Performance of dielectric liquids for land based and subsea environmental friendly transformers.

In large power transformers dielectric liquids are the main insulation used to separate the high voltage parts from tank and ground. When failing, small "lightnings" (called streamers) propagate through the liquid, establishing a conductive channel from high voltage to ground resulting in a destructive short circuit. Traditionally mineral oils have been alone in the market, but now - partly from environmental concerns - new liquids such as synthetic and vegetable esters, and synthetic gas-to-liquid oils arrive. The challenge is that both fundamental understanding of the breakdown mechanism of these liquids, and good standards for testing their functional properties lack. The project has as a goal both to investigate the fundamental breakdown mechanisms, and the functional properties of these dielectric liquids. If one understands the basic physical mechanisms and the material properties it is easier to establish sound dimension criteria and to develop new materials. Breakdown in a liquid occur after a lightning like channel has propagated through the liquid. The driving mechanism is in the tip of the propagating channel, where liquid is evaporated leaving a conductive plasma-filled channel in its wake. At the tip there will be a very high electric field that will ionize the molecules. Strong light emission that also may ionize is observed. In the high field the free electrons will multiply and develop into electron avalanches. The energetic electrons will be hitting liquid molecules, develop heat, evaporate the liquid and form a conductive plasma. The plasma channel has a high internal pressure. Depending on the gap distance and the applied voltage the streamers may either stop midway or cross the full insulation distance at the so-called breakdown voltage. A streamer channel propagates at around 2 mm/µs (equals km/s) up to breakdown voltage. If voltage is further raised then above a certain threshold called acceleration voltage, velocity will increase by a factor of hundred to 100-200 mm/µs (equals km/s). These fast streamers can result in breakdown for short duration impulses like lightning surges for insulation distances of 0,1-1 m that are relevant for a large power transformer. We have studied the behaviour of liquids in decimetre long gaps. Various mineral oils have different behaviours. It appears as if the behaviour is governed by presence of electronically active compounds as e.g. aromatics that have a low ionisation level. We have also seen how pressure influences propagation. This has relevance for transformers located at high altitudes and subsea. Increased pressure increased the breakdown voltage, while propagation speeds were largely unaffected. A small computer-operated small-scale setup was built for easy verification of functional properties of a liquid like streamer speed, breakdown and acceleration voltages. This offers an inexpensive way of a first screening of a candidate liquid, or the effect of an additive. Several liquids were tested and by statistical analysis it was found that the acceleration voltage was closely correlated to the first electronic excitation level. In the last phase of the project a test set-up for applying x-ray pulses synchronous with voltage application was built. The intention was to use this for ionizing the liquid and studying electron avalanches. However, these experiments have to be postponed to a later project. SINTEF has during more than a decade collaborated with NTNU on quantum chemical molecular modelling of dielectric liquids and additives to reveal properties as e.g. ionisation and excitation potentials. In this project they have been responsible for a PhD on modelling streamer propagation based on fundamental properties. Here molecular material properties are linked to macroscopic behaviour. The model is based on the calculated molecular properties of the liquid with or without additives and simulates how a streamer channel propagates in the medium, including how it may branch into several channels. While it is not an accurate model for streamer propagation at this point, it is well suited to study how key properties like the conductivity of the channel, the mobility of electrons in the liquid, photoionisation due to light emission from the streamer head etc. influence the streamer phenomena in dielectric liquids. SINTEF and NTNU has through this and earlier projects become established as an internationally leading group on liquid dielectrics. SINTEF is now leading an international working group where experts from industry and academia report on state of the art in the field of electric insulation for transformers.?

The results are communicated to a working group in CIGRE in Study Committees D1 and A2 that will produce a brochure (Report) on "dielectric performance of liquids". In this group manufacturers of transformers, producers of dielectric liquids, universities and Research institutes are present. The working group is lead my Lars Lundgaard. The Cigre study will be a basis for future revision of international IEC and IEEE standards

Driven by environmental concerns new vegetable based electrically insulating liquids enter a market dominated by mineral oils. The dielectric performance of liquids is the basis for design of liquid filled apparatus (e.g. transformers). Breakdown in a li quid occurs via a gas filled streamer propagating across the insulation: this propagation varies a lot depending on voltages and molecular properties of liquids (e.g. velocity may increase three orders of magnitude as voltage is increased). This gives liq uid dependent constraints for design. Today no validated and detailed physical model exists for the behavior of such liquids. Design is done based on benchmark models and return of experience on known materials Knowledge of liquid behavior under differen t conditions is essential for design, specification and testing of electric liquid filled apparatus. Good design criteria give possibilities for more compact and energy efficient designs. Especially an expected pressure dependence on breakdown voltages wi ll facilitate more compact and lighter designs for offshore and subsea transformers. The breakdown event in a liquid is governed by tip processes where a vaporization and formation of a plasma filled streamer channel is formed. The channels branch into a bush that limits the electric field at the channel tips. There is circumstantial evidence that the streamer formation is driven by electron avalanches in the high field region near the streamer tip. If this hypothesis can be verified it opens for establi shing scientifically based design criteria. The project will be split in three: -Small scale experiments focusing on quantum chemical streamer tip processes. -Large scale experiments focusing on channel propagation properties and branching. -Modeling bas ed on electron valance processes and energetic considerations