Skip to content

Smart Meters, Dumb Blackouts?

Smart load limits could help to keep the lights on through California’s wildfire season.

Today’s post was co-authored with Duncan Callaway.

California’s 2020 wildfire season is off to a blazing start. The burning ring of fire around the Bay Area is now mostly contained thanks to the super-human efforts of our firefighters. But we’re not out of the woods. Not even close. 

Although the current mega-fires were started by lightning, the most destructive fires in recent years have been caused by power lines or other electrical equipment.

One way to mitigate this kind of wildfire risk? Cut the power supply when the red flags go up. As we write this blog, the flags are up.

California’s largest utility is now warning of potential Public Safety Power Shutoffs (PSPS) to reduce the possibility that power lines and other equipment could spark new fires. Back in 2018, PG&E was sharply criticized for not de-energizing power lines prior to the devastating Camp Fire. Since then, utilities have been much quicker to call PSPS events: 

These De-Energization Data show a striking increase in PSPS activity.

These PSPS events have been controversial because power outages can impose high costs on homes and businesses. Last year’s outage costs were estimated at $10 billion. That’s a big number! But over time, these costs should come down as we adapt and adjust. 

There’s been plenty of discussion about adaptation via investments in back-up generators, micro-grids, battery storage. But what about re-thinking the PSPS events themselves? So far, PSPS events have been implemented using a blunt tool: planned outages.

PG&E 2019 PSPS map. Blue areas = planned outages
You know it’s an emergency when Cheeseboard Pizza goes dark.

This all-or-nothing approach leaves much to be desired. Last year, we were frothing our lattes and using our clothes drier while our neighbors a few blocks away went days without power. Ever since we’ve been wondering whether there’s a better way. 

Utilities in other parts of the world with frequent outages have been experimenting with better ways. One intriguing example can be found in South Africa. When there’s not enough supply available, this utility sends SMS messages to inform customers that they will be limiting load. Smart meters are used to detect which households are consuming below the prescribed limit. Households that comply remain connected. Non-compliers get remotely switched off (presumably with a chance to get back on when they reduce their demand).

Some California customers might balk at a request to limit load for days on end. But a “please limit-your-load” alert should be much preferred to this alert on your phone:  

Could load limits help California keep more essential loads connected through PSPS events? This depends on (1) whether there’s any power available to ration during PSPS events and (2) whether California’s smart metering infrastructure is up to the task. To get to the bottom of these questions, we’ll offer our spitball take. And then we’ll invite you, brilliant blog readers, to tell us if we’re on the wrong (or right) track.

1. Is there power to ration during PSPS events?

Let’s start with a cartoon illustration of how the grid is configured: Customers connect to distribution lines that connect to substations. On the other side of those substations sit transmission lines that criss-cross long distances to bring in power from remote sources.  

In distribution systems, there is typically only one path between a customer and their substation. If a distribution line is de-energized due to wildfire risk, there will be no power to ration. Everyone on that line is just 100% out of power.

On the other hand, in transmission systems, there are usually multiple paths from point A to point B (i.e., the system is “meshed”). If one transmission line needs to be taken out of service, but there is an alternative path for power to flow, we can still keep substations powered, along with the distribution systems they serve. If that alternative path can’t support all the power people are demanding, rationing could help to keep the proverbial lights on.  

So our question is…are PSPS events sometimes/often caused by de-energizing transmission lines?  

We think that transmission is definitely at play, though we don’t know how much of it is in meshed parts of the grid. Back in January, PG&E wrote that about 138,000 accounts in Northern California could have kept their lights on if they had installed backup generation at their substations. This means PG&E considered the distribution systems serving those customers to be safe. Furthermore, PG&E has reserved about 450 MW of temporary backup generation to keep customers powered in the 2020 fire season. Again, any distribution systems they power with those MW must not pose a risk if PG&E intends to power them. It must be risky transmission lines that are putting those customers in PSPS territory.

 2. How smart are California smart meters?

Even if there is potential for using load limits to keep essential loads afloat through PSPS events, this idea is not worth pursuing if the implementation costs are prohibitive. So this brings us to the next question. Could we leverage investments we’ve already made in smart meters and customer notification systems to implement load limits at relatively low cost?

We know that California utilities can remotely turn smart meters on and off. We also know that utilities can alert customers when a power outage is imminent. One of us is so excited about AMI that he’s read the Landis and Gyr manual for our PG&E meter cover-to-cover. This manual suggests our smart meter could have a “soft fuse” function. This would allow PG&E to program the meter to cut us off if our demand exceeds an amount that can be remotely adjusted. The meter could also restore power periodically to detect whether we’ve turned off enough things to keep us below the prescribed limit. 

The upshot? With some software development, existing AMI infrastructure could be up to the emergency load limiting task.

What are we waiting for?

Load limits are not the most elegant approach to allocating limited power supply. Compared to real-time pricing, load limiting is a blunt instrument. But it could offer a relatively low-cost improvement over the on-off switch we currently flip when PSPS events are called. Load limiting could complement some of the other investments utilities are already pursuing under the banner of wildfire adaptation, such as microgrids and grid sectionalization. It could also come in handy when electricity demand exceeds supply for other reasons, at least until we’ve made more progress on incorporating demand response. 

We are part of a larger research team that is working to understand how power system investments and operations should respond to mounting wildfire risk in California. Load limits seem like potentially low hanging fruit in this respect. If you disagree, tell us why! And if you share our nascent enthusiasm, let us know!  If this idea has potential, our team is game to help figure out how to implement it.

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Fowlie, Meredith. “Smart Meters, Dumb Blackouts?” Energy Institute Blog, UC Berkeley, September 8, 2020,


42 thoughts on “Smart Meters, Dumb Blackouts? Leave a comment

  1. I heard that the communications network used for AMI by the CA IOUs is relatively poor. I wonder if it will support the real time demand measurements needed to implement load limiting. Right now it does a very infrequent reading of the meter (hourly or daily, I think) for the same reason.

  2. There’s a different engineering approach for some of the TOs. PG&E did lots of “loop” transmission to provide some reliability, but that is now getting segmentation to they can cut power within some narrower areas that have the elevated fire risks. SCE, on the other hand, does more “hub & spoke” design, which means they could already isolate smaller elements. So that’s where PSPS impacts come from when transmission or subtransmission voltages have to be dropped for safety.

    That may not help when T lines have to get derated due to fires/smoke that’s nearby (or impacts them upstream).

    Supply sufficiency is whole different issue. With respect to load limiters via AMI for resi/small commercial–programs must be voluntary for various reasons, including the equity one. And while AMI could have been rolled out with dynamic retail rates, my take is that there’s a lot of regular resi customers who would see that as just a “utility ripoff” if there isn’t a non-CARE rate option that provides flat price or the static TOU periods. Then there’s the high usage surcharge based on climate zone averaging–which is some case like foothills would place someone with a load that would be “average” in a suburban setting “high usage” in Zone Z (PG&E) for example.

  3. This is a great idea. More broadly, we need of more of this approach of looking at the investments we have already made or those that we have the foundations to build on quickly rather than constantly looking for new systems and infrastructure. Right now CEC wants to create ANOTHER way of sending price signals over FM RADIO, rather than leveraging the smart meter investment and rather than leveraging all the work on DR potential and development of DR technologies. Our state gets tired of new toys about as quickly as a kid on Christmas morning eyeing the next package under the tree.

  4. I’m afraid I need to be a naysayer regarding this analysis. Public Safety Power Shutoff should not be conflated with requests for or restrictions on power delivery during periods of peak demand/resource inadequacy stress. They are two entirely different tools in the toolbox of reliability and resiliency. PSPS events are employed when there is a high threat of winds knocking down wires or tree/line contact. De-energize and reduce the threat of ignitions. It is very effective…too effective by analysis because the utilities do not count as a cost the social consequences (that $10 billion cited). It defeats the de-energization to maintain the line to provide power in the manner suggested –“Smart meters are used to detect which households are consuming below the prescribed limit. Households that comply remain connected. Non-compliers get remotely switched off (presumably with a chance to get back on when they reduce their demand).” A downed live line can still cause a fire, even at lower load.

    This may well be a good solution during the Flex Alerts and system emergencies called by the ISO as experienced in August, but not for the PSPS events of this week, or last year.

    Now, it is true that the Commission is trying mightily to reduce or eliminate the use of PSPS, only as a “last resort”. And the Commission is mightily trying to get the utilities to factor the social costs into their calculations. But the fact of the matter is last year PG&E documented the avoidance of 700 risk events that could have started fires, if not for PSPS.

  5. Such an approach will quickly bog down once one begins to grapple with some of the details.
    Equity is a big one. How do you establish the limit for different size houses and households? Do you base it on past usage or not? How do you establish the limit for different climate zones? What may make sense in Berkeley and South Africa may not in Stockton or (either) Brentwood.

    What time base will be used to determine a violation? A short time would be abusisve to those homes will variable loads (like all-electric homes or those with A/C); a short-time is also not necessary becasue of the diversity. A longer-time may make it too slow in responding.

    How will you treat homes with solar? Do you really want to cut them off when they will likely be adding to capacity at times? Do they get judged by net or gross consumption? Should those with battery storage be offerred incentives to use it in the public interest?

    While the general concept may sound attractive, I doubt that any specific implementation (i.e. answers to those questions) would be seen as a fair and workable consensus for a mandatory program. I could, however, see something working as a voluntary program.

    • Why focus on load limits when these meters were designed to support critical and real-time pricing. Price response eliminates the equity issues you’ve identified with utility-determined load limits.

      • I would agree that pricing options are superior to mandatory load limitations. Market forces are more likely to maximize load reduction with a minimum of outcry. Perhaps even letting customers sell power at CPP rates.

        • Thanks, Max, and everyone else for the great comments. Pricing options may be superior to mandatory load limits in many respects… but are they politically palatable/possible as a response to long PSPS events? Our thinking was that, in the case of a longer PSPS-induced outage, relying exclusively on price-based incentives could have some unacceptable equity implications – real or perceived- if higher-income households have no trouble paying to satiate their demand while income-constrained households go without for extended periods. Load limiting could offer an improvement over current PSPS practices – outages- and be preferred to RTP on equity grounds. That said, load limiting and real-time pricing could be construed as potential complements (versus substitutes). Devil is in the details, of course. Thanks for all the comments which are thought-provoking…..

          • The political tradeoff between mandatory utility initiated demand reductions and price response is very simple. Price response provides customer control which if insufficient can always be followed by utility mandatory demand reductions.

  6. I do not think this idea has legs. The transmission caused PSPS events should be a small subset of total events. If the utilities and CPUC were following current rules the risk of transmission line caused wildfires would be extremely small. The risk factors for distribution lines is much higher. So this solution only addresses a small part of the problem. Implementation would also be an issue. Determining the load limit would be complex and need to consider a number of health and safety issues that vary by household/meter. For example, while there is a database of medical equipment reliant on electric power, its accuracy is far from certain. The politics would make the CPUC proceeding daunting and I doubt the Energy Division would be willing to expend the resources necessary develop and obtain approval for the required regulations.

    On the other hand, if the implementation issues could be resolved, this idea could be a stepping stone to offering customers price differentiated levels of reliability. A major step forward in the efficiency of resource adequacy.

  7. Perhaps to address fairness issues, meters could be set in those tight times to limit each meter to say 2 kWh/hr inward flow (measured across the full hour, so as not to be a fast circuit breaker that interferes with electric cooking). It would encourage pre-cooling and moving EV charging, pool filtering, laundry, dishwashers etc out of the tight period. In sunny tight times ( like the recent days from 3-6 pm) the solar customers have a lot of slack since their behind the meter solar also reduces inward flow across their meter. I don’t think enlightened fairness would require the crippling of solar and shooting ourselves in the foot. In fact these meter inflow limits could add another value to the solar value stack and therefore free up solar value space to allow NEM reform that will help non-solar customers.

  8. The idea of using smart meters in real-time as a form of load limiter (Eskom) is one I’d not heard before. I’m sure there are dozens of creative solutions I’ve not heard.

    One example: When Burbank first activated their AMI system, they ran a program called SmartSaver, to combine data from their AMI system (interval data by meter) with data from their GIS system (which meter is connected to which transformer) to produce heat maps of each transformer on the system. They were able to identify transformers that were undersized, and upgrade them, avoiding outages in a hot spell. They also were able to downsize oversized transformers, reducing core losses. The line loss reductions alone paid more than half the cost of the AMI program.

    I’d love it if others would post examples of creative use of AMI data that we can all learn from.

    • This was the type of savings that we were pitched by PG&E, SCE and SDG&E back in 2006 from investing in AMI. And what have gotten from this? Yet another boondoggle in which the utilities tell us that, by the way, they haven’t gotten around to delivering the benefits that they used to shake us down for the investment in the first place. The problem isn’t in the technology–it’s in the incompetence of the managers who can’t deliver on their promises.

      • The fact is that the three IOU’s all objected to implementing smart meters. However, the CEC business case clearly identified clear overwhelming economic and operational benefits that the CPUC supported. The statewide pricing pilot likewise made an overwhelming case for price responsive demand reduction. Unfortunately, the CPUC failed to mandate and provide the followup ovesite to make sure the IOU’s actually implemented the pricing, billing, and other benefits included in the metering initiative. While IOU management resisted the related metering changes, it is the CPUC that failed to exercise its authority.

        • I think TURN would not agree with your characterization with who made the case for AMI–the IOUs saw $$$ in ratebase investment return. The CPUC was skeptical of the business case unless the pricing benefits were included. The IOUs claimed that the pricing was unneeded to create the benefits because they would gain “efficiencies” that have yet to appear in a substantial way.

          • Richard
            Regarding your latest comment, I disagree from a clear point of experience. I wrote the CEC smart meter business case. On behalf of and with authorization from the CEC and CPUC, I negotiated with the utilities, TURN, AARP, and numerous other intervenors to identify issues and create a path for implementation. As for the benefits, the business case clearly identified operational, pricing, demand response and numerous other benefits that all combined to compel implementation. Point of fact, the IOU’s did object to implementing smart meters, but as you know the IOU’s generally object to everything. With regard to pricing, my experience is that the CPUC and IOU’s competed neck-to-neck to see who could make the most convincing argument to NOT implement any form of dynamic pricing. So on this argument, you need to go back to the drawing board.

          • What happened at the CEC and then at the CPUC was interested case of how the IOUs first resisted a concept (I’m working on another where the utilities are converting privately owned MHP systems to utility ownership) and then took significant ownership in promoting when they realized how much money they could make. That’s what happened with the AMI systems. You were successful in one way (and I participated in those studies as well), but then the utilities took control in the rate cases and put in the gear to make money but left out the features that would benefit ratepayers.

          • “You were successful in one way (and I participated in those studies as well), but then the utilities took control in the rate cases and put in the gear to make money but left out the features that would benefit ratepayers.”
            I actually agree with you on this point, however I hold the CPUC, not the utilities, responsible for abrogating their duty to follow through on the CEC-CPUC effort to put in place a system that would address and prevent unnecessary demand-supply imbalances.

          • And I’m not sure that we’re disagreeing per se–it’s about what benefits were included in the business cases at what point in the regulatory process. You were at the outset when direct consumer benefits were included, and I was focusing on the implementation stage when the IOUs tried to make the case (which I’m not sure is valid) that just having the grid management benefits would be sufficient.

  9. Whether the existing inventory of California IOU smart meters have the capability to remotely limit demand is somewhat irrelevant.
    All of the existing smart meters have the capability to support critical peak and real-time pricing. Both rate forms use price signals that have proven very effective in motivating customer demand reductions. The business case developed by the CEC to justify smart meter implementation included price response benefits. No new hardware investment is needed, just get the CPUC to finally authorize implementation.

    • The utilities already have the authorization in CPP and demand response programs. It’s that the utilities are unwilling to compromise on important management issues for those programs that are impeding their adoption. Whether customers will be sufficiently price responsive when the market value of the generation component is only about one-sixth of the total rate is another big question.

%d bloggers like this: