Calibration of a large capacity gas flow meter may be accomplished wit
h a sonic nozzle bank as a reference system. For the design of a compa
ct movable prover which is made with many nozzles, the nozzles should
be installed as close as possible. The interference effect between nei
ghbouring nozzles can be an additional error source for prediction of
the mass flow of the assembly from the calibration of individual nozzl
es. A nozzle bank was assembled from seven toroidal throat nozzles wit
h a nominal throat diameter (d) of 4.3 mm in a hexagonal pattern, it w
as calibrated by a primary gas flow system at stagnation pressures fro
m 0.5 to 2 MPa and throat Reynolds number from 0.4 to 1.2 x 10(6). Thr
ee installation plates of sonic nozzles were fabricated with distances
between nozzles of 16-81 mm (L/d of 3.7 to 9.5). The average discharg
e coefficient of the seven nozzles in a nozzle bank was calculated fro
m the discharge coefficients of the individually measured sonic nozzle
s. The differences from the directly measured results were within +/-0
.05% for all test cases. Also, the proximity of the wall for the six o
uter nozzles located at 2.3 d from the pipe wall had no influence on t
he experimental results. The discharge coefficients as a function of t
hroat Reynolds number were in agreement with the recommended equation
of ISO 9300. (C) 1998 Elsevier Science Ltd. All rights reserved.