NEWS 2012

Posted September 9, 2012 (Updated October 10, 2012)

Event: October 24, 2012

GSE/CGS Cross Canada Lecture: Mike Jefferies, Golder Associates (UK) - Looking towards Beaufort Sea Development - Experience of Design, Regulation and Reality with the Molikpaq 1984-89

Location: Woodvale Community Facility, Millwoods Golf Course, 4540 - 50 Street

Time: 5:30pm Registration, 6:15pm Dinner, 6:45pm Presentation

Cost: $30 Members, $40 Non-members, $10 Students

Download in iCalendar format

The Molikpaq was a ‘bottom-founded barge’ drilling platform used for oil exploration at four locations in the Beaufort Sea 1984-91, before being deployed offshore at Sakhalin Island (Russia) as a production platform in 1997. These areas are ice covered for some (much) of the year, the ice is thick, the ice moves, and the moving ice causes intense cyclic loads as it crushes against the structure. In the case of the Molikaq, dynamic loading from ice crushing was roughly an order of magnitude more severe than any of the recent very damaging M7+ earthquakes. Ice-induced vibration brought the Molikpaq within minutes of platform-loss on 12 Apr 86.

The Molikpaq is an impressive steel structure, some 100 m square in plan and founded at nominally 20 m depth. This external appearance hides the main engineering feature – the ‘structure’ is simply a steel box to retain hydraulically placed sand, with effectively all the platform’s resistance to ice load coming from that sand. Geotechnical interest comes from the behavior of the sand under the ice loads applied to it. The ice loads itself is also interesting, with the platform stiffness influencing the ice loads realized.

When viewed in comparison to earthquakes, there is very little experience with the cyclic aspects of ice load. And when the Molikpaq was designed in the early 1980’s there was really rather scant knowledge of ice load magnitude – much of the knowledge was at small scale with uncertain extrapolation to real offshore platforms. Nevertheless, the design criteria did include cyclic ice loading scenarios. These scenarios were quasi-static, with no vibration of the platform expected.

Cyclic loading and hydraulically placed sand usually lead to specifying ground modification, and the initial design was configured around vibroflotation. But, vibroflotation would require significant additional time in the already tight construction window of open-water season. Based on the test data on the proposed construction sand, which showed dilatant behavior at very loose densities, the Molikpaq’s designers proposed centrifuge testing to confirm the adequacy of undensified hydraulic fill under the cyclic ice load scenarios – benign behavior was found. The acceptability of undensified fill was given further weight by the then-prevalent doctrine of ‘static bias’. Finally, ‘drainage time’, which relates to the rather slow period of load cycling, was viewed as a key performance parameter with considerable effort being placed on working with free draining (as opposed to silty) sand. These considerations lead to the Regulator giving a permit for deployment of the Molikpaq with an undensified, hydraulically-placed, sand. However, the uncertainties were recognized and the Molikpaq was equipped with some 600 sensors and a high-speed data acquisition system to measure loads on the platform, the platform response, and sand behavior. There was also a full-time monitoring team and near real-time assessment of the data – good use of the ‘Observational Method’.

Experience during the first winter season 1984-5 was generally in accordance with design predictions, although extended periods of ice-induced vibration at far greater frequency than the design scenarios were encountered. The benign sand behavior in this extended vibration led to some confidence, although ice thickness crushing against the platform was less than 0.6 m. The second winter 1985-6 involved rather thicker ice, with several interactions with 2 m – 3m thick ‘multi-year’ ice. One particular interaction, on 12 April 86 involved the continuous crushing of ice averaging about 3.5 m thick; this interaction extended over some twenty minutes, with ~1000 cycles of induced platform vibration > ~15% g. The sand fill showed rising pore pressure from cyclic mobility in the part closest to the ice, with the drainage limiting the increase in that pore pressure, until about ten minutes into the interaction; at that point the drainage lost control and full-cyclic mobility developed in a small part of the core adjacent to the ice loaded face of the structure. This zone of cyclic mobility continued to develop and expand in the following ten minutes as ice loading continued and became more extreme. The situation stabilized when the ice flow buckled and became grounded, which then damped out the cyclic loading.

Questions that arise from this incident include: How could this type of ice loading have been missed from the design scenarios? With hindsight, there was prior experience in Alaska that should have been given greater weight. Can the centrifuge be trusted for proof-of-concept? Actual sand fill conditions in the as-deployed confirmatory testing showed markedly better sand fill properties than in these tests, and the far better drainage time more than compensated for the change in ice load frequency. Is the ‘Observational Method’ appropriate for assessing liquefaction potential? The Molikpaq experienced numerous events at ~4% g for far greater durations than 12 Apr 86 but without any indication of potential cyclic mobility. And, is drainage an option to control liquefaction? The 12 Apr 86 sand behavior was partially drained and there is an indication that just slightly better drainage would have completely controlled the response to ice load.

Both ice and geotechnical factors will be discussed, as well as some suggested answers to the four questions posed above.

Mike Jefferies is a civil engineer with 35 years of experience, mostly in consulting but ten years of that with “owner” companies. It was this ten years with owners, and in the Canadian Arctic with Gulf Canada Resources in particular, that provided an enormous opportunity to “push the envelope” and which led to the most significant of his contributions to engineering (or, more accurately, engineering science).

A keynote speaker at international conferences on Arctic offshore engineering, hydraulic fill construction, and liquefaction, Mike has published some seventy-five papers ranging across ice loading of offshore platforms through to rock fracture grouting. But he is generally most known for the state parameter approach to soil characterization – an approach that has become one of the most cited innovations of the past twenty-five years of geotechnical engineering.

The state parameter work led to an invitation to write a book on soil liquefaction, now sold-out with a second edition pending. As will be evident from a quick glance at the book, Mike is an exponent of the heresy that geotechnical engineering must be based on applied mechanics, not geology, and that the critical state is fundamental, readily measurable, unique, and something every geotechnical engineer should appreciate.