The objective of this study is to propose an optimal plant design for blue hydrogen production aboard a liquefiednatural gas(LNG)carrier.This investigation focuses on integrating two distinct processes—steam methaner...The objective of this study is to propose an optimal plant design for blue hydrogen production aboard a liquefiednatural gas(LNG)carrier.This investigation focuses on integrating two distinct processes—steam methanereforming(SMR)and ship-based carbon capture(SBCC).The first refers to the common practice used to obtainhydrogen from methane(often derived from natural gas),where steam reacts with methane to produce hydrogenand carbon dioxide(CO_(2)).The second refers to capturing the CO_(2) generated during the SMR process on boardships.By capturing and storing the carbon emissions,the process significantly reduces its environmental impact,making the hydrogen production“blue,”as opposed to“grey”(which involves CO_(2) emissions without capture).For the SMR process,the analysis reveals that increasing the reformer temperature enhances both the processperformance and CO_(2) emissions.Conversely,a higher steam-to-carbon(s/c)ratio reduces hydrogen yield,therebydecreasing thermal efficiency.The study also shows that preheating the air and boil-off gas(BOG)before theyenter the combustion chamber boosts overall efficiency and curtails CO_(2) emissions.In the SBCC process,puremonoethanolamine(MEA)is employed to capture the CO_(2) generated by the exhaust gases from the SMR process.The results indicate that with a 90%CO_(2) capture rate,the associated heat consumption amounts to 4.6 MJ perkilogram of CO_(2) captured.This combined approach offers a viable pathway to produce blue hydrogen on LNGcarriers while significantly reducing the carbon footprint.展开更多
脱空和不密实是隧道衬砌最常见的两种病害。在这两种病害长期作用下会导致隧道出现破裂、渗漏水、钢筋锈蚀,最终造成隧道塌方等问题,严重威胁行车安全。采用探地雷达对隧道进行无损探测是发现这些病害或缺陷的常见方式,但大量雷达数据...脱空和不密实是隧道衬砌最常见的两种病害。在这两种病害长期作用下会导致隧道出现破裂、渗漏水、钢筋锈蚀,最终造成隧道塌方等问题,严重威胁行车安全。采用探地雷达对隧道进行无损探测是发现这些病害或缺陷的常见方式,但大量雷达数据的人工识别存在着工作量大、效率低、强烈依赖人员的专业素养等问题。本文提出一种基于深度学习的隧道衬砌缺陷的自动检测方法——自监督多尺度池化区域卷积神经网络方法(Self-monitoring Multi-scale ROI Align Region Convolutional Neural Network,SMR-RCNN),以提高缺陷识别的效率,并减少主观因素的影响。在雷达探测隧道衬砌的实践中,数据量巨大,但缺陷样本却很少,这对训练神经网络是一个相当大的挑战。为此,设计了一种数据增强的方法来增加缺陷的样本数量,且使用一种自监督对比学习的网络模型来提取雷达数据的特征,然后将其迁移到改进后的Faster-RCNN网络模型中;最后,使用有标签的样本对改进的Faster-RCNN网络进行细调训练。实验结果表明,相较于传统的Faster-RCNN方法,本文提出的算法增强了神经网络对脱空和不密实两类缺陷的自动识别能力,在检测精度上得到了显著提高,mAP值提升了12%。展开更多
文摘The objective of this study is to propose an optimal plant design for blue hydrogen production aboard a liquefiednatural gas(LNG)carrier.This investigation focuses on integrating two distinct processes—steam methanereforming(SMR)and ship-based carbon capture(SBCC).The first refers to the common practice used to obtainhydrogen from methane(often derived from natural gas),where steam reacts with methane to produce hydrogenand carbon dioxide(CO_(2)).The second refers to capturing the CO_(2) generated during the SMR process on boardships.By capturing and storing the carbon emissions,the process significantly reduces its environmental impact,making the hydrogen production“blue,”as opposed to“grey”(which involves CO_(2) emissions without capture).For the SMR process,the analysis reveals that increasing the reformer temperature enhances both the processperformance and CO_(2) emissions.Conversely,a higher steam-to-carbon(s/c)ratio reduces hydrogen yield,therebydecreasing thermal efficiency.The study also shows that preheating the air and boil-off gas(BOG)before theyenter the combustion chamber boosts overall efficiency and curtails CO_(2) emissions.In the SBCC process,puremonoethanolamine(MEA)is employed to capture the CO_(2) generated by the exhaust gases from the SMR process.The results indicate that with a 90%CO_(2) capture rate,the associated heat consumption amounts to 4.6 MJ perkilogram of CO_(2) captured.This combined approach offers a viable pathway to produce blue hydrogen on LNGcarriers while significantly reducing the carbon footprint.
文摘脱空和不密实是隧道衬砌最常见的两种病害。在这两种病害长期作用下会导致隧道出现破裂、渗漏水、钢筋锈蚀,最终造成隧道塌方等问题,严重威胁行车安全。采用探地雷达对隧道进行无损探测是发现这些病害或缺陷的常见方式,但大量雷达数据的人工识别存在着工作量大、效率低、强烈依赖人员的专业素养等问题。本文提出一种基于深度学习的隧道衬砌缺陷的自动检测方法——自监督多尺度池化区域卷积神经网络方法(Self-monitoring Multi-scale ROI Align Region Convolutional Neural Network,SMR-RCNN),以提高缺陷识别的效率,并减少主观因素的影响。在雷达探测隧道衬砌的实践中,数据量巨大,但缺陷样本却很少,这对训练神经网络是一个相当大的挑战。为此,设计了一种数据增强的方法来增加缺陷的样本数量,且使用一种自监督对比学习的网络模型来提取雷达数据的特征,然后将其迁移到改进后的Faster-RCNN网络模型中;最后,使用有标签的样本对改进的Faster-RCNN网络进行细调训练。实验结果表明,相较于传统的Faster-RCNN方法,本文提出的算法增强了神经网络对脱空和不密实两类缺陷的自动识别能力,在检测精度上得到了显著提高,mAP值提升了12%。
