Praxair to build new world-scale SMR hydrogen plant in Louisiana
BMW Group electrified vehicle US sales down 42.7% in February; X models account for 55.5% of BMW brand sales

Study finds twin demand peak for residential time-of-use EV charging; implications for grid operators

A researcher at the University of San Diego has used energy meter-level data from the San Diego region to analyze the energy load profiles of residential customers under the time-of-use (TOU) rate with and without EV charging requirements.

Unlike previous forecasts on the effects of EV charging loads, the energy load profile of TOU customers with EVs shows a “twin demand peak” where there is a peak demand during the evening hours and another at midnight. The results, published in a paper in the journal Energy Policy, reveal potential issues for grid operations with greater EV adoption and the importance of careful TOU rate design.


City of San Diego average hourly load for the EVTOU residential rate group (Q1: Jan–Mar, Q2: Apr–Jun, Q3: Jul–Sep, Q4: Oct–Dec). Kim (2019)

Governments around the world have taken direct action to promote the adoption of the plug-in electric vehicle (EV) using new legislation, tax incentives, and other policy instruments. California’s Governor Brown has initiated the zero-emission vehicle (ZEV) mandate that targets the deployment of 1.5 million ZEVs by 2025. Rapid integration of such a large number of EVs will inevitably cause uncertainty and variability on the operation of the existing electric power system. There is high uncertainty on not only the speed and scale of EV adoption but also the EV energy and power requirements. Furthermore, there is an additional unknown human factor in the anticipated mass adoption of EVs that clouds the forecasted charging load requirements: when will EV drivers actually charge their vehicles?

These uncertainties have contributed to a condition where there is no clear roadmap on building the appropriate EV charging infrastructure that must be strategically constructed to foster a rapid, seamless transition to transportation electrification.

… The vast majority of the EV charging is expected to take place at people’s homes (i.e., residential). Therefore, the effect from residential EV charging (L1 & L2) is expected to be significant in both the total energy load and the shape of the load profile. One key takeaway from this study is that the forecast of EV charging loads primarily hinges on the assumption that EV drivers charge their vehicles based on travel patterns. That is, most EVs begin charging when EV drivers arrive at home after work during peak hours (approximately 5–8 p.m.). This study attempts to shed light on this assumption using actual meter-level energy data from EV owners. Strong evidence against such charging behavior assumption would have significant impact on the validity and accuracy of the EV charging load forecasts, which would consequently have an impact on energy policy.

—Kim (2019)

For the study, Jae Kim focused on the San Diego Gas & Electric (SDGE) service territory. SDGE is one of the three major investor-owned utilities (IOU) in California and supplies power to roughly 1.4 million businesses and residential customers in a 4,100 square-mile service area.

There are roughly 30 different billing rate groups within the residential customers in the SDGE service territory. The majority of the residential customers fall under the “DR” rate group—the default for residential customers. The DR plan is a tiered-pricing system with increasing rates with increasing usage. Each customer is given a monthly baseline usage allowance based on their location. If a customer exceeds the baseline allowance by a certain threshold, then the rates increase from Tier 1 to Tier 2. There is a range of allowable usage within Tier 2. Once a customer exceeds the Tier 2 range, then the rates increase to the high usage charge (HUC) level.

The baseline allowance and the high usage charge threshold vary depending on the region and time of year.

SDGE also offers time-of-use (TOU) pricing plans to its residential customers. In a standard TOU plan, each day is broken into on-peak and off-peak time zones with energy costing less during the off-peak hours. Similar to the standard DR plan, the TOU plans are also tiered plans so if a customer exceeds the baseline allowance by a certain threshold, the rates increase.

Within the TOU group, there are also specific rate groups that are designated explicitly for residential customers who charge an EV(s) on the premise (EVTOU).

Kim found that in the standard TOU load profiles, there is a single peak load during evening hours across all regions similar to the region’s non-TOU load profile. However, in the EVTOU load profiles, there are twin peak loads with one of the peaks occurring at midnight when the super off-peak period begins.

The load also decreases rapidly from midnight to a leveled minimum load within a couple of hours. The explicit choice of midnight as the time when the super off-peak period begins has a direct impact on the occurrence of the second peak demand. Since most EV chargers are controlled easily using a smartphone app, it is highly plausible that EV owners would simply shift the charging activity to the changes in the TOU rate structure. For example, if SDGE were to change its super off-peak period start time from midnight to 2 a.m., then the demand peak would simply shift as well to 2 a.m. Therefore, careful consideration of the TOU rate design is critical to shaping the load profile as more EVs are adopted and charged at homes. Poor TOU rate structure design would lead to unintended consequences such as new peak demand and rapid ramping requirements.

… Results from this study clearly show the importance of TOU rate structure design with respect to EV charging loads. Poor design of the rate structure can lead to the rise of new energy peaks and other unintended consequences. The necessary next research step is the modeling and simulation of the EV charging loads as a function of different TOU rate structure design and technology adoption rate.

For example, the SDGE service territory is expected to experience a significant growth of EVs over the next decade. As more EV drivers are on the road, there will be greater energy demand at the residential sector. An analysis on the potential energy load profiles based on different TOU rate structures across regions and EV charging load requirements would give insights on how to best design the rate structure to induce the best charging behavior for better grid operations.

—Kim (2019)


  • Jae D. Kim (2019) “Insights into residential EV charging behavior using energy meter data,” Energy Policy, Volume 129, Pages 610-618 doi: 10.1016/j.enpol.2019.02.049



Most chargers have a start time, mostly "now".
What you need is a "charged by" time, say 6.30 am and some central control system to stagger the start times, or else you just take the serial number of the charger and get the start (or end) time from the serial number modulus 120 or 60.
You might want to suck in a few KwH initially, in case you need a short run, but you could do the bulk of the charging from 12-6 or 7am.
If there is a lot of solar, you might want to charge at work during the middle of the day.

Steve Reynolds

How about this as a future utility strategy to let EV charging help rather than hurt demand peaks:
The utility posts real time (say every minute) TOU pricing on a web portal. This price reflects supply and demand.
EV owners have an app that chooses when to charge based on getting the best price. Different apps may use different algorithms, but all will do most charging when the price is low.
There should be no privacy concerns about demand data, since the utility does not even need to collect individual customer TOU data (it stays inside the meter).


I have every confidence an automated market will work this out very well over time. Average daily charge requirement is likely well less than 15 kwh per day and the average amount of time on the road is less than 2 hours. So depending on the circumstances normal drivers will have 20+ hours to recharge for 2 or 3. There will probably need to be some charging availability in the work parking areas but I'm sure that will be feasible. Its a great challenge for city planners, utility operators and software designers. I'm sure they are up for it.


It is obvious that e-energy supply and demand will have to be better managed to satisfy users and suppliers.

1. More clean renewable energy (REs) will have to be produced and stored to replace polluting CPPs and eventually NGPPs.
2. Homes, offices, schools, industries and transport vehicles will have to be electrified using clean electricity sources.
3. Recharging PHEVs/BEVs have to be mostly carried out outside peak demand periods to reduce impacts on the energy grids.
4. BEVs with very large battery packs should be allowed to sell energy to the grids and buy it back at higher price during off peak periods. BEV owners should determine/set the price/level/speed and time of energy sale and purchase.
5. Regional and national energy saving programmes should implemented to liberate some of the clean e-energy required for the electrification programmes.
6. It is all doable if we want to reduce pollution and GHGs and climate changes.

The utility posts real time (say every minute) TOU pricing on a web portal. This price reflects supply and demand.
This would lead to a whipsaw effect as many chargers switched off and on as the TOU rate went above their threshold one minute and below it the next.  This is obviously an inefficient and even harmful outcome.
What you need is a "charged by" time, say 6.30 am and some central control system to stagger the start times

The ideal situation would be "in-fill" demand management which centers charging loads around the nightly minimum in other loads.  This is probably best handled with a central allocation system of some kind which spreads the demand across the available time window to level the demand and operate generation at maximum efficiency.  To do this, the allocator needs to know minimum and maximum demand and total energy needed (or minimum and maximum amounts).  This can fit in a rather small data packet.

The J1772 charging system is fairly flexible.  The charger generates a signal to the vehicle which gives the maximum allowed charging current.  I've seen nothing in the spec which states how fast the vehicle must respond to changes in this current limit, but it's doubtful that they don't do it at all.  This allows the chargers to control power demand in real time with a fairly high level of precision.  A central allocation system could use this to keep demand flat all night.  Contrast everyone's charger kicking on right at 1 AM!


A fully managed grid will remotely select charging times and charging rates to match energy distribution (demand) with supply.


You don't want fully central control with everyone's charger being controlled by a master controller, it is too brittle in terms of being hacked and it is a single point of failure.
Better to give signals to the local grid to indicate when is a good time to charge.
typically, this would be a price signal, up to 48 hours (for instance) in advance. This could be based on demand and availability of cheap renewable power.
You also want to avoid the "everyone switching on at 1am" scenario, so you might give different groups different hour start times, and everyone would different minute in hour start (and end) times.
If people want to charge "right now", they can do so, but they'll pay for it if it is a peak time, else, they can have it charged by 7am and let the algorithms and pricing data figure out the details.
I see security and resisting hacking as the biggest problems.


It will be difficult to recharge your EVs, at any price, when the grid does not have enough energy left. Essential leveling could be achieved with automated control from suppliers. Priority system, coupled with various prices, could work well to satisfy users and suppliers.

The new digital remote power meters installed in the last 5 years or so could be upgraded to do it.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Your Information

(Name is required. Email address will not be displayed with the comment.)