There's been lots of talk lately about how utility business models need to change. And as the distribution end of the electric grid gets more complex, we're seeing a lot of interesting point solutions emerge for addressing some aspects of that complexity.

But these various point solutions will need to be integrated into a holistic system. And while the utility business model does need to change, that doesn't specify what will need to happen at a technology level. So what will the overall distribution grid look like when it is eventually modernized? Here's an overview of the picture I'm working off of these days, for what it's worth.

First, a couple of basic principles.

The distribution end of the grid is becoming much more complex. We are introducing distributed and community-scale generation, energystorage and even new significant loads (EVs, for example). This could potentially be quite destabilizing for the substation-and-upward utility at high saturation levels, hence the (in my opinion, much overwrought) concerns about rooftop solar. But these new factors are all potentially networkable and thus controllable, or at least able to be monitored and adapted to in real time.

The utility distribution grid workforce is becoming older over time, and is also under increasing pressure from more erratic and severe weather events. Backfilling the workforce accordingly would be extremely expensive without relying a lot more on automation. So there will be more automation.

Finally, as the overall utility grid operator is increasingly under pressure to keep everything balanced while attempting to manage an aggregation of increasingly variable yet oft-correlated distribution nodes, they will turn even more to pricing and incentive schemes that encourage load stabilization at the node level, as well as capacity-on-demand when needed from one node to another.

By "node," of course, I'm referring to the distribution grid from the distribution substation through to the customer premises. It's oversimplified (for one thing, the right scale may be a geographic cluster of such "nodes"; I'm just shorthanding here), but for the purposes of this column, it's easiest to think of the utility as managing the supply of power (i.e., both capacity and transmission hardware) to a collection of such nodes, each of which isn't connected to one another, only to the utility grid. The utility owns some of the node, the customer owns the edge of the node (e.g., the actual demand-producing equipment), but we need to increasingly think of them as an integrated whole, connecting intranode via networked automation as enabled and incentivized by utility-framed economics and control.

So with all that as background, what will the grid eventually look like?

The microgrid

To the extent practicable, each building becomes a grid-tied microgrid, with on-site distributed generation and load control, and, in the larger buildings, storage. "Microgrid" means a lot of things to a lot of people, but for me what it implies is that the building is (again, to the extent practicable) self-sufficient and under a single, intelligent control system.

Of course, many buildings will not be appropriate for solar on the roof. But where natural gas is available and the need for thermal energy is relatively significant, combined-heat-and-power (CHP) systems will be utilized. CHP systems, in fact, have an advantage over solar DG in that they can be fired when needed, whereas with solar, the sun shines or it does not (hence all the interest lately in storage). A rare few buildings owners will even opt for other forms of distributed generation such as solid-oxide fuel cells or small-scale wind. Point being, there are a wide variety of DG options that will be increasingly placed on-site. But at the same time, not every building will be "net-zero energy" or such. That's OK, but overall, there will be a lot more generation capacity at the building level.

But whether or not the facility has distributed generation, it will have a networked load monitoring and control system (NLMC) integrated into as many intelligence-enabled loads as possible within the building. What are loads? Same as always (with one big exception). Lights, air conditioning and heat, industrial equipment (in the case of commercial & industrial, or "C&I"), appliances, and entertainment/IT. For the most part that's about it, which seems simple, but it actually hides a huge amount of variation in terms of what loads look like, how readily they can be adjusted as needed, and how they should be prioritized.

The goals of this NLMC are to 1) keep the building's overall impact on the node as stable as possible (with few spikes and drops in net demand); and 2) enable sharing excess capacity back to the overall node when called upon to do so (often described as "automated demand response," or ADR).

So let's break out C&I from residential-type buildings. C&I buildings, of course, come in all shapes and sizes, but what they all share in common is that they are all operated for a purpose other than just habitation. This means there are times when you need to run a piece of equipment even if the electricity cost is high. This is simplest to think about in office buildings, where there needs to be sufficient light and comfort and plug-level capacity for me to sit here on a cloudy day and bang out this column on a computer in a well-lit room without sweating all over the keyboard. But it gets really complicated when you take that same principle to a metal foundry or a food processing plant. If the goal is to stabilize facility impact on the node, this is hard to manage in such environments, even before you put solar on the roof and have to deal with what happens when clouds pass over -- much less addressing how to control an on-site battery.

Thus, what is needed in C&I is a robut NLMC that can handle a really wide variety of loads, integrate into whatever DG is on-site, even integrate into on-site storage. Able to handle not just simple HVAC controls, but a huge variety of industrial loads as well. And to smooth it all out on a real-time basis, via an ability to control loads in a buildings-owner-specified prioritization scheme. And to utilize that same prioritization scheme to feed capacity back into the node for ADR with minimal impact on facility operations. One single system that can be rolled out across just about all C&I facilities, so there's a consistent solution for the utility to tie into. I'm biased of course, but I think this unnoticed development from Powerit Solutions is actually a really big deal in this regard. 

On the residential side, it's a similar story, except that the loads are smaller and even more varied. There's also the need for an NLMC for the same purpose, but it will need to be super-streamlined and standardized so that the "smart home" can plug right in and start performing these energy-related activities. Sorry, Apple, but I think this is probably best done via an open standard so as to capture as many homeowner-purchased devices as possible. But overall, the goal of the NLMC-connected smart home will be the same: to present a very stable load profile to the node, while providing ADR capacity.

The mesogrid

To the utility, stabilization at the individual customer site is a nice goal (and may be encouraged by shifting to pricing even residential customers on a demand-charge basis in addition to total consumption). But stabilization within a full node is even better. Thus, coordination across meter connections -- so that one customer in a node can cover for another customer in the same node -- will eventually be implemented.

This is a big change from where we are today. It requires very specific information on each meter and facility within the node on a real-time basis, and some control capability to trigger actions inside the meter. Right now the closest is full-territory ADR, which is done more broadly and most often only on very selective loads. But as more capabilities get pushed into the substation and distribution grid in general, utilities will be able to leverage NLMCs for this real-time intra-node balancing act. Eventually, it might take the form of credits on individual facilities' monthly bills, where the credits are paid to those whose NLMCs bid in capacity from lower-priority loads as configured by their individual facility managers. Set the parameters in your NLMC (this load is moveable, for this amount of time, if I'm paid this amount per kilowatt) one time, and the system runs without need for managers to intervene. Basically, the intelligent substation controller (whether embedded in the actual hardware at that level, or more likely applied across the utility enterprise) serves as master to a lot of opted-in, incentivized "slave" NLMCs.

Looking at this same meso-scale, we also can now include community-scale generation (mostly solar farms and, less frequently, smaller wind turbines) as a supplement to DG at metered facilities. Not only does this allow those building owners with poor onsite DG options to "get in the game," but it also allows each node to come closer to being not only stable, but net-zero demand on a run-rate basis. Which of course means there is more available capacity for ADR when capacity is needed elsewhere across the utility territory. Furthermore, with intelligent node "master" devices now in play, if the node is sufficiently large (admittedly, this is often not the case right now), the entire "mesogrid" could possibly remain up even if the backhaul utility grid goes down for some reason.

Energy storage

There are important roles for storage across the entirety of this concept -- storage on-site at buildings, storage at the community level, storage directly connected to DG. However, as you can see by now, that storage needs to be thought of in the context of all of the other ways to intelligently deal with loads.

Shifting a low-priority load is cheaper than utilizing storage for peak-shifting purposes. Storage on-site in this context, when every facility is still tied to the central grid as a source of "backup" capacity, really makes the most sense when it is sufficient to perform on-site uninterruptible power when the grid goes down (as will increasingly happen in many regions). Right now, using batteries alone for this is highly uneconomic, as you would need way more battery capacity than you would need for 99.9 percent of other times. But incorporating on-site storage into the NLMC allows for simultaneous shutting down of all non-mission-critical loads and firing up smaller batteries so that you could get useful (one to two hours) backup power out of onsite storage systems, even without DG. And with DG, the NLMC would allow these smaller storage systems to keep their microgrids up for a whole lot longer (which might be very important, say, after a hurricane hits).

But this means the design of the controls must be led by an understanding of how to shift loads, rather than starting with the battery. The battery is probably the easiest part of that equation to control. To make on-site storage economic in most circumstances will require an NLMC that can work with the wide variety of heterogeneous loads it will need to control. Otherwise, you leave a lot of the less expensive capacity untouched and you need to make up for it by over-installing expensive battery capacity. To date, most of what I've seen in the marketplace is emerging battery-specific controls. In my opinion, this is insufficient.

Implications

So what does this vision require to succeed?

  • A short list of NLMC systems that utilities are able to plug in to and that can comprehensively cover load shifting, DG reaction, storage release/charge, and ADR functionality. Just having one vendor for these systems will probably be insufficient, because of the many types of customers out there, but a short enough list of broad-category players that are familiar to utility systems. 
  • Utility distribution automation, enabled by intelligent "node" devices (at the transformer level most likely) and smart meters for real-time monitoring.
  • Changes to the current utility revenue model. This has been covered better by others elsewhere, but simply put, utilities will need to earn revenue by managing the grid, not selling centralized-generation kilowatt-hours. New open standards will be required to enable new on-site hardware to easily connect with and be prioritized/controlled by the applicable NLMC.
  • Strong new security to protect all of this intelligence-enabled automation from hackers.

Another implication here is that intelligent controls can and should put an end to this whole "utilities versus rooftop solar" debate. Or at least the "It'll destabilize the grid!" side of that debate. The revenue considerations are still real, but intelligent load control under the roof should be able to go a long way toward compensating for variation in generation conditions on top of the roof, cheaply and with a lot of other system benefits.

This is just my own vision, and I've left out some of the details and implications. But I offer it up in hopes of kicking off further conversations.