Development of a Visualization Tool for the ARIES Systems Code

Slides:



Advertisements
Similar presentations
First Wall Heat Loads Mike Ulrickson November 15, 2014.
Advertisements

ASIPP Zhongwei Wang for CFETR Design Team Japan-US Workshop on Fusion Power Plants and Related Advanced Technologies February 26-28, 2013 at Kyoto University.
Who will save the tokamak – Harry Potter, Arnold Schwarzenegger, Shaquille O’Neal or Donald Trump? J. P. Freidberg, F. Mangiarotti, J. Minervini MIT Plasma.
September 15-16, 2005/ARR 1 Status of ARIES-CS Power Core and Divertor Design and Structural Analysis A. René Raffray University of California, San Diego.
1 Lane Carlson ARIES Pathways Project Meeting Gaithersburg, MD, Oct 13-14, 2011 Substantiating the ASC & Implementing the DCLL Blanket.
Overview of the ARIES “Pathways” Program Farrokh Najmabadi UC San Diego 8 th International Symposium on Fusion Nuclear Technology Heidelberg, Germany 01–
January 8-10, 2003/ARR 1 Plan for Engineering Study of ARIES-CS Presented by A. R. Raffray University of California, San Diego ARIES Meeting UCSD San.
April 27-28, 2006/ARR 1 Finalizing ARIES-CS Power Core Engineering Presented by A. René Raffray University of California, San Diego ARIES Meeting UW, Madison.
June 14-15, 2007/ARR 1 Trade-Off Studies and Engineering Input to System Code Presented by A. René Raffray University of California, San Diego With contribution.
29 July Lane Carlson, Charles Kessel Mark Tillack, Farrokh Najmabadi ARIES-Pathways Project Meeting Washington, D.C. June 29-30, 2010 Exploring the.
Contributions of Burning Plasma Physics Experiment to Fusion Energy Goals Farrokh Najmabadi Dept. of Electrical & Computer Eng. And Center for Energy Research.
Page 1 of 14 Reflections on the energy mission and goals of a fusion test reactor ARIES Design Brainstorming Workshop April 2005 M. S. Tillack.
1 Lane Carlson, Mark Tillack, Farrokh Najmabadi, Charles Kessel University of California, San Diego & Princeton Plasma Physics Lab US/Japan Workshop on.
Proposals for Next Year’s MFE Activities C. Kessel, PPPL ARIES Project Meeting, Sept. 24, 2000.
Development of the New ARIES Tokamak Systems Code Zoran Dragojlovic, Rene Raffray, Farrokh Najmabadi, Charles Kessel, Lester Waganer US-Japan Workshop.
Optimization of a Steady-State Tokamak-Based Power Plant Farrokh Najmabadi University of California, San Diego, La Jolla, CA IEA Workshop 59 “Shape and.
Overview of ARIES Compact Stellarator Study Farrokh Najmabadi and the ARIES Team UC San Diego US/Japan Workshop on Power Plant Studies & Related Advanced.
Progress on Determining Heat Loads on Divertors and First Walls T.K. Mau UC-San Diego ARIES Pathways Project Meeting December 12-13, 2007 Atlanta, Georgia.
Physics Issues and Trade-offs in Magnetic Fusion Power Plants Farrokh Najmabadi University of California, San Diego, La Jolla, CA APS April 2002 Meeting.
Highlights of ARIES-AT Study Farrokh Najmabadi For the ARIES Team VLT Conference call July 12, 2000 ARIES Web Site:
ARIES Systems Studies: ARIES-I and ARIES-AT type operating points C. Kessel Princeton Plasma Physics Laboratory ARIES Project Meeting, San Diego, December.
March 20-21, 2000ARIES-AT Blanket and Divertor Design, ARIES Project Meeting/ARR Status ARIES-AT Blanket and Divertor Design The ARIES Team Presented.
Creating a Calibration Measurement Monitoring System for Many, Ever-changing, Complex Instruments John Wilson Software QA Engineer Agilent Technologies,
The Pursuit for Efficient S/C Design The Stanford Small Sat Challenge: –Learn system engineering processes –Design, build, test, and fly a CubeSat project.
Re-Examination of Visions for Tokamak Power Plants – The ARIES-ACT Study Farrokh Najmabadi Professor of Electrical & Computer Engineering Director, Center.
Progress in ARIES-ACT Study Farrokh Najmabadi UC San Diego Japan/US Workshop on Power Plant Studies and Related Advanced Technologies 8-9 March 2012 US.
A new approach for exploration of tokamak power plant design space Farrokh Najmabadi, Lane Carlson, and the ARIES Team UC San Diego 4 th IAEA Technical.
Systems Analysis Development for ARIES Next Step C. E. Kessel 1, Z. Dragojlovic 2, and R. Raffrey 2 1 Princeton Plasma Physics Laboratory 2 University.
Managed by UT-Battelle for the Department of Energy Stan Milora, ORNL Director Virtual Laboratory for Technology 20 th ANS Topical Meeting on the Technology.
San Diego Workshop, 11 September 2003 Results of the European Power Plant Conceptual Study Presented by Ian Cook on behalf of David Maisonnier (Project.
Hardware. Make sure you have paper and pen to hand as you will need to take notes and write down answers and thoughts that you can refer to later on.
NSTX-U NSTX-U PAC-31 Response to Questions – Day 1 Summary of Answers Q: Maximum pulse length at 1MA, 0.75T, 1 st year parameters? –A1: Full 5 seconds.
1 Lane Carlson ARIES Pathways Project Meeting San Diego, CA Jan 23-24, 2012 Updating the SCLL Design & ASC Documentation.
Thoughts on Fusion Competitiveness Initiative Farrokh Najmabadi, George Tynan UC San Diego University Fusion Initiatives Meeting, MIT 14-15, February 2008.
ARIES-AT Physics Overview presented by S.C. Jardin with input from C. Kessel, T. K. Mau, R. Miller, and the ARIES team US/Japan Workshop on Fusion Power.
Systems Code – Hardwired Numbers for Review C. Kessel, PPPL ARIES Project Meeting, July 29-30, 2010.
1 Lane Carlson 1, Mark Tillack 1, Farrokh Najmabadi 1, Charles Kessel 2 1 University of California, San Diego & 2 Princeton Plasma Physics Lab US/Japan.
1 Lane Carlson ARIES-Pathways Project Meeting Gaithersburg, MD, July 27-28, 2011 Generalization and Blanket Updates to the ASC.
Summary and Closing Remarks Farrokh Najmabadi UC San Diego Presentation to ARIES Program Peer Review August 29, 2013, Washington, DC.
Compact Stellarator Approach to DEMO J.F. Lyon for the US stellarator community FESAC Subcommittee Aug. 7, 2007.
Fusion Nuclear Science - Pathway Assessment C. Kessel, PPPL ARIES Project Meeting, Bethesda, MD July 29, 2010.
Outline Introduction Research Project Findings / Results
Systems Analysis Development for ARIES Next Step C. E. Kessel 1, Z. Dragojlovic 2, and R. Raffrey 2 1 Princeton Plasma Physics Laboratory 2 University.
Assessment of Fusion Development Path: Initial Results of the ARIES “Pathways” Program Farrokh Najmabadi UC San Diego ANS 18 th Topical Meeting on the.
Development and Scope of ARIES Systems Code Zoran Dragojlovic A. René Raffray Farrokh Najmabadi ARIES-“TNS” Project Meeting April 3 and 4, 2007 University.
Discussion Slides S.E. Kruger, J.D. Callen, J. Carlson, C.C. Hegna, E.D. Held, T. Jenkins, J. Ramos, D.D. Schnack, C.R. Sovinec, D.A. Spong ORNL SWIM Meet.
1 Lane Carlson, Charles Kessel, Stephen Efthyvoulos ARIES-Pathways Project Meeting Bethesday, MD April 4-5, 2011 Finalized Systems Code Modifications &
Comments on ARIES-ACT 1/2011 Strawman L. El-Guebaly Fusion Technology Institute University of Wisconsin-Madison
ARIES Pathways Project 05/29/08
PLM, Document and Workflow Management
GUI Design and Coding PPT By :Dr. R. Mall.
Pilot Plant Study Hutch Neilson Princeton Plasma Physics Laboratory ARIES Project Meeting 19 May 2010.
System Design Ashima Wadhwa.
X.R. Wang, M. S. Tillack, S. Malang, F. Najmabadi and the ARIES Team
Improvements to power flow modeling in the ARIES system code
The European Power Plant Conceptual Study Overview
Trade-Off Studies and Engineering Input to System Code
University of California in San Diego
Progress on Systems Code Application to CS Reactors
Farrokh Najmabadi Professor of Electrical & Computer Engineering
Comments on ARIES-ACT 10/20/2010 Pre-Strawman
Near-term plan for the current ARIES project
Status of the ARIES Program
CS 8532: Advanced Software Engineering
Systems analysis of ACT2 design space
Visualizing the Attracting Structures Results and Conclusions
Overview of Workflows: Why Use Them?
Analysis of Technical and Programmatic Tradeoffs with Systems Code
ACT-1 design point definition
University of California, San Diego
Presentation transcript:

Development of a Visualization Tool for the ARIES Systems Code Stress work of others - myself just a small part. Lane Carlson ARIES-Pathways Project Meeting UCSD May 19, 2010 19 May 2010

Systems code has evolved into a design space scanning tool Rather than optimize about a specific design point, new approach scans a wide operating space for a range of possible design points. It becomes necessary to visualize slopes - is it a steep optimization point with no leeway or at a shallow point with a relaxed constraint? A larger design window may open when a constraint is slightly relaxed, with substantial improvement in COE. Other times, moving away from the “optimized” point gives a more robust and credible design with minimal impact on COE. START) Over the past few years, the systems code has evolved. 1)Constrained too tightly at optimal operating point, backed into a corner. APPLICATIONS 1) Increased scanning capability and resolution with cluster. 3) Detailed hand and CAD analysis done for credibility. Code is SELF CONSISTENT - it will fulfill its own equations but needs to be verified. FLOW: Large scan Identify area of interest Narrow down search Specific point study (From F. Najmabadi) 19 May 2010

Large databases can be processed Have capability to parallel process the databases on a cluster with hundreds of nodes. Princeton Plasma Physics (PPPL) computer cluster: 200+ processors Quad-core CPUs Up to 16 GB RAM 1 gigabit ethernet START) I mentioned that we might have 10^6-10^7 points to process. CAN run on a laptop, just takes a long time. Large systems scans are possible (106 - 107 points) Targeted systems scans around a region of interest Operating point search and sensitivity scans, supported by detailed analysis 10^6 = 60 days on laptop vs 1 day with 20 nodes, or less with more nodes 2 GHz Core2 Duo laptop AMD Quad-core node ~700 points/hr ~2000 points/hr 105 points = 142 hrs = 50 hrs/node 106 points = 1428 hrs = 500 hrs/node 10^6 = 60 days with laptop vs. 1 day with 20 nodes PPPL Cluster 19 May 2010

It is a challenge to visualize large data sets Visualizing large datasets is a difficult task, almost an art 106 points overwhelm monitor real estate “Lots of numbers don’t make sense to ‘low-bandwidth’ humans, but visualization can encode large amounts of data to gain insight.” - San Diego Super Computer Center SDSCC Earthquake Simulation NOAA (National Oceanic and Atmospheric Admin.) data-logging buoys 19 May 2010

NOAA Data Visualization 19 May 2010

More examples of visualizing large data sets European Airspace Reboot: During volcano activity (April 17, 2010) World Data from Gapminder.org Income per person vs. life expectancy European Airspace Reboot: After volcano activity (April 20, 2010) 19 May 2010

We are developing a visualization tool to utilize the scanning capability of the new systems code VASST - Visual ARIES Systems Scanning Tool Working to visualize the broad parameter space to extract meaningful data and uncover new relationships Graphical user interface (GUI) permits 2D and 3D plots of any parameter Purpose: to give the user more visual interaction and explorative power to bring to light new relationships 19 May 2010

(Visual ARIES Systems Scanning Tool) VASST GUI Highlights (Visual ARIES Systems Scanning Tool) Number of points in database Blanket database used Auto-labeling Pull-down menus for common parameters Color bar scale Constraint parameter set by user Data cursor displays info on mouse-click Correlation coefficient Save plot as TIFF, JPEG, BMP, PNG… 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=8.5 T SiC blanket 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=7.5 T SiC blanket 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=7.0 T SiC blanket 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=6.5 T SiC blanket 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=6.0 T SiC blanket 19 May 2010

Example: Qdivinb vs COE, CC: B, constraint: B=5.0 T SiC blanket 19 May 2010

Extra: Pnelec (unrestricted) vs COE, CC: COE SiC blanket Possible attractive power plant designs in the 500 MW range 19 May 2010

Extra: B vs COE, CC: R SiC blanket 19 May 2010

Extra: Btmax vs COE, CC: R SiC blanket 19 May 2010

An additional visualization feature might include “linking and brushing” the data Highlighting any parameter space would show real-time effect on other parameters 19 May 2010

Three important elements are required to operate VASST A intuitive visualization tool A capable, experienced user In-depth chronicle of details    param COE 4 m 30 5 m 35 6 m 40 7 m 45  19 May 2010

A chronicle/history is required for record keeping purposes What input parameters were used? What version of the systems code was used? (Subversion control) What blanket was implemented? What were the assumptions applied in the code? What filters were implemented? (Pnetel, Qdiv, B, etc.) What costing algorithms were used? Every result/picture/graph should be backed up with specifics of its origin 19 May 2010

Revision control is necessary and helpful Needful to control revisions to the code to keep track of changes made by multiple users. Subversion (SVN) revision control software keeps the code centralized in a server repository. Users can “check out” the code, modify it, then “commit” it back to the central server as the most recent updated version. START) I mentioned that we might have 10^6-10^7 points to process. CAN run on a laptop, just takes a long time. Large systems scans are possible (106 - 107 points) Targeted systems scans around a region of interest Operating point search and sensitivity scans, supported by detailed analysis 10^6 = 60 days on laptop vs 1 day with 20 nodes, or less with more nodes 19 May 2010

Clarification of assumptions Btmax vs. SC current = ? Three references are an order of magnitude apart. START) I mentioned that we might have 10^6-10^7 points to process. CAN run on a laptop, just takes a long time. Large systems scans are possible (106 - 107 points) Targeted systems scans around a region of interest Operating point search and sensitivity scans, supported by detailed analysis 10^6 = 60 days on laptop vs 1 day with 20 nodes, or less with more nodes 19 May 2010

Clarification of assumptions From Jan 2009 ARIES meeting, Z.D. Need clarification of specifics between DCLL and SiC blanket. Any other assumptions that need concrete definitions? Basis for physics and engineering design? Scaling factors from other designs? START) I mentioned that we might have 10^6-10^7 points to process. CAN run on a laptop, just takes a long time. Large systems scans are possible (106 - 107 points) Targeted systems scans around a region of interest Operating point search and sensitivity scans, supported by detailed analysis 10^6 = 60 days on laptop vs 1 day with 20 nodes, or less with more nodes From paper, “An advanced computational…” by Z.D. 19 May 2010

We would like to fill in the operating space Are there pertinent or interesting areas the ARIES team would like to look at? - Increase Qdiv? Elongate plasma? More aggressive or relaxed B? Comments, suggestions, improvements? ARIES-AT physics DCLL blanket ARIES-AT physics SiC blanket Aggressive in physics START) We have been talking about pushing the limits and scanning a design space, so how can we do this and what have we done so far? Prior designs constrained us. For the first time we are able to move around in this space. Abscissa (X), ordinate (Y) Aggressiveness means tradeoff between risk and reward. High-tech and more risky or low-tech and safer. (E.G.) ARIES-I, DCLL blanket: AGGRESSIVENESS is not just in the blanket, as shown here, but also in other systems such as higher magnetic materials and fields, higher thermal conversion systems, pushing limits of power on the divertor, plasma elongation capabilities, and so on. Pushing the limits may be harder to achieve but we don’t want to be pushed into a tight design corner. 2) Use the systems code….the payoff is a fuller understanding of tokamak design. ARIES-I physics DCLL blanket ARIES-I physics SiC blanket Aggressive in technology 19 May 2010

Summary The systems code is operational and able to scan and process a large design space. A first version of VASST GUI is operational and ready to look at pertinent system scans for the ARIES team. Beginning chronicle of details and specifics of the different blanket modules. Payoff of filling the operating space is a fuller, richer understanding of tokamak design. Live VASST demo? 19 May 2010

Future work Re-examine the TF and PF coil j vs. B relationships. Prior SC magnets may be too optimistic (3x) - re-examine lower B fields for possible solutions. Develop solutions to the power balancing within the limits of materials/coolants - critical to accessing smaller power plants. After a broad scan, fill in the operating space of the four corners with sufficient resolution. START) some areas that have surfaced recently in which we can use the systems code scanning ability are… 1) Revisit TF, PF coil j vs B relationship to make better projections in the future. We now have a design point for magnet coils from ITER, which we expect will be built. 2) High temp SC magnets look interesting but have some fundamental issues with fields perpendicular to the tape that degrade their performance. Therefore we may look for points that do not require very high fields at the magnets. 3) [Radiated power fraction = (0.24-0.32)] 19 May 2010

Extra Slides To discuss individually: Rev number for costing algorithms? 2 blanket types: SiC and DCLL (C.K.) DCLL ΔFW = 0.038 m Δblkt = 0.50 m ΔVV = 0.31 m Δshld/skel = 0.35+0.075xIn(<Nw>/3.3) m ηth ~ 42%, Ppump ~ 0.04xPfusion SiC ΔFW = 0.0 m Δblkt = 0.35 m ΔVV = 0.40 m Δshld/skel = 0.24+0.067xIn(<Nw>/3.3) m ηth ~ 55%, Ppump ~ 0.005xPfusion START) some areas that have surfaced recently in which we can use the systems code scanning ability are… 1) Revisit TF, PF coil j vs B relationship to make better projections in the future. We now have a design point for magnet coils from ITER, which we expect will be built. 2) High temp SC magnets look interesting but have some fundamental issues with fields perpendicular to the tape that degrade their performance. Therefore we may look for points that do not require very high fields at the magnets. 3) [Radiated power fraction = (0.24-0.32)] 19 May 2010

ARIES systems code consists of modular building blocks Systems code integrates physics, engineering, design, and costing. Systems Code Analysis Flow 1. PHYSICS Plasmas that satisfy power and particle balance 2. ENGINEERING FILTERS APPLIED 3. ENGINEERING & COSTING DETAILS Power core, power flow, magnets, costing, COE Systems code INTEGRATES and provides a detailed build with costs of components and COE. PHYSICS: submit an input file (~25 parameters) specifying the range of constraints/parameters you want to vary and the resolution (finer takes longer) PHYSICS FILTERS: to remove all improbable physics, P CD < Pheat, f bs < 1, Tau He/Tau e < 4. May start out with millions of physics points, but after filtering out unreal solutions, may get hundreds to tens of thousands of possible design points. ENG FILTERS: 1) 6-18 T, limited by material and plasma confinement requirements 2) 5-8 MW/m^2 for AT, LiPb, SiC, 10-12 for CS, He-cooled 3) AT <6 MW/m^2 4) Near 1000 MW net electric power DETAILS: Blankets: blanket module allows different blankets to be tested: dual-coolant, lead-lithium (DCLL), silicon carbide (SiC), ARIES-AT = SiC faced with W, cooled w Pb-Li ALSO, different blankets can be compared against each other. I.e. DCLL w/ and w/o manifold. Can compare the thermal conversion efficiencies of the blankets. Costing includes direct and indirect costs, will explain more later. Filters include: Modules include: Toroidal magnetic fields Heat flux to divertor Neutron wall load Net electric power Blankets Geometry Magnets Power flow Costing DCLL SiC ARIES-AT 19 May 2010