Reconstructed Discontinuous Galerkin System Design Description

This template follows INL template TEM-140, "IT System Design Description."

commentnote

This document serves as an addendum to Framework System Design Description and captures information for SDD specific to the Reconstructed Discontinuous Galerkin module.

Introduction

The MOOSE Reconstructed Discontinuous Galerkin module is based on the MOOSE framework and thus inherits the unique features and base characteristics of the framework, as outlined in the Framework System Design Description. Specific details unique to the module are outlined in this document.

System Purpose

The Software Design Description provided here is description of each object in the system. The pluggable architecture of the underlying framework of the Reconstructed Discontinuous Galerkin module makes MOOSE and MOOSE-based applications straightforward to develop as each piece of end-user (developer) code that goes into the system follows a well-defined interface for the underlying systems that those object plug into. These descriptions are provided through developer-supplied "markdown" files that are required for all new objects that are developed as part of the Reconstructed Discontinuous Galerkin module. More information about the design documentation for MOOSE-based applications and like the Reconstructed Discontinuous Galerkin module can be found in Documenting MOOSE.

System Scope

The purpose of this software is to provide capability to MOOSE-based applications to use a second-order, cell-centered finite volume method (FVM). This module provides a systematic solution for implementing all required components in a second-order FVM such as slope reconstruction, slope limiting, numerical flux, and proper boundary conditions. Additionally, this module provides an implementation of these components for the scalar advection equation.

Dependencies and Limitations

The Reconstructed Discontinuous Galerkin module inherits the software dependencies of the MOOSE framework, with no additional dependencies. The design of this module is motivated by the needs of its client applications.

Definitions and Acronyms

This section defines, or provides the definition of, all terms and acronyms required to properly understand this specification.

Definitions

  • Pull (Merge) Request: A proposed change to the software (e.g. usually a code change, but may also include documentation, requirements, design, and/or testing).

  • Baseline: A specification or product (e.g., project plan, maintenance and operations (M&O) plan, requirements, or design) that has been formally reviewed and agreed upon, that thereafter serves as the basis for use and further development, and that can be changed only by using an approved change control process (NQA-1, 2009).

  • Validation: Confirmation, through the provision of objective evidence (e.g., acceptance test), that the requirements for a specific intended use or application have been fulfilled (24765:2010(E), 2010).

  • Verification: (1) The process of: evaluating a system or component to determine whether the products of a given development phase satisfy the conditions imposed at the start of that phase. (2) Formal proof of program correctness (e.g., requirements, design, implementation reviews, system tests) (24765:2010(E), 2010).

Acronyms

AcronymDescription
APIApplication Programming Interface
DOE-NEDepartment of Energy, Nuclear Energy
FEfinite element
HITHierarchical Input Text
HPCHigh Performance Computing
I/OInput/Output
INLIdaho National Laboratory
MOOSEMultiphysics Object Oriented Simulation Environment
MPIMessage Passing Interface
SDDSoftware Design Description

Design Stakeholders and Concerns

Design Stakeholders

Stakeholders for MOOSE include several of the funding sources including DOE-NE and the INL. However, Since MOOSE is an open-source project, several universities, companies, and foreign governments have an interest in the development and maintenance of the MOOSE project.

Stakeholder Design Concerns

Concerns from many of the stakeholders are similar. These concerns include correctness, stability, and performance. The mitigation plan for each of these can be addressed. For correctness, Reconstructed Discontinuous Galerkin module development requires either regression or unit testing for all new code added to the repository. The project contains several comparisons against analytical solutions where possible and also other verification methods such as MMS. For stability, the Reconstructed Discontinuous Galerkin module (located within the MOOSE repository) maintains multiple branches to incorporate several layers of testing both internally and for dependent applications. Finally, performance tests are also performed as part of the the normal testing suite to monitor code change impacts to performance.

System Design

The Reconstructed Discontinuous Galerkin module inherits the wide range of pluggable systems from MOOSE. More information regarding MOOSE system design can be found in the framework System Design section. The rDG module home page provides an overview of the various systems used by this module. Documentation for each object, data structure, and process specific to the module are kept up-to-date alongside the MOOSE documentation. Expected failure modes and error conditions are accounted for via regression testing, and error conditions are noted in object documentation where applicable.

System Structure

The architecture of the Reconstructed Discontinuous Galerkin module consists of a core and several pluggable systems (both inherited from the MOOSE framework). The core of MOOSE consists of a number of key objects responsible for setting up and managing the user-defined objects of a finite element simulation. This core set of objects has limited extendability and exists for every simulation configuration that the module is capable of running.

BCs

DGKernels

Materials

Postprocessors

UserObjects

The MooseApp is the top-level object used to hold all of the other objects in a simulation. In a normal simulation a single MooseApp object is created and "run()". This object uses its Factory objects to build user-defined objects which are stored in a series of Warehouse objects and executed. The Finite Element data is stored in the Systems and Assembly object while the domain information (the Mesh) is stored in the Mesh object. A series of threaded loops are used to run parallel calculations on the objects created and stored within the warehouses.

MOOSE's pluggable systems are documented on the MOOSE website, and those for the Reconstructed Discontinuous Galerkin module are on this webpage, accessible through the high-level system links above. Each of these systems has a set of defined polymorphic interfaces and are designed to accomplish a specific task within the simulation. The design of these systems is solid and is managed through agile methods and ticket request system either on GitHub (for MOOSE) or on the defined software repository for this application.

Data Design and Control

At a high level, the system is designed to process HIT input files to construct several objects that will constitute an FE simulation. Some of the objects in the simulation may in turn load other file-based resources to complete the simulation. Examples include meshes or data files. The system will then assemble systems of equations and solve them using the libraries of the Code Platform. The system can then output the solution in one or more supported output formats commonly used for visualization.

Human-Machine Interface Design

The Reconstructed Discontinuous Galerkin module is a command-line driven program. All interaction with the Reconstructed Discontinuous Galerkin module is ultimately done through the command line. This is typical for HPC applications that use the MPI interface for running on super computing clusters. Optional GUIs may be used to assist in creating input files and launching executables on the command line.

System Design Interface

All external system interaction is performed either through file I/O or through local API calls. Neither the Reconstructed Discontinuous Galerkin module, nor the MOOSE framework, nor the other MOOSE modules are designed to interact with any external system directly through remote procedure calls. Any code to code coupling performed using the framework are done directly through API calls either in a static binary or after loading shared libraries.

Security Structure

The Reconstructed Discontinuous Galerkin module does not require any elevated privileges to operate and does not run any stateful services, daemons or other network programs. Distributed runs rely on the MPI library.

Requirements Cross-Reference

  • rdg: Diffusion
  • 10.3.1The system shall be able to solve a simple diffusion problem.

    Specification(s): test

    Design: Diffusion

    Issue(s): #108

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • rdg: AuxAccumulator
  • 10.6.1The system shall support the accumulation of point values as if they were point sources into an auxiliary field

    Specification(s): test

    Design: AuxAccumulator

    Issue(s): #9

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • 10.6.2The system shall report a reasonable error when accumulating point values into an auxiliary field when the system is not properly finalized

    Specification(s): error

    Design: AuxAccumulator

    Issue(s): #9

    Collection(s): FUNCTIONALFAILURE_ANALYSIS

    Type(s): RunException

  • 10.6.3The system shall support mapping data from rays to an aux variable and reset the aux variable to 0 on each time step

    Specification(s): charge_mapping

    Design: AuxAccumulator

    Issue(s): #9

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • 10.6.4The system shall support mapping data from rays to an aux variable and reset the aux variable to 0 on each time step on a 2 dimensional mesh

    Specification(s): charge_mapping2D

    Design: AuxAccumulator

    Issue(s): #9

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • 10.6.5The system shall support mapping data from rays to an aux variable and then solve differential equations based on the data mapped from rays

    Specification(s): simple_potential_solve_aux

    Design: AuxAccumulator

    Issue(s): #9

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • rdg: ResidualAccumulator
  • 10.6.6The system shall support the accumulation of point values as if they were point sources into a nonlinear field

    Specification(s): test

    Design: ResidualAccumulator

    Issue(s): #16

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

  • rdg: VariableSampler
  • 10.6.7The system shall allow a ray to sample the value of field variable at any point in space as it moves through space

    Specification(s): variable_sampling

    Design: VariableSampler

    Issue(s): #12

    Collection(s): FUNCTIONAL

    Type(s): Exodiff

References

  1. ISO/IEC/IEEE 24765:2010(E). Systems and software engineering—Vocabulary. first edition, December 15 2010.[BibTeX]
  2. ASME NQA-1. ASME NQA-1-2008 with the NQA-1a-2009 addenda: Quality Assurance Requirements for Nuclear Facility Applications. first edition, August 31 2009.[BibTeX]