A model construction for predicting the inflow rate to a dam was developed taking into accuracy of rainfall prediction. Based on artificial rainfall prediction data which include quasi prediction errors, an order of the model for the inflow rate into the dam was determined so as to minimize the inflow rate errors. We evaluated the proposed method for model structure determination by use of actual data of the inflow rate of Kawago dam in Kyushu electric power company and the rainfall at the Kawago area. The model structure determined by use of the proposed method showed a reduced order comparing with a model ignoring the prediction error of rainfall. The constructed model gave a satisfactory prediction accuracy of inflow rate into the dam. The proposed method for model construction proved to be effective in determining the order of the model which included errors of input data.

View full abstract

This paper proposes a new method for the prediction of daily maximum electric load by using a reccurent neural network. Although algorithms like back-propagation are usually utilized in learning of recurrent neural networks, there are two serious problems, which one is that learning algorithm is inadequate to achieve optimum weight vector, i.e., local minimum problem, the other is that there is no method of settling structure of network. Proposed method can overcome these two problems by applying genetic algorithm to learning of recurrent neural networks. Two types of crossover operators are newly introduced into genetic algorithm for the optimization of the network structures. Therefore the number of hidden units is automatically fixed and local minimum can be avoided. Computational experiments show that the proposed method can produce recurrent neural networks with high abilities of prediction.

View full abstract

Recently, electric power demand is rapidly increasing and will be in the future. In the large power systems with complex configuration, undamped power swing with low frequency tends to occur. Therefore several stabilizing control schemes, e. g. a power system stabilizer (PSS), have been investigated so far.
On the other hand, superconducting magnetic energy storage (SMES) has been expected as an effective apparatus in power systems since a SMES located in power systems is capable of leveling load demand, compensating for load changes, maintaining bus voltages and stablilizing power swings. The effectiveness of each function, however, depends upon the location of the SMES in the power system because the output power from the SMES is distributed according to the impedance ratio of the transmission line at the SMES location. Therefore, it is rather difficult for realize a SMES to server for two different purpose at the same time.
This paper proposes a combination of SMES with a high speed phase shifter (HSPS). The HSPS, which consists of a phase shift transformer and a set of power converters, is capable of controlling the power flow of the transmission line by adjusting the phase angle of phase shift transformer. Therefore, it is expected that the combination of SMES and HSPS can realize a highly effective controller independent of its location. Numerical examples demonstrate that the proposed apparatus located far from a generator in a long distance bulk power transmission system is capable of stabilizing the power swing as effectively as the SMES located at the generator terminal. In addition, the effectiveness on the load change compensation as well as on the power system stabilization is confirmed numerically.

View full abstract

This paper proposes a construction method of a neural network for auto-tuning of feedforward signal for the boiler control of thermal power plant. The neural network costructed by the proposed method produces an appropriate feedforward signal by tuning weights of the network. The weights are adapted effectively by using PI control output of teaching signal. The proposed neural network was evaluated not only by use of the simple model which expressed crucial features of the thermal plant, but also by the detail simulator which faith fully represented the features of the actual power system, and proved to have appropriate control per-formances.

View full abstract

This work is a computational approach of transport (electrical & thermal conductivities) and ther-modynamic (enthalpy & density) properties of SF₆ gas polluted with Al vapor, which are the basic data necessary for the prediction of the behavior of high power arcs in GISs (Gas Insulated Switchgear) or between ACSR conductor. These calculations provide also a profile of electric field vs arc current for wall-stabilized arcs. Calculations have been done for a wide range of the temperature from 15, 000 K to 30, 000 K at a pressure of 0.4 MPa with the vapor admixture ratio from 0% to 50%. The results show that the electron density and the electrical conductivity are enhanced by Al vapor with relatively lower ionization potential even though a large number of electrons are expended to form F^- through recombination, especially at temperature 4, 000_??_8, 000 K. For instance, at 5, 000 K electrical conductivity σ is 0. 16 S/cm while it is 0.88 S/cm and 1.8 S/cm for 10% and 50% of Al vapor contamination, respectively. But below or near 3, 000 K, number of free Al atoms are nearly zero as they all remain in the molecular form of AlF and AIF₃ resulting almost no effect upon the electrical conductivity. The thermal conductivity is found to change only under heavily, more than 10%, contaminated condition of Al vapor. It decreases below 5, 000K due to waning of SF6 molecule whereas it increases from 5, 000K to 8, 000K due to the reaction thermal conductivity by the breakdown of AlF. The effect of Al vapor inclusion upon enthalpy and density is almost negligible. In order to provide a typical example for using obtained data, a study has been performed for prediction of the behavior of wall-stabilized SF6 gas arcs using the transport properties of the gas contaminated with Al vapor. Results indicate that the electric field intensity and axis temperature of arc at a certain value of arc current decrease for Al vapor inclusion. These reductions in the electric field intensity and the axis temperature are the reflections of input power contraction and of pressure suppression inside the tank, respectively.

View full abstract

La_1-x Sr_x MnO₃ is usually applied as cathode material of SOFC. Since conductivity, porosity and mechanical strengh are required to this cathode, selection of the diameter of the raw powder as well as sintering at appropriate conditions are indispensable. In this paper, sintering characteristics is analysed by using thermal dilatometer, for the purpose of obtaining the fundamental characteristics for several material. Influence of the sintering temperature, time and powder diameter on the sintering characteristics was clarified. And this sintering characteristics measured by dilatometer was shown to express the actual one well. Moreover, stability of the sinterd body was also investigated at 1, 000°C and large shrinkage was observed for non-stoichiometric material. As a result, sintering shrinkage measurement by thermal dilatometer was revealed to be useful in these analysis.

View full abstract

This paper investigates the dynamic behaviour of tap changing operations in Tap Change Under Load (TCUL) transformers. Interactions among multiple TCUL controls will cause oscillatory behaviour in tap changing actions, which unnecessarily increases tap operations before reaching equilibrium point. Conditions for the interaction are derived and features of the phenomena are examined through theoretical analysis as well as through numerical simulation. It is shown that various types of oscillations exist depending on the obtained conditions. The validity of the proposed analysis is verified through numerical simulations in an example system having three TCUL transformers, where sustained and transient type oscillations are observed as predicted by the derived stability conditions.

View full abstract

This paper describes an optimal load-shedding policy based on quadratic programmings using only a part of state variables of the power system after severe generation outages. When a large amount of generation outage occurs, the imbalance between supply and demand powers causes a declining frequency. Generators of utilities can not be operated excessively beside off normal frequency. So a part of load must be shed to prevent the system's damages. So far, the optimal load-shedding policies have been studied using all of the state variables of power system. However, in the real power system operations, it is difficult to obtain remote informations in a fraction of a second after generation outages.
In this paper, firstly, an optimal load-shedding method is constructed using the Quadratic Programming (QP) under the assumption that all the state variables in a power system can be available. However, these state variables can not be easily accessible as mentioned above. Secondly, to cope with these difficulties, a suboptimal load-shedding scheme based on the local state variables is studied. The proposed load-shedding method is based on an ideal of the above mentioned optimal one. The change of line-power flows and the amount of the generation power outage are used as the accessible data at each load point. Incorporating these local informations into the optimal load-shedding method based on the QP method, the proposed load-shedding method is established in this paper.
The effectiveness of this proposed method is illustrated by two examples, and simulation results on a model power system show that the method is encouraging.

View full abstract

For the evaluation of voltage stability of electric power systems using power vs voltage curve (P-V curve) in a real time voltage monitoring system, many cases of load flow calculation are required. To reduce computing time, parts of network in lower voltage level are required to be reduced into small equivalent networks. This paper describes a new method reducing parts of power system in lower voltage level maintaining the same voltage characteristics as those in whole network before reduction. Thus, P-V curve obtained at the network where some parts are reduced, corresponds to the one obtained at the whole network.
A new procedure to formulate an equivalent network is shown. Especially, in order to make the two P-V curves correspondent as accurate as possible, the reactive power of load in the equivalent network is determined as the function of active power of load and a new procedure compensating the resistance of the equivalent network is also introduced.

View full abstract

Authors already developed a model which describes the behavior of co-generator with and without wheeling. Wheeling enables co-generators to send electricity produced by them to remote customers through the transmission lines of utility companies. In the previous study, we theoretically analyzed its optimal behavior from the viewpoint of maximizing the co-generator's benefit and numerically evaluated the effects of wheeling on the energy efficiency. In this study the model is expanded so as to investigate its behavior under time-of-use-pricing of electricity. While the previous model is of a single time period, new model has two periods, i.e. peak and off-peak periods. It is interesting to model and analyze the behavior of co-generator under time-of-use-pricing because some modes of operating co-generation depend upon the price of each period.
Expanded criterion of the behavior under time-of-use-pricing is formulated and optimized analytically based on the Kuhn-Tucker condition. The condition under which wheeling occurs and that under which electricity is purchased are also clarified. In comparison with the behavior of auto-producers which supply only electric power, co-generator expands the range in operation with respect to time-of-use-pricing. The impacts of wheeling on the energy efficiency estimated in numerical examples imply that it is socially beneficial because it increases the efficiency of the system with co-generation plants.

View full abstract

The exact locations of 52 ground flash points of winter lightning were determined by VTR images taken from two observation sites. These optical observations of lightning flashs by VTR cameras had been carried out for about 40 days from November to December every year from 1985 to 1992 in Kahoku-Gata, a reclaimed flat land situated in the northern suburbs of Kanazawa, Japan. This land lays close to the Sea of Japan. Two kinds of meteorological radar echo diagrams are obtainable in this district, one being recorded by the Tojimbo radar of Meteorological Agency and the other by the Hotatsu-San radar of Ministry of Construction. The former is located at about 70km southwest and the latter at about 20km northeast of Kahoku-Gata. It is pointed out as a conspicuous feature that positions of the 52 ground flash points are all found on fringe of radar echo patterns obtained by the Tojimbo radar. As for 29 cases among them, the radar echo diagrams observed simulta-neously by both radar sites have been obtained. The echo distributions in the corresponding two echo-diagrams are similar to each other, but the details are different. Nevertheless, the ground flashes around Kahoku-Gata occurred on the fringe of radar echo patterns. Thus, when edge of the radar echo approaches to the important substation sites or transmission routes, any lightning flash to the ground might strike those power apparatus.

View full abstract

In the area having heavy volcanic ashfall, the distribution line trouble has been caused. It seems owing to the increase of the leakage current on the insulators polluted by volcanic ashes. We examine the factors which increase the leakage current and decrease the fluctuations of it in order to abate the distribution line trouble caused by the volcanic ashfall. For this purpose, to begin with, the electric characteristics are in detail examined, for example, the chemical compositions of the volcanic ashes and the electric conductivity, the ion concentration etc. of the volcanic ashes solution. Then on these basic data, the dependence of the leakage current on the various conditions of insulators polluted by volcanic ashes is experimentally investigated. As the result, the following matters are elucidated:
(1)The leakage currents are increased and in addition the fluctuations of them are increased involving the generation of corona only when the ashes accumulated in the ribs of insulators become wet.
(2)The anti-pollution type insulator is useful because it decreases the leakage current and its fluctuation.
(3)The leakage current on the solid-core pin insulator is not increased by the volcanic ashes.
In addition, the policies for abating the troubles also are got such as the determination of the most suitable direction, numbers and insulator types.

View full abstract

The effect of the ground conductivity on lightning-induced voltage waveforms on an overhead wire influenced by inclined return-stroke channels is studied based on an outdoor experiment with a reduced-scale model. It turns out that the ground conductivity affects the induced voltage waveforms associated with inclined return-stroke channels, and the effect is dependent on the three-dimensional configuration of the channel relative to the line.
Moreover, experiments verified induced voltage waveforms calculated by solving the Telegraphers' equations in combination with the numerical calculation of electric fields.

View full abstract

In back flashover analysis, it is necessary to clarify how the grounding of transmission line tower and electric facilities responses to lightning surge. Although the grounding of transmission line tower has several ohms in low frequency range (dc-power frequency) as the ordinary grounding resistance, we adapt several tens ohms in lightning surge analysis generally. Even if the grounding impedance is reduced less than several ohms in low frequency range for high soil resistivity, it is not proved that the reduction of the grounding impedance is effective in high frequency range as lightning surge. We investigated the surge response of the tower grounding by numerical electromagnetic field analysis and time domain reflection method under the condition that the ground has infinite or finite conductivity. A cylindrical tower model showed almost zero grounding surge impedance for perfectly conductive ground. A four-flames tower model, however, of which flames extend from the top to the base showed the capacitive grounding surge impedance about 20 Q even if the ground is perfect conductor. This is caused by the leaning current paths formed by flames. The grounding mesh structure made by horizontal electrodes also had the surge impedance about 20 Ω and the capacitive surge response. These results show that a grounding system has a lower limit of the surge impedance when we try to reduce the impedance by wayy of the the prevention of back flash over for lightning surge.

View full abstract

In order to discuss the UHV GIS test voltage, the deterioration characteristics of GIS insulating spacer model was studied by repeating lightning or switching impulse voltage. Deterioration factor in which indicates the decrease of lightning or switching impulse breakdown voltage up to 1000 times applications was found to be 35.7/45.5. And, the relation between partial discharge inception voltage and breakdown voltage was studied after 1000 times impulse application. As the result, it was found that partial discharge inception voltage was increased in spite of breakdown voltage decrease due to the impulse voltage.
Moreover, in order to make clear the deterioration phenomenon, breakdown characteristics of models with various boundary conditions between electrode and epoxy resin was studied and the relation between the boundary conditions.
Furthermore micro-discharge was also studied. From the investigated results, the deterioration by repeating voltage was caused by micro-discharge due to electric field intensification on micro protrusion of the electrode surface.

View full abstract

Effect of spherical metal particles on partial discharge in oil Immersed Insulating system was investigated theoretically and experimentally. Weakest link theory was applied to calculating the breakdown probability and the average breakdown field using the Welbull parameters of clean oil. Spherical copper particles of the same size were introduced to the oil immersed Insulating system by counting the number exactly. The partial discharge inception field was measured with the precise partial discharge monitoring system. The experimental results were consistent with the results analized theoretecally. It is thought that partial discharges are initiated at the triple junction of oil, pressboad and particles where the electric field is the highest In the oil immersed insulating system.

View full abstract

Superconducting technology is regarded to break through an obstacle of the future electric power transmission, because of its highly densified and large transmission capability. We propose a concept of the future power system composed of various superconducting apparatuses. A prototype model system named to “PROMISE (PROspective power transmission Model system Integrated under Superconducting Environment)” is constructed to prove the realization of the above concept. PROMISE is composed of superconducting transformer(60Hz, 6/3kV, 1000kVA class), superconducting fault current limiter(6kV, 200A class) and superconducting power cable(5m, 6kV, 650A class). This paper describes that PROMISE realized the transmission of the electric power of 3.8kV-46OkVA (50Hz). This is the first achievement of the world. The voltage-current synthetic test verified that PROMISE can withstand AC voltage of 6kV while it is carrying AC current of 170A (60Hz). The AC loss of superconducting cables, the heat leak of cryostat and the core loss of superconducting transformer are measured to estimate the transmission loss of PROMISE. These fundamental performances of PROMISE may indicate the feasibility of the future introduction of the superconducting technology for electric power system.

View full abstract

The fluctuating output power has been observed for a disk-type MHD generator in the FUJI-1 exper-imental facility for closed cycle MHD power generation. The fluctuation increases under the high loading conditions achieved with high seed fraction or high load resistance. In coincident with a big fluctuation of power output, the static pressure at the nozzle oscillates with frequency of about 500Hz. The electron temperature of the plasma fluctuates in wide electron temperature range between the partially ionized seed and the fully ionized seed.

View full abstract