Mathematical optimisation of diver ascent profiles at a constant risk of decompression illness
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Divers use decompression schedules that provide a stepped ascent to the surface from their maximum depth to help prevent the occurrence of decompression illness. The risk of decompression illness resulting from these schedules varies across different dives and the models used to generate them. The diver is unaware of this variance in risk. This thesis describes an investigation into the feasibility of producing optimised isoprobabilistic decompression schedules that minimise the time it takes for the diver to reach the surface from maximum depth. In particular, 1.3 bar constant partial pressure of oxygen in helium dives are considered. The US Linear Exponential Multi-gas (LEM) model is used to describe the risk of decompression illness for a given dive. The Sequential Quadratic Programming (SQP) method is used to minimise the ascent time given non-linear risk constraints and a maximum dive time constraint. Two approaches to describing the ascent profile have been investigated. The first scheme finds the stop times at each possible stop depth to produce optimised schedules. The total time for decompression is a function of the sum of the stop times. The second scheme defines the ascent profile as a three parameter hyperbolic tangent equation. The SQP method finds the three parameters that produce optimised decompression schedules once the curve is converted to a schedule of decompression stops. The schedules produced by the SQP method, using a curve to describe the ascent profile, show that it is feasible to produce optimised iso-probabilistic tables that are operationally practical given an acceptable physiological risk model. Comparison with the QinetiQ 90 tables with a nominal 2% operational risk of decompression illness show that the method could provide reductions in the ascent time subject to manned testing.