Giulio
Ciaravino, Paola Gualtieri and
Guelfo Pulci Doria
Dipartimento di Ingegneria Idraulica ed
Ambientale Girolamo Ippolito
Università degli Studi di Napoli Federico II -
Via Claudio 21 - 80125 Napoli
Tel. 0039-81-7683460-Fax 0039-81-5938936-E-mail: pulci@unina.it
Abstract: Cavitation arising in hydraulic machines is an
important problem which now is becoming more important because of the large
diffusion of pumped hydroelectric storage plants, with the presence of pumps and
turbines. With reference to a pump the presence of cavitation can be
experimentally detected through the lowering of the performance of the pump. But
the presence of cavitation can be detected without hydraulic measurements too.
Some experiments have been carried out in Naples in order to test the acoustic
method of detecting cavitation. The main results obtained are described in this
paper.
Keywords: centrifugal pump, cavitation, acoustic method
Cavitation
arising in centrifugal pumps is often an important problem to which it is always
necessary to pay attention. Normally the presence of cavitation in a pump at a
definite value of its flow-rate can be experimentally detected through the
lowering of the head (H) of the pump itself with respect to the head value fixed
at that flow-rate by the characteristic (Q,H) curve of that pump (at least 1% or
3% lowering according to the chosen cavitation arising standard). The presence
of cavitation can be detected without hydraulic measurements in the plant too.
Some experimental methods indeed can serve this purpose. Among these ones the
acoustic method is well-known: this method detects the cavitation presence by
the increasing of the sound power uttered by the flow and appears more sensitive
than other ones especially in order to detect cavitation arising; this method
also, through suitable sound analysis, can reveal deeper particulars of the
engine working. Consequently, since it has been observed through a number of
experiments (refer for instance by [1] [2] [3] [4]) that the analysis of noise
allows to detect cavitation in an operating pump without interfering either with
the flow or with the working unit and yet providing higher sensitivity as
compared with traditional methods, several experiments on cavitating pumps have
been carried out in Naples in order to test the acoustic method [5] [6]. These
experiments confirm of the effectiveness of the acoustic method because it was
really possible to show the cavitation arising even before that the performance
of the pump lowered and furnish other deeper particulars about the engine
working. The main results obtained through these experiences are described in
the following paragraphs.
As well known,
generally speaking, the working point of a pump can be represented in a (Q, NPSH)
plane (Q is the flow-rate and NPSH is the available Net Power Suction Head). In
the same plane a cavitation curve can be defined, which divides the points of
this plane characterised by cavitation presence from the other ones
characterised by no cavitation presence. Through the acoustic method it was
possible to experimentally state the features of the cavitation curves of two
different similar pumps. The generic shape of these pumps is represented in
Fig.1.
Fig. 1 Generic shape of the pumps
In Fig.2,
indeed, the cavitation curves of these pumps are represented. The pumps differ
the one from the other in that the optimum flow-rate is 2,67 dm3/s
for the first one and 6,00 dm3/s for the second one; in that the
angle of attack b of the
impeller is of 40° for the first one and of 18° for the second one; and
obviously in that the input section of the impeller is wider (11,72 cm2)
in the first one with respect to the second one (5,95 cm2). In spite
of these differences the two non dimensional cavitating curves are very much
alike.
Fig. 2 Cavitation curves of the pumps
Always through the acoustic method it has been possible to show that in a cavitating pump, cavitation can arise in different points of the engine, and yet in the volute. This fact was detected through a suitable spectral analysis of the cavitation noise, with a microphone settled close to different points (A, B, C, in Fig.1) of the cavitating pump. In Fig.3, 4, 5, the results of this spectral analysis are shown. "A" position (Fig.3) is close to the impeller inlet: here a main frequency of 6.30 kHz appears; “B” position (Fig.4) is close to the volute: here a main frequency of 16.00 kHz appears; “C” position (Fig.5) is close to a little well near the output section: here a main frequency of 10.00 kHz appears. The different main frequencies show undoubtedly that in points “A”, “B”, “C”, different cavitation nuclei were arising: they could be indeed in detail detected through the acoustic method.
Fig. 3 Results of the spectral analysis("A" )
Fig.4 Results of this spectral
analysis are ("B" )
In Fig.6 are shown, always in the (Q,NPSH)
plane, in a suitable scale, twelve curves of the sound power uttered by a pump
inserted in a closed circuit, at the increasing of the circulating flow-rate,
each one relative to a fixed value of the upstream depression; these curves
represent the sound power compared with the same one in the same hydrodynamic
conditions but without cavitation: for each curve, the separation from its
reference line shows evidently a sound power increasing and then the cavitation
arising. The separation locus is the cavitation curve. But it is interesting to
observe that it is possible to distinguish three fields of flow-rates: a first
field in which the cavitation is always absent, a second field in which the
cavitation is always present, and a band in which the cavitation noise is
characterised by an intermittent presence. This means that the cavitation noise
is present in this band in a percentage of time presence. This means that the
cavitation noise is present in this band in a percentage of time increasing from
0% at the beginning of the band to 100% at the end of the band. So it is shown
that the cavitation curve is really a cavitation band, owing to the fact that it
is necessary to distinguish the arising of the cavitation (with few and sporadic
cavitating nuclei) from the full presence of the cavitation (with a continuous
presence of cavitating nuclei). In Fig.6 this behaviour is clearly shown with
reference to a pump yet different from the two ones previously referred to.
Fig. 5 Results of this spectral analysis are shown. ("C")
It has been
shown that it is possible to get interesting and important information about the
cavitating pump also through the only listening to the cavitation noise. In this
connection it is consequently possible to draw in the (Q,NPSH) plane other
curves (or bands) which are relative to points of that plane characterised by
different types of cavitation noise. In Fig.7 such curves are drawn with
reference to the first pump related to in this paper. It is possible to
distinguish the following curves:
