输运计算软件Attila介绍

Attila软件是一个输运计算软件。采用Sn + 间断有限元,采用的网格为线性四面体单元,可以计算非结构几何。2007年被接收为ITER工程设计软件,在聚变以及辐射屏蔽方便有广泛的应用。Attila的发展历程如下

  • 1995年: Los Alamos CIC3-Group开始开发
  • 2014年8月5日:瓦里安医疗(varian)宣布收购Transpire公司部分资产,包括Acuros软件产品和Attila软件产品。瓦里安医疗(varian)是全球最大肿瘤放射设备制造商,目前市值100.38亿美元(20171206)
  • 2017年10月6日:发布Attila9.1。最新版集合了Attila与MCNP6,叫Attila4MC 。可以在同一个GUI上使用CAD直接建模,用Attila做设计计算,MCNP做验证计算。MCNP6来自Los Alamos National Laboratory。Attila来自Varex Imaging。SpaceClaim来自 ANSYS, Inc。

参考

attila-founder

Greg Failla, Chief Executive Officer of Transpire, Inc., has plenty of reasons to be proud of what the company has accomplished in the past six years. In 2002, Radion Technologies (later reincorporated as Transpire, Inc.) was founded by Failla and two former Los Alamos National Laboratory (LANL) scientists, Drs. John McGhee and Todd Wareing. Drs. McGhee and Wareing launched the startup company while on an entrepreneurial leave of absence from LANL, where they worked as scientists. They were joined soon after by Dr. Allen Barnett, who previously worked as a shielding engineer in the U.S. Navy’s Naval Nuclear Propulsion Program. Through a licensing agreement with LANL, the company built on core technology that originated at the laboratory to develop a complete radiation transport software product, Attila, that can predict how radiation behaves in a broad range of applications faster and more accurately than just about anything else. Since the first official release of Attila in January 2004, interest has grown rapidly. Attila is now being used in over seven countries for applications as diverse as radiation shielding, radiotherapy, medical imaging, fusion research, homeland security, spacecraft design and reactor analysis. In addition, the company has received numerous Small Business Innovation Research grants, including two from the National Cancer Institute for medical imaging and radiotherapy, which total almost \$2 million. In 2007, Transpire generated close to \$1 million in revenue from software and training alone, and anticipates exceeding this in 2008. Because of these revenues and the large number of grants, Transpire will be able to broaden the software for additional markets. The software has recently been added to the short list of validated codes for International Thermonuclear Experimental Reactor (ITER) neutronics analyses. ITER is a joint international research and development project that aims to demonstrate the scientific and technical feasibility of fusion power and involves partners from all over the world. The company also has a multi-year project with Pacific Northwest National Laboratory to develop a scenario analysis tool to detect radiological threats at U.S. ports of entry. The software has been licensed by leading healthcare companies involved in both radiotherapy and medical imaging. Additionally, Transpire has active collaborations with the University of Texas M.D. Anderson Cancer Center for radiotherapy and Baylor College of Medicine

ATTILA: A three-dimensional, unstructured tetrahedral mesh discrete ordinates transport code

Many applications of radiation transport require the accurate modeling of complex three-dimensional geometries. Historically, Monte Carlo codes have been used for such applications. Existing deterministic transport codes were not applied to such problems because of the difficulties of modeling complex three-dimensional geometries with rectangular meshes. The authors have developed a three-dimensional discrete ordinates ($S_{n}$) code, ATTILA, which uses linear-discontinuous finite element spatial differencing in conjunction with diffusion-synthetic acceleration (DSA) on an unstructured tetrahedral mesh. This tetrahedral mesh capability enables the authors to efficiently model complex three-dimensional geometries. One interesting and challenging application of neutron and/or gamma-ray transport is nuclear well-logging applications. Nuclear well-logging problems usually involve a complex geometry with fixed sources and one or more detectors. Detector responses must generally be accurate to within 1%. The combination of complex three-dimensional geometries and high accuracy requirements makes it difficult to perform logging problems with traditional $S_{n}$ differencing schemes and rectangular meshes. Hence, it is not surprising that deterministic $S_{n}$ codes have seen limited use in nuclear well-logging applications. The geometric modeling capabilities and the advanced spatial differencing of ATTILA give it a significant advantage, relative to traditional $S_{n}$ codes, for performing nuclear well-logging calculations.

3D unstructured-mesh radiation transport codes

Three unstructured-mesh radiation transport codes are currently being developed at Los Alamos National Laboratory. The first code is ATTILA, which uses an unstructured tetrahedral mesh in conjunction with standard Sn (discrete-ordinates) angular discretization, standard multigroup energy discretization, and linear-discontinuous spatial differencing. ATTILA solves the standard first-order form of the transport equation using source iteration in conjunction with diffusion-synthetic acceleration of the within-group source iterations. DANTE is designed to run primarily on workstations. The second code is DANTE, which uses a hybrid finite-element mesh consisting of arbitrary combinations of hexahedra, wedges, pyramids, and tetrahedra. DANTE solves several second-order self-adjoint forms of the transport equation including the even-parity equation, the odd-parity equation, and a new equation called the self-adjoint angular flux equation. DANTE also offers three angular discretization options: $S_{n}$ (discrete-ordinates), $P_{n}$ (spherical harmonics), and $SP_{n}$ (simplified spherical harmonics). DANTE is designed to run primarily on massively parallel message-passing machines, such as the ASCI-Blue machines at LANL and LLNL. The third code is PERICLES, which uses the same hybrid finite-element mesh as DANTE, but solves the standard first-order form of the transport equation rather than a second-order self-adjoint form. DANTE uses a standard $S_{n}$ discretization in angle in conjunction with trilinear-discontinuous spatial differencing, and diffusion-synthetic acceleration of the within-group source iterations. PERICLES was initially designed to run on workstations, but a version for massively parallel message-passing machines will be built. The three codes will be described in detail and computational results will be presented.