System-of-Systems Architecture
Physical View
The INTERSECT AAM system is deployed at Oak Ridge National Laboratory’s (ORNL’s) Manufacturing Demonstration Facility (MDF), Spallation Neutron Source (SNS), and Oak Ridge Leadership Computing Facility (OLCF) as a cross-facility instrument-science workflow. The INTERSECT AAM's Physical View contains the following components (Fig. 150):
The 3D metal printer with its control computer, both located at the MDF.
The OLCF Advanced Computing Ecosystem (ACE) testbed computer used for the thermomechanical simulation in the Distributed Experiment Steering architectural pattern’s feedback loop of the 3D metal printing process.
The SNS with its neutron beam experiment and corresponding neutron diffraction sensors.
The SNS Analysis cluster computer used for the analysis of the neutron diffraction data that is used for validation.
The domain expert computer used for experiment planning of the 3D metal printing process.
These physical components interact with each other as follows (Fig. 150):
The experiment plan is created by a domain expert on the domain expert computer.
The experiment plan is transferred from the domain expert computer to the 3D metal printer control computer, crossing the ORNL to MDF facility/network boundary.
During the 3D metal printing process, thermocouple sensor data and infrared images are transferred from the 3D metal printer control computer to the OLCF ACE testbed computer used for the thermomechanical simulation in the Distributed Experiment Steering architectural pattern’s feedback loop, crossing the MDF to OLCF facility/network boundary. The data transfer is automated and the execution of the thermomechanical simulation is triggered by the data transfer.
During the 3D metal printing process, thermomechanical simulation results that modulate the printing process are transferred from the OLCF ACE testbed computer to the 3D metal printer control computer in the Distributed Experiment Steering architectural pattern’s feedback loop, crossing the OLCF to MDF facility/network boundary. The data transfer is automated and the modulation of the printing process is triggered by the data transfer.
The 3D printed metal structure is physically transferred from the MDF facility to the SNS for neutron beam experiments to collect ex-situ neutron diffraction data. In an alternative experimental setup, the 3D printing process is performed at the SNS in the live neutron beam, collecting in-situ neutron diffraction data.
The neutron diffraction data generated by the SNS facility is transferred to the SNS Analysis cluster computer as part of the Distributed Design of Experiments architectural pattern’s feedback loop. The data transfer is automated.
The neutron diffraction analysis results are transferred from the SNS Analysis cluster computer used for analyzing the neutron diffraction data to the domain expert computer used for experiment planning of the 3D metal printing process as part of the Distributed Design of Experiments architectural pattern’s feedback loop, crossing the OLCF to ORNL facility/network boundary. The data transfer can be automated and the execution of the analysis can be triggered by the data transfer.
Fig. 150 Physical components of the INTERSECT AAM system and their interactions
Logical View
In the INTERSECT AAM's Logical View, the physical components are abstracted as iteracting INTERSECT infrastructure systems as follows (Fig. 151):
Additive Manufacturing System: The 3D metal printer with its control computer at the MDF.
Experiment Steering Analysis System: The OLCF ACE testbed computer used for the thermomechanical simulation.
Spallation Neutron Source System: The SNS with its neutron beam experiment and corresponding neutron diffraction sensors.
Design of Experiments Analysis System: The SNS Analysis cluster computer used for the analysis of the neutron diffraction data.
Experiment Planning System: The domain expert computer used for experiment planning.
Fig. 151 Infrastructure systems of the INTERSECT AAM system and their interactions
The INTERSECT AAM's Logical View also maps Systems, Subsystems, and Services to the infrastructure systems, with logical systems spanning over one or more infrastructure systems and the intersection of logical and infrastructure systems providing services as follows:
Logical \ Infrastructure System |
Additive Manufacturing System |
Experiment Steering Analysis System |
Spallation Neutron Source System |
Design of Experiments Analysis System |
Experiment Planning System |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Data View
The INTERSECT AAM's Data View consists of the Operational Data necessary to function in the INTERSECT ecosystem and the Experimental Data needed for performing and resulting from the AAM experiment.
Operational Data
In general, INTERSECT operational data is organized according to the Entity-Relationship Data Model in the INTERSECT Data View. The INTERSECT AAM system’s resources, users, experiments, and campaigns are part of the experiment-specific operational data model.
For example, adding a user to the INTERSECT AAM system results
in the addition of instances of the User and User Profile entities. The
domain expert may create a number of related experiments; the relationship
between them might be captured as one or more instances of the Campaign
entity, with the eventual creation of appropriate Campaign Result,
Campaign Error, and/or Campaign Template instances.
The physical components utilized in the INTERSECT AAM system,
such as the 3D metal printer and its controller computer, are defined as
resources with corresponding instances of the Resource entity. The
INTERSECT AAM system interacts with other, previously-defined
resources within the ecosystem, such as the OLCF ACE testbed
computer and the SNS Analysis cluster computer that have been defined
by their administrators/owners and are available to the AAM system as
part of the INTERSECT ecosystem.
Experimental Data
The following data artifacts are generated and/or managed as experimental data by the INTERSECT AAM system:
- Experiment Design Plan
The experiment design plan describes the goal, which is the validated 3D printing of a metal part with predetermined structural stresses. It is modified and improved as part of the Distributed Design of Experiments architectural pattern’s feedback loop. The initial and subsequent plans are stored in a repository.
- Experiment Plan
The experiment plan is the sequence of predetermined steps and associated parameters necessary to 3D print the metal part. The parameters include the targeted structural stress and the options for changing the laser parameters, such as temperature and speed. It is a comma-separated values format (CSV) file with a file size in the kilobyte range. The experiment plans are stored in a repository.
- IR Data
The IR data are processed IR camera images with IR data points spatially mapped to the 3D printed structure. Each processed image is a CSV file with a file size in the kilobyte range.
- Thermocouple Data
The thermocouple data are the measured temperature values at the base of the 3D printed structure and CSV files with a file size in the kilobyte range.
- Sample Metadata
This data describes the printed sample for an experiment.
- Neutron Diffraction Data
The neutron diffraction data is stored in a X-ray diffraction (XRD) file. It is available on the SNS Analysis cluster computer after the experiment is finished.
Operational View
The INTERSECT AAM's Operational View describes tasks and procedures from the viewpoint of real-world operations stakeholders. The intent is to capture practical constraints and procedures for the operation and use of the additive manufacturing equipment.
The INTERSECT AAM system has two experimental setups. The primary setup runs the AM process at the MDF and uses the SNS to collect ex-situ neutron diffraction data. The secondary setup runs the AM process at the SNS in the live neutron beam, collecting in-situ neutron diffraction data, using a more mobile 3D printer and a containment environment. Both setups employ thermocouple sensors and IR cameras. These two setups include the following operational aspects that impact the AM process depicted in Fig. 152:
The IR camera can be moved from the primary setup at the MDF to the secondary setup at the SNS, requiring spatial recalibration.
The thermomechanical simulation can run on a variety of hardware platforms, using either local or remote (to the 3D printer) computational resources.
Beam time at the SNS requires specific resource allocations and availability is limited.
The thermo-mechanical simulation (i.e., ADAMANTINE) is often run using process containers (e.g., Docker), which helps to streamline the setup and execution. These simulations can run on the edge computer at the MDF, on the OLCF ACE testbed, or even on the SNS Analysis cluster computer. The autonomous adaptive control (via ADAMANTINE) steers the operational parameters for the 3D printing process, based on information gathered from the thermocouple sensors and IR camera. The Robot Operating System (ROS) software environment on the 3D metal printer control computer controls the printing process and coordinates the overall set of actions. The visualization/dashboard runs via ROS and offers real-time information. An operator (human) is responsible for monitoring the printing process for safety reasons, with phsical “kill switches” to abort if needed.
The Okuma MU-8000V Laser EX device used in the AM process includes both, a laser for additive procedures (i.e., for melting and depositing metal powder) and a milling tool for subtractive machining (Fig. 153). The device can only use one mode at a time, additive or subtractive. A close-up view of the platform, camera, and laser head are shown in Fig. 154.
Fig. 153 The Okuma MU-8000V Laser EX device and its control station. The milling platform in the center is in a level position but can rotate as needed. The laser head is to the right (gold tip) and the thermal camera is to the left (plastic cover).
Fig. 154 Closer view of the milling platform, laser head (right - gold tip) and thermal camera (left - plastic cover).
User View
The INTERSECT AAM's User View defines the roles of human actors, the processes implemented by the INTERSECT architecture to support those roles, and the user interfaces necessary to support those roles.
Roles
The primary human actor in the INTERSECT AAM system is the domain expert (DE). This actor is responsible for conceiving the autonomous experiment, creating the experiment design plan and experiment plan, and submitting the plans to the corresponding systems. Depending on the structure of the AAM project team, the DE may fill multiple User Roles:
- User
The DE plans and executes a number of different AAM experiments.
- Maintainer/Operator
The DE may be responsible for maintenance of the domain expert computer, the 3D metal printer, or the 3D metal printer control computer. Other resources, such as the ACE testbed and the SNS Analysis cluster computer have their own designated maintainers/operators.
- Administrator
The DE may be responsible for approving new resources (e.g., replacing the domain expert computer with a more recent or capable machine). Other resources, such as the ACE testbed and the SNS Analysis cluster computer have their own designated administrators.
- Owner:
The DE may be an owner of resources for the purposes of accounting or other fiscal responsibilities. Other resources, such as the ACE testbed and the SNS Analysis cluster computer have their own designated owners.
- Provider
The DE does not fill this role. Other resources, such as the ACE testbed and the SNS Analysis cluster computer have their own designated providers.
Processes
The User Processes implemented by INTERSECT AAM system may include some combination of the following, among others:
- Login
As part of conducting an INTERSECT AAM experiment, the DE logs into the INTERSECT user interface.
- Compile DAG
The DE uses the INTERSECT user interface to assemble the necessary resources to accomplish an AAM experiment.
- Request resources
The physical resources depicted in Fig. 150 are represented in INTERSECT as resources and need to be allocated as part of the experiment setup.
- Trigger workflow
When the resources for the experiment have been allocated and mapped to the created DAG, the INTERSECT AAM experiment workflow is triggered and executed.
User Interfaces
Multiple user interfaces will be used by the DE as part of executing AAM experiments. Administrative user interfaces will allow the DE to coordinate other users and resources. In the operator role, the DE will complete tasks such as monitoring resources, setting up resources, and updating resource information and configuration. In the owner role, the DE will make use of interfaces to edit resource configurations, manage user permissions and roles, and view the resources that the DE owns. The user role will expose the largest selection of user interfaces to the DE, with tasks ranging from applying for an INTERSECT account, to editing user profiles, to creating, starting, and steering a campaign of AAM experiments.
Standards View
The INTERSECT AAM's Standards View consists of Internal Standards and External Standards.
Internal Standards
The INTERSECT AAM's Standards View consists of the following Internal Standards:
- Logical View
- Systems, Subsystems, and Services
Minimum requirement: Infrastructure Management System
Minimum requirement: User Management System
Minimum requirement: Orchestration System
Minimum requirement: Data Management System
Minimum requirement: Campaign Management System
Minimum requirement: Communication System
Minimum requirement: Apapters
- Error and Failure Notification and Handling
Minimum requirement: Detection
Minimum requirement: Notification
Minimum requirement: Handling
- User View
Minimum requirement: User Roles
External Standards
The INTERSECT AAM's Standards View consists of the following External Standards:
Requirement
Okuma MU-8000V Laser EX: The Okuma MU-8000V Laser EX is a 3D metal printer combines the latest laser additive technology with subtractive machining capabilities. It is a multitasking computer numerical control (CNC) machine that implements laser metal deposition (LMD) technology with the ability to cut unique parts of many different sizes and shapes. LMD supplies powder from nozzles and performs laser melting and bonding to the parent material.
Requirement
Robot Operating System (ROS): ROS [B97] is a set of Linux-based software libraries and tools for robot applications. It is deployed on the 3D metal printer control computer.
Requirement
OLCF ACE Testbed: The ACE testbed [B103] is a unique OLCF capability that provides a sandboxed area for deploying computing and data resources and facilitating the evaluation of diverse workloads, including for INTERSECT and IRI. It is used to run the ADAMANTINE thermomechanical simulation.
Requirement
ADAMANTINE: ADAMANTINE [B104] is open-source sofware to simulate heat transfer for additive manufacturing. It is deployed on the OLCF ACE testbed computer used for the thermomechanical simulation.
Requirement
SNS VULCAN Engineering Materials Diffractometer : The SNS VULCAN Engineering Materials Diffractometer [B105] is designed for deformation, phase transformation, residual stress, texture, and microstructure studies. It is an instrument that has unique scientific capabilities, its own control software, and generates the needed neutron diffraction data in its own data formats.
Requirement
SNS Analysus cluster computer: The SNS Analysus cluster computer is located at the SNS and specifically designated for analyzing neutron diffraction data from the SNS.