56 lines
2.3 KiB
Markdown
56 lines
2.3 KiB
Markdown
# An autonomous control framework for advanced reactors notes
|
|
|
|
## First Pass
|
|
**Category:**
|
|
This is a vision-like paper.
|
|
|
|
**Context:**
|
|
Dr. Wood is a professor of nuclear engineering at the University of Tennesse.
|
|
He started working at UT after working at ORNL for 20+ years where he was
|
|
a senior researcher. He works on common cause failures.
|
|
|
|
**Correctness:**
|
|
There are not many citations, and a lot of the paper seems like it is off-the-cuff.
|
|
This was an invited article though in the *Nuclear Engineering and Technology*
|
|
journal. He's also got quite the resume.
|
|
|
|
**Contributions:**
|
|
Dr. Wood gives a good overview of where we are (automated control), and
|
|
where we need to go (autonomous control). Comparisons are made to other
|
|
industries, and how they've addressed these challenges. Clarifications are made
|
|
with respect to how the nuclear industry is unique.
|
|
|
|
**Clarity:**
|
|
Well written and clear, but a bit wordy at times. Examples are used but
|
|
sometimes it just feels like a mouthful.
|
|
|
|
## Second Pass
|
|
**What is the main thrust?**
|
|
Wood talks a lot about what it means for a system to be autonomous. He
|
|
actually starts with a etymological argument- "*automatos* means self-acting,
|
|
whereas *autonomos* means independent." The point of an autonomous system
|
|
is that it must be able to make intelligent decisions that a human would
|
|
otherwise make. Knowing how to handle degraded or faulty sensors or
|
|
actuators, switching controller trains, and knowing when scramming is
|
|
necessary are all decisions that an autonomous system would make, that
|
|
a simple automated system could not. The key is in the decision making.
|
|
|
|
**What is the supporting evidence?**
|
|
Dr. Wood compares a lot to NASA control systems for the Mars rovers. These
|
|
systems use two tiers of controllers: a decision layer and a functionality layer.
|
|
The functionality layer does what you think--it manages the physical
|
|
functioning. The decision layer is different. It decides on controller modes and
|
|
works instead with operational level goals.
|
|
|
|
**What are the key findings?**
|
|
The biggest findings are that an autonomous control system for a nuclear
|
|
power plant is possible, but by far the biggest roadblock is a high assurance
|
|
decision making system.
|
|
|
|
## Third Pass
|
|
**Recreation Notes:**
|
|
|
|
**Hidden Findings:**
|
|
|
|
**Weak Points? Strong Points?**
|