Failure Grading System

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328187416

Failure Mode Assessment of the Generator Stress Grading System

Conference Paper · June 2018


DOI: 10.1109/EIC.2018.8481040

CITATIONS READS
0 1,915

2 authors:

Christian Staubach Thomas Hildinger


Siemens Voith
22 PUBLICATIONS   98 CITATIONS    28 PUBLICATIONS   82 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Nano Insulation View project

Stress Grading Fundamentals View project

All content following this page was uploaded by Christian Staubach on 19 October 2018.

The user has requested enhancement of the downloaded file.


Failure Mode Assessment of the Generator Stress
Grading System
Christian Staubach Thomas Hildinger
University of Applied Science Voith Hydro
Hannover, Germany Sao Paulo, Brazil

Abstract— In this paper different kinds of electrical driven containing miscellaneous issues not caused by the insulation
failure modes in the end-winding region, are discussed and system itself, such as water ingress, contamination foreign
classified. Failure types related to the stress grading system are objects, mechanical impact and others. This will not further be
analyzed via numerical simulation models and root causes are analyzed in this work.
given for the different defects based on the calculation results.
Typical pictures of various findings are presented and explained.
Phase split discharge phenomenon is mostly caused by
In a last step possible diagnostic measurements to identify issues insufficient clearance and therefore a design-related problem.
related to the stress grading are discussed and some PD-pattern Although insulation cracking is mainly a result of mechanical or
are presented exemplarily. thermo-mechanical overstressing of the main insulation, this
issue is closely linked to an electrically driven failure type.
The aim of this work is to give the basis to correlate findings in Typically, a mechanical damaged insulation is also weakened
the end-winding and especially the highly stressed stress grading electrically.
area with their most likely cause of defect. In this paper we will concentrate on the other 3 failure types,
OCP (CAT) and ECP (SGT) deterioration as well as bar to
Keywords—Stress grading, FEM, Assessment, Hydro Generator,
Stator bar ground discharges, which are closely related to the overall
generator stress grading system.
I. INTRODUCTION
Electrically driven failures within the generator winding II. CLASSIFICATION INTO GLOBAL AND LOCAL DEFECT
insulation have still a major share in reported forced outages of CAUSES
power plants. Well known failure mechanism, such as slot
discharges and vibration sparking are occurring in the slot region When it comes to the causes for electrically driven failures
[1], whereas other phenomena are limited to the end-winding within the stress grading system, a classification into overall
region. global issues, such as inadequately tailored OCP or ECP-
For the end-winding region several insulation related issues resistance characteristic, and local defects, like voids or
reported are reported and pictures of the findings are published mechanical damages are reasonable. Therefore, the individual
[2-6]. However, detailed assessment about the root causes and failure mechanisms are analyzed based on this distinction.
recommendations for risk mitigation as well as corrective and
preventive measures are not given often. A. Global defect Causes
A common cause for failures linked to the stress grading
system is an inadequate ECP, OCP or combination of ECP and
OCP resistance characteristics. The influence of the resistances
on the resulting electrical (kV/mm) and thermal (°C) stress can
be determined by means of modern simulation tools based on
the Finite-Element-Method [7].

Fig. 1: Five groups of electrically driven failure types in the end-winding region

Basically, electrically driven failure types in the end-


winding region can be divided in 5 different groups. While the
five major groups highlighted in Fig. 1 are caused by wrong
design, material selection, etc. there is an additional group
Fig. 2: Model of the geometry used in the simulation model

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE


For the following evaluation and assessment, a model of a F2 Too high Too high OCP
hydro generator stator bar in the involute region together with F3 Too low Too low OCP/ECP
a core section is used, see Fig. 2. F4 Too high Too low End ECP
In the model defined boundary condition, such as applied
voltage, copper temperature, forced convection and material If the OCP has a too high OCP-resistivity ρOCP (F1 and F2), the
properties are set. Only the OCP and ECP-resistivity ρOCP and electrical shielding function is not given anymore.
ρECP are parameterized. To take the non-linearity of the ECP In that particular case the OCP behaves like an additional
into account, its resistivity is described by the equation insulating material. As a result, high electrical stress is
occurring in the OCP-region close to grounded elements, such
ρECP[E] = ρECP∙e-βE, as clamping fingers or laminated sheets. There will be no
influence of the ECP resistance characteristic anymore, because
where E is the absolute value of field strength and β a there is no electrical contact between OCP and ECP, see Fig. 4.
coefficient describing the non-linearity. For the fundamental
parameter study the temperature dependency of the stress
grading material ρECP[T] and ρOCP[T] are neglected. In general,
only a little influence of the temperature on the stress grading
is observed if the temperature is not exceeding a defined
threshold value.
To investigate the influence of OCP and ECP-resistivity on
the stress grading functionality, the maximum electrical stress
Emax occurring along the bar surface (path in Fig. 2) is
calculated for various ρOCP and ρECP combinations.

Fig. 3: Resulting maximum electrical stress in kV/mm depending on various ρOCP Fig. 4: Path plots of voltage and field strength along the bar surface in case of
and ρECP combinations the failure types F1 and F2 together with the reference R

In Fig. 3 the surface plot of Emax is plotted over various If OCP and ECP-resistivity ρOCP and ρECP are both very low,
combinations of ρOCP and ρECP. The data is obtained by means an electrical field enhancement is resulting at the ECP end (F3).
of several hundreds of calculations for ρOCP and ρECP For a constant ρOCP and increasing ρECP the maximum field Emax
combinations over a range of up to 10 decades of their is moving towards the OCP-ECP overlapping. For a too high
resistivity values. resistivity ρECP the resulting electrical stress in this area will
The 4 corners of this surface, F1 – F4, highlights the different become too high (F4).
failure types related to the stress grading system, whereas R The corresponding Path plots of voltage and field strength
represent a proper working system. These 4 failure types a along the bar surface for these two failure scenarios F3 and F4
characterized by the following OCP and ECP-resistance are shown in Fig. 5.
behavior summarized in Table 1. In addition, the location of the
maximum electrical stress Emax is given.

Table 1: Overview about the different failure types caused by global defects
OCP-resistance ECP-resistance Location Emax
R Proper Proper OCP/ECP
F1 Too low Too high OCP
While the failure type F1/ F2 will most likely be detected by HV-
testing during machine assembly, stator bars exhibit to failure
type F4 will often pass HV-testing.
If the ECP-resistance for a given ECP-length is too low
(F3), corona will occur at the ECP-end. An example is shown in
Fig. 7 for a parallel ring with ECP configuration.

Fig. 7: Corona as a result of field enhancement at the ECP end caused by too low
ECP resistance (F3)

III. LOCAL DEFECT CAUSES


Sometimes it may be difficult to distinguish between global
and local defect causes for a given visual finding. For instance,
the arcing from bar surface to the clamping finger in the top
picture from Fig. 6 can also be caused by a small local defect
of the OCP, such as tape or paint variation or tiny erosion spots
Fig. 5: Path plots of voltage and field strength along the bar surface in case of where the resistivity becomes too high or low.
the failure types F3 and F4 together with the reference R The following Table 2 summarizes the three failure types
F5, F6 and F7 caused by local defects and analyzed in this work.
The pictures in Fig. 6 show possible findings in machines
caused by the failure types F1/ F2 and F4 discussed above. On the Table 2: Characteristic of three failure types caused by local defects
left hand picture field concentration in the air gap between bar OCP and ECP resistance Local defect location
surface and clamping finger due to the missing OCP shielding F5 Proper OCP
function results in PD or arcing. The right hand picture shows F6 Proper OCP overlapped by ECP
PD withening indicating field concentration at the OCP/ECP- F7 Proper ECP
overlapping due to high-ohmic ECP resistance characteristic.
Typically, failure type F5 can be causes by a local spot in
the OCP-region either be low resistive caused by conductive
particle agglomeration or very high ohmic because of missing
particles or deterioration via PD erosion [9]. For the following
analysis only the OCP-layer is shown, compare to Fig. 2.
Additionally, the clamping fingers are not considered here.
Their influence will be discussed later.
Fig. 8 gives the results for electrical stress (kV/mm) and
temperature (°C) for a high and low ohmic spot within the OCP.
The location of the defect is highlighted with the dashed
rectangular. As boundary conditions the line-to-ground voltage
(11,54 kV) is assigned for the electrical analysis, whereas
ambient temperature is assumed for the thermal analysis. Due
to this thermal hot-spots in the OCP-region can easily be
Fig. 6: Possible erosion pattern caused by the failure types F1/ F2 (left) and F4
identified. Low-ohmic spot means the defect location is far
(right) in the stress grading area [8] more conductive than the residual OCP; the high-ohmic defect
is several decades higher in resistivity. In case of a low-ohmic
Failure types F1/ F2 and F4 can be a result of a poor tape or local defect no influence on the electrical stress as well as
vanish quality, wrong choose of resistivity or may be temperature is obtained. A significant field enhancement is
introduced during the manufacturing process of the stator bar. resulting, if the local defect becomes too resistive. The linked
field strength will most likely result in PD-activity during
operation.
Fig. 10: Possible erosion pattern caused by the failure types F5 (left) and F6
(right) with local high-ohmic defects [10]

Fig. 10 shows typical erosion pattern caused by the failure


types F5 and F6. Continuous internal PD-activity in small voids
direct underneath the OCP will deteriorate the material
resulting in a loose of its conductivity. Once the OCP-erosion
reaches the OCP surface, the high-ohmic small defect,
simulated and discussed previously, will sustain the PD-activity
and further destroy the OCP-material. A very early stage of the
erosion is presented in the left picture.
Fig. 8: Results for field and temperature distribution in the OCP-area with a local Normally, the OCP-erosion is not that critical and
defect (F5) generators are known to run for several years without any
corrective measures. However, if the high-ohmic erosion spot
In Fig. 9 the results for field and temperature distribution for is located in the clamping finger region and clearance is quite
a local defect in the OCP/ECP-overlapping region are given. low, the electrical field enhancement can result in arcing
The results are comparable with the previous once in Fig. 8. between the bar surface and the finger [9]. This arcing is critical
High electrical field strength is only obtained if the local defect because it is quite intensive and the energy flow is
becomes very resistive, i.e. will perform like an insulation spot. perpendicular towards the insulation surface. This mechanism
An influence on the temperature distribution is not observed. will result in faster deterioration of the main insulation and
Although, not presented here, the defect size itself will have an could lead to a ground fault finally.
impact on the electrical stress. Fig. 11 summarizes these situation an also presented some
example of correlating PRPD-pattern.

Fig. 11: Influence of clamping finger on erosion mechanism caused by a local


high resistive spot in the OCP-area (F5)

In Fig.12 results for a local defect within the ECP-region are


presented. The location of the defect is highlighted with the
dashed square. Shown only is the 2. ECP-layer, see Fig. 2. In
general, these results will be strongly influenced by the axial
Fig. 9: Results for field and temperature distribution in the OCP-area with a local position of the defect spot, the magnitude of the spot resistance
defect in the OCP/ECP-overlapping region (F6)
deviation and its size.
In both cases there is a field enhancement observable.
However, more severe is the high-ohmic defect. An influence
on the temperature distribution is hardly visible.
higher voltages due to its non-linear behavior. However, a lot
of factors, such as design, used material, etc. may distort the
results and accuracy of the measurement. Additionally, local
defects, will not influence the tanD-measurement and not
suitable for defect detection.

Fig. 15: Comparison of PRPD-pattern for ECP-deterioration (F4) from different


literatures [9]

On- or offline PD-measurement can illustrate global or local


Fig.12: Results for field and temperature distribution in the ECP-region with a
failure types related to the stress grading system if PD is
local defect (F7)
present. Different failure types will result in specific PRPD-
Especially, the high-ohmic local defect can result in pattern, see Fig 11. [6,9]. Accuracy of the measurement will
continuous PD-activity in operation. The left picture in Fig. 13 depend on design, failure type and location within the winding,
shows the typical whitening as sign of ECP-deterioration degree of erosion, measuring setup, etc.. Therefore, the same
caused by local electrical stress enhancement. The right picture failure mechanism can result in different PD-measurements,
shows local PD-activity via an UV-camera indicating the local compare to Fig. 15.
defect position. V. SUMMARY
Electrical failures within the stress grading system can be
grouped in global or local defect causes. In most cases global
defects are caused by insufficient tailoring of OCP and/or ECP
resistance characteristics, whereas local defects are driven by
variations in resistance distribution.
By means of numerical simulations these failure types can
be identified and corrective or preventive measures addressed.
Diagnostic measurements, such as tanD or PD, may point
Fig. 13: Possible erosion pattern caused by the failure types F7 (left) and PD- possible problems related to the stress grading system out.
activity observed via UV-camera indicating local defect in ECP-area
However, assessment of measurement results is difficult and in
many cases not comparable with other units.
IV. DIAGNOSTIC MEASURES
REFERENCES
Common measurements to identify potential issues related
[1] M. Liese, M. Brown: “Design-Dependent Slot Discharge and Vibration
to the stress grading system are dissipation factor (tanD) and Sparking on High Voltage Windings”
partial discharge (PD) measurement. In special situations [2] C. Maughan: “Problems with newer Generators”, http://ccug.users-
global defects within the stress grading system can be evaluated groups.com/AnnualMeetings/2013/Presentations/Problems-with-newer-
by means of the tanD-results. Fig. 14 gives an overview about generators-3%20-%20Maughan.pdf.
the theoretical influence of the different global defects on the [3] http://www.ccj-online.com/generators/
reference tanD-curve of a single bar. [4] G. Klempner. I. Kerszenbaum: „Operation and Maintenance of Large
Turbo-Generators, John Wiley & Sons, 2004
[5] W. H. Moudy, “Deterioation Mechanism in Recent Air Cooled Turbine
Generators”, ASME Power, Atlanta, GA, USA, 2006
[6] C. Hudon, M. Bélec, “Partial Discharge Signal Interpretation for
Generator Diagnostic“, IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 12, No. 2, April 2005
[7] C. Staubach: “Comprehensive electrical and thermal analysis of the stress
grading system of a large hydro generator”, IEEE Electrical Insulation
Magazine, 2018
[8] G. C. Stone, et al.: “RECENT PROBLEMS EXPERIENCED WITH
MOTOR AND GENERATOR WINDINGS”, Petroleum and Chemical
Industry Conference, 2009
[9] C. Staubach: “Electrical Failure Mode Assessment of the Insulation
System in the Generator End-Winding Region”, 11. Essener Tagung
Fig. 14: Influence of the different global defects on the reference tanD-curve Turbogeneratoren in konventionellen Kraftwerken, Windparks und
Wasserkraftwerken, Essen, Germany, 2018
Basically, the initial tanD value is dependent on the OCP- [10] F. Pohlmann, J. Weidner, D. Schattauer: “Design and 25 Years
resistance, whereas the ECP will mainly influence the tanD at Experience of Turbine Generators with Global VPI Stator Windings”,
EPRI generator workshop, Rome, Italy, 2013
View publication stats