Treatment of half cycles in rainflow counting

I believe the treatment of half-cycles in the rainflow counting by concatenation, by and large, results in them being "doubled" or treated as full cycles. While this approach is sometimes recommended for long, "well-behaved" time series (for example, in Guidelines for Design of Wind Turbines − DNV/Risø), IEC 61400-1 recommends treating half-cycles as half-cycles (contributing half the damage).

I believe this is because some fatigue-generating time series (e.g., a braking event) are not so well-behaved, i.e., they result in large half-cycles that never close. At a minimum, an option to treat them as half-cycles could be useful.

Cheers!

The issue names no files or tests. Start by locating the rainflow-counting implementation and review how unclosed half-cycles are represented, then compare the behavior with IEC 61400-1's half-damage rule. Done should include an agreed treatment or option for half-cycles and tests covering a braking-like series.

Written by the indexing model from the issue text.