Saturday, February 27, 2010

New Epiphany: Smart Grids Require Real-Time all-IP Networks

It is clear to me now that Smart Grids require real-time all-IP networks and management tools.

We started our journey with the use case that customers would need meter data 24 hours delayed and that we would collect it every hour or so (with the most aggressive case being every 15 minutes). We also had the use case for a few amount (many hundreds) of sensors for Distribution Automation.

As the last 24 months have shown, the market has been accelerating and it is now demanding possible capabilities around Time-of-Use pricing, and Real-time pricing plus the addition of advanced demand response and advanced load management under the same network (i.e. more sensors for Distribution Automation, smart appliances, electric vehicles, solar PV panels, and energy storage devices).

So, if you are thinking about building for the minimum use case, like we did five years ago, please stop and re-assess. Typical Smart Grid / AMI networks using unlicensed spectrum and limited spectrum can only move data from 20 kbps to 150 kbps at the very most (note: 2G and 3G public networks with licensed spectrum have the same problems). Would narrowband speeds be enough for collecting meter data every 15 minutes and real-time data from a limited number of sensors? Sure. We are managing fine our 410,000 meters and 1,000 Distribution Automation sensors. But we have very tiny growth room to add more Distribution Automation sensors, smart appliances, electric vehicles, solar PV panels, and energy storage devices..

So what would you do if the meter data needs to be collected every 5 minutes, or every minute, or real-time? And what would you do if you realize that adding more Distribution Automation sensors is one of the most powerful Smart Grid benefits and need to now read thousands of sensors real-time? And what would you do when you have to deal with thousands, or tens of thousands, of solar PV panels, electric vehicles, smart appliances, and energy storage devices on the customer side of the meter to support advanced demand response and load management programs?

And have you considered the additional data bandwidth requirements to have the adequate cyber-security, the adequate network management performance tools to maintain the desired service level agreements, and the needed quality of service guarantees?

An RF-Mesh with limited IP support, unlicensed frequency, limited spectrum, limited cyber-security, and no guarantees in quality of service is not the right answer for the new use cases. And 2G and 3G public networks with limited spectrum, limited cyber-security, and no guarantees of quality of service are also not the right answer either.

On the other hand, Fiber, 4G wireless (Wimax and LTE), and Broadband Over PowerLine technologies are the better answer. 3G with HSPA 7.2 and 3G with EVDO Rev A might be adequate for small Smart Grid networks. Any current and new investment should be as future proof as possible!

If we were collecting real-time data from the 500,000 devices on our network, we would be generating about 40 petabytes of data per year from 100 terabytes today. Of course, we would have moved away from the relational data base model to run operations and moved to the time-series data base paradigm by then. Even with that improvement move to time-series databases, the amount of data that we would need to collect and keep might be close to 10 petabytes annually vs. 40 petabytes. That amount of real-time collection, analysis, and decision making can only be achieved with a real-time all-IP network.

If you are in the middle of deployment, you will need to find an upgrade strategy sooner rather than later. If you decided but not yet deployed your Smart Grid / AMI choice, you still have time to switch to the right technology and partners. If you have already made your decision and deployment, your partner(s) needs to give you an upgrade path at a very reasonable price.

I want an upgrade path to a real-time all-IP network with the adequate cyber security, the needed quality of service guarantees, and sophisticated network management performance tools to ensure my required service level agreements.


Andy Bochman said...

I hope this message, which has been simmering in recent months, gets out to all the folks who need to hear it loud and clear ... and fast!

Todd Hoff said...

100 percent correct. When I sat through the smart grid presentations and talked to the representatives I was floored. The grid was dumb. It was nearly a batch system. No feedback loop of real-time cost data could be fed back to the users while they are using the resource, which makes the system much less useful. It's such a shame to do so much work and spend so much money for something that was out of date 5 years ago.

Trung Le said...

I definitely agree with you about the need for an universal IP network (IPv6 in particular).

IP standard might not directly deal with the data flood problem, but it's the most practical and surest way to build a scalable information layer for the smart grid that facilitates interoperability and network management (think about how many computers and network devices are running and being managed on the Internet, the biggest IP network of all). Take a step further, we would also want to be able to not only identify any single device in the smart grid but also navigate from one to another (think about the World Wide Web and how useful hyperlinks are).