Rate of penetration has been considered as an important factor in the entire drilling industry, which can largely determine the overall costs of drilling a well. This paper proposed a novel real-time prediction of rate of penetration by combining the Attention-based Bidirectional-Long Short-Term Memory and Long Short-Term Memory (Att-Bi-LSTM-LSTM). Eight parameters, which are total vertical depth, weight on bit, revolutions per minute, mud flow rate, density, viscosity, drill-bit outer-diameter, lithology, and rate of penetration, are adopted as datasets. The drilling speed of the well is trained and validated through the drilling data while a sliding window is introduced for the real-time update. In addition, the presented prediction model is compared with other traditional prediction methods. Finally, the prospect of field application and further study is discussed and suggested. The results indicate that the proposed model shows good accuracy and robustness. Moreover, compared with the traditional methods, the model exhibits good superiority with smaller absolute and relative errors. For field applications, the model proposed in this paper attempts to provide a solution to the prediction of real-time rate of penetration. The results are expected to provide guidance for the further study on the increase of drilling speed and reduction of well costs.

Bed level dynamics at the interface of the salt marsh and tidal flat have been highlighted as a key factor connecting the long-term biogeomorphological development of the marsh to large-scale physical forcing. Hence, we aim to obtain insight into the factors confining the location of the marsh edge (i.e., boundary between tidal flat and salt marsh). A unique data set was collected, containing measurements of daily bed level changes (i.e., integrative result of physical forcing and sediment properties) at six intertidal transects in the North Sea area. Moreover, various biophysical parameters were measured, such as sediment characteristics, waves, inundation time, and chlorophyll-a levels. The data show that both bed level change and waves decreased from the lower intertidal flat toward the marsh edge and further diminished inside the marsh. However, no direct general relation was found between waves and bed level change. Bed level change inside the marsh was always small, regardless of wave energy. By combining the data sets, we demonstrate that the location of the lower marsh edge is restricted by two interacting factors: inundation time and bed level change. For vegetation establishment to withstand longer inundation stress, which slows down plant growth, more stable bed levels are required so that plants are not heavily disturbed. Conversely, to withstand more dynamic bed levels that disturbs plant growth, lower inundation stress is needed, so that plants grow fast enough to recover from the stress. The results suggest that bed level change is important in determining the position of the marsh edge.