Voices: IBM's Jeffrey Taft: Bringing intelligence to the power grid

Jeffrey Taft is chief intelligent-grid architect for IBM’s Global Business Services. As such, his background in digital-signal processing is vital as he heads up IBM’s effort to become the purveyor of the software that runs the enormous distributed-processing network underlying the intelligent-power grid. EDN asked Taft about what the intelligent grid entails, what the benefits are to businesses and consumers, and how digital-signal processing plays a role in delivering power to consumers.

Just what is the intelligent grid?

First, let’s talk about the grid as it exists right now: You have a distributed infrastructure that includes power-generating stations, high-voltage transmission lines to carry bulk power out to delivery points, substations to break that power down to lower voltages, feeder circuits that run out to neighborhoods, and then distribution transformers to step the voltage down some more so that it’s suitable for your house. Currently, your utility probably finds out about a power outage only after enough users call in. An intelligent power grid adds instrumentation to the grid devices, to the substations, and to the lines themselves, and it collects massive amounts of data and then processes the data so that it can be acted upon in an automated fashion.

What does all this intelligence mean to the residential or industrial consumer?

There are several dimensions to the benefits of having an intelligent grid; some have a direct impact, and some are behind the scenes. Let’s talk about customer impact: Would you like to be able to know when your power is more or less expensive and be able to optimize your usage based on that visibility? Would you like to know that, during certain hours of the day, your power is more expensive? Would you like to have your utility call you and tell you that the power is off at your home or facility and allow you to take early steps? Would you like to not even experience an outage because the utility had a way to head it off before it ever even happened? These capabilities are all available with an intelligent grid.

Why do the utilities want to move to the smart grid?

Once the utility adds distributed sensors to the transmission-and-distribution system, it can automatically analyze all of that data for control purposes, for asset monitoring, for power-quality monitoring, and for outage intelligence. Rather than rely on customers to alert it to power outages, the utility will know exactly where the outage is, what equipment is affected, and what the root cause is and automatically dispatch the repair crew. But, even before the crew leaves, you’d like to isolate that fault with automatic switching and get as many customers as possible back online by rerouting power around the problem. The intelligent grid can do all that.

You’re an electronics engineer with a strong background in digital-signal processing. How does that background help you build a smart power grid?

There are a lot of parameters that you have to derive from basic waveform data. The power-distribution system is a three-phase system with complex connections. Its waveforms should nominally be simple sine waves, but, of course, in practice, they’re not, and you have to do sophisticated signal processing just to extract the parameters that tell you what’s going on. Through digital-signal processing, we get the information that tells us the real and reactive power flow that detects and locates problems in fault signatures and event correlations. We can tie that data together across multiple sensors and multiple power lines.

For example, an ordinary circuit breaker can operate multiple times to see if a fault is going to “self-heal.” Let’s say a branch falls and shorts out two power lines. The short may be temporary and burn away; you don’t want to trip a circuit breaker and lose power due to a temporary glitch. So, the circuit breaker opens up the circuit, waits, closes it, and sees if power is there. It can repeat this process several times, and, every time the breaker closes, it causes another surge of current through the breaker, the transformer, and the lines, causing serious stress on the equipment. If you could tell immediately from the power waveforms that it was going to be a permanent fault, you wouldn’t do the reclose cycling. Other power scenarios may play out over weeks or months.

Awareness of the power performance all begins with the number crunching of digital-signal processing. Digital-signal processing is at the very heart of the intelligent grid, and advanced signal understanding via analytics transforms signal data into information that can be acted upon both by automated systems and by people. Ultimately, the intelligent grid allows the utility to measure and maintain power quality and reliability, perform sophisticated grid control, maximize asset utilization, and respond quickly—and, in many cases, automatically—to grid problems.