Smart Distribution/SE4SG/CIM Subset
Supplementary material for the SG2SG workshop paper "A Common Analysis Framework for Smart Distribution Networks Applied to Survivability Analysis of Distribution Automation" by Anne Koziolek, Lucia Happe, Alberto Avritzer, and Sindhu Suresh.
Details on Technical Realization of our Framework
The CIM can for example be implemented in MOF or Ecore from the Eclipse Modeling Framework, so that the Graphical Modeling Framework can be used to generate the model editors. Model editors for CIM models, as sketched on the left of Fig. 1 in the paper, can be implemented in the Graphical Modeling Framework from Eclipse. Model transformations into different analysis models shall be created, e.g. relying on the QVT-R model transformation language.
As a survivability model, the approach uses the survivability models from [3]. The metamodels for communication network simulation, power flow, and demand response are yet to be selected.
The remainder of this paper explains the mapping of CIM to a survivability model as a specific example for an analysis enabled by our approach. Other analysis tools can be connected in a similar manner.
[3] A. Avritzer, S. Suresh, D. S. Menasche, R. M. M. Leao, E. de Souza e Silva, M. C. Diniz, L. Happe, A. Koziolek, and K. Trivedi. Survivability models for the assessment of smart-grid distribution automation network designs, International Conference on Performance Engineering (ICPE'13), 2013. To appear.
Details on used CIM Subset
In this work we use CIM concepts from the core standard IEC 61970-301:2011 [2] (UML model version 13v19, packages Core, Topology, Wires, SCADA, Meas) and the extension for distribution management, IEC 61968-11:2010(E) [1] (UML model version 10v31, package Assets as available from IEC.
Additionally, we use the yet unstable Informative::InfAssets::ReliabilityInfo concept, which is available in the IEC 61968-11 UML model version 10v31 [4] but not described in the official IEC publication.
Because the CIM metamodel is extensive and has complex inheritance hierarchies, we cannot present the metamodel itself here due to space limitations. We refer the interested reader to the CIM user group, where the CIM UML models are available [3] (current models are only available for member organizations and for private members, older model versions are publicly available). Instead, we focus on extracting the concepts relevant for survivability analysis of a distribution circuit and illustrating them in the context of this work using an example model shown in Fig. 1.
As described in Section III of the paper, we consider the distribution network between two Core::Substations, one of which powers the distribution circuit while the other serves as a backup power source (cf. Figure 2 in the paper). In between the Substations, the distribution circuit is assumed to be a sequence of sections separated by Breakers (i.e. reclosers). Within a section, any type of other Core::ConductingEquipment can be placed: ConductingEquipment is the superclass for "parts of the power system that are designed to carry current or that are conductively connected therewith" [2], for example Wires::ACLineSegments, which represent cables, or LoadModel::ConformLoads, which represent energy consumers that follow a daily load change pattern. The lower right of Fig. 1 shows an example, where a section consists of three Wires::ACLineSegments connecting households as an EnergyConsumer to the distribution circuit.
The described sequence of Breakers and other Conducting Equipment fully defines the topology of a distribution circuit. We mark the elements that together form a section and the circuit itself with the Informative::InfOperations::CircuitSection and Informative::InfOperations::Circuit model elements. This information can be automatically derived from the topology and does not have to be manually modelled, but is needed to annotate properties of the sections.
The ConductingEquipment in the distribution circuit is connected using ConnectivityNodes and Terminals in the CIM, as illustrated in the figure with the notation from [2] (Also used in CIM University slides). At this stage of research, the ConductingEquipment in the distribution circuit is only connected electrically to the two aforementioned Substations and must not be connected electrically to anything else outside the distribution circuit.
In the CIM, communication links are represented by SCADA::RemoteUnits and SCADA::CommunicationLinks. On the one hand, a RemoteUnit is connected to the Terminals of a Breaker to enable the Breaker to communication with the control center (the connection is not direct, but realized over the concepts RemotePoint, RemoteSource, MeasurementValue, and Measurement). One RemoteUnit models the central control unit (remoteUnitType = ControlCenter). For sake of simplicity, we assume in this section that RemoteUnits at the Breakers are connected with one CommunicationLink each to the control center. In reality, however, CommunicationLinks can be more complex networks with possibly redundant and meshed communication links.
To express properties of the logical CommunicationLink, it is annotated by an Assets::ComMediaAsset model element, which for examples specifies the type of communication link used (such as such as fibre optic cable, power-line, telephone, etc.). For more detailed analysis, for example for detailed communication simulation, more properties could be annotated here, possibly by extending the CIM. For this work, we express the reliability of the ComMediaAsset with the Informative::InfAssets::ReliabilityInfo concept. A ReliabilityInfo annotates any asset with a failure rate (as expected number of failures per year momFailureRate) and a mean time to repair (in hours as mTTR). This reliability information could be derived, for example, by running detailed communication simulations for a communication link.
References
[1] IEC 61968 Application integration at electric utilities - System interfaces for distribution management - Part 11: Common Information Model (CIM), July 2010. Edition 1.0.
[2] IEC 61970 Energy management system application program interface (EMS-API) - Part 301 Common Information Model (CIM) Base, Aug. 2011. Edition 3.0.
[3] IEC TC57. WG Draft Documents, 2012. http://cimug.ucaiug.org/CIM%20Model%20Releases/Forms/AllItems.aspx accessible for CIM User Group members.
[4] IEC TC57 Working Group 14: System Interfaces for Distribution Management (WG14). IEC 61968 subpackages of the CIM, version 10v31, Jan. 2010. accessible for CIM User Group members.
Summary of Section IV and Notes
In section IV, we have presented the mapping of a CIM model of a distribution circuit D to most input parameters of the survivability analysis described in Section~\ref{sec:apps:models:surv}, namely
- p(i): The probability that communication has failed given that section i fails
- q(i,d): The probability that sufficient backup power is available given that section i fails at any time during a day of type d.
- r(i,d): The probability that demand response is effective given that section i fails at any time during a day of type d and that not enough backup power is available.
- \gamma_i: The average repair rate to restore communication if section i fails.
- ES_s(i,d): The energy not supplied in state s of the model given that section i fails at any time during a day of type d.
With this mapping, we can calculate survivability measures of a distribution circuit given that a section i fails at a day of type d. If we assume that every section i fails with equal probability, we can furthermore derive a single survivability measure for a distribution circuit by averaging the survivability measures of the sections. Similarly, the model parameters could be further aggregated over different types of days by calculating a weighted average.
Currently, we do not derive the remaining rates \alpha (automatic restoration rate), \beta (demand response rate), and \delta (manual repair rate) from the CIM model. Selecting the appropriate model elements or even extending the CIM to provide this information is subject to future work. In this paper, we assume fixed rates as described below.
Parameter with Description and Value as a rate in events/hour:
- 1/\sigma: mean time for the reclosers to isolate failed section is approx. 0
- \alpha: automatic restoration rate is 30
- \beta: demand response rate is 4
- \delta: manual repair rate is 1/4
Detail on Case Study
Design options to consider
- Install another breaker to split the section 6: Installing another breaker allows more fine-grained failure isolation. In our case, we consider to split section 6 is split into two new sections 6a and 6b. If a failure occurs in 6a, 6b can be restored together with the surrounding sections, thus leading to shorter outages in 6b and consequently to lower cumulative energy not supplied, and vice versa.
- Improve communication reliability: Communication reliability could be increased by adding backup cell phone network communication to existing 900 MHz radio communication. The resulting redundancy will improve communication reliability.
- Increase available elastic load: The available elastic load could be increased by offering higher compensation for customers. Let us assume that this offer will convert 15% of the fixed load into elastic load for the sections that in principle support demand response.
These three options each represent a class of design decisions: Power topological changes, communication network changes, and demand response mechanism changes.
An excerpt of the resulting model for option 1 (to install a breaker) is shown in the figure below.
