No abstract

The U.S. Navy has developed a dynamic penetrometer called the expendable Doppler Penetrometer XDP) to determine in situ soil stren@. The XDP was originally designed to provide an undrained shear strength profile in SOR cohesive sediments by measuring the instantaneous velocity of a sound source probe as it falls through the water column and penetrates the seafloor. The Navy recently has added and verified a capability to determine strength properties of cohesionless soils. Introduction Seafloor characterization is an ever-present need and challenge in the ocean engineering world. In situ soil strength is a vital aspect of seafloor engineering in the offshore environment. However, determining this soil strength is very time consuming and expensive with present conventional methods. The expendable Doppler Penetrometer (XDP) was developed to aid in satisfying the Navy's continuing need for expeditious means for seafloor characterization. This dynamic penetrometer was originally designed to provide undrained shear strength profiles by measuring its deceleration as it penetrates the seaftoor after falling through the water column. The Navy recently has added the capability to determine the strength of cohesionless soils. Controlled laboratory tests were conducted simulating XDP deployments in a highly dense, completely saturated cohesionless soil. The results of these tests were used to formulate relationships to determine the relative density and undrained angle of internet friction of cohesionless soils from the XDP. The results of the laboratory tests and the newly developed relationships are presented. These new relationships were used to estimate the soil strength of two field sites, one a calcareous sand site offshore Key West, FL and the other a dense quartz/mineral sand offshore Biloxi, MS. The results from the XDP tests and an established in situ test, the piezocone, at these field sites are presented and conclusions are drawn regarding the capability of the XDP to measure soil strength in cohesionless soils. Expendable Doppler Penetrometer Features and Use The expendable Doppler Penetrometer (XDP) system (fig. 1) is designed to determine the soil strength of various types of ocean seafloors. The system initially measures the instantaneous velocity of a probe with a sound source as it free-falls through the water column and penetrates the seafloor. Utilizing the Doppler Shift principle whereby the apparent frequency is a direct function of the probe's velocity away from a receiving hydrophone, the velocity of the probe is determined up through the point that it comes to rest in the seafloor. The velocity-time curve that is produced enables estimates to be made of the undrained shear strength ofcohesive soils and the penetrability of seafloor soils in general. The XDP probe (fig. 2) is a hydrodynamically shaped body made of aluminum and steel. The probe has a body of diameter of 3.5 in. and weighs 59.8 lb. in air and 47.6 lb. in sea water.

Modern data processing systems require optimization at massive scale, and using machine learning to optimize these systems (ML-for-systems) has shown promising results. Unfortunately, ML-for-systems is subject to over generalizations that do not capture the large variety of workload patterns, and tend to augment the performance of certain subsets in the workload while regressing performance for others. In this paper, we introduce a performance safeguard system, called PerfGuard , that designs pre-production experiments for deploying ML-for-systems. Instead of searching the entire space of query plans (a well-known, intractable problem), we focus on query plan deltas (a significantly smaller space). PerfGuard formalizes these differences, and correlates plan deltas to important feedback signals, like execution cost. We describe the deep learning architecture and the end-to-end pipeline in PerfGuard that could be used with general relational databases. We show that this architecture improves on baseline models, and that our pipeline identifies key query plan components as major contributors to plan disparity. Offline experimentation shows PerfGuard as a promising approach, with many opportunities for future improvement.