Transitioning to the power sector’s Smart Grid (SG) involves delivering the full continuum of functionality for SG applications, as follows:
- Automation, and
Of these functionalities, most of the SG applications today deliver sensing, monitoring (i.e., sensing plus communication), and some level of diagnosis which may require additional post-processing. The remaining three functionalities -- control, automation, and optimization -- are much more difficult to implement for a variety of reasons, not the least of which is the dearth of algorithms available to support these applications in a grid that is often operating in ways for which it was not originally designed and in electricity markets that are not fully integrated.
The Importance of High-Performance Sensors
But sensing, while basic on the surface, is not “a walk in the park” for a grid that is undergoing major physical and market structural changes even as it moves towards increasing automation.
Traditionally, we have used static optimal power flow (OPF) models for state estimation of the nodes in the distribution system and for designing power security protection schemes. With the computer processing capability available today, these models have become quite fast, with state recalculations for the entire distribution system now possible in seconds using blade servers. Today a full simulation can be completed well within a typical SCADA cycle. But the results are still “estimates” of the state of the distribution system.
While better OPF models are necessary, they are not sufficient to address the upcoming challenges of the SG because the distribution system is being operated in ways never contemplated when it was designed, due to the introduction of:
- Two-way power flows associated with distributed resources, micro-grids, and virtual power plants
- Intermittency introduced by PV installations in distribution systems
- Electric vehicle charging
- Automated demand response, and
- M2M appliances in end-user facilities
The most challenging impacts of these changes are:
- Volatility of distribution power flows is increasing significantly
- The rate of change of power flow metrics is accelerating, and
- Short-term load forecasts of “net demand” are exhibiting a broader range of uncertainty than before, creating challenges for operators to schedule production that meets minute-to-minute demand
As a result, even the fastest OPF models need to be supplemented (as well as re-calibrated) with real-world data that reflect these new operating regimes and uncertainties.
Bottom line: to maintain acceptable reliability, security, and stability levels as the physical grid and its associated markets are re-structured, we need to have a much more granular, real-time situational awareness of the distribution system.
How do we address this need for heightened situation awareness as we continue to implement interoperable SG applications?
Introducing the Über-Sensor Platform
Let’s assume that we have installed a set of ultra-accurate, ultra-fast, sensor platforms in our distribution system. The sensors are synchronized to the same “very exact” nano-second accuracy clock. All sensor measurements are time-stamped and concurrent.
What do the sensor platforms consist of? Each sensor consists of a high accuracy analog front-end (AFE) and a flexible, high-performance field programmable gate array (FPGA) integrated as a system-on-module (SOM), connected to, or embedded in, power measurement devices or intelligent electronic devices (IEDs).
Using technology available today, combined with some proprietary technology configured in a novel manner, the AFE can be highly accurate – providing, for example, a steady noise floor of ~-150 dB all the way out to 105 Hertz and perhaps ~-120 dB at 106 Hertz. The AFE can be supplemented by fast digital signal processors (DSPs) and hardware accelerators. Each sensor could then provide wide spectrum capture, power system hardware DSP, programmable DSP cores, programmable response, a customizable IED firmware stack, and the ability to communicate securely in milliseconds. It is, in effect, a platform that can support all six of the SG functionalities discussed in the opening paragraph of this dialog.
A sensor platform with these performance characteristics could be used in point-on-wave, phasor, power/energy, and power quality measurement applications. Electricity waveforms could be sampled with high accuracy at high frequency. For example, a 2 µ-second transient could be captured and displayed with high fidelity. For distribution system voltage levels, total vector error would need to be in the range of ~<0.2%, which is attainable today, but not without some difficulty.
With an Über-Sensor, We Can Equate “When” with “Where” in the Distribution System
Imagine what we could do with this Über-sensor platform/network in the electric distribution system.
Here’s an example: an incipient fault, or a 2 micro-second transient created by a lightning strike, has created an aberration in the wave form emanating from a particular location in the distribution system.
It spreads like the ripples from a stone cast into a pool. “I felt a great disturbance in the Force”, says Obi-Wan Kenobi. In Gridiopolis, the Engineer Ultra (υπερ-και για το μηχανικό) feels the hair stand out on his Greek neck and he reaches quickly for his krater. But seriously……
In the power system, just like in “The Force”, we know that everything is connected to everything in dynamic meta-stable equilibrium – it is a network of inter-dependent nodes in constant motion, maintained in equilibrium by grid operators.
Emanating outwards from the location of the above transient event, associated disturbances in the wave-form propagate across the distribution system and are sensed by the distributed über-sensor platforms. Because all of the sensor measurements are synchronized and concurrent, the time taken for the propagation to reach an individual sensor is known, and thus the disturbance can be traced back to a particular location by a form of triangulation between the sensors. (Of course, such a locational algorithm is yet to be developed, and we would also need pattern recognition and out-filter algorithms to eliminate the modulations of the disturbance introduced by intermediate devices.)
Yes, “When” Can Equal “Where” in the Distribution System
In effect, if we know very precisely the “when” of a disturbance, then, with a high-performance set of sensor platforms, we should be able to determine the “where” (the origin of the disturbance). In addition, with pattern recognition, the type and likely cause of the disturbance could also be determined.
With a network of über-sensor platforms, we would not need an OPF model to estimate states of the distribution grid, because we would “know” the states in real-time. We could archive and use this operations knowledge to re-calibrate the OPF model for off-line planning applications and post-mortems of disturbance events and remediation schemes.
The above application would be useful in terms of two high-value SG applications:
- High-speed fault detection, location, and causality
- Intelligent replacement of aging power system infrastructure through the identification of incipient failures
Availability of Über-Sensors Today
Über-sensor platforms like this are available today as “evaluation boards”. They go beyond ARM capability to add all of the additional functionality above, with a much-reduced chip-count per module. After they’ve being thoroughly tested and refined in upcoming field applications, they will ultimately be turned into high volume, low-cost ASICs, embeddable in IEDs commonly used in distribution systems.
Why Do We Need Über-Sensors?
Policy-driven structural changes are occurring, and are expected to occur, in the power grid and its associated markets. These changes will have profound operational implications for the grid. The transition to “Smart Grid” applications is being driven by the need to cope with these operational challenges.
In this context, if we are to deliver the same or better reliability of service at an affordable cost to electricity customers, we will need the above high-performance/low-cost sensors as fundamental building blocks to support our ultimate goal -- an interoperable, automated, optimized SG capable of maintaining resilient dynamic equilibrium while under constant fire from millions of continuously-occurring potentially destabilizing events.
As always, comments are welcome and appreciated.