Pile/shaft designs using artificial neural networks (i.e., genetic programming) with spatial variability considerations.

Details

• Creators:
  McVay, Michael C. ; Klammler, Harald ; Tran, Khiem ; Faraone, Michael ; Dodge, Nathan ; Vera, Nilses ; Yuan, Le
• Corporate Creators:
  University of Florida. Dept. of Civil and Coastal Engineering
• Subject/TRT Terms:
  • Bridge foundations
  • Piles (Supports)
  • Skin resistance
  • Data collection
  • Load tests
  • Prestressed concrete
  • Genetic algorithms
  • Limestone
  • Strain gages
• Publication/ Report Number:
  BDK75-977-68
• Resource Type:
  Tech Report
• Geographical Coverage:
  Florida
• Corporate Publisher:
  Florida. Dept. of Transportation
• Abstract:
  The work focused on the improvement of FB-DEEP’s prediction of skin and tip resistance of concrete ; piles and drilled shafts in Florida. For the work, data from 19 concrete pile sites and 18 drilled shaft sites were ; collected. This included 458 standard penetration test, SPT, borings on the pile sites and 815 borings on the ; drilled shaft sites. A total of 64 static pile load tests and 66 drilled shaft tests were acquired. For the piles, 48 ; tests reached Davisson Capacity, of which 28 had separation of skin and tip resistance. All of the drilled shafts ; were instrumented with strain gauges from which unit skin transfer (T–Z) was assessed for Florida limestone. ; All of the data were uploaded into the FDOT online database based on position (i.e., station + offset, or GPS). ; In the case of piles, the data (e.g., boring vs. measured skin friction) were analyzed with a genetic ; program (GP) algorithm to construct equations for unit skin friction and tip resistance based on soil type (USCS) ; and SPT N values. The resulting GP skin friction curves were found to be similar to FB-DEEP; the tip resistance ; curves had higher unit tip resistance vs. blow count values, as well as being only averaged 4 diameters/widths ; beneath the piles. In addition, the practice of setting SPT N to zero for N< 5 was found to be conservative, and ; the use of N=5 for N< 5 is recommended. For both current FB-DEEP and GP curves, load resistance factor ; design, LRFD , were obtained for borings within 100 ft. In the case of borings outside this distance or for sitespecific ; conditions, method error (CVm) for FB-DEEP and the GP curves is presented from which LRFD  may ; be found. ; In the case of drilled shaft, the GP algorithm a developed normalized unit skin friction vs. displacement ; curve for limestone, which were similar to Kim (2001). In the case of ultimate skin friction in limestone, the ; GP algorithm was used to validate the FDOT relationship between unit skin friction and rock strength ; (unconfined compression, split tension).
  More +
• Format:
  PDF
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/27185/dot_27185_DS1.pdf
• File Type:
  [PDF - 1.11 MB]
• Collection(s):
  US Transportation Collection
• Main Document Checksum:
  urn:sha256:76cd7a3cfdf914efe56356210a41bbaf2b21b06efe91d5bfef851eab3ccd06e2