Online state estimation in modern grids

Combining online estimation and lifetime monitoring means maintenance costs can be reduced and the reliability of a network increases, write Dr Ivana Mladenovic and Dr Thomas Werner


Today’s grids are facing new and challenging issues

Credit: Bert Kaufmann

The liberalization and opening of markets has, in the past two decades, seen power systems undergo numerous structural changes, many driven by environmental and economic factors.

The integration of offshore and onshore wind farms, Germany’s ‘Energiewende’ and increased integration of renewables in lower voltage levels have changed the face of the power system. A network that was not designed for bidirectional power flow with decentralized, stochastic and volatile loading and generation has had to face new challenging issues.

As a result, today’s grids have to develop according to the following influences: increased volatile and fluctuating loading of system components; identification/detection and avoidance/elimination of bottlenecks; ensuring strong ancillary services and appropriate allocation of the responsibilities between transmission and distribution system operators; enhanced functionality of protection systems; improvement of measurement, monitoring and communication systems; and life-cycle management of networks and assets.

However, these challenges are not always of the same importance for different voltage levels. Mainly, distribution grids are affected, since 96 per cent of renewables are integrated in medium- and low-voltage networks.

In general, these grids are not equipped with measuring, communication and condition monitoring systems to the same extent as transmission networks. Therefore, in today’s era of automatization and digitalization, development and realization of suitable accompanying systems for grid management play a key role.

Flexible, dynamic, automated

Modern power networks can be characterized by flexible, dynamic and automated operation. Their structure differs from conventional grids in that self-organizing subsystems also carry responsibility for the local system’s efficiency, supply stability and reliability.

The modern power system can also be seen as a multimodal energy system, where all the parts individually contribute to the stability and efficiency of the complete energy system.

The foundation for optimal operation is knowledge of the state of these energy systems and their components. It allows the use of advanced applications and decision-making for improved network and asset management.

Figure 1. Partial view over ICAAS

Credit: Siemens

In modern power systems, there are two approaches for state assessment:

ࢀ¢ Online state estimation describing the actual network state with respect to electrical values like voltage and active and reactive power; and

ࢀ¢ Lifetime or condition monitoring of assets such as transformers and cables, which provides relevant information for asset management, for example to describe the aging of an asset.

In the last decades, computer-based applications have been developed for both tasks, especially for use in transmission and primary distribution grids.

Compared to secondary distribution grids, these grids encompass a lower number of nodes, branches and assets. This allows for more or less complex applications with intensive engineering.

In secondary distribution networks with a high number of equipment and assets, the use of traditional methods will not provide the right cost-benefit relation. Because of this, new concepts are in development for online state estimation and condition monitoring.

Online state estimation

With increasing network requirements, it becomes more important to have a consistent view of the state of the distribution network. Applications like load flow, voltage-var-control or optimal feeder configuration need a set of input data which describe the network state with minimal error, and which is self-consistent. State estimation is an established method for calculation of the complete network state on the basis of a set of measurements, and represents an essential part of any modern network’s control centre.

The state-of-the-art estimation which is used in power control systems is based on a comprehensive data model of the network, an extensive installation of measurement devices and stable communication channels from the control centre to the stations.

Transferring the same state estimation concepts to the secondary distribution networks is very costly: acquisition and maintenance of network model data is time consuming, the number of measurement devices installed in the field is too low for the conventional algorithms, and a lack of private communication lines requires the use of public communication networks with all of the related communication costs.

However, the cost-benefit ratio of state estimation improves if its results can be used not only for operation control of the distribution network, but also in the asset management of its components.

By combining online state estimation with online lifetime monitoring, the network operator’s awareness of asset conditions increases and allows for the change from regular maintenance of the assets to condition-based maintenance.

As a result, maintenance costs can be reduced and the reliability of the network increases, as does the satisfaction of the network customer.

Asset-management and maintenance strategies can be classified into groups relating to the condition and/or importance of the component.

Even-oriented (no significant maintenance before failure) and time-dependent (fixed time intervals for maintenance, importance included) strategies do not consider component condition. In many utilities today, there is a movement to condition-based strategies, which can also be upgraded to reliability-based strategies wherein, in addition to condition, risk-management and prioritization assessments of the equipment are made.

However, for condition determination of the components, off-line diagnostic methods are mostly used. These require special diagnostic equipment, disconnection of the component from the grid, and trained personnel to perform diagnostic measurements. This takes much time and mostly gives only classification of the actual condition.

In order to reduce the costs for the diagnosis, increase operation reliability and get an overview of the remaining lifetime of the components in the entire grid, online diagnostic methods are required.

Moreover, for the purpose of estimating remaining lifetime, life models are needed to estimate the equipment’s probable remaining life dependent on actual and expected load occasions.

With this in mind, an experiment on MV paper-insulated, lead covered cables was carried out. A fully automated and integrated cable accelerated aging system (ICAAS), that realizes artificial but realistic aging by freely definable aging parameters and load profiles, was developed, realized and verified.

In Figure, 1 a partial view of some of the aging and protection components of the ICAAS system and the cable samples are shown.

After verification of system functionality and determination of the most suitable aging conditions within a pre-test, the main aging experiment began. Test objects included several groups of MV PILC cables from different generations: besides brand new cables, there were also samples that had been in field operation for up to 45 years, and both of these were artificially aged.

During the artificial aging, monitoring of diagnostic parameters as well as of aging conditions were calculated with an accurate measuring system in regular time intervals.

In addition, initial measurements – parametric studies (PS), which were also repeated during the aging process – have been carried out to determine the influence of variable test conditions, such as voltage or temperature, on the dielectric parameters. In this way a sophisticated databank with over 300,000 measurements and 1 TB of data was built up.

In the next stage, numerous studies and analysis were performed, resulting in:

ࢀ¢ Development of methods and approaches for data processing and interpretation;

ࢀ¢ Determination of the physical dependencies and the critical levels of important diagnostic parameters (aging condition, test parameters, environmental parameters);

ࢀ¢ Description of the aging processes through life/aging models;

ࢀ¢ Characterization of the condition, aging behavior and remaining lifetime of PILC cables.

In this way, condition estimation is improved and prediction of the remaining lifetime of PILC cables is enabled.

Integration of measurement and communication techniques in distribution grids offers the possibility for their improvement to modern power grids, characterized by flexible, dynamic and automated operation.

For control of the operation, online state estimation methods can be used. By supplementing them with life models of network components, it is possible to introduce online condition monitoring with active and dynamic prediction of components’ remaining lifetimes.

This way not only overcomes the disadvantages of offline diagnostic systems, but also ensures that no additional measurement equipment or personnel are needed to get an overview of the remaining lifetime of the components in the entire network.

Ivana Mladenovic works in Siemens’ Switching and Power Grid group. In 2012 she won the John Neal Award of European Electrical Insulation Manufacturers.

Thomas G Werner is innovation manager and principal key expert for distribution automation at Siemens.


No posts to display