In Australia, the implementation of anchors is generally governed by specifications specifically developed for dam anchors based on the experience of specialists, owners and regulators. In other regions, guidance on design, construction and testing of anchors is usually derived from one, or a combination, of the above- mentioned sources. Overall Stability – Free Length Design The pr inc ipl e for s a t i s fying the requirements of overall stability in anchored dams is that the free tendon length should be sufficiently long so that during loading the weight of rock mobilized in the volume of a notional inverted cone is adequate to resist the imposed forces. Conventional practice dictates that the apex of the inverted cone extends from the proximal end or half way along the fixed anchor length. The angle of the inverted cone varies between 60 degrees for soft or heavily fractured rocks to 90 degrees for competent rock masses. Geometry of mobilized inverted rock cone for overall stability analysis (BS8081:1989) Internal Stability – Fixed Anchor Design Despite the acknowledgement within the industry that the distribution of bond stress within the fixed anchor of a grouted tendon is highly nonlinear, conventional design practice for anchors in dams is still done based on average bond stresses and the assumption of uniform distribution of bond throughout the fixed anchor. Therefore, PTI DC35.1-14 and BS8081: 1989/2015 recommend a limit of 10 m (32.8 ft) on the bond length, arguing that lengths greater than this are less efficient. In relation to dam anchors, this phenomenon has been demonstrated in full-scale trial anchor studies. In Australia, full-scale gun barrel testing on an instrumented 20 m (65.6 ft) long, 91 strand tendon with 10 m (32.8 ft) bond length indicated that an applied static test load of 18.5 MN (4,158 kip) dissipated exponen- tially to zero at 4.5 m (14.8 ft) from the proximal end of the fixed anchor (Cavill, 2000). In Germany, Feddersen (1997) reported that instrumented 55-strand trial anchors were used on the EDER Dam (famous for having been substantially damaged by the bouncing bombs during World War II) indicated that load transfer only occurred along 2.5 to 3.0 m (8.2 to 9.8 ft) of a 10 m (32.8 ft) long bond length when subjected to a load of 12.5 MN (2,810 kip). Despite this, anchors are generally designed on the basis that load is pro- portional to fixed length, where the ultimate load that can be resisted by the fixed anchor is given by P ult = p • d • L • t ,ult where P ult = ultimate anchor load; L= fixed anchor length; d = drillhole diameter; and t = ultimate bond stress at the ground/ ult grout interface. In practice, tult is derived from three principal methods: 1. Calculation using geotechnical para- meters (e.g., internal angle of friction of the rock in the founding stratum) 2. Estimation from desk studies where bond stresses have been back-analyzed from anchoring works in similar ground (e.g., Table 24 in BS8081:1989) 3. Full-scale in-situ tests undertaken in the field It is emphasized that t is not only a ult function of the ground conditions but also the construction techniques employed by the specialist anchor contractor. Therefore, investigation tests should be performed, where possible, by the same contractor that will be engaged for the production anchor works, using the same drilling technique, equipment and same drillhole diameter. During the raising of Hazelmere Dam in South Africa (Mothersille, 2017), the world’s largest sacrificial trial anchor program was undertaken using four 61- strand anchors installed in 355 mm (14 in) diameter drill holes to depths of about 45 m (148 ft) in jointed quarzitic sand- stone. The trial anchors with bond lengths of 3 m (9.8 ft) were used to quantify in-situ bond stresses at geologically representative locations throughout the length of the 473 m (1,552 ft) long structure. The information gained from this extensive program was used to optimize the fixed anchor designs for production anchors, which varied from 6 to 9 m (19.6 to 29.6 ft) for tendons with 49 to 91 strands (working loads varying from 6,835 to 12,694 kN [1,536 to 2,854 kip]). 82 • DEEP FOUNDATIONS • SEPT/OCT 2018