Home Back


Project title: DIELECTRIC SPECTROSCOPY - An innovative technology for the diagnosis of electric power equipment insulation
Contract: 22100/01.10.2008
Contracting Authority: National Centre for Programme Management
Contractor: ICMET Craiova
Project Manager: Eng. Dorin Popa
Program 4 – Partnerships in Prioritary Domains
Thematic field: Energy
Project type: PC
Proposal acronym: SD
Project development period: 01.10.2008– 29.07.2011
Total value of project: 1.400.000 lei
From budget: 1.240.000 lei
Co-financing: 160.000 lei
Project Partners:
Project leader - National Research Development and Testing Institute for Electrical Engineering– ICMET Craiova
Project manager: Eng. Dorin Popa
Partner 1 – University of Craiova
Project responsible: Conf. Dr. Ing. Florian Stefanescu
Partner 2 – University of Pitesti
Project responsible: Conf. Dr. Ing. Dumitru Cazacu

Project abstract:
The electrical insulation of the electric power equipment is naturally subject to ageing and degradation following the operation in thermal and electric stress conditions.
Ageing leads to the decrease of the lifetime affecting in this way the safe operation of the electric power transmission system. That is why the researches and practical activities were directed to a better understanding of the ageing processes and to finding technologies for insulation diagnosis and remaining lifetime estimation. At world level, several maintenance types for the electric power equipment were applied with a view to ensuring the safety of electric power system with as low as possible costs. A first step was to select those insulation condition investigation methods that bring the most useful information. From the many known methods there were selected: oil dissolved gas analysis for the prevention stage and partial discharge (PD) measurement to render evident the serious faults. The project approaches a new method namely analysis of insulation response in the frequency range to render evident the incipient faults as well as to assess the remaining lifetime. In comparison with the methods analyzing the insulation response in the time range this one has the advantage of compensating temperature influence on the response. By computer aided modeling and laboratory experiments on infested and/ or artificially aged insulation samples, a certain fault type will be associated to the tan delta curve in the frequency range. The diagnosis procedure will be validated by measurements on power and instrument transformers and on electric cables in laboratory and on-site conditions.

Project achievement diagram:
Execution stage No.1: Study of the measuring techniques for dielectric analysis
Execution period of stage No.1: 01.10.2008– 27.02.2009
Total value of stage No.1: 160 500 lei
From the budget: 100 500 lei
Co-financing: 60 000 lei
Budget distribution was the following:
CO - ICMET Craiova : 32 500 lei from the budget 60 000 lei co-financing
P1 - Universitatea din Craiova: 49 500 lei from the budget
P2 - Universitatea din Pitesti: 18 500 lei from the budget

Execution stage No.2: Drawing up of insulation system experimental models with different infestation and thermal and/ or electric ageing degrees
Execution period of stage No.2: 01.03.2009 - 30.10.2009
Total value of stage No. 2: 59 930 lei
From the budget: 59 930 lei
Co-financing: 0 lei
Budget distribution was the following:
CO - ICMET Craiova : 59 930 lei from the budget.

Execution stage No.3: Tests on experimental models of insulation systems
Execution period of stage No.3: 30.10.2009 - 30.11.2010
Total value of stage No. 3: 191 442 lei
From the budget: 151 442 lei
Co-financing: 40 000 lei
Budget distribution was the following:
CO - ICMET Craiova : 32 500 lei from the budget and 60 000 lei co-financing
P1 -
Universitatea din Craiova: 49 500 lei from the budget
P2 - Universitatea din Pitesti: 18 500 lei from the budget

Execution stage No.4: Demonstration of functionality and utility of insulation investigation technology model based on dielectric spectroscopy
Execution period of stage No.4: 30.11.2010 – 30.09.2011
Total value of stage No. 4: 371 052 lei
From the budget: 311 052 lei
Co-financing: 60 000 lei
Budget distribution was the following:
CO - ICMET Craiova : 259 193 lei from the budget
P1 - Universitatea din Craiova: 20273 lei from the budget
P2 - Universitatea din Pitesti: 31586 lei from the budget
P3 – Retrasib Sibiu : 30000 lei co-financing
P4 – Electroputere Craiova: 30000 lei co-financing

Execution stage No.1: Study of the measuring techniques for dielectric analysis
The drawn up study had the following structure:
• Introduction
• Known measuring techniques and their contribution to performing an objective diagnosis
The conclusion of this chapter was that the diagnosis methods have developed rapidly in the last years and credible results were obtained but they are influenced by the following factors:

- Insulation temperature changes strongly the calculated moisture content
- The ageing products dissolved in oil and stored in pressboard and paper increase the conductivity of the whole insulation and stimulates water role
- In the cases when the insulation has a non-usual geometry the accuracy of the moisture content calculation decreases
- Dielectric response is influenced in a major way by barriers and oil channel areas. Therefore, dielectric response indicates the moisture from the main insulation and is not so sensible to the moisture content from the insulation around conductors.

• The theoretical bases of the dielectric response and the computation of dielectric response function under the form of current, voltage and losses.
In this chapter, dielectric response function was explicited for different proposed model types and the theoretical basis for releasing current analysis (PDC - polarization and depolarization current), recovered voltage analysis (RVM - recovery voltage measurement) and power frequency factor analysis in the frequency domain (FDS - frequency domain spectroscopy) was presented.
• Published results of dielectric spectroscopy application in time and frequency domains
The conclusions of this chapter are:

- RVM analysis software strongly depends on oil conductivity. At the same time, insulation geometry and temperature have an influence on the results though the paper moisture content was constant during the measurements.
- PDC analysis is little influenced by insulation geometry and has a weak dependence on temperature. The moisture content increases when oil conductivity increases though in fact it remained constant.
- Frequency domain analysis (FDS) provides the best compensation of insulation geometry. According to FDS analysis results when the temperature rises the cellulose gets dry faster than in real case. This tendency reveals an imperfect temperature compensation.

• Considerations on power transformer insulating system simulation
Selection criteria for the software programs used to simulate the phenomena own to dielectric spectrography

Execution stage No.2: Drawing up of experimental models for insulation systems with different infestation and/or thermal ageing degrees
Execution period of stage No.2: 01.03.2009 - 30.10.2009
Total value of stage No. 2: 59 930 lei
From the budget: 59 930 lei
Co-financing: 0 lei
Budget distribution was the following one:
CO - ICMET Craiova : 59 930 lei from the budget.

Stage objectives:
- Ascertainment of the physical – chemical factors which influence the operational condition of the paper-oil insulation
- Design of experimental models for paper-oil insulation
- Achievement of experimental models
- Drawing up of the program for model preparation with a view to performing the measurements for dielectric losses determination depending on the frequency of the voltage stressing the dielectric.

Abstract of execution stage No.2
In “Introduction” chapter, the project launching reason was substantiated namely the introduction of a new evaluation technique for the water content from the paper-oil insulation of transformers. Traditionally, this evaluation is performed directly analyzing the oil because of oil sampling easiness. Unfortunately, the relation between the water in oil content and water in solid insulation content depends on temperature. Therefore, the moisture content from the solid insulation can be ascertained by the so called equilibrium curves. Since the equilibrium condition is rather rare at a transformer in operation it was raised the problem of finding new methods for the ascertainment of the moisture content from the solid insulation of the transformer based on dielectric response analysis. The importance of developing these measurement methods results from the fact that water accelerates cellulose ageing process with direct consequences on transformer safety in operation.
In chapter “Moisture distribution in solid insulation of transformer” the paper-oil insulation was analyzed taking as criterion the moistening capability.
The following structure types were identified:

- Thick structures representing about 50% of the total cellulosic insulation mass. Though they contain a significant moisture quantity they have a small contribution to moisture migration to the total insulating system due to the high constants of the diffusion process; for example a few years at normal operation conditions (70-80°C)
- Thin cold structures are considered the insulating components operating at oil temperature as pressboard barriers, end fittings etc. They represent 20-30% of the total cellulosic materials mass. These components retain a great amount of water during the daily temperature cycles.
- Thin warm structures are the ones operating at temperatures close to conductor temperature.

About 5% of the cellulosic material mass operates at high temperatures. Moisture migration is the fastest in this area. However, the moisture content of the components from this group is significantly smaller than the one from the “thin cold structures” and therefore the contribution to the total moisture migrating inside and outside the oil is small.
The conclusion of this chapter :The test specimens must be achieved from the thin cold structure category.
In chapter “Moisture distribution depending o the temperature areas” the transformer insulation is classified depending on the temperature areas. This classification is important because if the environmental and loading conditions of the transformer do not remain constant the moisture inside the transformer is distributed depending on temperature. Taking into account that inside it, the transformer is divided into temperature areas and appreciating that water in oil distribution is uniform, moisture distribution in the transformer can be determined based on the values from the equilibrium curves.
Conclusion of this chapter: Moisture content distribution is opposite in comparison to temperature because cellulose capability to absorb water decreases with temperature. In the thermal treatment process of the insulation samples it is important to establish the temperature domain.
In chapter „Moisture content equilibrium“ three aspects are treated:

- thermodynamic equilibrium
- equilibrium diagram utilization
- moisture migration in the main insulation of the transformer.

In the dynamic operation of a transformer, moisture distribution tends to equilibrium so that a stationary regime to be achieved. The equilibrium condition depends on insulation temperature, geometry and moisture content The transformer will be considered as being formed of several insulating structures with different thicknesses and temperatures which tend to equilibrium having oil as exchange medium between them.
Taking into account that water solubility in oil and cellulosic material capability to absorb water depend on temperature, a change of temperature will have as a result a change of paper and oil moisture. With increasing temperature, the water in oil solubility increases too while cellulose absorption capability decreases so that the equilibrium process forces the water molecules to migrate from cellulose in oil. With deceasing temperature, the cellulose materials absorb the water molecules from oil. The water is retained in oil by its aromatic components or by impurities. It can be also absorbed to hydrate the polar components resulted from ageing.
Moisture distribution in transformer coils is generated by temperature distribution on axial and horizontal directions. The result of this phenomenon is a low moisture content in the areas where the temperature is high. Moisture equilibration is reached faster between the outer layers of cellulosic insulation surface and surrounding oil than in the total volume of the materials with high thickness. At low temperatures, the moisture migration process is focused to the layers with small thickness.
Therefore, oil moisture does not give information about the solid insulation system moisture. On the other hand, it does not vary during transformer operation with a constant load. Under these circumstances, the conclusion could be that the measurement of the whole solid insulation system moisture can be considered correct even if a total thermodynamic equilibrium is not reached.
The procedure of equilibrium diagram use to obtain the cellulose moisture is affected by the following error types:

- Moisture infestation of the sampled oil
- Equilibrium diagrams are valid only in thermodynamic equilibrium conditions
- Temperature differences smaller than 30°C resulted from temperature distribution inside transformer are not taken into consideration
- The equilibrium curves are specific to cellulosic material with known physical- chemical properties
- Aged cellulose absorbs moisture more hardly than new cellulose.

The paper presents a mathematical model for moisture diffusion in cellulosic materials based on the equations of the 2-nd Fick's laws of diffusion using an independent temperature value for the diffusion coefficient. The proposed mathematical model is useful for operations as coil installation in transformer tank or on-site repair works to assess water contamination of a dry, oil impregnated or in contact with air insulation.
In chapter „Design and achievement of experimental models” the requirements of the standards referring to ageing by maintaining a high temperature (IEEE C57-100) as well as the specimen dimensions and nature were explained. The procedures for each specimen preparation were drawn up based on the experience acquired at experiments. Taking into account that the deviations from the specimen sampling and preparation technique lead to results that cannot be used, the working procedures were optimised until results reproducibility was obtained experimentally. Comparative measurements were performed on different instrument types so that to appreciate the error deviation for each quantity to be measured.
Conclusion of stage 2: The conditions for starting the experiments are met.