As a vital part of transmission and distribution systems, transformers are built and expected to be unfailingly reliable. Nevertheless, internal faults like partial discharges can occur, and the problem with such faults is that if left un-corrected, they can eventually morph into catastrophic faults that can result in power outages and even end-user property damage.
Transformer Monitoring: What's Involved
* Data acquisition
* Sensor development
* Data analysis
* Development of links between measurements and failures
Easily Prevented
Preventing disasters of this nature is actually quite simple, and involves transformer monitoring. Monitoring transformers and spotting problems before they turn into unmanageable incidents can prevent faults that are costly to fix and may result in a loss of service. Transformer monitoring mainly involves data acquisition, sensor development, data analysis, and the development of causal links between measured values and failures of transformers.
Installing monitoring equipment on transformers is usually done for two reasons:
1. Monitoring important transformer functions can help detect developing faults before they lead to a catastrophic failure
2. Monitoring transformer functions can allow for a change from periodic to condition-based maintenance
Monitoring Equipment
Monitoring equipment is permanently mounted on the transformer and is online 24/7. Reliable, low-cost monitoring is thus a necessary condition. Failure rates of transformers are usually low (0.2 - 2% per transformer/year), and high-cost failure prevention systems cannot thus be justified, especially when redundancy is available and the consequential costs are thus limited. To keep within this cost barrier, some compromise on the functionality of the
monitoring equipment is necessary.
Transformer Monitoring: Parameters
* Oil temperature
* Moisture levels
* Operation of cooling fans
* Electrical load levels
In a majority of cases, it is enough to supply a reliable warning signal without online analysis and diagnosis, provided that manual or automatic diagnostic methods are available to follow up the alarm. Specifically with regard to power distribution networks in the US, a majority of the transformer population is aging, and most emerging faults can be expected from these units.Monitoring equipment should thus be designed for field installation on operational transformers that might date back a few decades.
Detection of Developing Faults
The main transformer parts that need monitoring are insulation quality, winding temperatures, oil quality, and mechanical moving parts such as on-load tap changers (OLTC). Monitoring the windings and insulation systems for gas-in-oil and partial discharge are crucial; temperature and load monitoring on the other hand, is regarded as base information and should be included in any type of transformer monitoring.
OLTC failures are typically caused by mechanical faults with bearings, springs, shafts and drive mechanisms, closely followed by electrical faults such as burnt transition resistors, choked contacts, and insulation problems.
Some parameters of transformer monitoring, and the sensors best applicable are discussed below:
* Dissolved Gas Analysis (DGA)
An established diagnostic method, gas-in-oil analysis involves analyzing the types, concentration and production rates of generated gases. Different types of gases are produced based on the types of faults; for example, overheated cellulose leads to the formation of carbon oxides, but arcing leads to the generation of acetylene.
Depending on the criticality of each unit, oil samples are taken manually at regular intervals (between 12 and 24 months) and the gasses are extracted from these samples. On-line gas sensors are typically the first choice in designing full time monitoring systems for DGA; simply because the technique of analysis is well established and accepted, and the sensor is truly capable of detecting a wide range of failure types.
Partial Discharges (PD)
Partial discharge testing of de-energized transformers is a valuable tool for evaluation of overall transformer integrity, however, on-line, real
time, partial discharge monitoring is, in general, more expensive and complicated. For example, glass fiber rods acting as wave-guides inside the main tank have been applied to large transformers, but the cost and complexity of installation has made such a system unsuitable for online monitoring.
There are several advantages associated with electrical PD monitoring, but it has been difficult to design field applications thanks to the difficulties in separating internal and external PD sources. Sensors that are being tested and developed currently include externally fitted acoustical sensors, which are more cost-effective but are susceptible to disturbances from rough outdoor substation environments.
* Temperature
The load capability of a transformer is limited by the hot spot of the windings. The hot spot is typically calculated indirectly from measurements of oil temperatures and load current. An alternative method involves fiber-optic temperature sensors that are installed in the winding during the manufacturing process. These sensors come in two varieties - fibers which measure the temperature at a single point, and distributed fibers that measure the temperature along their length. All of these systems involve high costs; in particular, the distributed fiber sensor is the most expensive to install and can only be applied to new transformers.
New Techniques
The condition of the insulation can be judged from other parameters as well, such as moisture levels and particulate content. Data interpretation with these parameters is not straightforward, but new techniques are being developed using software to analyze the large body of historical data available and identify patterns of progression towards failure. If similar deterioration is detected for a transformer in service, remedial action can be taken.
Other types of online sensors have also been investigated. Examples of such systems are online measurements of the moisture content of the oil, static charge in oil, optical sensors and pump monitoring.
Also on-line measurements of the moisture in the cellulose by optical fiber techniques are being studied. In general, these systems do not have a strong coupling to important and frequent failure modes.
