Embedded Files
Next Generation Power Conversion
Next Generation Power Conversion
Obtaining a Compressor Map
•Based on the test result of a newly installed S-CO2 TAC compressor
•No experiments were performed to precisely steady-state the compressor inlet conditions to reflect real gas effects
•Plan to obtain a dimensionless performance curve of the compressor with precise valve regulation
•Scheduled to experiment with compressor with inlet conditions inside phase dome
Surge and Post-surge Behavior
•Compressor surge and post-surge behavior is studied experimentally
•Post-surge pressure frequency and recovery process is investigated
•Inlet temperature and system CO2 inventory affect surge behavior of compressor
Magnetic Bearing Lubrication Analysis
ABC loop - magnetic bearing test
Controlled CO2 are injected into the AMB test system
AMB test results : shaft trajectory data from AMB feedback sensor
Autonomous Brayton Cycle (ABC) loop
Shaft trajectory data (vacuum)
TAC with magnetic bearing
Shaft trajectory data (S-CO2 )
Test Results Analysis
Reynolds’ equation is used as governing equation
Lubrication analysis with random position
Reynolds’ equation
Lubrication analysis flow chart
Lubrication force vs. CO2 condition
Lubrication analysis for each test
Precooler Transient Analysis
•Based on the nuclear system analysis code (e.g. MARS-KS and GAMMA+), precooler transient analysis model is modified.
•Effect of fluid properties change and heat exchanger component thermal resistance is added.
Development of Fast Heat Exchanger Transient model
•Design model based on LMTD method with correction factor for control the precooler
•Build model based on precooler mechanism
Precooler System Analysis for S-CO2 System Control
•Validating the developed model with experimental results and system analysis code
•Off-design model developed based on on-design model and correction factor
Compressor Inlet Control Based on Optimal Control Theory
Compressor inlet temperature controller is designed based on fast precooler model and optimal control theory.
Verifying various controller based on control theory by experiment
Inlet temperature is maintained lower than 0.5%, which meet the ASME criterion.
Machine Learning based S-CO2 System Control
•Reinforcement learning (RL) based control research is being conducted to elevate the level of autonomy of the S-CO2 system.
•Time series surrogate model has been established using simulation data produced by the system analysis code.
•RL agent has been trained based on proximal policy optimization (PPO) algorithm.
•Trained agent in the simulation environment will be tested and validated on a real hardware facility, the ABC test loop.
Pre-training and Transfer Learning
Page updated
Google Sites
Report abuse