Deep reinforcement learning for methane slip reduction in hybrid-powered liquefied natural gas marine vessels
Bibliographic record
Abstract
• Methane slip offsets CO 2 reductions achieved by using LNG as a marine fuel. • Managing methane slip is key for achieving climate-neutral marine shipping. • DRL-based energy management systems are proposed for methane slip reduction. • Performance of the proposed systems is evaluated using real-world data. Hybrid-powered liquefied natural gas (LNG) vessels offer a viable pathway for reducing greenhouse gas (GHG) emissions in maritime transport by lowering CO 2 emissions. However, these benefits can be offset by CH 4 emissions, which are typically highest at low engine loads due to methane slip. Efficient engine operation is therefore essential for achieving overall GHG reductions. Using a hybrid LNG-battery powertrain provides additional degrees of freedom to modify the operation and mitigate GHG emissions; however, the optimal power allocation between the LNG engine and the battery system must be developed specifically to the operation of the vessel. This paper investigates the use of three deep reinforcement learning (DRL) algorithms, namely twin delayed deep deterministic policy gradient, soft actor-critic, and proximal policy optimization, to develop intelligent energy management systems (EMSs) for hybrid-powered LNG vessels. The proposed DRL-based EMSs aim to minimize cumulative GHG emissions from sailing trips by optimizing power allocation between the powertrain components. The proposed strategies are evaluated against the peak shaving and load levelling (PS-LL) EMS currently used on the vessel under study. The DRL-based EMSs reduced GHG emissions by up to 14.6% more effectively than the PS-LL EMS, when compared to a baseline of operations simulated without hybridization and measured relative to an offline optimization benchmark. This improvement is primarily due to better engine load management, which reduces methane slip.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".