Demand Charges — The Hidden Cost (and dirty secret) of EV Charging for Businesses

[aesculus]

So in ME you pay a different cost on the rate versus the time? In my utility I can can consume at any rate (up to the max load of my system at 200 amps) but the price varies wildly by the time of day. So I can get my 25kwh of power any way I want it, just as long as I stay within a time window.

 

[mknox]

So in ME you pay a different cost on the rate versus the time? In my utility I can can consume at any rate (up to the max load of my system at 200 amps) but the price varies wildly by the time of day. So I can get my 25kwh of power any way I want it, just as long as I stay within a time window.

Commercial customers often pay a variable per kWh charge (sometimes in TOU buckets, sometimes based on the hourly market price) plus they pay a charge based on the largest peak demand (kW) reached in the month. Residential and small commercial customers typically do not have the demand charge component.

 

[miimura]

Residential and small commercial customers typically do not have the demand charge component.

I just noticed that PG&E specifically disallows setting up a new small commercial meter expressly for the purpose of metering EVSEs. They evidently want this usage to add to the energy and demand charges of an existing meter, or go on a demand metered schedule.

The EV charging sentence was put into effect June 11, 2015.

 

[David_Cary]

The Hidden part of demand charges is an interesting topic. Of course they aren't hidden but probably are neglected. Demand can be controlled and when it represents 50% of the bill, it should be controlled.

The Sunday church example. The a/c should turn on very early and be done by the time the lights get turned on (or the EV charging occurs).

I have a residential demand rate (Only heard of these in NC and AZ). I have trimmed my demand significantly. Our rate here is based on peak 15 minute periods during a peak rate period. So typically my night time peaks are 15 kw and daytime are 5 (obviously car charging a big part of that - but I only charge at 5kw). I have my lower element on my hot water at 120V - so my peak attributable to that is 1.2 kw instead of 4.5 kw (I have solar here also but the cloudy days still get you). I also have staged a/c that I can lock into low stage during peak periods. Then of course, don't run the dryer during peaks. Demand management works very well so businesses should not be neglecting these things as "cost of doing business".

But of course, demand management doesn't really help the public charging issues. You could probably sell the concept of free charging when the sun is out or at night and potentially cut out the demand charge issue. The expectation in the future is free public charging during sunny days - we'll see. But it does make a lot of sense.

 

[Zapped]

It would be interesting to understand how the "peak 15 minute periods" work. In cold climates the larger parking lots have car block heaters outlets that are on a cycle timer. ( ie only 25 outlets of 100 outlets on at a time )

So if one could do the same with charging a car for example ( 16kW for a 208V-80 Amp charge station on a 14 minute cycle ) one might avoid the peak charge.

Question is how is the 15 minute peak measured by the meter ? How long would the "OFF" time have to be ?

 

[dgpcolorado]

It would be interesting to understand how the "peak 15 minute periods" work. In cold climates the larger parking lots have car block heaters outlets that are on a cycle timer. ( ie only 25 outlets of 100 outlets on at a time )

So if one could do the same with charging a car for example ( 16kW for a 208V-80 Amp charge station on a 14 minute cycle ) one might avoid the peak charge.

Question is how is the 15 minute peak measured by the meter ? How long would the "OFF" time have to be ?

While demand rate structures may be different elsewhere, this is how it works here:

A conventional power bill would have a charge for energy used — measured in kWh — plus a basic access charge for infrastructure (maintaining power lines, administration, and the like). A demand rate has three components:
1) An energy charge, which is lower than the conventional rate.
2) A demand charge, based on the highest power (kW) used in any 15 minute window during the month.
3) An access charge, usually somewhat higher than the standard rate access charge but still small compared to the other charges.

I had to do the rate comparisons for our local library because the power co-op was giving us the option to choose between conventional and demand rate structures (for truly large power users, say those above a 30 kW demand, demand rates are generally required). For our area the rates are $0.131586/kWh for conventional, non-demand, billing plus $29.50/month access charge for a business. The demand rates are $0.080029/kWh, plus $14/kW for highest demand, plus $40/month for access charge. This is how the comparison looked for the months of June and December of 2014:

As you can see, the conventional rate structure was cheaper than the demand rate structure so we changed to conventional billing. We figure that it will save us about $900 a year. Note that our demand billing was quite a bit higher in June than in December. Why? Because when all the AC units kick on at the same time we get a spike in the kW we draw from the power co-op and the demand rate is based on the highest power draw in any fifteen minute period during the entire month. Since power companies have to supply however much power is needed at any given time, large power users, generally those that sometimes spike above a 30 kW draw, have to pay demand rates because it costs the power company more to generate or buy enough capacity to supply all the users during peak usage.

If all this sounds complicated, it is! Took me awhile before I understood how demand rate structures really work. But I hope that this real world example helps.

 

[Zapped]

Just to clarify...
My question is.. does the meter have to "see" the peak kW for a period of time before it registers. (ie: 15 minutes would register, 1 minute would not )
If so, I assume one could cycle the power on a particular load (with some sensing device ) to avoid the meter from registering the peak ? ( ie 10 minutes ON, followed by xx minutes OFF so the meter "resets" so to speak )
At our office we actually have an interlock so that the AC units cycle through and only one AC unit is allow to run at a time for no more than 10 minutes.

For a car charge station it would be nice to cycle the power to avoid the utility meter from registering that peak load.
This is all unnecessary of course when the Power Wall becomes available and some peak shaving can take place.

 

[miimura]

The meters used for demand and TOU metering are almost certainly all digital Smart Meters now. I would expect them to register peak demand values for loads as short as one second. Starting a synchronous motor takes many times more power than constant running. I am not in the field, but I would think that using soft start or Variable Frequency Drive to start large motors would significantly reduce demand charges in large facilities that use many such motors.

 

[Cottonwood]

Just to clarify...
My question is.. does the meter have to "see" the peak kW for a period of time before it registers. (ie: 15 minutes would register, 1 minute would not )
If so, I assume one could cycle the power on a particular load (with some sensing device ) to avoid the meter from registering the peak ? ( ie 10 minutes ON, followed by xx minutes OFF so the meter "resets" so to speak )
At our office we actually have an interlock so that the AC units cycle through and only one AC unit is allow to run at a time for no more than 10 minutes.

For a car charge station it would be nice to cycle the power to avoid the utility meter from registering that peak load.
This is all unnecessary of course when the Power Wall becomes available and some peak shaving can take place.

It is my understanding that peak loads are averaged over 15 min intervals. For example 150 kW for 1 minute, 30 kW for 5 minutes, and 10 kW for 15 minutes would all show as a 10 kW peak averaged over 15 minutes and an energy usage of 2.5 kWh (10 kW for 0.25 hour).

 

[miimura]

It is my understanding that peak loads are averaged over 15 min intervals. For example 150 kW for 1 minute, 30 kW for 5 minutes, and 10 kW for 15 minutes would all show as a 10 kW peak averaged over 15 minutes and an energy usage of 2.5 kWh (10 kW for 0.25 hour).

I don't think this is true. If it was, they would just use a 15 minute interval meter and take the highest kWh for any interval and multiply by 4. I'm pretty sure that PG&E uses a different meter that specifically records the peak demand for each interval. This implies that it is not the average of the interval.

 

[Cottonwood]

I don't think this is true. If it was, they would just use a 15 minute interval meter and take the highest kWh for any interval and multiply by 4. I'm pretty sure that PG&E uses a different meter that specifically records the peak demand for each interval. This implies that it is not the average of the interval.

If that were true, why would they specify 15 minutes? Your description would give the same result if the period were 15 minutes or 30 days, it's just the peak.

Here is a description from an Xcel Energy page on billing, http://www.xcelenergy.com/staticfiles/xe/Regulatory/Regulatory%20PDFs/rates/SD/SDBusRates.pdf

The demand charge is applied to each month’s billing kW, which
is the maximum 15-minutes average actual kW demand measured
during the billing month, rounded to the nearest whole number.
Billing demand is adjusted upward if the power factor shown on
the bill is below 90 percent, by dividing actual kW by the power
factor and multiplying by 90 percent. However, monthly billed kW
will never exceed the kW value determined by dividing billed kWh
sales by 75 hours, based on a 30-day month. If the billing demand
is less than 25 kW for 12 consecutive months, you have the option
to transfer to Small General Service.​

Here is a description from a PG&E document http://www.pge.com/tariffs/tm2/pdf/ELEC_SCHEDS_E-19.pdf

Definition of Maximum Demand: Demand will be averaged over 15-minute intervals
for customers whose maximum demand exceeds 499 kW. “Maximum demand” will be
the highest of all the 15-minute averages for the billing month. If the customer’s use of
electricity is intermittent or subject to severe fluctuations, a 5-minute interval may be
used. If the customer has any welding machines, the diversified resistance welder
load, calculated in accordance with Section J of Rule 2, will be considered the
maximum demand if it exceeds the maximum demand that results from averaging the
demand over 15-minute intervals. The customer’s maximum-peak-period demand will
be the highest of all the 15-minute averages for the peak period during the billing
month. (See Section 6 for a definition of “Peak-Period.”) See Section 14 for the
definition of maximum demand for customers voluntarily selecting E-19. ​

As you can see there is precedent for going to 5-minute periods for very peaky customers, but is still averaged over the 5-minute periods.

 

[Zapped]

Thanks for the clarification. Looks like there's no easy way to avoid that demand charge.

 

[miimura]

@Cottonwood. You got me. I was too lazy to go read the tariff. Maybe they don't need a special meter after all.

 

[dgpcolorado]

Just to clarify...
My question is.. does the meter have to "see" the peak kW for a period of time before it registers. (ie: 15 minutes would register, 1 minute would not )
If so, I assume one could cycle the power on a particular load (with some sensing device ) to avoid the meter from registering the peak ? ( ie 10 minutes ON, followed by xx minutes OFF so the meter "resets" so to speak )
At our office we actually have an interlock so that the AC units cycle through and only one AC unit is allow to run at a time for no more than 10 minutes.

For a car charge station it would be nice to cycle the power to avoid the utility meter from registering that peak load.
This is all unnecessary of course when the Power Wall becomes available and some peak shaving can take place.

With caveat that other utilities may do it differently, I asked the question of my contact at our local power co-op. His response:

The demand is calculated by taking the peak power draw in three consecutive 5-minute time windows and averaging these three values. That is, the demand reading is an average of three readings over a 15-minute time frame. The highest average demand reading in the billing cycle is what is recorded by our meter and then billed for, if the account is on the demand rate tariff. Note, the 5-minute windows are rolling.

So, one data point on how demand rates are measured at one utility. Seems similar to the PG&E example above but here it applies to customers with lower power draws (presumably because the co-op pays demand rates to its power supplier).

 

[Ulmo]

Is there any discount for supply peak smoothing? I mean, e.g., PG&E tells such a connected customer when they can draw x power (kW) for y period (milliseconds) for z cost ($), where x power is more than they were at other times (theoretically), and z is less than the regular cost. Or, is there some type of pass through available of both energy and power costs from the variable cost energy market? I would foresee this operating on a very short time basis for sun and wind supplies (I can Internet ping UCLA from Monterey Bay in under 27ms and even Anchorage, Alaska is only 71ms, so tip of state to tip of state would take under 80ms, and Internet delay in sending sun and wind variance information for the relevant collection regions would almost always be under 80ms), but it takes time for car charging loads to respond; what is the minimum time to cut a car charging load, measured in approximate milliseconds or so?

By the way, this Demand Charge billing practice by utilities seems completely uncorrelated to the utilities' actual costs: if everyone peaks at different times, then that can be absorbed by the utilities for a fraction of the cost of everyone peaking at the same time. The only reason it seemed to make sense before is that it happened to correlate to actual simultaneous peak use due to widespread weather and work hour synchronized use across all people in a region and basically no highly diversely variable loads. I thought about it, and dumb car charging that is a standard 6kW across an entire parking garage would be the same thing as that, but if the car charging was not dumb and was supply controlled, then that would no longer be the case at all.

Basically, peak demand charges to customers are an antiquated billing method that does not fit the variable supply and variable load capable modern age, using solar, wind, and the many variable load capable items such as electric vehicle charging. The kW capability for car charging could be doubled or tripled to 10kW-30kW, and car charging could become a huge 1GW sink to put all the extra mid-day solar spikes, and eat up all the otherwise "economic" curtailments of solar and wind energy available for dispatch, as well as other supply variances (such as regional HVAC low load periods).

I may be missing the point: parking garage EV loads are probably in negotiated rate plans that we never see, and it's screw all the rest of us for top dollar. In that field, the only respite we can get is what we can generate and store ourselves. States like California may be intentionally not passing through solar and wind power peak supply cost benefits to large users in order to encourage them to buy their own solar and battery solutions and implement their own solar supply variable load capabilities. The largest of these users probably already purchase their own chunk of efficient rural solar & wind supplies in other parts of the state and negotiate transmission through established utilities, bypassing both the standard rate plans and the construction costs of local solar & wind farms in urban environments. Those solar & wind farms are usually so far away they can be put in a collection of almost-always-sunny-or-windy-somewhere pools that average out to something they figure they can get away with not even using variable load capable devices. In this view, their main cost to utilities is transmission, and their rate plans for unwindy dark days and nights would hit the power markets.

I still say the best way to support solar and wind power and energy in this State and probably anywhere is making rate structures much more specific to the charges they really represent. All this averaging worked back when the utility was the sole provider of electricity, used antiquated sources, and total use was fairly inexpensive for most users, but now, that's different.

According to Buying Clean Electricity | Department of Energy most of the "clean energy" costs more. What if the "clean energy" costs less during certain times of day, and people want that lower cost reflected in their use patterns, such as enabling of EV charging without exorbitant demand charges based upon supply generation costs + grid rather than just grid costs + peak supply sources?

 

[Ulmo]

Is there any discount for supply peak smoothing?

So, I write and write.

I Google, I search, I seek out.

I follow many dozens of sites about the topic.

I look at graphs from CAISO, try to read some of their myriad hard to comprehend bureaucracy. I look at the documents at Tesla and many threads about chargers.

I yell at virtual rooftops about this over, and over, and over, and OVER again!

And yet, plugging along quietly, all along since 2010, without anybody here telling me about it, is exactly what I was asking for:

Home

Demand Response (DR) programs help utilities maintain grid reliability and enable customers to realize significant value. Unfortunately, existing proprietary solutions add unnecessary cost and complexity. The OpenADR Alliance was created to standardize, automate and simplify DR to enable utilities to cost-effectively meet growing energy demand, and customers to control their energy future. Together we are creating the future of demand response today. Join us!

About the OpenADR Alliance
The OpenADR Alliance (now with over 130 member companies) was formed 2010 by industry stakeholders to build on the foundation of technical activities to support the development, testing, and deployment of commercial OpenADR and facilitates its acceleration and widespread adoption.

A person from a local University doing parking lot charging systems knew about it. But, nothing about it was mentioned, anywhere, until I wrote them specifically about it.

I forgot to uncheck "search this forum only" (stupid default). Here's the search results after this post for all forums, showing only this post for that word:

To me, this means that the Ivory Towers are being given exclusive access to the businesses of the world; I can ask everyone, thousands of people, hundreds of times, over half a dozen years, about this, and no one even bothers to mention that it already exists! That good old Affirmative Action at work, which denied me time and time again the opportunity to join the ranks of the college educated, once again destroy my purpose of being, my ability to work in the area of my interests, and my ability to make a decent wage. Instead, I'm left with bread crumbs in a dead end career that I loathe and hate, never think of at night or daydream about, and still think of the things I'm interested in despite it all.

This is both very angering, and a relief that someone is supposedly working on it, in real life. According to my contact, this is already all done. I'd like to see it in action, of course, and know what the numbers really are. Solar Power and Wind are huge in California, and should be able to make great use of such a program if it is complete enough. Not only did I have no idea it actually existed, but I don't see any examples of it in action.

Ok, now the next step is acknowledging its existence and wondering what its progress is, what is being done. Once again, this is a case of the Ivory Towers of Corporations that have access to this standarization being able to make use of it without the rest of us knowing about it. Can we be reimbursed for charging our cars using OpenADR at home, or at our places of work that aren't part of the Fortune 500?

Me finding out about this now when it has existed for 7 years is both useful and useless at the same time. ("bitter and sweet" to use an awful phrase from long ago.)

Without knowing the "key words", the "made up names", I didn't know what to search for. "Variable Cost", "Energy Market", "Dynamic Cost Electricity" don't come up with the actual implementation name when I Google those things. So this overloaded term of "Demand Response", which isn't the right English to describe the concept (since it isn't demand, it is a marketplace, it is a negotiation of prices and possible products (called electricity, of different types)), has its own little standard called OpenADR. So there it is. There we go. There's the key. God damn it.

Even if I knew of "Demand Charges", that name is already in use, so I didn't think to search for that. Also, Google has been pretty unhelpful, as usual, researching any new topic that I don't already have all the information for. If ever there was a case of reinforcing perceptions, Google is it. DuckDuckGo isn't better in this regard.

California is literally "curtailing" (wasting without use) Gigawatts of solar and wind power every sunny day (and even more so days that are both windy and sunny) during times when PG&E charges exorbitant "demand charges" to small businesses and upwards of $0.50/kWh to smaller businesses and homes without letting us use that energy when it is most plentiful, with the solar power generators (and some hydro generators) literally paying the utilities to use the energy, and all the while right under our noses the big companies have been using the systems that get this energy cheaper. Maybe they just installed it this week, or maybe they installed it years ago, but all the while, we could have been using it ourselves, and not one EVSE I know of yet has this feature on it (although it is already said that they already do, so they must use the feature hidden from our knowledge). Tesla's HPWC doesn't support it, and I've never seen it at any parking lot or PlugShare site that I've ever used.

 

[Ulmo]

Greenlots supports SCE's EV Charger tests

Greenlots

 

[Ulmo]

Wow; check out this DoubleSpeak:

“By embracing workplace demand response programs, utilities gain new opportunities to engage their customers and avoid disintermediation by third parties,” said Brett Hauser, CEO of Greenlots.

Translation: Utilities don't want to let end users without knowledge of or access to these features the ability to ramp up use during high electricity availability and ramp down use during low electricity availability in order to make a case for getting contracts for supplies that are priced to the supplies instead of exorbitant "demand charges" during peak electricity availability (from sun and wind) during peak cost periods from the utilities (charging the most for their electricity and delivery during the time when they have the cheapest electricity). I.e., mechanized theft.

Don't tell Mr. Hauser I said that; instead, go ahead and implement his systems in your companies' parking lots, and get the money for yourselves. The electricity is out there. Check out my screenshot from today in CAISO:

And although CAISO doesn't give curtailment figures for today, a recap of an article already referenced on TMC:

Spring Oversupply Lifts CAISO Curtailments | RTO Insider:

80 GWh Curtailed in March
CAISO curtailed about 80 GWh of renewable generation in March, nearly double the curtailments during the same month last year. So far this year, curtailments have occurred in 31% of all five-minute dispatch intervals, compared with 21% last year and 16% in 2015, the ISO estimates.

And then, the utilities "avoid disintermediation by third parties" so they can gouge customers during mid-day for exorbitant "Demand Charges" and "Peak Time of Use Rates" that are unfathomable for most of the country exactly when they have the most cheapest energy. Thanks, Gov. Davis!!!!!

And:

CAISO: Renewable energy curtailment could hit 8,000 MW this spring

 

[AWDtsla]

Just did some napkin math, it seems like the demand charges alone for the current supercharger network could be over $70 million per year? If they double the number of sites, with double the number of stalls, then $280 million per year in ~2018? This is a huge sum. Granted my rates table only goes up to 1000kW per meter, and I'm just applying that number globally.

I barely see how independent L3 charging is profitable.

edit - I was off by a factor of 2, but still huge numbers. It feels like the costs of supercharging are not being accurately reported in Tesla financials.

 

[KarenRei]

Nothing about this surprises me. And the reason is that the above calculations budget 50x 30kWh charges per month. Range-wise, that's the equivalent to a gas station selling five gallons of gas per day. I can assure you, that gas station would go out of business in short order. You can't amortize your capital costs with such low usage figures (and the "demand charges" are just that, the power company amortizing their capital costs).

As others have mentioned, if you have such rare visits, batteries are the obvious solution to buffer the high charging draws. You could have a supercharger run on a 15A 120V socket if the visits were rare enough Of course, a battery buffer transfers the capital costs to you rather than having them on the power company's end...