Situated where the North Sea meets the Atlantic Ocean, the Orkney Islands offer high potential for wind, wave and tidal-based power. But more than a decade ago, energy entrepreneurs there met a common obstacle: the grid's cables, transformers and other physical infrastructure could handle only incremental new production. Governments (both local and national) and the area's Scottish & Southern Energy Power Distribution (SSEPD) utility were bullish on renewables. But could they justify the cost – estimated at some £30m – needed for upgrades that would enable the islands to become an energy exporter?
Today the Orkney Islands have accommodated new energy production, adding almost 20MW to an at-capacity grid, without the brute strength (and brutally expensive) infrastructure upgrades previously believed necessary. The Orkney Smart Grid project, which has enabled the growth, serves as a early showcase for active network management (ANM), a technology premised on adding a new layer of high-precision grid management to help power companies tackle grid constraints, integrate more distributed energy sources and respond to growing demand for electricity in a more affordable, timely and sustainable way than building new grid infrastructure.
The Orkney Islands project results from collaboration between SSEPD, the University of Strathclyde's Institute for Energy & Environment, and Smarter Grid Solutions (SGS), a smart grid technology company that originated from the university several years ago, to commercialise ANM concepts developed there. In both its ideas and its personnel, SGS maintains close ties to the University of Strathclyde and its energy institute: Bob Currie, the company's co-founder and technical director, wrote the doctoral thesis on which the Orkney Islands project is based. Another co-founder, Graham Ault, now the company's development director, holds a chair in electric power systems.
With ANM, SGS aims to give power networks a level of automated supervision, control and data-based decision-making already common in other highly complex systems, such as aerospace and defence. Power grids typically lag these other fields, and a major focus has been on moving away from generic, out-of-the-box system components (such as automation controllers designed for use across multiple industries), and creating its own technology infrastructure that is designed specifically for the challenges of power grids. To this end, the company has introduced a family of applications to handle ANM tasks including regulating the amount of power produced or consumed by generators, loads or storage devices; scheduling and coordinating power supply and demand, based on operational goals (such as increasing use of renewables); monitoring real-time weather data to determine overhead conductors' carrying capacity; and improving grid visibility by reducing errors and estimating electrical parameters where physical measurements are not available.
Underlying these applications is sophisticated platform technology, which runs on commodity server hardware and performs critical tasks including communications, interfacing with and controlling field devices, and providing a common database to facilitate the applications' and platform's work. ANM is data intensive, and becomes more so as the concept is extended to larger power grids, comprising larger and more complex pools of data sources, for example voltage and current transformers, device controllers and external systems.
This data management is inherently real time. Values in the database represent critical constraint locations on the grid as well as controlled devices. The data can change frequently and must be updated and monitored within tight time constraints and with a high level of fault tolerance, to identify control actions needed to maintain the grid within operating limits.
SGS's earlier implementations of ANM relied on the data management built into commercially available automation controllers' software. But in its new-generation product, the company determined it would gain greater performance, resilience and interoperability by licensing code from a vendor specialising in database management systems (DBMSs), particularly if that provider offered expertise in mission-critical, real-time systems.
After evaluating some half-dozen database products, it chose the Extreme DB Cluster technology from McObject, a US company. Founded more than a decade ago, McObject is a specialist in in-memory database systems (IMDSs), which manage records in main memory to eliminate the performance latency inherent in persistent (disk- or flash-based) storage. McObject has also added, in its IMDS technology, the ability to designate persistent storage for selected record types.
SGS chose Extreme DB Cluster based on closely examining the technology, but was also influenced by the vendor's track record and reputation: McObject pioneered the use of IMDSs in embedded systems; millions of instances of the Extreme DB software are deployed in commercial applications ranging from consumer electronics to mission-critical combat jet avionics, industrial control systems and telecoms and networking devices. In addition, it could point to existing customers in the power generation and distribution field.
Extreme DB Cluster edition is a distributed database system that manages databases across multiple hardware nodes, enabling two or more servers to share the workload (see Fig. 1). In an Extreme DB Cluster deployment, any process on any node can update its local database, and the clustering software replicates the changes to other nodes. By incorporating this, SGS determined it would gain the following desired characteristics and capabilities:
• Enhanced performance: - To provide the highest speed, Extreme DB Cluster relies on its core IMDS design. By storing records in main memory, an IMDS eliminates the overhead of disk and file IO, cache management and other functions found in traditional on-disk DBMSs. Higher responsiveness and predictability also stem from the in-process architecture: the database system is embedded in the application process, eliminating the latency of inter-process communications (IPC) between client and server modules.
• Fault tolerance: SGS must meet power network operators' stringent fault-tolerance requirements. Distributing the database system across multiple hosts ensures continuous availability in the event of a failure on one node. In addition, it supports the same acid (atomic, consistent, isolated and durable) database transactions offered by the non-clustering editions, making it an attractive choice for applications such as those that demand integrity of distributed data. Its shared-nothing architecture eliminates reliance on a shared san or other storage resource.
• Scalability: The ANM technology must scale from small to large deployments, to accommodate utility customers of varying sizes. An appeal of clustering database technology is its ability to scale out economically by adding low-cost (that is commodity) servers, rather than having to scale up by moving to higher-power server platforms.
• Interoperability: To accommodate SGS customers with diverse operating environments, the database system needed to run on (at a minimum) Windows Server, Linux and Solaris. Extreme DB Cluster does this, and it also provides multiple APIs) including C/C++ API and a Java Native Interface (JNI) used by SGS. The JNI technology, in particular, was seen as an asset, because it enables access via the widely used Java language to core database functions, while the database software itself executes in compiled C/C++, which is typically seen as providing higher performance. It also streamlines coding by enabling developers to access Extreme DB while working entirely with plain old Java objects (Pojos). SGS also makes use of Real-Time Java to retain determinism within its smart grid algorithms.
This ANM method uses four synchronised database nodes: one main and one standby database to support SGS Core, the platform component that hosts real-time smart applications, represents critical constraint locations and devices, and identifies needed control actions; and one main and one standby database to support the SGS Comms hub, the platform software that acquires and manages data from external grid management systems and controlled devices. The platform with its four nodes is designed to run autonomously in locations ranging from regional control centres to local electrical substations.
"Extreme DB Cluster is at the heart of our server platform," said Neil McNeill from SGS. The database system serves to provide "a single representation of all the network elements in real-time" and to control those elements, he said.
McNeil noted that its role is similar to that of a master record in supervisory control and data acquisition (scada) systems, which are well-known in the power generation and distribution industries, except that scada is mostly observational and controlled by an operator, whereas ANM runs autonomously and exercises a high degree of control over the network and its components. In its time constraints and autonomy, McNeil said that ANM technology edged away from scada and closer to the category of real-time protection, automation and control (PAC) systems, as illustrated in Fig. 2.
Like PAC, ANM responds synchronously to events and conditions on the power grid, whereas scada systems are asynchronous, McNeill said. The required time threshold for the overall ANM system response, including database processing, is approximately 800ms. However, SGS has tested the system with Extreme DB Cluster successfully down to 20ms, McNeill said.
"That's very good and we're pleased with that level of performance," he said.