The reversible inlet-outlet device of a vertical pipe is a major type of the bidirectional intake-outlet structures adopted by pumped storage power stations. Backflows usually occur at the bottom of its orifice in the outflow mode due to the structural characteristics

but the backflow velocity and its spatial extent are relatively small

requiring a very sensitive velocimeter and a careful arrangement of the gauge positions. Such backflows could not be detected in the model tests if the measurements were inappropriate. This paper compares the differences in the measurements of velocity distributions at the orifice section

produced by adopting different velocimeters or different gauge arrangements. The results show that the backflow occurs in our test cases

and they are not detectable unless the velocimeter produces interference low enough and keeps effective when the gauge is very close to the orifice bottom. The backflow was missed in the previous studies that adopted high-interference velocimeters

failed to measure the positions close enough to the bottom

or arranged only a small number of gauge points over the cross-section. We find that significant differences in the measured velocity at the same position by adopting different types of velocimeters that are different in the structures and measuring principles

especially in the low flow velocity cases

reaching the range of 75.0% - 150%. For a cross section with uneven velocity distribution

multiple gauge lines should be arranged; each line should have enough points

such as more than 7 - 9. The velocity pattern and hydraulic parameters cannot be obtained accurately unless the complete vertical profiles of flow velocity are observed. This study would be useful for further experimental study of the flow patterns in the reversible inlet-outlet device.