292
W. W. WILLMARTH AND S. S. LU
FIG. 1. Sketch of artangemcnt of hot wimS for ~ ~ . J U ~ L I U
of 4. u , . , and u2. (with
permission of J . Fluid Mech.).
the 18 different criteria were addcd topthcr in 18 separate storage regions
and an average value computed for each point in the sampling window for
each of the 18 criteria. Our initial results showed sliehtly larger contributions to Reynolds strew in the middk of the sampIiag window for the criteria
with u, < 0 with nogative slope.
With this encouraging rcsult we began a search for improved criteria. An
important improvement was obtained by pawing the sigqal representing u,
through a thid-mdcr Butterworth low parrs filter with half power point at
80 Hz. The filter removed hi& frtquoncy components and apparently
caused a reduction in the number of ropeotod or spurious samples. At any
rate the contribution to Reynolds stress in t b ~ m p h
oonmponding to
uv
IL
L
FIG. 2. Sketch of conditional sampling method {with permission of J . Fluid Mech.).
W. W. WILLMARTH AND S. S. LU
FIG. 1. Sketch of artangemcnt of hot wimS for ~ ~ . J U ~ L I U
of 4. u , . , and u2. (with
permission of J . Fluid Mech.).
the 18 different criteria were addcd topthcr in 18 separate storage regions
and an average value computed for each point in the sampling window for
each of the 18 criteria. Our initial results showed sliehtly larger contributions to Reynolds strew in the middk of the sampIiag window for the criteria
with u, < 0 with nogative slope.
With this encouraging rcsult we began a search for improved criteria. An
important improvement was obtained by pawing the sigqal representing u,
through a thid-mdcr Butterworth low parrs filter with half power point at
80 Hz. The filter removed hi& frtquoncy components and apparently
caused a reduction in the number of ropeotod or spurious samples. At any
rate the contribution to Reynolds stress in t b ~ m p h
oonmponding to
uv
IL
L
FIG. 2. Sketch of conditional sampling method {with permission of J . Fluid Mech.).