Patrik Hilber

Multiobjective Optimization Applied to Maintenance Policy for Electrical Networks

A major goal for managers of electric power networks is maximum asset performance. Minimal life cycle cost and maintenance optimization becomes crucial in reaching this goal, while meeting demands from customers and regulators. This necessitates the determination of the optimal balance between preventive and corrective maintenance in order to obtain the lowest total cost. The approach of this paper is to study the problem of balance between preventive and corrective maintenance as a multiobjective optimization problem, with customer interruptions on one hand and the maintenance budget of the network operator on the other. The problem is solved with meta-heuristics developed for the specific problem, in conjunction with an evolutionary particle swarm optimization algorithm. The maintenance optimization is applied in a case study to an urban distribution system in Stockholm, Sweden. Despite a general decreased level of maintenance (lower total maintenance cost), better network performance can be offered to the customers. This is achieved by focusing the preventive maintenance on components with a high potential for improvements. Besides this, this paper displays the value of introducing more maintenance alternatives for every component and choosing the right level of maintenance for the components with respect to network performance.

Vulnerability Analysis of Power Distribution Systems for Cost-Effective Resource Allocation

This paper proposes a method to allocate resources in power distribution planning and also introduces a new reliability index category, RT, flexibility to adjust to different laws or distribution system operator (DSO) policies of long outages. Possible legal consequences for distribution system operators are first identified and studied. A vulnerability-analysis method is introduced, including a statistical validation. The overall idea is to identify and evaluate possible states of power distribution systems using quantitative reliability analyses. Results should thus indicate how available resources (both human recourses and equipment) could be better utilized, e.g., in maintenance and holiday scheduling and in evaluating whether additional security should be deployed for certain forecasted weather conditions. To evaluate the method, an application study has been per- formed based on hourly weather measurements and about 65 000 detailed failure reports over eight years for two distribution systems in Sweden. Months, weekdays, and hours have been compared and the vulnerability of several weather phenomena in these areas has been evaluated. Of the weather phenomena studied, only heavy snowfall and strong winds, especially in combination, significantly affect the reliability. Temperature (frost), rain, and snow depth have a relatively low or no impact.impact.

Identifying Critical Components for Transmission System Reliability

This paper presents a method to quantify and rank transmission system components by their importance for system reliability under different load scenarios. Each component is ranked by three separate importance indexes based on its expected outage rate and impact on: 1) system security margin; 2) load supply; and 3) generation units. By studying these three interests individually, a more complete view of the risks to system reliability can be assessed. The method is demonstrated on a detailed power system model (7000 components) of a significant part of the Great Britain transmission system at 400 and 275 kV. The results show how sensitive the component indexes are to the load scenario. The method provides an input for decision-making when planning maintenance and new investment and can be used as a complement to deterministic criteria.

A method for evaluating the impact of electric vehicle charging on transformer hotspot temperature

The expected increasing market share of electric vehicles is a response to the combination of new technological developments, governmental financial control, and an attitude shift of residents to a more environmentally friendly lifestyle. The expected capacity required for charging, imposes changes in the load to the already existing components in the electric power grid. In order to continue managing these existing assets efficiently during this load change, it is important to evaluate the impact imposed by the battery charging.

Prerequisites for transformer lifetime modeling towards a better understanding

The transmission transformer represent a significant asset in the electrical network. The transformer is expensive to manufacture and it is costly to replace. The cost of the transformer replacement is approximately 4 million EURO which is larger than the average component replacement activity. Therefore it is desired to make the replacement both timely and smooth to reduce unnecessary costs. Life time modeling is a tool for achieving such cost efficient replacements.

Potential of dynamic rating in Sweden

Power system owners are facing major challenges, for example with changed electricity consumption and production patterns; more distributed generation and increased demand of cost efficiency while maintaining high reliability. The concept of dynamic rating can act as a part solution and implies that the capacity of a component is dynamically varying as a function of external parameters such as weather and load history. This hence implies that a component can be better utilized. This can have both direct economic benefits for the utility, but also benefits for the society and the environment by lower tariff levels, faster and cheaper connection of local environmentally friendly electricity production and less climate impact associated with component production and installation. This paper gives a brief introduction to the concept of dynamic rating applied to power systems. Furthermore, results and conclusions from two workshops during 2013 are summarized, where representatives from Vattenfall and Fortum (distribution system operators), ABB (producer and developer that work with dynamic rating solutions), Swedish national grid and several persons from the academia contributed. This was complemented by interviews with the involved companies. The result is a mapping of knowledge, research, development interests, current situation and future visions. All parties show a great interest and see potential, but there are also challenges to be solved.

Transformer hot-spot temperature estimation for short-time dynamic loading

On the transformer, the effect of thermal stress is aging of the solid insulation. Excessive thermal stress could damage the solid insulation. To avoid this, transformer condition monitoring systems could use thermal models to forecast the operating temperatures during dynamic loading. There are several thermal models of varying complexity, including the thermal models stipulated by the IEC and IEEE standards. In this paper, the thermal model which is referred to by IEC Std. 60076-7 as the differential method, is modified, so it accounts for the oil viscosity dependence on temperature. The model is validated using hot-spot temperature measurements from a 40-MVA transformer, OFAF-cooled, 21/115 kV which is located in the subarctic climate region.

Reliability centered asset management

This paper describes three projects aimed at the development of optimal maintenance strategies in power systems. A poster will be compiled from the information herein.

Model of capacity demand under uncertain weather

Load forecasting is important in the operation of power systems. The characteristics of the electrical energy consumption are analyzed and its variation as an effect of several weather parameters is studied. Based on historical weather and consumption data received from a distribution system operator (DSO), numerical models of load forecasting are suggested according to electrical power consumption and on daily peak power respectively. Two linear regression models are presented: simple linear regression (SLR) with one input variable (temperature) and multiple linear regressions (MLR) with several input variables. The models are validated with historical data from other years. For daily peak power demand a MLR model has the lowest error, but for prediction of energy demand a SLR model is more accurate.

Influence of Ambient Temperature on Transformer Overloading During Cold Load Pickup

This paper proposes a method to investigate the socioeconomical aspects of transformer overloading during a cold load pickup (CLPU) in residential areas. The method uses customer damage functions to estimate the cost for their power interruption and a deterioration model to estimate the cost for transformer wear due to the CLPU. A thermodynamic model is implemented to estimate the peak and the duration of cold residential load. A stochastic differential equation is used to capture the volatility of the load and to estimate the probability for transformer overloading. In a numerical example, an optimal cold load pickup for a two-area system is demonstrated where transformer overloading is allowed. In this example, an ambient temperature threshold is identified, where transformer overloading is socioeconomically beneficial.