Creating a MicroGrid Market: Using a Frequency Driven Pricing Curve To Dispatch Load and Embedded Distributed Generation And To Charge and Pay for Participation

Posted on July 03, 2013
Posted By: Mark Lively
 
The electric industry around the world is increasingly enamored with the concept of MicroGrids, distribution networks with embedded distributed generation.  MicroGrids often have enough distributed generation to be self sustaining when the distribution network is disconnected from the transmission system.  In some respects, the MicroGrid concept is a throwback to the nineteenth century when utilities were first formed.  Indeed, in the nineteenth century there were only self sustaining MicroGrids.  It wasn’t until the twentieth century that holding companies built transmission lines to join together the MicroGrids in an attempt to lower the unit cost of utility operations.

The principal difference between the initial MicroGrids of the nineteen century and today’s MicroGrids is ownership.  The distributed generators on the initial MicroGrids were all owned by the utility.  Today, distributed generators are owned by hundreds, thousands, or tens of thousands of competing economic interests.  Indeed, I look at the term “distributed generation” almost as a secret code for “independently owned power plants.”  The diverse ownership of distributed generation plants has effectively disenfranchised the utility from selling electricity generated by utility owned plants.  This diverse ownership has created a need for a new way to operate a MicroGrid Market, a concept much different from the financial procedures of the nineteenth century.

When a utility owns the generation on a MicroGrid, the utility’s control room personnel can operate with a command and control mindset.  The operators of the individual generators get their paycheck from the same company paying the system controllers.  The system controllers presumably have the imprimatur of the utility to determine the least cost way to operate the system, making the system controller’s word to be law within the utility.  In contrast, when independent power producers provide the vast majority of the generation on the MicroGrid, there needs to be some other way to get these independent power producers to raise and lower their generation levels.  Hence the need for a MicroGrid Market.


Figure 1 is a Pricing Curve.  As is suggested by its name, Figure 1 can be used to set the price for electricity delivered to the MicroGrid, as well as for energy taken off the MicroGrid.  Figure 1 is also a dispatch curve, in that independent power producers can decide how to dispatch their individual units against the Pricing Curve in Figure 1.

Figure 1 provides a price in $/MWH that is applicable for any system frequency.  Figure 1 begins with a hyperbolic sine.  In this case, the independent variable for the hyperbolic sine is the frequency error divided by minus 0.005 Hertz.  The horizontal axis of Figure 1 is plotted in Hertz.   The actual Pricing Curve in Figure 1 is $30/MWH above the hyperbolic sine.  At 60 Hertz, the frequency error is zero and the hyperbolic sine of zero is zero.  Thus, the Pricing Curve at 60 Hertz is $30/MWH, the size of the offset.  The prices at significant other points along the Pricing Curve are shown in Table 1.


 

Development of Pricing Curve

Table 1

Frequency

Frequency

Error

Hyperbolic

Sine

Price With

Offset

59.970

-0.030

$201.71

$231.71

59.980

-0.020

$27.29

$57.29

59.990

-0.010

$3.63

$33.63

60.000

0.000

$0.00

$30.00

60.010

0.010

-$3.63

$26.37

60.020

0.020

-$27.29

$2.71

60.030

0.030

-$201.71

-$171.71

 

The Pricing Curve in Figure 1 effectively provides an alternative to having a system operator with command and control capability.  The price goes up and down with system frequency.  Independent power producers can dispatch themselves based on the concurrent price given by the Pricing Curve.  Indeed, if there are any utility owned distributed generators, those generators may also be able dispatch themselves against the Pricing Curve.  However, as utility owned generation, there may be some obligation to dispatch to 60 Hertz independent of the profitability of such action.



Mayor Gray of the District of Columbia has a vision that half of the electricity in Washington, DC, shall come from renewable generation, presumably roof top solar.  Roof top solar has no operating costs, making its dispatch point $0.00/MWH.  Whenever the Pricing Curve in Figure 1 shows a positive price, roof top solar has an incentive to produce electricity.  This point occurs when system frequency is about 60.02047 Hertz.  The formula for the Pricing Curve and a system frequency of 60.02047 Hertz produces a price of $0.018/MWH.  At 60.02048, the formula for the Pricing Curve produces a negative price of -$0.041/MWH.  Thus, absent other considerations, independently owned roof top solar plants should use their electronics to sense the concurrent frequency and shut down when system frequency climbs above 60.020 Hertz.  Figure 2 shows the right portion of the Pricing Curve, in the area where it starts to dip below zero.

Continuing on Mayor Gray’s vision, many renewable energy projects are able to sell Renewable Energy Credits (RECs) or earn Production Tax Credits (PTC) for any energy they produce.  The level of RECs and PTCs vary.  For purposes of an example, I will use $50/MWH as the sum of the various credits available to independently owned solar plants.  A system frequency of 60.02537 Hertz produces a negative dispatch price of ‑$49.903/MWH.  A system frequency of 60.02538 Hertz produces a negative dispatch price of ‑$50.063/MWH.  Thus, absent other considerations, a roof top solar plant that is earning RECs and PTCs that total $50.00/MWH should use its electronics to sense the concurrent frequency and shut down when system frequency climbs above 60.025 Hertz.

The Pricing Curve in Figure 1 would be used for pricing both (1) independent power producers and (2) load, though there should be a slight differential to provide some revenue to the owner of the MicroGrid.  Ideally, the price differential should reflect the marginal losses and constraints on the MicroGrid, just as independent system operators differentiate the price on their transmission systems based on marginal losses and constraints.

Many SCADA (Supervisory Control And Data Acquisition) systems collect generation data every four (4) seconds.  A four second pricing period provides the incentives for generators to respond quickly to frequency error.  Some SCADA systems collect generation data every second.  Later I will use an example with pricing periods that are one (1) minute long.

The SmartGrid is supposed to be capable of collecting billing information that can be used with the Pricing Curve in Figure 1.  Ideally, the pricing intervals for loads would be comparable to the pricing intervals discussed previously for generation.  Figure 3 shows the Pricing Curve when frequency is in the range between 59.990 Hertz and 60.010 Hertz.  Figure 3 gives prices that vary only minimally in this range, and the pricing interval is less critical.  When frequency moves outside this range, the price changes much more rapidly.  The slope of the Pricing Curve outside this range provides an incentive for the billing interval to be small enough to catch the price differences.



Many electricity networks attempt to keep system frequency in the range between 59.980 Hertz to 60.002 Hertz.  The -.005 Hertz divisor for the frequency error produces incentives for generators and loads to operate within this range.  At frequencies above 60.020, the Pricing Curve produces prices that are less than $2.71/MWH, as shown in Table 1.  Few fossil fired generators can operate profitably at such prices, and will attempt to shut down.  At slightly higher frequencies, the Pricing Curves sets an increasingly negative price, providing incentives even for non-fossil renewable plants to shut down.  Conversely, at frequencies below 59.98 Hertz, the Pricing Curve produces prices that are above $57.29/MWH.  When the frequency drops further below 59.98 Hertz, most fossil fired power plants will have an incentive to produce electricity.  Further, customer loads, which are notoriously price inelastic, will see prices climb enough such that loads will shut down.  For instance, at 59.96 Hertz the Pricing Curve produces a price of $1,520.48/MWH, which should be high enough to see the effect of price elasticity on many loads.

The Pricing Curve in Figure 1 is merely an initial estimate.  The $30/MWH offset from the hyperbolic sine should change while the MicroGrid is operating.  Consistent operation away from 60.000 Hertz should cause the offset to moderate in a manner that encourages participants in the MicroGrid to move the system back toward 60.000 Hertz.  For instance, assuming a grid that is dominated by solar distributed generators, the system might be expected to operate at 60.020 Hertz or higher.  At 60.020 Hertz, the solar generators are still earning $2.71/MWH, which is an order of magnitude smaller than most prices today.  A mechanical clock run by system frequency would run faster than a GPS synchronized clock.  Any accumulated time difference could be used to change the $30/MWH offset.  When the mechanical clock runs fast, as is true for a period of excess generation, the positive time error would reduce the $30/MWH.  Conversely, when the mechanical clock runs slow, as is true for a period of excess load, the negative time error would increase the $30/MWH.

 

Changing the Offset to Drive Frequency Error to Zero

Table 2

Time

Period

Frequency

Frequency

Error

Cumulative

Error

Offset

Hyperbolic

Sine

Pricing

Curve

1

60.020

0.020

0.020

$30.00

-$27.29

$2.71

2

60.020

0.020

0.040

$29.40

-$27.29

$2.11

3

60.020

0.020

0.060

$28.80

-$27.29

$1.51

4

60.020

0.020

0.080

$28.20

-$27.29

$0.91

5

60.020

0.020

0.100

$27.60

-$27.29

$0.31

6

60.020

0.020

0.120

$27.00

-$27.29

-$0.29

7

60.020

0.020

0.140

$26.40

-$27.29

-$0.89

8

60.019

0.019

0.159

$25.80

-$22.34

$3.46

9

60.019

0.019

0.178

$25.23

-$22.34

$2.89

10

60.019

0.019

0.197

$24.66

-$22.34

$2.32

11

60.019

0.019

0.216

$24.09

-$22.34

$1.75

12

60.019

0.019

0.235

$23.52

-$22.34

$1.18

13

60.019

0.019

0.254

$22.95

-$22.34

$0.61

14

60.019

0.019

0.273

$22.38

-$22.34

$0.04

15

60.019

0.019

0.292

$21.81

-$22.34

-$0.53

16

60.018

0.018

0.310

$21.24

-$18.29

$2.95

17

60.018

0.018

0.328

$20.70

-$18.29

$2.41

18

60.018

0.018

0.346

$20.16

-$18.29

$1.87

19

60.018

0.018

0.364

$19.62

-$18.29

$1.33

20

60.018

0.018

0.382

$19.08

-$18.29

$0.79

 

One approach to adjusting the price offset is presented in Table 2 and is illustrated in Figures 4 and 5.  Table 2 presents a time series of prices while the offset is being adjusted to push system frequency toward zero.  The basic assumption is that there is sufficient zero cost generation to match load and to exceed load if there is not an economic incentive to achieve that matching.  Initially the frequency is 60.020 Hertz, a frequency that produces a slightly positive price when the offset is $30.00/MWH.  The frequency error is 0.020 Hertz, creating a cumulative error at the end of the minute of 0.020 Hertz-Minutes.  The offset has an initial value of $30.00/MWH.  As time increases, the cumulative frequency error increases, so long as the frequency error itself is not zero.  The increase in the cumulative frequency error is used to decrease the offset.  Figure 4 shows the frequency error and cumulative frequency error for 70 time periods, as opposed to the 20 time periods in Table 2.  For each time interval, Figure 5 shows the offset, the hyperbolic sine, and the price.  The goal of pushing the frequency error to zero is used to slowly change the offset, which in turn achieves the goal of a zero frequency error.






The hyperbolic sine which is the basis for Figure 1 was chosen for the characteristics of having a slight slope around the targeted frequency and very large slopes when there is a large frequency deviation.  The negative -0.005 Hertz divisor created knees in the Pricing Curve when the deviation was about 0.02 Hertz, high or low.

A MicroGrid Market will produce widely varying prices, as is common for many commodities, including electricity.  Recently I analyzed the hourly prices on PJM for the year 2012, which I present in Figure 6.  The prices have been sorted for presentation purposes, with the lowest prices on the left of Figure 6 and the highest prices on the right.  To some extent, Figure 6 looks much like the reverse of Figure 1, though with significant differences.  Primarily, Figure 6 is actual data for PJM.  Figure 1 is a Pricing Curve, how to set prices at various system frequencies when there is a specific price offset of $30/MWH.



I would expect that the prices produced by Figure 1 might indeed have a distribution similar to the distribution of Figure 6, unless Mayor Gray’s vision is achieved.  Under Mayor Gray’s vision of a huge deployment of solar generation, for most of the day light hours the system would be operating on the right side of Figure 1.  The result will be a huge shift in Figure 6, with many more hours when the price was negative, zero, or very low.  Conversely, during the night, there would be almost no renewable energy available and the system would be operating to the left of Figure 1.  The result would be a shift in Figure 6 with many hours when the price is very high.  The combination of these two dynamics would mean a very small period in the middle of Figure 6 when the prices were moderate and very large left and right tails when prices were extremely low or extremely high.

Thus, the growth in independently owned distributed generation is lessening the applicability of the utility concept of a system operator with command and control functionality over the generation in the utility’s footprint.  The reduction in the command and control functionality and the growing diversity in economic ownership of the distributed generation together require a different method to control those independently owned distributed generation.  The independent ownership of those distributed generation suggests that the needs to be appropriate financial incentives to get the distributed generation to operate in a manner that will keep the lights on.  A MicroGrid Market with prices based on system frequency, a la the Pricing Curve in Figure 1, provides a financial mechanism that appropriately incentivizes the independently owned distributed generation.


For further reading on this topic:

A light hearted discussion is presented in “Microgrids And Financial Affairs - Creating A Value-Based Real-Time Price For Electricity,” Cogeneration and On-Site Power Production, September, 2007.
http://www.cospp.com/articles/article_display.cfm?ARTICLE_ID=307889&p=122

A discussion of time varying pricing of distribution wires is presented in “Dynamic Pricing: Using Smart Meters to Solve Electric Vehicles Related Distribution Overloads,” Metering International, Issue 3, 2010, which is available at www.LivelyUtility.com

 
 
Authored By:
Mark Lively is a utility economic engineer who develops financial models generally dealing with electricity and natural gas.  Mr. Lively earned a BS in Electrical Engineering from the Massachusetts Institute of Technology (MIT) in 1969 and a MS in Management from the Sloan School of MIT in 1971.  He has been a registered Professional Engineer in the District of Columbia since 1989.

Mr. Lively began his utility career with
 

Other Posts by: Mark Lively

Related Posts

 
 

Comments

July, 03 2013

Steve Drazga says

-

July, 04 2013

Len Gould says

Excellent presentation of an interesting idea. I have two concerns with the system before I would alter IMEUC to offer it as an alternative to the modified SMD market which I originally used.

1) Is there loss of necessary / critical information for the grid operator (and large slow-reacting generators such as coal and nuclear) by having no 24 hour forward plan of load to work with? Can this market provide a replacement for any such need, if required?

2) Who will organize and arrange needed transmission from distant generation, when and if needed? (e.g. California). Under IMEUC, with the SMD market locked in 24 hours in advance, and responsibility for arranging transmission entirely up to the individual generator entities, and with sever penalties for non-delivery including due to lack of transmission capacity, distant providers have a window of time, e.g. from midnight to start of peak hours, to organize required transmission or to purchase alternative supplies locally. This market system would require suppliers or etc. to work only with estimates, implying (excusable?) failures to deliver from a distance due to congestion, which it seems to me could open up opportunities for sharp operators to game the system, such as California in the 1990's .

July, 09 2013

Mark Lively says

Len,

In regard to 1), the concept is applicable to unscheduled load and generation. If the grid operator has already issued his schedule and the generation/load differs from the schedule issued by the grid operator, then something like this would apply. For instance, when a microgrid is operating on an islanded mode (which is when frequency goes wild) then the grid operator generally doesn't have the control that most people would want him to have.

In regard to 2), when the microgrid is isolated, scheduling power from a distant generator hardly is feasible. This method provides a market price for non-performance instead of the penalty that you mention. As a market price, it punishes those who do bad and rewards those who do well. In the case of the non-delivery that you mention, such a price would provide cash to those who helped keep the system operating because of the non-delivery.

July, 09 2013

bill payne says

Many SCADA (Supervisory Control And Data Acquisition) systems collect generation data every four (4) seconds. A four second pricing period provides the incentives for generators to respond quickly to frequency error. Some SCADA systems collect generation data every second. Later I will use an example with pricing periods that are one (1) minute long.

Beware of hackers.

http://www.prosefights.org/lenovog580/8toxp.htm#fawkes

July, 10 2013

Len Gould says

Mark. I notice that both your responses are aimed toward "isolated micro-grids", whereas in IMEUC - Independent Market for Every Utility Customer - Preliminary Business Case

IMEUC - Independent Market for Every Utility Customer - Part 2 - Market Operation I am addressing normal operation of any typical grid anywhere, large national or isolated micro.

July, 12 2013

Mark Lively says

For me a MicroGrid is islanded from the rest of the grid. To the extent that one uses a larger definition of a distribution system that is connected to the grid, then the market price of the grid is important. If a DG is higher cost than the grid price, we want to ramp it down, paying it only the grid price. Conversely, if a DG is lower cost than the grid price, we want to ramp it up, again paying it the grid price. The more interesting situations are when the interconnection to the grid is overloaded, either into or out of the connected MicroGrid. Then the extent of the overload can be used to increase or decrease the price within the connected MIcroGrid in a way to decrease the overload to a tolerable level. Another interesting situation is when the grid price is for a time period that is longer than the operating period desired for balancing the MicroGrid. Again, the imbalance on the time line can be used to moderate the grid price. I will look at your two links later. Need to get to a meeting.

July, 12 2013

Len Gould says

Agreed all Mark. I'm actually more interested in the potential to use your price-setting mechanism rather than SMD to establish the market price universally on the main grid for both buyers and sellers. Perhaps with a "predictive window" which, in certain circumstances, and communicated to everyone, could move the "target frequency" up or down slightly to help manage unusual circumstances predicted.

July, 16 2013

Mark Lively says

Len, Yes, the pricing mechanism I describe could be used to set the price on the grid, but it must be geographically differentiated. For most grids, that means that there would be a different price at each interconnection point between balancing authorities. Settlement at those prices would cash out both inadvertent interchange and loop flow. There would also be prices within each balancing authority, location by location. Most of the ISOs have such prices now for each generation plant. But my concept would allow even shorter time windows than are used by ISOs.

As for the predictive window that you seek, the offset mentioned in the article provides something akin to what you seek, almost. The price varies from the offset based on system frequency. The interaction of system frequency with the marginal cost of producers will indeed result in a target frequency that is slightly different from the standard frequency. But there needs to be a way to adjust the economics to push the target back to standard. Further, an individual can try to forecast how system frequency is likely to move on his own.

Part 2 of your presentation is more about hedging, how a customer could contract with a third party to get a fixed rate, perhaps in exchange for agreeing to interrupt so much. Certainly third parties would be allowed to assume the responsibility of paying the balancing authority the real time price determined under my system and then charge the customer some agreed upon price. Enron and other marketers have long done such things. The third party could hedge himself by generating like amounts of electricity at that site. Or could take the chance that the geographic difference between the customer and the generator will not get too far out of balance. Or could buy the electricity from some generator. But I don't see the detailed "reliability" in Part 2 as coming to pass.

June, 02 2014

ram 14 says

SRSG incorporated in 1997 with a vision to provide the technology of Apple.
System integration and Macbook air service center in Kolkata

July, 29 2014

Mkt Bhavish says

Artemis Hospital, spread across 9 acres, is a 350 bed, state-of-the-art multi-speciality hospital located at Gurgaon, in the Capital Region, India. Artemis Hospital is the first JCI and NABH accredited hospital in Gurgaon.
Cervical spine surgery in gurgaon And Physiotherapy hospital in dwarka

Add your comments:

Please log in to leave a comment!
back to top


Sponsored Content