Big Data

The Smart Grid Big Data Challenge

The Smart Grid Big Data Challenge

If we want to achieve ongoing smart grid optimization, we can’t do so without actionable intelligence. It’s needed to correct deficiencies throughout the power delivery system, not just to the customer side of the meter, but to transmission, substation and distribution systems. All the real-time data points being collected across it — generated by millions of wireless sensors and other smart interconnected devices — is creating one of the biggest Big Data challenges out there. So what’s the plan?

Contributor

    • Mark Wyatt, Vice President for Grid Modernization and Distribution, Duke Energy

Download Podcast

Apple Podcast, Google Podcast, Spotify, Pandora, iHeartRadio, SoundCloud, TuneIn, and Stitcher. Find other syndication channels here or search CIO Talk Network podcast on any other app.

Summary

One of the biggest challenges utilities face with the implementation of Smart Grid is how to manage the influx of Big Data that will come with it. Most of the networks around the United States are in need of updating to a digital format because the data that is returned to engineers is often days, weeks or months old when it is processed.

Getting digital access to real-time or near real-time data provides great business value in terms of catering to the customer and reducing costs. Empowering customers with information about their energy usage, being able to predict and respond timely to power outages, improving the accuracy of meter readings and reducing the need to send employees out into the field are just some of the benefits that can be attained through Big Data.

Part of the challenge comes from updating a process built around using old data to using real-time data and putting analytics around it. But beyond that, how much data needs to be brought back from meters for it to be useful? Most utilities will tend to shoulder the cost of bringing back all data and piecing through the data that is relevant, which often is less than five percent of the collected information.

Understanding up front what data is necessary to make a critical decision and reducing the amount of data that needs to be collected can help make Big Data much less overwhelming. Utility CIOs are looking to other industry verticals to determine how Big Data has worked for their industry, as the challenges are not that different.

The raw processing capability of the existing hardware and software is currently there in terms of being able to do what utilities need the technology to accomplish, so the concern is less of a technology challenge and more of one to the business organization in asking, “How do I turn that into real value?”

Transcript

Sanjog Aul: Welcome listeners this is Sanjog Aul, your host, and the topic of conversation today is “The Smart Grid Big Data Challenge,” and I have with me Mark Wyatt. Mark is the vice president for Grid Modernization and Distribution with Duke Energy. Hello Mark, thank you for joining us. Now talking about Big Data is a huge trend right now, but it was important to talk about in this context, because with Smart Grid, figuring out the plan for solving this problem is arguably step number one. So what problems are faced by utilities, including transmissions, substations and distribution systems that in theory Big Data would be able to correct? Which factors do you think are greater problems than the others?

Mark Wyatt: If you think about the broad ecosystem of the transmission and distribution network that electric utilities are dealing with, most of all the networks across the United States today are fairly aged and are in significant need of replacing, going from a more electromechanical, analog infrastructure to a more digital enabled infrastructure. And today we’re operating that system well. We’re able to provide good reliability and availability to customers, but the challenge we have is a lot of the data that we get off of the system today before it’s digitally enabled is either days, weeks or months behind when we actually get it in from the field to our engineers and other key folks within the organization.

“The challenge we have is a lot of the data that we get off of the system today before it’s digitally enabled is either days, weeks or months behind when we actually get it in from the field to our engineers and other key folks within the organization.”

By putting the digital technology out on the system, whether it’s intelligent transformers, re-closures, capacitors or smart meters, we are able to get more realtime or near real-time information from these various devices to help us improve reliability and availability, give better insight to our customers in terms of how they use our products, enabling them to have more control over how they use our products and when they use our products to help manage the personal lifestyle that they look at. It also helps us in terms of minimizing costs on the system. We have the ability to do a lot more remote activity, from a control standpoint and enabling us to get service to customers more quickly and more reliably and probably the biggest area of improvement is around how we respond to power outages in particular. We’re one of the last utilities out there where customers in most cases have to contact us as a utility to tell us when their power is out.

With this digital capability we have better insight in terms of when outages are likely to happen and when they do happen and can be more proactive in dispatching crews out and getting power restored to the customer center and having a more proactive experience. So with this massive amount of data that we’re collecting, that’s going to enable us to do those things that I have just referenced. The challenge that we have is ensuring that we bring the data back in a timely manner. We know how we’re going use the data, what tools we’re going to process against that data and how we are going to enable our engineers and customers services reps to use the data. This is a massive amount of data, and if we bring it back in a timely fashion and only bring it back when we need it, there’s value that we see to the utility of that customer. That helps to manage this overall volume of data that comes in, and it becomes less of a technical challenge and more of how we get our employees equipped to actually leverage and use that data.

“This is a massive amount of data, and if we bring it back in a timely fashion and only bring it back when we need it, there’s value that we see to the utility of that customer.”

So while we talk about The Big Data Challenge, if you manage it well, it doesn’t become overwhelming. But you really have to have good processes and good tools in place to gradually bring it back as values identified on the system.

Sanjog: Where do people have to be cautious as they go about implementing Big Data in terms of process optimization and using the right people, process and technology?

Mark: The real challenge is that a lot of the data that we get today off the analog side of our system or the electromechanical side is either days, weeks or months old, so our employees, our engineers and our customer service reps are used to taking very old data and trying to predict what’s going to happen based on older types of data. So you’re being very reactive in terms of how you use that data. The challenge we have now is our processes are not really established or set up to have data coming in on a more real-time or near real-time basis. So our employees are not well prepared, or well skilled or well equipped to know how to actually bring in that data, put analytics around it and begin to have more near real-time analysis of what’s happening.

“The challenge we have now is our processes are not really established or set up to have data coming in on a more real-time or near realtime basis.”

A good example could be as a customer calls in and requests a new service. Right now, it takes us anywhere from five to 10 days to make an appointment and set up the appointment to connect that customer’s services. The technology that we’re deploying now enables the customer to actually go out and see when we have service connection availability and schedule that appointment themselves, and it does not require a manual intervention. But all of our back end processes have to be established and set up so that if and when the customer takes the ownership of scheduling that appointment, all off our back end processes can recognize that so that we get the efficiencies out of it. So it’s more about, how do you look at all the steps and the processes knowing that you now have more real time data available versus offering very, very old data.

Sanjog: Big Data promises effective manifestation and subsequent optimization of the Smart Grid, but both are fairly young on the maturity curve. How can we make sure these two ultimately unproven technologies meet our expectations in saving us money and allow us to reach the end goal?

Mark: If you look at other business sectors, retail, banking, finance or industries that have really bridged the gap in terms of how they leverage the mass of amounts of data that they’re bringing into their business to adjust and adapt their business model, I think the tools, the techniques and the processes that they use are not different than the variables that we need to use as a utility. So we reach out on a quite regular basis to other industry verticals outside of utilities to say, “How have you addressed this problem if you had access to more granular, more real-time or near real-time information,” to begin to help us think about how we manage process.

You need to have the systems and the tools in place when this data comes back off of the Grid, and we have centralized control centers today that are set up to be able to manage and operate against that data. So the challenge for us is how do you ensure that you’re bringing it back on the timely basis, that your algorithms can process it and that you can filter only those elements that are important? Because if you think about a substation, as an example, you can bring back hundreds of data points every second associated with that substation, but in most cases, you only need four or five key points coming back that gives an engineer or a control operator what they need to make good decisions. So it’s a lot about looking at all the data elements, planning which ones you bring back, making sure the software can process it and only bring back additional data when needed.

“It’s a lot about looking at all the data elements, planning which ones you bring back, making sure the software can process it and only bring back additional data when needed.”

So I think a misnomer or a kind of a falsehood that we’re talking about now is that you have to have all of that data coming back before you can make intelligent decisions. In my opinion, you only need to bring a subset of that data back, and as you learn more about that, you can then add additional data elements as you get comfortable with those few that you do bring back.

Sanjog: What sorts of insights are we looking for from Big Data? What tangible, hard dollar value can we derive from it? What do you think utilities should be doing?

Mark: There are really two aspects that I would say we look it as we deploy Grid modernization technology: what data and information can help our employees, our engineers and our land technicians to be able to deliver service more quickly, more efficiently and more cost effectively for customers, and how do we interface with customers today? In most cases, customers call us for two reasons: (1) to initiate service or (2) if they have an issue or problem, and that issue or problem is either their power is out or they have a question about their bill.

The real value for us is giving that customer a more real-time view, of how they’re using energy on a daily, weekly and monthly basis, that helps them in terms of planning for how much money they need to budget to pay their bill. They may need to do more efficiency things in their homes, so giving them good consumption information, time of day, and providing it to them when they want it and where they want it. Do they want it on the mobile device, do they want it on a laptop, do they want it pushed to them or do they want go to a site to pull it? So it’s really around consumption information, weather information and other things about what’s happening on the system. So that really helps them get in better control. When you go back to the internal side of equation, the value that we see is we’re looking to optimize the operation of our system. We have a certain amount of energy or electricity that is generated by our base load power plants everyday, every morning, every week.

“The real value for us is giving that customer a more real-time view, of how they’re using energy on a daily, weekly and monthly basis, that helps them in terms of planning for how much money they need to budget to pay their bill.”

And if we can look to how we optimize the distribution system to take that energy into the Grid, we can more efficiently bring that energy and supply it to the customer when they need it. We also can do reliability improvements. We’ve spend a good bit of capital and a good bit of operations and maintenance expense on as set management, maintaining our assets or maintaining re-closures, capacitors, transformers and substations. So if we can see the history of how that piece of equipment operates, we can better levelize and predict when we need to do maintenance on those assets and dramatically reduce the cost of having to send folks to the field to look at doing more predictive maintenance, and also it reduces the cost of the capital to actually have to go out and replace some of those devices. So it helps on the efficiency side as well as giving the customers more transparency, and at the end of the day it’s all about cleaner more cost effective energy that we deliver to the customer.

Sanjog: What proof points do we have in showing that Big Data is a success in terms of ROI or improving operations for utilities on either side of the meter?

Mark: Great question. Let’s talk specifically about the data that we are able to get from smart meters today. If you think about where we’ve come from in terms of meters and the available information, most utilities today have two types of meters that they have installed at customers’ homes, one that’s purely an electromechanical analog meter that you have to have an individual go out and read to be able to get the information needed to bill that customer to look at a voltage issues and look at reliability issues.

“If we can see the history of, how that piece of equipment operates, we can better levelize and predict when we need to do maintenance on those assets and dramatically reduce the cost of having to send folks to the field to look at doing more predictive maintenance.”

The second evolution is something that we have an automated meter. An automated meter reading is where you go out and you install a meter where you have a one way communication mechanism, or you can drop by and use radio technology to read that meter, basically getting the same data associated with what you would get manually reading that meter. This provides efficiency in terms of how you’re actually reading the meter or reducing labor cost and improving the predictability and time of getting an accurate read. While those have served us well and we’ve been able to give accurate reads to customers, now that you have the smart meter, we can actually go out and read that meter with whatever frequency we want to. We can read it every 15 minutes, we can read it every hour, we can read it once a week or we can read it once a month.

I think that’s a very important thing to understand. While the meter is capable of being read on a very timely frequency, you need to understand the business process that you need to put around it to leverage that data. If you bring back 15 minute intervals, we’ll say three channels of data on a 15 minute interval for a typical population of about 100,000 meters, and you multiply that over a year’s period of time, that’s the same volume of data that we’ve historically collected on about four million meters. So the approach we’ve taken is this: only bring back the data that we need to accurately bill the customer, to accurately give the customer inside info in terms of how they are using their energy and also give us the ability to remotely connect and disconnect that meter as customers want to move in and move out.

Also, through the meters that we’ve deployed in various jurisdictions, we have seen substantial reductions in labor costs associated with reading meters, we have improved the reliability and predictability for getting a bill out on time and an accurate bill. We are no longer estimating bills. In one of our jurisdictions, once we’re fully deployed with all of our automated meters, we will eliminate approximately one million estimated bills a year, that means there’s one million bills that will go to the customer where they will have an accurate read and accurate account of what they owe us, which makes the conversation between us and the customers that much more predictable. They also can go out on our portal every morning and see their previous day’s usage, and they can track that over whatever period of time, and they can interact with us when they have questions about how that’s done. That’s improving the relationship with us and the customer, reducing the number of reactive calls that come to our call center.

“Only bring back the data that we need to accurately bill the customer, to accurately give the customer inside info in terms of how they are using their energy.”

Last but not least, when the customer wants to move in and move out of a particular facility, sometimes it takes five to 10 days to schedule and do that connect or disconnect. Now that customers can go online and call a service rep within whatever period of time they want to have the connection and disconnection, we schedule it to their convenience. We no longer have to send a takeout, and that meter can be connected and disconnected literally within a 15 to 30 minute period of time. Your satisfaction goes up, and we’re reducing labor cost by not having to go out and read that meter. Multiply that by the volume of meters and reasons that we have, without getting in to specifics, you really get hard dollar savings, you get customer satisfaction and it improves dramatically.

Sanjog: When it comes to mission critical decisions, do you think we are at a point where Big Data can actually be trusted, because things could go wrong at the time of collection of data, the processing and finally at the interpretation stage too?

Mark: The short answer is yes, we are well positioned to do that. Once you get enough volume of new infrastructure to pull it, let’s say a critical mass of smart meters or a critical mass of distribution automation devices, where you’re actually able to start collecting the data on whatever frequency you need to, if you think about the technology enablers today, whether it is the hardware, the software or the telecommunication’s network, we have the raw processing capability and the power to bring that data back on a very timely, reliable basis. Do you need five nines on the reliability on the data? Do you need four-nines reliability? At least from the Duke perspective, and on the telecommunication side, instead of us continuing to build out our own private networks, we are relying on publically available networks like AT&T and Verizon, so we’re leveraging those networks and they have a higher degree of liability.

“We have the raw processing capability and the power to bring that data back on a very timely, reliable basis.”

The real key is how often, how soon and how quick do you need to bring that data back? If you schedule it appropriately and then you bring it in and you have it aggregated, it helps us better predict where we’re going to have low growth issues and where we’re going to have picking issues on the system. And so we can bring this data back, give it to engineers, give it to planners, and they get to feel some effects of what’s happening on the system. The other real key to us, and we’re not there yet, is the system is able to do it, but we’ve got a cultural change around. For a lot of the tools that we use today when we go bill out a new piece of infrastructure or we go and put in a meter on the customers home, we have automated systems to capture the activity and come back and include it in our system, but there is a natural delay based on human interaction when you’re in each step of that process.

This equipment can self-discover; we know when a meter is set, and we know when a transformer is installed. If you think about the analogy of when you get a new cell phone, when a cell phone comes in from whatever provider that you have, you get instructions to tell you how to activate that cell phone and transition it to a new plan. These systems have the ability to self discover and bring it back into our base system so that you take the human interaction out of it. The quality of our data improves and we take care of a lot of manual interaction. But then again, you’ve got to be very mindful. Big Data is not as big as most people think. You can bring back hundreds or thousands of points, but you need to limit the points that you bring back when there is a business value to use it. And if you do it well, it does not overwhelm your system, it does not overwhelm your employee, and you have a much higher probability of getting value on that data.

“Big Data is not as big as most people think. You can bring back hundreds or thousands of points, but you need to limit the points that you bring back when there is a business value to use it.”

This equipment can self-discover; we know when a meter is set, and we know when a transformer is installed. If you think about the analogy of when you get a new cell phone, when a cell phone comes in from whatever provider that you have, you get instructions to tell you how to activate that cell phone and transition it to a new plan. These systems have the ability to self discover and bring it back into our base system so that you take the human interaction out of it. The quality of our data improves and we take care of a lot of manual interaction. But then again, you’ve got to be very mindful. Big Data is not as big as most people think. You can bring back hundreds or thousands of points, but you need to limit the points that you bring back when there is a business value to use it. And if you do it well, it does not overwhelm your system, it does not overwhelm your employee, and you have a much higher probability of getting value on that data.

Sanjog: How will our utility challenges compound, and how are we supposed to keep ourselves at pace with the things changing around us?

Mark: While a lot of the capability or a lot of the technology we’re deploying today has the ability to bring back large volumes of data, the key is only bringing that data back when you need it and leveraging it when you need it. I think a lot of companies, utilities and non-utilities, say, “Let’s bring all of it back, and we will figure it out over time how we leverage it.” And that overwhelms the IT department, that overwhelms the CIO, and they are now busy trying to bring back and manage terabytes or petabytes of data, and my history tells me that less than five percent of that data is ever truly used.

“It is not as big of a challenge to the technology organization, but more of a challenge to the business organization to say, “How do I turn that into real value?”

So you bring back a massive amount of data and there’s expense associated with processing it. You’re overwhelming the system, and your employees and the users of that data don’t know how to filter through it. So it’s more about putting discipline up front in terms of telling me what data you need, how you’re going use it, how are you going to pay for it and how you’re going to turn it to a useful insight. Once you do that in a very deliberate fashion, you’ll pick up momentum and then you’ll start bringing back more and more of that data. And if you plan it that way, it is not as big of a challenge to the technology organization, but more of a challenge to the business organization to say, “How do I turn that into real value?”

So I really hope that we’re starting to balance the conversation around Big Data. It’s not as scary as most people think it is. If you put good, solid business discipline around it up front, you can manage both the technical and the business side of it, and it will not be nearly as difficult as most people play it out to be.

“I really hope that we’re starting to balance the conversation around Big Data. It’s not as scary as most people think it is.”

Contributors

Mark Wyatt

Mark Wyatt, Vice President for Grid Modernization and Distribution, Duke Energy

Mark Wyatt serves as vice president of grid modernization for Duke Energy. He is lead executive for the company’s grid modernization function, which will deliver enhanced operational efficiencies for the company’s transmission and distr... More   View all posts
Add Comment
Click here to post a comment

Advertisement

Cigniti-Kobiton-Webinar-364776190-mobile-centric3-MPU
Mark Wyatt

Login


Not Member Yet?
Register

Register

  • Name

  • Contact Info

  • About Yourself

  • Minimum length of 8 characters
  • Upload
  • Location

  • Professional Background

  • Other Social Profiles

  • Areas of Interest