Steps Toward a Smarter Transmission System Operation

 By Jorge L. Jardim and Luiz Claudio A. Ferreira

The Brazilian power grid is characterized by high energy transfers among the country's regions over very long high voltage transmission lines, both AC and DC. Such development has aimed at maximizing the use of renewable resources, mainly hydro generation that has been progressively located farther from load centers. Consequently, disturbances in critical transmission paths can cause large power imbalances and risk the system electrical security. These characteristics increase the chances of occurring frequency, angular or voltage instabilities.

Analyses of the load interruptions above 1,000 megawatts shows that the main causes are associated with a combination of natural phenomena and equipment or protection system failure, which are difficult to predict and have occasionally triggered cascading events. This article presents a brief description of the existing facilities, ongoing developments and planned efforts aiming at improving the Brazilian system operation procedures and security.

The maximization of power transfer among the regions, often necessary for economic reasons, relies on a comprehensive system of special protection schemes, including generation and load shedding and islanding protection systems. The available transfer limits are currently computed by extensive operational planning and online security assessments. An intermediate stage, under development, is an automatic near-real-time security assessment. Of course, the focus of these three assessment stages is on situation awareness, which is not a new concept, but of non-trivial realization. The special protection scheme setting points are determined offline through extensive simulations, but it is possible to envision that, if necessary, some of them can be calculated near real time in the future. Obviously, such applications require and rely on online data with reasonable accuracy, i.e., sufficient measurement redundancy, correct parameters and effective state estimation.

The security assessment approach is through security region per transmission bottleneck. The security transfer limits are computed as a function of generation dispatch in three separated areas specifically determined for each bottleneck. Each of the five system operator (ONS) control centers monitors the security of a set of transmission paths. Currently, 16 of these are being evaluated online. Similar effort is done in the operational planning stage, but uncertainties in load levels, outages and generation dispatch are expected to be relatively high at this stage. Therefore, a much larger set of cases representing several combinations of such conditions is evaluated, which generates a huge amount of simulation scenarios. The computation of a security region typically demands from 800 to 1,000 detail time domain simulations. With the use of efficient numerical methods and high performance computation this can be completed in less than two minutes for the currently used network and dynamic models. The computation efficiency requirement for planning assessments is much less stringent, but efficiency is still desirable because of the much larger set of scenarios.

The near real time security assessment can be very helpful in preparedness for either expected or unexpected grid conditions (e.g., planned or urgent outages, respectively), allowing more time for decision making. Based on generation and outage scheduling and load forecast, a near-real-time scenario can be constructed with a relatively low level of uncertainties. The power flow solution for these scenarios must be flexible enough to “tune” dispatch and voltage control setting points in order to keep voltages and flows within operational limits.

Phasor Measurement Units (PMUs) are being implemented and have been successfully used for post-mortem analysis. PMU applications for model validation and detection of sources of undamped oscillations are being researched and developed. PMU applications are also expected to help with state estimation and system restoration.

Security assessments both in operational planning and real time do not consider uncertainties in models and disturbances. Considering that several large load interruptions resulted from combined events including equipment failure, there is a growing interest in performing risk analyses to evaluate similar conditions and likelihood of cascading effects.

Development of applications to help with system restorations is also being planned. So far, the strategy used has been quite successful and consists of an independent phase in which restoration actions are based on local information only and a coordinated (by the system operator - ONS) phase for regional interconnections. However, as expected, real system conditions during the restoration quite frequently deviate from those defined in operational instructions, making the operators decision more difficult and therefore resulting in unsuccessful actions and undesirable delays. Again, there is a great potential for the use of high performance applications to assist operators in this process.

Traditionally, the decision making support are mainly conceived from the technological view point in which system operators must process a significant amount of information. As technology evolves, more measurements, application results and displays are provided to them. Often, the problem is not lack of information, but to find the most relevant for each particular context. Therefore, there have been initiatives to conceive support systems centered on operator situational needs and attempting to provide high level situational information.

In the decision process, most of the time and effort are spent identifying and classifying the situation. This work can be facilitated by helping operators to keep focused and by providing an effective short- and long- term knowledge base, considering the conditions to which they are subjected (time pressure, conflicting objectives and uncertainties).

Improvement in operator training methodology and technology are also necessary. They must be exposed not only to routine procedures, but also to extraordinary situations. Important aspects are the improvement of mental models and means of keeping focused. The level of realism of the training environment (simulator) needs to include dynamic models able to represent transient stability, cascading effect and dynamic phenomena during restorations. Protection system must be modeled for most of the equipment. In addition to the model requirements, part of the supervised actions must also be automated to represent the actions of third parties. This is particularly necessary for reproducing large perturbation scenarios.

In summary, the steps towards a smarter transmission system operation have been focused in situation awareness to enable quick and effective operation actions either automatic or assisted. The increasing operational complexity and the recognition that large disturbances have been originated from causes difficult to predict do point to the need of improved predictive facilities and operator’s abilities and resources. Some of these are traditional objectives, but with new technological perspectives.

Contributors