Korto Cavitation Services

➔   Home   Introduction   Tests   Monitoring   Models   Information   Archive   Contact
➔   Plant_managers   Turbine_manufacturers   Monitoring_manufacturers

Cavitation Monitoring System

• This is patent pending

This monitoring system is a one-sensor implementation of the Multidimensional method presented in the Introduction. It is patent pending under the title High-sensitivity unbiased one-sensor monitoring of cavitation in hydro turbines - PCT/IB2022/062093. The basic qualities of this solution:
 (a) the system reacts to all cavitation segments equally, and
 (b) it has an extremely high sensitivity,
make it superior to other available one-sensor or other simple cavitation monitoring systems.
The system is applicable on hydro turbines of all types, inclusive Pelton.

• Description

The system consists of a vibro-acoustical sensor and a special signal and data processor, both installed on the turbine shaft, and an additional unit in a still space. This unit inductively supplies power to the rotating processor, organises the synchronisation with the runner rotation and receives the data from the processor. It then processes the data further and sends the result to the central plant monitoring system or displays it directly. The basic result consists of:
• J(Ο) - mean global cavitation intensity, and
• J(Ο,Φ) - instantaneous global cavitation intensity,
which are described in the Introduction; here Ο stands for parameter(s) specifying the operation point, and Φ for runner´s instantaneous angular position. The result further includes some details related on the rotor/stator interaction (see the Introduction); these are obtained by the analysis of J(Ο,Φ).

The system is especially sensitive to the cavitation at the runner and close to it, and, within this, especially to cavitation causing erosion on the runner. The vibro-acoustical signals from such cavitation segments travel from their locations on the runner blade via the blade body and the turbine shaft to the sensor. This transmission is equally efficient for all the locations on the blades. This ensures the quality (a) mentioned above.

The common method for assessing cavitation intensity counts on the mean value of a selected signature of the signals received by the sensor(s). The processing resulting in the J(Ο,Φ) deviates from this and, preserving consequences of the rotor/stator interaction, results in the quality (b). Here, traces of different segments of this interaction are assessed independently. This makes it possible to detect, with full sensitivity, the rise of cavitation caused by a single cavitation segment, which otherwise, in the circumferential mean value, would be suppressed by a factor typically reaching the value of the product of the number of guide vanes and the number of runner blades.

The monitoring procedure starts with the introductory calibration performed on a dense set of operation points of the turbine in a good state. At each point, the automatic or manual segmentation of the Φ-scale into the rectangles as illustrated above is performed on the raw estimates of J(Ο,Φ) (the grey curve). The segmentation consists of the following steps:
• the peaks caused by the rotor/stator interaction are found by looking for the local maxima,
• their amplitudes are taken for the amplitudes of the rectangles, and
• the first local minima below and above the peak´s location in Φ are taken for the limits of a rectangle.
The resulting data on the rectangles in the calibration process are memorised as the references. They document the fingerprint of the rotor/stator interaction as seen by the sensor on the shaft.
When processing the running monitoring data, the reference lying closest to the running operation point is found, the running J(Ο,Φ) is cut into the intervals as in that reference, and the peak values of the running J(Ο,Φ) within these intervals are taken for the amplitudes of the running rectangles.

While the described segmentation which distinguishes between different contributions to a given J(Ο,Φ) can simply be made, additional analysis is necessary to identify the element or elements of the rotor/stator-interaction which is responsible for each contribution, i.e. for each rectangle. A typical result achieved in a case that such an identification was successfully made is illustrated in the following graph. It follows the findings noted under Consequences of the rotor-stator interaction in the Introduction.

The data displayed in this and all the illustrations of the monitoring results shown below were obtained on the same turbine as in the illustration of the rotor/stator interaction in the Introduction and in the case described in the Test. There were 4 runner blades in the turbine, and out of 24 guide vanes, considerably strong cavitation was found only after 7 of them.

The background of the quantitative data on the identified cavitation segments shown is a suitable choice of the signal signature used in the processing; it makes contributions from different cavitation segments additive.

The segmented form of the J(Ο,Φ) is displayed in the large graph, G. The green or green/red rectangles present the values of cavitation intensity J found within respective intervals of Φ, where the scale of Φ, instead of by values within the 0-360° range, is specified by the notation showing the angular position of the respective element of the rotor/stator interaction which causes the intensity presented by a rectangle.

In this example, as shown in D, the operating points, Ο, are presented by the three qualities, coded by the two colours as in G; the reference is also indicated by its serial number, A.

The remaining displays B, C, and E, present, graphically and numerically, the results of the analysis of J(Ο,Φ) based on the recognition of the responsible turbine parts. These results are based on the additivity mentioned above, and yield, also for both the running operation and the reference, estimates of the intensity of cavitation on the runner blades, C, estimates of the intensity caused by the pronounced 7 guide vanes, E, as well as an estimate of the global cavitation intensity in the turbine, B. All the intensity values in this group and in G, are expressed in an ad-hoc unit for cavitation intensity, ci.

The described algorithm enhances traces of cavitation at the runner and close to it and presents them with high sensitivity as separate contributions to J(Ο,Φ). The other cavitation mechanisms, having no repetitive variations or having repetitions not synchronised with rotation (e.g. free cavitation behind the runner), are included in the results as constant components of J(Ο).

• Simpler algorithms

There are two alternative processing versions which are easier to organise:
One ... No trial is made to identify the rotor/stator interaction sources; the J(Ο,Φ) graph is displayed above the simple Φ-axis.
Two ... No cutting of J(Ο,Φ) into intervals is made, and thus original versions of J(Ο,Φ) are shown.

The version One is illustrated in the graph above. The data related to the runner blades, C, and guide vanes, E, are not available; only the mean global intensity, J(Ο), is known.

The version Two - the graph below. Instead of comparing two sets of rectangles, red and green ones, which shows clearly the difference between the running cavitation and that in the reference, the direct comparison of the running and reference J(Ο,Φ) makes the difference less visible. In the shown case, the match of the Φ-positions of the running and reference traces is perfect; this reduces the difference to the values close to the peak, which is good. In most cases, however, the two Φ-positions are not as good as that. This may make the related running and reference traces only partially overlap or even lie next to each other. With the display with the rectangles such mis-match, which is not essential but attracts attention, would be hidden. The full processing as presented above in the Description is thus highly recommended. It requires an additional work in the calibration test.

• Resolution

High resolution with respect to turbine elements has its direct diagnostic value, and it also guarantees high sensitivity in cavitation detection. The second graph in the Description above and the three graphs below demonstrate these effects. They show the resulting displays for four different situations with respect to cavitation spatial spreading - with running cavitation intensity passing over the reference by a rather small amount of 10 %:
- on only one runner blade behind only one guide vane;
- on all runner blades behind only one guide vane;
- on one runner blade behind all guide vanes;
- on all runner blades behind all guide vanes.

Comparison of the mean-intensity displays and the big graph with the instantaneous cavitation intensity demonstrates everything clearly. A small rise of cavitation in one isolated point of the machinery (e.g. on one runner blade behind one guide vane) is clearly visible in the big graph. There, the rise is shown in its full scope. In the global mean intensity display the rise is hardly noticeable; in reality it might be hidden in the random fluctuations of the result. It is a bit better in the other mean-intensity displays. In them, however, such a small and well-localised cavitation is visible via 1/4 or 1/7 of its intensity value. With rise of cavitation which involves more turbine elements the visibility in the mean-intensity displays is improved but it is still the best presented in the detailed display on the right.

All the displays shown in this illustration are a part of what one has on the cavitation-monitor screen, directly or in the cavitation channel of the central plant monitoring system. Further displays include time log over a selected past time interval, running display of the cavitation-intensity dependence on operation parameters, an option for focusing on the past details, and the accumulated erosion display.

• On the use of the monitoring results

Early detection of damage even if well-localised is guaranteed by the high spatial selectivity and the resulting high sensitivity.

Ageing effects and other systematic changes in turbine cavitation characteristics are visible in all the displays. The differences between different turbine elements are also visible. If calibration was fully made, the elements can also be identified.

Optimisation of turbine operation can be organised via plant SCADA in a way to avoid or suppress cavitation in a turbine, or to minimize the total cavitation intensity in turbines of a plant provided all the turbines are equipped for cavitation monitoring.

Optimisation of turbine maintenance can be organised via the log of the accumulated erosion. As noted in the Optimisation of turbine maintenance in the Tests, the erosion rate can be estimated by J 2.46, where J is the mean global cavitation intensity, J in J(Ο). The same holds true for the intensities related to the runner blades or guide vanes. A vector consisted of all these erosion-rate estimates should be read in equidistant moments while the turbine is operated and added to the running sum. The resulting vector of the accumulated erosion estimates should be calibrated as described in the Tests. Once calibrated for J(Ο), the same proportionality constant should be used for the estimates for the runner blades and guide vanes. Such final estimates of the accumulated erosion reached till the actual moment can then be used to decide on the overhauls. Here, the role of each turbine element can also be reliably assessed.

• Services offered to the monitoring-system manufacturers

• Delivery of a prototype of the cavitation monitoring system
• Alternatively, delivery of the software & detailed specification of the hardware
• For both: Licence for the use of the patented solution

➔   Top   Home   Introduction   Tests   Monitoring   Models   Information   Archive   Contact
Plant_managers   Turbine_manufacturers   Monitoring_manufacturers