Acoustic detection of boiling in LMFBRs: An estimate of sensitivity derived from experiments during the commissioning of PFR

This paper describes a programme of experiments undertaken during the commissioning of the 600-MW PFR at Dounreay. The objective of the programme was to obtain data on reactor background acoustic noise levels and to estimate the sensitivity of the installed acoustic boiling detection system.

The boiling-noise detection system installed in PFR consists of seven waveguides. The waveguides are solid metal rods which extend from the reactor top to just above the core top level. Five are arranged on a 0.75m radius circle with two at the edge of the core. The sensing elements are standard accelerometers.

For the commissioning experiment three rigs were installed in the core locations below three of the waveguides. Each rig was over 12m long and carried three sodium-proof microphones at levels corresponding to core top, mid-core and core bottom. Each rig also carried an electrically-powered vapour generator which could inject a stream of sodium vapour into the surrounding 250°C sodium, so producing a boiling type noise. In addition to these nine microphones in the core, microphones were also installed at the inlet and outlet of each of the three pumps. To handle the data for this experiment a system of amplifiers and three fourteen-track tape recorders was used. To check amplifier gains a calibration signal was automatically injected at the end of each record. On line one-third octave and narrow-band analysis was also available, but the bulk of the analysis was done from the tape records.

The commissioning programme was in two parts: a measurement of background noise and an investigation of the detection sensitivity at the waveguides of the noise from the vapour generator (this implicitly included the transmission loss). In the background measurement programme the noise level at each microphone was recorded at pump speeds from 200 to 960 rev/min, full speed, in approximately 10% steps. To determine the sensitivity, the signal at each transducer was recorded with each vapour generator operating in turn. The power input to the vapour generators was 1100W, of which some 800W required to balance the heat losses. The power available for boiling consequently was only 300W; nevertheless the signal was detectable at pump speeds up to 750 rev/min on waveguides radially 0.75m from the source.

Since the noise output of the vapour generator was measured by a microphone close to the source, this result could be used, with data obtained on background noise and measurements of sodium boiling noise obtained on force convection rig experiments, to deduce that a highly sub-cooled boiling source involving 40kW of thermal energy would be detectable in PFR with a margin of 10dB. The conclusion rests on the assumption that there is no major adverse change in the transmission losses at the time detection is required, such as might be caused by a large quantity of free gas in the sodium.