Everyone’s familiar with the mechanical shutter noise made by a camera. We’re so conditioned to this sound because it signifies “picture taken”. We continue to demand it even though modern digital phone cameras are solid-state, with no moving parts, and play a recording of that sound just for our benefit.
The “equivalent top-load” curve derived from a bi-directional static pile load test is a bit like that. It’s an afterthought to a test method that doesn’t actually produce a single top-down load curve.
Bi-directional testing was first conceived of by various researchers and practitioners perhaps as early as the 1960’s (see J.M. Amir’s paper, “Interpretation of Load Tests on Piles in Rock”) More widespread documented use first appeared in the 1980’s in Brazil, Japan and the United States, as described in Dr. Bengt Fellenius’ paper “Analysis of results of an instrumented bidirectional-cell test”.
In the early days, a bi-directional static load test (BDSLT) was typically designed with the hydraulic jack placed at the shaft base, directly separating the end bearing from the shaft friction. This is rare today. Now that many thousands of load tests have been performed around the world, it has become apparent that in most situations friction capacity is much greater than bearing. The optimal location for the jack is at the balance point, where the capacity of the end bearing plus friction below the jack is equal to the friction capacity above. Locating the jack properly at this balance point is essential since only in doing so can the maximum overall shaft load be determined without the need for extrapolating either the upper or lower capacity. Determining this point is an art, a combination of detailed geotechnical knowledge, previous experience and often a bit of luck. The end bearing and various soil layers’ individual contributions to total resistance is then determined using strain gages.
The results of the test are best presented as separate unit capacity end‑bearing (q‑z) and side friction (t‑z) curves. However, engineers were used to seeing the traditional test top-load (TL) curve, and demanded the same from the BDSLT. So, in the late 1990’s, Dr. John Schmertmann developed a simple analysis method for deriving the Equivalent Top Load (ETL) curve from BDSLT data. Essentially, it’s a first‑order simplification of the finite-difference soil-spring t-z analysis as used in many foundation design and analysis programs. He never published his results outside of the explanatory appendices of data reports, and others independently derived & published essentially the same method. A good overview is presented in the paper “Assessment of methods for construction of an equivalent top loading curve from O-cell test data” by Seo, Moghaddam and Lawson.
The ETL analysis has many sources of uncertainty, some of which are obvious and some which may be more subtle. Significant assumptions have to be made to construct the curve. Does the soil have the same resistance loaded upward as opposed to downward? Can the BDSLT upward load distribution curve be inverted to compute the additional elastic compression of a top-down load? Is subtracting the buoyant weight (twice!) too conservative? And these are just the obvious ones. On the flip side, a TL curve from a traditional load test will by necessity include the effects of the reaction piles or reaction mass supports, while an ETL curve is constructed from a truly isolated foundation element. Having said all that, the few direct comparison studies done seem to show good agreement, within the acceptable range of precision for adjacent elements which nevertheless are not identical units – see for example “Comparison of the Bidirectional Load Test with the Top-Down Load Test” by Kwon et. al.
At LTC we are very familiar with all aspects of BDSTL analysis and ETL construction, including accounting for conditions like deep cut-off, anticipated scour, tensile vs compressive loads, etc. The optimal use of a BDSLT test is as a source of q-z and t-z curves, which can be used to improve / validate computer models such as FB-MultiPier. Using the ETL curve as a primary criterion for judging if a shaft “passed” the test is not recommended due to all the uncertainties enumerated above. The curve has utility as a visual aid and to benchmark against other tests, because even though it’s a derived result and not a direct measurement, it helps make the test results “click”.