2148-2152
,共5页
邱有恒%邓力%李百文%赵英奎
邱有恆%鄧力%李百文%趙英奎
구유항%산력%리백문%조영규
蒙特卡罗方法%重要性函数%伴随输运%权重窗
矇特卡囉方法%重要性函數%伴隨輸運%權重窗
몽특잡라방법%중요성함수%반수수운%권중창
Monte Carlo method%important function%adjoint transport%weight window
权重窗技术是最常用的降低方差技巧之一,合理运用权重窗技术能大幅提高计算效率。MCNP程序具备基于栅元的伴随权重窗发生器功能,但对一些含重复结构栅元或非对称结构栅元的模型,由于MCNP程序不能计算其中某些栅元体积,伴随权重窗发生器失效。本文在MCNP程序基础上设计了基于M ES H技术的伴随权重窗自动生成方法,该方法不改动真实模型,采用一套虚拟的相空间覆盖真实模型。将新方法应用于重复结构算例,计算效率大幅提升。对光子散射基准实验模拟表明,新方法的计算结果与实验值吻合得非常好,相对偏差仅1.2%,计算效率约为直接模拟的8倍。
權重窗技術是最常用的降低方差技巧之一,閤理運用權重窗技術能大幅提高計算效率。MCNP程序具備基于柵元的伴隨權重窗髮生器功能,但對一些含重複結構柵元或非對稱結構柵元的模型,由于MCNP程序不能計算其中某些柵元體積,伴隨權重窗髮生器失效。本文在MCNP程序基礎上設計瞭基于M ES H技術的伴隨權重窗自動生成方法,該方法不改動真實模型,採用一套虛擬的相空間覆蓋真實模型。將新方法應用于重複結構算例,計算效率大幅提升。對光子散射基準實驗模擬錶明,新方法的計算結果與實驗值吻閤得非常好,相對偏差僅1.2%,計算效率約為直接模擬的8倍。
권중창기술시최상용적강저방차기교지일,합리운용권중창기술능대폭제고계산효솔。MCNP정서구비기우책원적반수권중창발생기공능,단대일사함중복결구책원혹비대칭결구책원적모형,유우MCNP정서불능계산기중모사책원체적,반수권중창발생기실효。본문재MCNP정서기출상설계료기우M ES H기술적반수권중창자동생성방법,해방법불개동진실모형,채용일투허의적상공간복개진실모형。장신방법응용우중복결구산례,계산효솔대폭제승。대광자산사기준실험모의표명,신방법적계산결과여실험치문합득비상호,상대편차부1.2%,계산효솔약위직접모의적8배。
The weight window is one of the most widely used variance reduction tech‐niques . When used judiciously , it can save substantial computer time . Because the MCNP program can not calculate the volume of some special cells ,for example ,repeat‐ed structure cells or asymmetrical cells ,the adjoint weight window generator of MCNP is almost unavailable for a complex geometry model . A mesh‐based adjoint weight window generator was provided in the paper .It utilized an extra virtual mesh model where original model is enclosed ,and no modification was made on the original model . It can be used for almost all models .The new method was successfully used to a repeat‐ed structure model , and the FOM value greatly increased . A satisfactory simulation result of a photon skyshine experiment model was obtained by the new method .Com‐pared with the experiment result ,the relative difference of the new method is 1.2%only .The FOM of the new method is almost eight times as large as that of the direct Monte Carlo simulation .