Asset Health Index and Risk Assessment Models for High Voltage Gas-Insulated Switchgear Operating in Tropical Environment

Andreas Purnomoadi

Research output: ThesisDissertation (TU Delft)

345 Downloads (Pure)


Following deregulation in the energy sector during the 1990s, which was also triggered by the ageing of infrastructure and the increasing demands from regulators and customers, many network utilities adopted the Asset Management (AM) in the hope to earn more, have better credit ratings and gain from stock prices. In line with this fact, the emergence of the AM international standard, such as the ISO 55000 series in 2014, gained rapid acceptance among network utilities around the globe.

AM has its core in the asset decision-making process. This activity lies simultaneously at the strategic, tactical and operational level of AM, over the lifecycle of the asset. In such an environment, the asset managing department should not only focus on the reliability of the asset but also on balancing costs, risks and asset performance. Regarding maintenance, the money spent on every maintenance task should benefit the company’s business values.
This thesis focuses on the development of decision-making tools for maintenance of high voltage AC (HVAC) gas-insulated switchgear (GIS) operating under tropical conditions. GIS has been chosen because of its critical role in the transmission network. Any GIS breakdown is usually expensive and requires an extensive outage. Moreover, under tropical conditions, this study observed GIS failure rates over twice the value reported by CIGRE’s survey of 2007. The study was conducted in this research’s case study termed the Java Bali (JABA) case study. The latter consists of 631 CB-bays of 150 kV and 500 kV GISs located in Java and Bali of Indonesia.

Today’s AM decision-making tools for electrical power grids are generally based on Asset Health Index (AHI) and risk assessment (RA) models. These models assist the asset manager in answering the following questions:
1. What is the condition of each GIS in the network?
2. Which one is more likely to fail compared to the others?
3. Which one is more critical compared to the others in terms of making a possible impact on the company’s business such that the mitigating action is prioritised?
4. What optimal action(s) is/are needed to be taken?

Developing the above-mentioned models requires sufficient knowledge of the characteristics of GIS operating under tropical conditions. To that purpose, both statistical analysis and forensic investigations in the JABA case study have been undertaken to find the critical condition indicators for the AHI model. The results are as follows:
1. The tropical conditions have influenced both directly and indirectly the performance of GIS. Corrosions at the exposed GIS parts were seen to have a common direct influence of tropical conditions. They can trigger leakages, secondary, and lead to driving mechanism subsystems’ failures, which reduce the GIS’ performance. The intensive and frequent lightning in tropical conditions is a so-called Failure Susceptibility Indicator (FSI), indicating that a failure mode is expected to initiate more likely than for the same GIS in other environments, especially if the surge arrester fails to protect. Moreover, the GISs outdoor and from the older generation are more susceptible to breakdown under tropical conditions.
2. A high amount of humidity was found in the non-CB enclosures of GIS from lower voltage class (i.e. Class 2 GIS with a voltage level of 150 kV). The origin of this humidity mainly comes from the desorption of moisture from the spacer or internal GIS surfaces during operation.
3. The critical failure modes in GIS operating under tropical conditions are as follows: dielectric insulation breakdown, loss of mechanical integrity in the primary conductor and failing to perform the requested operation due to driving mechanism failure.

Following this study’s findings, laboratory tests in the HV Laboratory of TU Delft were conducted to investigate the influence of high humidity content on the spacer flashover in GIS. The results confirmed without condensation, humidity has no impact on the withstanding strength of the insulation system under AC, LI+/- and SI. Our model also showed that the breakdown voltage under LI+ due to condensation at the surface of a solid insulator is lower than that due to a 2 mm metallic particle attached on the identical solid insulator at 3000 ppmV.

We applied the findings from both field investigation and laboratory tests into our models in the following ways:
1. In the AHI model:
a. Statistical and JABA lab case studies were performed to assess the system’s vulnerabilities and normative levels, in particular, the humidity content in GIS the non-CB enclosure as long as the value was far from the possibility of condensation.
b. The likelihood of failure is determined by so-called condition scale codes reflecting the deterioration of the subsystems.
c. The failure susceptibility indicators (FSI) flag deviating circumstances, such as heavy environmental conditions, operation and maintenance records and the inherent/design factor of GIS. The FSI are just an expectation that is not based on evidence as in a condition indicator. Therefore, the FSI work as warning flags for the decision-maker.
2. In the RA model:
a. Risk is defined as the likelihood of failure times the consequences. The result of the AHI defines the likelihood of failure in the RA model.
b. On the other hand, the consequences consist of seven business values of a transmission utility from the JABA case study, namely, safety, extra fuel cost, energy not served, equipment cost, customer satisfaction, leadership and environment.

We have successfully implemented these models on a GIS example from the JABA case study. Evaluation of possible risk treatments was also done using multi-criteria analysis (MCA) to optimise three parameters: cost, time-to-finish treatment and residual risk.

In practice, transmission utilities face more complex situations with more types of equipment in the network. The methodology discussed in this thesis, however, can be the cornerstone for the development of decision-making tools for other assets at the tactical level of AM as well.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Delft University of Technology
  • Smit, J.J., Supervisor
  • Rodrigo Mor, A., Supervisor
Thesis sponsors
Award date13 Jan 2020
Print ISBNs978-94-6384-098-9
Publication statusPublished - 2020


  • Asset management
  • Asset Health Index
  • Risk assessment
  • Gas insulated switchgear
  • Tropical environment
  • Decision-Making

Cite this