Atmosphere and Energy Research Group (AERG)

Cristina ArcherNiranjan Ghaisas

Shengbai XieChi Yan

Yang Pan

High-Performance Computing Symposium, University of Delaware, 28 January 2015

Outline

• Overview:1. Our wind energy research2. Computational tools3. Why Mills?4. Lessons learned

• Individual talks:– Shengbai: Single-turbine simulations;– Niranjan: Multi-turbine simulations;– Chi: Theoretical framework;– Yang: Turbines-hurricane feedbacks.

1. Our wind energy research

• Wind turbine wake effects;• Turbines-hurricane interactions;• New theoretical framework for studying

atmospheric turbulence.

Single- and multi-turbine wakes

In-house code WiTTS (Wind Turbine and

Turbulence Simulator)

Single- and multi-turbine wakes

In-house code WiTTS (Wind Turbine and

Turbulence Simulator)

OpenFOAM-based SOWFA (Software for Offshore/onshore

Wind Farm Applications)

Both are Large-Eddy Simulation (LES) codes

Turbines-hurricane interactions

Turbines-hurricane interactionsNO TURBINES

WITH TURBINES

Non-incompressible, non-Boussinesq (NINB) framework

• Incompressible assumption:

• Boussinesq assumption:

Non-incompressible, non-Boussinesq (NINB) framework

• Incompressible assumption:

• Boussinesq assumption:

Non-incompressible, non-Boussinesq (NINB) framework

• Incompressible assumption:

• Boussinesq assumption:

• NINB Navier-Stokes equation:

2. Computational tools vs. topics

• Wake effects

• Turbines-hurricane

• NINB framework

2. Computational tools vs. topics

• WiTTS • SOWFA/OpenFOAM

• WRF

• Wake effects

• Turbines-hurricane

• NINB framework

2. Computational tools vs. topics

• WiTTS • SOWFA/OpenFOAM

• WRF

• Wake effects

• Turbines-hurricane

• NINB framework

WiTTS

• Wind Turbine and Turbulence Simulator;• Under development in-house;• FORTRAN 77, MPI parallelized in 3 directions;• Combines an atmospheric boundary layer (ABL)

solver and an actuator line model for the wind turbine blades;

• Tested for incompressible ABL under neutral, stable, and unstable conditions with and without a single turbine.

SOWFA

• Software for Offshore/onshore Wind Farm Applications;

• Developed by the National Renewable Energy Lab (NREL) using OpenFOAM’s C++ libraries;

• LES code for atmospheric boundary layer (ABL) with advanced treatment of wind turbines;

• Website: https://nwtc.nrel.gov/SOWFA

WRF

• Weather Research and Forecasting model;• Most widely used mesoscale numerical weather

prediction model (>25,000 users);• Developed by NCAR/NOAA/FSL/NRL and

supported by NCAR for free; • Advanced physics, dynamics, numerics;• Global to small (~1 km) scale;• Website: http://www.wrf-model.org/index.php

3. Why Mills?

• All models need large disk space and memory, thus multi-processor, parallel environment necessary;

• Complex setup, with multi-users and multi-versions of codes, libraries, compilers;

• Post-processing and data display with graphic software require large memory and disk space too;

• Need support for maintenance, upgrades, security, etc.

4. Lessons learned

• Valet packages are great;• Involve IT early;• System versions are better than user versions;• Adopt early a file and directory naming convention;• Request to be notified about node failures or

reboots;• Name log files with $JOBID so they do not get

overwritten;• Be afraid of, and ready for, big system upgrades!