Recent developments of monte-carlo codes molflow+ and synrad+

Molflow+ and Synrad+ are Monte Carlo simulation tools for ultra-high vacuum and synchrotron radiation, respectively. Over the years they have become a common tool for designing and analysing the vacuum system of particle accelerators. This contribution gives a short summary about new features added since the last IPAC contribution [1]. Synrad+ now supports low flux mode, a weighted Monte Carlo technique where the represented number of photons is reduced at every reflection, providing significantly better statistics at low flux regions. As for Molflow+, angle maps allow recording the molecules’ directional distribution at any point, and then desorb a reduced gas quantity according to the recording. In linear systems, this allows iterative simulations that have been proven to treat systems up to 7 orders of magnitude of pressure difference. Without the new technique the computing time would be prohibitively slow on desktop computers, which is what most users of the two codes use. Both codes now have a built-in geometry builder that allows creating simple models through a set of 3D operations and modifying those imported from CAD tools. Molflow+ has been extended with additional diagnostic tools, such as a logger that records properties of all hits on a scoring surface, and histogram plotters that visualize the distribution of the number of bounces, the distance to absorption and the time of flight of the gas molecules. The codes have recently become open source, and it has been made compatible with, and tested on different versions of Linux and macOS. CODE OVERVIEW Molflow+ is a simulator for ultra-high vacuum that has been written in the 1990s and ported to modern C++ based code in 2007 [2]. It uses the test-particle Monte Carlo method, tracing the trajectory of virtual gas molecules from source (a gas injection or thermal outgassing location) to absorption (typically a vacuum pump). The geometry is represented as vacuum boundaries (walls) with polygons. These polygons, extended with physical properties (temperature, sticking and outgassing, sojourn time, etc.) are referred to as facets. As the simulation is running, several counters record hits that belong either to entire facets, or cells of post-processing entities called textures or profiles. Since the number of counters is defined before the simulation is launched, the memory requirement remains the same throughout a run. Using these counters, physical quantities such as pressure, density and impingement rates are calculated and updated on the screen every second. The calculated values and the color-coded textures and profile plots fluctuate with each screen update, but as the statistical error of testparticle MC simulations decreases with the square root of the number of hits [3], they converge to the solution over time. This allows the user to decide when the results are accurate enough to stop the run, at which point they are visualized internally or exported for further post-processing. Synrad+ is a code forked from Molflow+, using the same ray-tracing engine. Instead of gas molecules, it traces photons originating from magnetic accelerator elements. These elements, typically dipoles, quadrupoles or periodic elements like wigglers, are referred to as magnetic regions. With user input defining beam properties, starting point position and direction, they are represented as a number of trajectory points, each of which can generate a virtual photon representing a certain photon flux. These virtual photons are then traced through the geometry, hitting wall facets. Upon a hit, reflection, absorption, and optionally backscattering and transmission probabilities depend on the wall material, roughness, incident angle and photon energy. Such probability tables are included for a few metals and can be defined for new materials by the user. The two codes share file formats and their interface is similar. A coupled usage of the two codes would be first simulating flux absorption with Synrad+, then converting it to dynamic outgassing in Molflow+ and finally proceeding with a vacuum simulation, as demonstrated in [4]. NEW TOOLS FOR THE GEOMETRY Earlier versions of the codes imported the geometry through the STL file format, which is extensively supported by CAD programs due to its popularity in 3D printing. That format describes solid bodies’ surfaces by a list of triangles, which Molflow+ merges by detecting coplanar and adjacent triangles, colinear sides and shared triangle vertices. Nevertheless, simplifying a real 3D model to a vacuum geometry is a non-trivial process: ideally an experienced mechanical engineer removes non-relevant parts (screws, flanges, mechanical supports), inverts the volume by converting voids in the structure to solid parts, then simplifies curved parts (which can only be described by a large number of planar facets) and joins walls to prevent leaks. Even if done correctly, many CAD programs introduce small rounding errors during the conversion to STL format, and the meshing of surfaces to triangles is arbitrary, making Molflow+ post-processing (orienting textures, etc.) difficult. Geometry Editor Due to the issues above, and since many users of the codes are physicist without access to professional CAD tools, a geometry editor was added to the codes, that allows ___________________________________________ † Corresponding author: roberto.kersevan@cern.ch † 10th Int. Partile Accelerator Conf. IPAC2019, Melbourne, Australia JACoW Publishing ISBN: 978-3-95450-208-0 doi:10.18429/JACoW-IPAC2019-TUPMP037 MC7: Accelerator Technology T14 Vacuum Technology TUPMP037 1327 Co nt en tf ro m th is w or k m ay be us ed un de rt he te rm so ft he CC BY 3. 0 lic en ce (© 20 19 ). A ny di str ib ut io n of th is w or k m us tm ai nt ai n at tri bu tio n to th e au th or (s ), tit le of th e w or k, pu bl ish er ,a nd D O I