Panda is a concept that Electricité de France (edf) has been using and developing for 10 years. It began towards the end of the 1980s as a reaction to complaints about the burden of the complex monitoring systems of its hydroelectrical structures. In answer, EDF developed software on Macintosh for acquiring and processing monitoring data, intended to meet the needs of users in the field.

This software, known as TAM-TAM, achieved the following in one tool: simplified acquisition; immediate local check; automatic real time transfer of data to external centres; and archiving of these data in secure databases. As well as reducing operating costs, it was quickly found that the programme made the system more reliable. For example, it solved the problem of errors caused and time wasted as a result of the multiple entry of data and recovery of files in different formats by the various people involved in monitoring at EDF.

The installation of the software went hand in hand with a revalidation of dam monitoring professionals, whose responsibilities increased as the quality of data acquired improved.

From 1996, on the strength of the experience gained with TAM-TAM, and in order to make the software compatible with the new remote measurement systems that had been installed from 1990 onwards, EDF set up the Panda software (Progiciel d’Acquisition et d’aNalyse des Données d’Auscultation, or monitoring data acquisition and analysis software) using a PC Windows environment. This programme, which uses the functions of TAM-TAM with more modern ergonomics, is now being used in more than 150 implementations in France, representing approximately 350 structures (250 hydraulic and 100 nuclear).

A new market in Argentina

HINISA and HIDISA are two Argentinean companies producing electricity from hydroelectric stations. These two companies operate six power plants and seven dams located at the foot of the Andes cordillera in the Mendoza province on the Atuel and Diamante rivers.

Both HINISA and HIDISA were privatised in 1994, and since that time Argentinean and French investors, including EDF, have acquired more than 50% of the companies’ capital.

EDF was chosen as operator of the hydro power plants. As operator, it installed the Panda software in February 1998 to satisfy a real need to make the monitoring system more reliable.

Before installing Panda, a series of different data collection and analysis systems had followed one another: from an entirely manual system, extremely time consuming and involving a high risk of error (multiple transcriptions) to computerised systems. The computerised systems certainly ensured time savings, but they did not fully satisfy HINISA-HIDISA’s wish to make the system secure and reliable.

In fact, the software used until 1997 lacked checks as far upstream as possible in the measurement chain. This sometimes allowed major errors to slip in, which was costly in the time required for their correction and detrimental to the confidence of local authorities. In addition, discrepancies in the databases of the different sites appeared frequently.

The problem was quite similar to that encountered on the other structures monitored by EDF some years previously. It was logical, therefore, for EDF to propose to install an organisation based on its experience in structure monitoring. Roque Paolantonio, responsible for monitoring at HINISA-HIDISA, summarises the advantages of the system as follows:

The actual ‘skeleton’ of Panda is based on two independent chains of measurement operation. The first is the responsibility chain of the HINISA-HIDISA operator, whose main role, being as close as possible to the structures, consists of:

•Taking regular measurements on site and making installation inspection rounds.

•Carrying out the first level of immediate diagnosis: examining the results of the measurements and producing graphs of the rough physical phenomena.

•Archiving the results.

•Sending results to other sites that have a central role or that have responsibility for safety.

The second chain of operations is one of expert analysis. This is performed by EDF-DTG, which carries out:

•Collection and long term archiving of the measurements provided by the power plant operator.

•Completion of a second level diagnosis, using ‘advanced and specific methods’. This includes systematic drafting of anomaly sheets for all measurements outside tolerance ranges.

•Writing regular summary reports, including a detailed interpretation of the measurement results and an opinion on the structure’s behaviour.

Installation of the Panda software enabled this organisation to be put in place quickly. The Panda software was flexible enough to enable the databases of the 12 structures being monitored to be simply compiled. This represented management of more than 1000 sensors and of 20 to 40 years of data logs.

In a useful extra benefit, re-evaluation of the whole monitoring system provided an opportunity to negotiate with the local authorities to abandon sensors that were obsolete, and to install new ones.

Paolantonio described the system architecture chosen and explained its advantages: ‘The shared headquarters of the companies HINISA and HIDISA are located approximately 300km from the dams which we monitor.

‘Each valley, Diamante and Nihuil, has a Panda station with the databases of the respective structures. The monitoring staff record their measurements using the Portable Data Registration unit, and they are responsible for the quality of their measurements.

‘The software helps them at all stages of self-checking. Once the data have been checked and validated, they are sent to Mendoza and to France by modem via Internet.

‘In Mendoza, I perform the second level inspection on my Panda station, which contains the databases of all the structures. Sorting tools enable me to view all or part of the data (for example, I can visualise only measurements incurring comment). The support of EDF, which checks the data in real time from France, is a very reassuring element for us. The data are checked a third time and analysed using statistical tools. For any measurement outside the tolerance range, EDF sends us a fax to indicate the anomaly and process it. On occasions, this third check has enabled us to detect problems that we had not detected ourselves, but in general, the multiple checks imposed by the software enable us to avoid all errors and omissions. The configurable report printing part (in Word format) is also very useful, as we have to submit graphs every month to the inspection organisation (ORSEP Cuyo) for all the devices for which measurements are taken.

‘Another very important aspect for us is the security of the system: access to the software is managed by passwords associated with the levels of authorisation. These authorisations enable us to limit access to the database configuration. Chance manip-ulations, whose consequences may be serious for the durability of the database, are thus avoided. We have now been using the software for over a year and have saved a considerable amount of time and eliminated all the error which occurred quite frequently before installation of this system.’

Paolantonio summarises the system’s positive points as follows:

•There is monitoring quality control, thanks to the validation of measurements, by different levels of validation, on the basis of pre-established procedures.

•It is a modular tool. It can be configured by the user and is applicable to civil engineering structures of all sizes.

•It allows for dynamic management of alert thresholds during calculation of physical phenomena, in the event there are measurements outside configurable limit values.

•Monitoring is secure, but it can be interfaced with all types of sensor and remote measurement.

Panda 1999

A new version of Panda is due for release in April 1999. The main functions of this new version are:

•Measurements can be recorded on site, by a person using the keyboard manually, or the PDR, or remotely by the central operator via an automatic data acquisition system.

•There is assistance for ‘immediate’ diagnosis. Readings are automatically transformed into physical measurements. A set of graphic functions helps diagnose the measurement: the functions are chronological graph, super-imposition of several years, correlation and evolution profile. It is possible to define alert thresholds in the event of measurements outside limit values. Points outside the limits are then monitored specifically. To simplify the power plant operator’s

first level check, EDF has developed a rough measurement validation based on the display of a dispersion strip. This dispersion is centred on a value obtained from an appropriate statistical model.

•A structure knowledge database can be saved. Each database is compiled by system device constants, which are chronological series of measurements with their associated comments. It is enriched with every new measurement. Computer protection systems ensure safe storage of the data.

•Immediate transmission is available to the inspection and expert analysis centres. Measurements are transmitted by modem (telephone network, Internet) or floppy disk to the various inspection centres. They can also be transmitted to the expert analysis centres by telematics.

•Printing of regular reports can be programmed. The flexibility of the application offers the possibility of editing graphs of measurement logs, associated with diagrams, tables of numeric values and comments. Common models are provided, and each user can create models suited to his or her own needs (documents in Word format, for example). At each stage (in the field, during inspection), the user can add comments to the measurement, which will be archived with the measurement.
Related Articles
Spotlight on… South America