Content uploaded by Farbod Farzan
Author content
All content in this area was uploaded by Farbod Farzan on Jan 31, 2015
Content may be subject to copyright.
52 ieee power & energy magazine july/august 2013
1540 -797 7/13/$31.0 0©2013IEEE
A
Digital Object Identifier 10.1109/MPE.2013.2258282
Date of pu blication: 19 Jun e 2013
Microgrids for
Fun and Profit
By Farnaz Farzan, Sudipta Lahiri,
Michael Kleinberg, Kaveh Gharieh,
Farbod Farzan, and Mohsen Jafari
A VISION SHARED BY MANY EXPERTS IS THAT FUTURE COMMUNITIES
(residential and commercial developments, university and industrial campuses, military instal-
lations, and so on) will be self-sufcient with respect to energy production and will adopt
microgrids. With power generation capacities of 10–50 MW, microgrids are usually intended
for the local production of power with islanding capabilities and have capacity available for sale
back to macrogrids. A typical microgrid portfolio includes photovoltaic (PV) and wind resources,
gas-red generation, demand-response capabilities, electrical and thermal storage, combined heat
and power (CHP), and connectivity to the grid. Advanced technologies such as fuel cells may also
be included. This article describes the problems encountered in analyzing prospective microgrid
economics and environmental and reliability performance and presents some results from the
software tools developed for these tasks.
Integration of Microgrid Operation and Investment
The value of a microgrid portfolio depends on its projected return on investment and the potential
growth in its operating income. For a portfolio of nancial assets, valuations are based on projec-
tions of the market prices of those assets, using historical
data about prices, industry trends, and futures prices as a
basis for the projections. For a microgrid, the investment
payoff is directly linked to the operation of the physical
assets, and return on investment depends on how these
operations will be optimized and utilized in the short term.
The long-term value of a microgrid depends on when (in
terms of market conditions) investments were made and
also on the amount of the investment and its nancing
costs. Grid energy and fuel costs, the price of the neces-
sary technology (e.g., PV equipment, wind turbines, or storage), state incentives, and parameters
such as nance charge rates, nance terms, and the relationship between the nance rate and the
discount factor could all affect the optimal investment decision.
Typical investment models for infrastructure assets utilize assumptions about the short-
term average performance of the assets and further assume that the underlying system operates
The Economics of
Installation Investments
and Operations
july/august 2013 ieee power & energy magazine 53
optimally (as computed on the basis of some average func-
tion). Operational dynamics driven by endogenous factors
(i.e., asset reliability and degradation, demand prioritization,
resource allocation, and risk management) and exogenous
factors (i.e., weather forecasts, natural gas prices, and exter-
nal demand) are usually ignored or captured only on the
basis of discrete choices or simple variance analysis. At the
same time, any modularized approach to long-term invest-
ment, where assets are acquired and generation capacity
is increased in stages, would affect short-term operational
dynamics. Short-term returns from the microgrid will in
turn inuence long-term decisions about when to invest and
what to invest in.
Figure 1 illustrates a model developed at DNV KEMA
for evaluating investments in different congurations of a
microgrid, taking into account economic and environmental
metrics. This model simulates the microgrid operating opti-
mally in parallel with the grid. It also simulates operation in
islanded mode when the grid is down, when maximizing reli-
ability criter ia is key. On the resource side, different generation,
storage, energy efciency, and automation technologies are
considered. On the demand side, buildings and respective
end-use load are modeled in detail.
Energy Economics
The operation of a microgrid is closely tied to energy econom-
ics. This includes both the nancials of interacting with the
utility and macrogrid and the cost of self-generation. Various
resources contribute to the economic benets of a microgrid:
✔ Energy efficiency upgrades on equipment will lower
the overall load baseline.
✔ On-site generation, possibly in conjunction with energy
storage, can be utilized to avoid peak energy costs and
even create revenue streams by selling energy back to
the grid once price signals justify it economically.
✔ Enrollment in demand response programs can be
regarded as a means not only to reduce energy costs
but also to generate revenue by reducing load on the
grid. Demand response can be provided by both self-
generation and curtailable end-use load.
✔ While grid energy transactions and fuel costs domi-
nate the economics, microgrid participation in
capacity and ancillary services markets can also be
important incremental revenue drivers.
✔ The reliability improvements obtained through island-
ing capability and sufficient local resources can be
valuable—quite valuable, depending on the mission
© artville
54 ieee power & energy magazine july/august 2013
of the facility and the critical load served during
islanded operations.
The marketplace in which the microgrid is playing signi-
cantly affects the amount of savings realized. For example,
many commercial end users are charged time-of-use rates,
and switching to a different tariff scheme—such as one based
on real-time data or on hourly locational marginal pricing
(LMP)—could be benecial to them. The decision to switch
might not be a trivial one, however, as different schemes
impose different risks on the microgrid, based on price vola-
tility and penalties for failing to deliver energy to the grid as
scheduled if local resources, especially renewables, come up
short. It should be noted that a microgrid has limited options
for mitigating its risks. The options vary from relying on self-
generation to locking in both its cost and revenue streams by
means of long-term agreements with respective parties.
An example of a microgrid’s daily control process is shown
in Figure 2, where decisions regarding energy purchase, CHP
production, and the use of battery storage are optimized
against the day-ahead electricity price and the available PV
production. Less expensive on-site generation is utilized not
only to avoid the higher cost of purchasing energy from the
grid but also to gain revenue by selling energy back during
morning peak times. The annual savings from self-generation,
efciency upgrades, and demand response participation is
shown in Figure 3. Finally, Figure 4 demonstrates cash ows,
reecting investments in various technologies and the sav-
ings generated by the microgrid. Note that the energy bought
from the grid is, in this case, extremely nonconforming, as
the microgrid optimizes around renewable production, native
demand, and grid hourly prices. This is not an extreme case;
on an annualized basis, there is no net sell-back to the grid. In
this example, load grows over time as occupancy of the facili-
ties increases despite the energy efciency measures imposed.
The cumulative cash ow for energy investments and opera-
tions shows an approximately eight-year payback.
Interaction of the Elements
in Overall Energy Economics
While individual resources contribute to a microgrid’s bene-
ts, broader value can be achieved from the interaction of indi-
vidual elements. Renewable resources introduce uncertainties
in operations due to intermittent availability, for example as
a result of varying patterns of wind speed and solar irradia-
tion. These uncertainties become important because they can
cause shortages or excesses of energy compared with what
was planned for, and they therefore can lead to variation in
costs and revenues. Adopting appropriate strategies to mitigate
the risks associated to such uncertainties requires operational
decision-making tools that account for such uncertainties
while scheduling different generation and storage resources.
Energy storage devices (either thermal or electrical) can be
considered as buffers within the system that enhance exibility
in responding to uctuations due to renewable resources. But
the effectiveness of such storage applications depends heavily
on how the devices are controlled. The control strategy should
figure 1. Overview of the DNV KEMA model.
Optimal
Investment
Microgrid Elements
Building Dynamics, Energy Profiles, and Tenant Behavior
Campus Microgrid Operations and Economics
Energy
Efficiency and
Bulding
Automation
DG: PV, WT,
and CHP
Storage:
Electric and
Thermal
Particiaption
in DR
Energy
Efficiency and
Building
Automation
DG: PV, WT,
and CHP
Storage:
Electric and
Thermal0
Particiaption
in DR
Reliability
Module
Daily Operations
Optimization
Simulation
Across the Year
Available
Investments
and Cost
Annual Economics,
Emissions, Zero Net
Energy, Tenant
Satisfaction
Tenant
Utility
Function
Temperature
Sunlight
Day Ahead
Energy Prices
8760 Data
july/august 2013 ieee power & energy magazine 55
figure 2. (a) PV production, (b) CHP production, (c) day-ahead prices, (d) energy purchased, (e) energy sold, and (f) bat-
tery storage state of charge for a sample microgrid.
PV Production
Day-Ahead Price
Energy Sold
CHP Production
Energy Bought
Battery State of Charge
100
30
450
180
0
20
40
60
80
100
120
140
160
400
350
300
250
200
150
100
50
0
25
20
15
10
5
0
kWCents/kWh
18
16
14
12
10
8
6
4
2
0
kW
kWkWkWh
010 20
300
250
200
150
100
50
0
15 5101520
5101520
5101520
5
10 20155
10 20155
20
40
60
80
(a) (b)
(c) (d)
(e) (f)
be designed to make sure the storage devices are available when
they are expected to provide service. Moreover, the storage con-
trol needs to take into account the cost of energy for charging
and technical constraints around charging, discharging, and
performance degradation of the device. In the model results
shown, the storage resource is co-optimized with energy pro-
duction and demand response in a mixed-integer programming
formulation. The examples shown are based on “real data” in
the sense that typical Los Angeles–based building information,
energy prices, and renewable performance are used. In this
instance, investments in energy efciency are the single most
valuable option. Electrical storage is still too expensive to make
sense on its own, but when coupled with signicant investments
in PV generation, storage starts to show benets. Thermal stor-
age is economical purely in terms of shifting air-conditioning
load from peak to off-peak. The investment portfolio optimi-
zation is complicated by current policies around rebates and
tax incentives for different energy efciency investments and
renewable technologies. For instance, if a continued decline in
PV costs is projected, it may make sense to delay major PV
investments until the last year incentives are available.
Noneconomic Benefits
The benets from microgrids are not only economic.
Microgrids can be viewed as a means of creating zero-
net-energy communities and meeting other environmental
goals established by states or regulatory agencies. More-
over, microgrids can operate in islanded mode and sus-
tain the power supply in the event of a grid outage. This
is in particularly crucial in order to resume the operation
of critical infrastructure such as military facilities, hospi-
tals, ports, public transportation, and emergency-response
facilities. With the aging of grid infrastructure and restric-
tions on new investments in transmission and distribution
networks, microgrids can serve as an alternative solution to
intense investment in the centralized grid. Figure 5 shows
how a microgrid can supply a portion of load during a grid
outage. In this conguration, PV and battery storage (BS)
are sufcient to supply all critical and uninterruptible load
for each hour of the outage, and storage level decreases with
the duration of the outage. Uninterruptible load experiences
a momentary outage (not represented on plots), because
no uninterruptible power supply (UPS) is installed and
56 ieee power & energy magazine july/august 2013
insufcient resources are available to
supply noncritical load.
Stochastic Operation
Traditionally, the wholesale electricity
market uses reserves and load-following
capacity to hedge against shortage risks
and load variations. Moreover, the size
and abundance of generation resources
and ancillary services such as LMP
protect macrogrids against market vola-
tility in prices, demand, and generation
capacity. Microgrids are quite vulner-
able to these risks, however, due to their
smaller size and the volatility of their
internal generation resources. Their
only hedging mechanisms against
shortage risks are to purchase energy
from the grid at spot prices, which can
be quite high at times of peak load or
in an emergency, or to contract with
energy service companies.
A typical microgrid will most likely
be owned by a community or small
group of public and private investors.
The investment on microgrids will be
very different from a traditional power
grid since, due to their size and distrib-
uted nature, a small- to medium-sized
investment will be more common. Fur-
thermore, to be attractive for private
investors, a faster return on investment
compared to the traditional grid will
be expected. It is also very likely that
these investors are motivated by the
energy and cost savings that can be real-
ized from the local generation of power
and by the security and reliability that
microgrids can offer, especially at times
of peak loads and during unusual events
like natural or man-made disasters. Like
any other nancial investment, risks will
play major role in the operation and con-
trol of microgrids. The risk is present in
both the design of a microgrid and its
daily operation. By appropriately sizing
the microgrid and minimizing the risks
from energy economics, the microgrid’s
owners and investors will be able to
maximize their savings while ensuring
higher levels of energy security and reli-
ability. By doing so, the microgrid will
also be able to help mitigate the risks
of the larger grid, especially at times of
emergencies and high peak loads.
figure 5. Operation of a microgrid in islanded mode during a grid outage.
USP SOC (kWh)
BS SOC (kWh)
BS
CHP
PV
Noncritical Load
Critical Load
Uninterruptible Load
Noncritical Load
Critical Load
Uninterruptible Load
Total Load
Load Served
Total Resources
Storage SOC
Total Load
Load Served
Total Resources
Storage SOC
Total Load
Load Served
Total Resources
Storage SOC
0
123
Outage Duration (h)
50
100
150
kW
200
250
300
figure 4. Sample microgrid cash flow diagrams over 15 years.
6,000
Finances
0
US$
2,000
–6,000
2013
2015
2017
2019
2021
2023
2025
2027
–4,000
–2,000
4,000
Balance
Savings
Cash Outflow
BackNext
figure 3. Sample microgrid energy economics.
9,000
Energy Balance
0
2013
2015
2017
2019
2021
2023
2025
2027
1,000
2,000
3,000
MWh
4,000
5,000
6,000
7,000
8,000
NextBack
Sellback
Projected Demand
Efficiency Savings
Internal Production
Net Demand
july/august 2013 ieee power & energy magazine 57
To account for risks in microgrid operation, uncertain-
ties should be formulated using a stochastic optimization
model. It is interesting that as we move from deterministic
to stochastic models, the planning decision moves toward
more prior commitments and less spot purchasing, leading
to lower expected cost and variance. (This is a function of
expected volatility in energy prices.) As expected, the dif-
ference in the way deterministic and stochastic models make
decisions depends on several factors, including a microgrid’s
conguration and the variability of its resources. Therefore,
a careful examination of existing settings will help the deci-
sion maker choose the appropriate model for planning and
operation so as to make sure that ignoring uncertainty (i.e.,
choosing a deterministic model) does not have an adverse
impact in terms of increasing the variation of planning deci-
sions and to make sure that taking uncertainty into account
does not lead to a more complicated and costly model with-
out creating a noticeable benet for the decision maker.
As a microgrid’s on-site capacity increases, its cost distri-
bution (in terms of both the average and standard deviation)
becomes less sensitive to risks and uncertainties. Risks and
uncertainty in cost distribution increase with more renew-
able penetration and decrease with more fuel-red genera-
tion within the microgrid.
Stochastic Investment
Models developed at Rutgers University are able to balance
the risks and outcomes associated with microgrid investment
figure 6. Deterministic investment model.
5
4
3
MW
2
1
0Year
1
Year
2
Year
3
Year
4
E [GF]
E [PV]
E [WT]
E [ST]
figure 7. Stochastic investment model.
5
4
3
MW
2
1
0Year
1
Year
2
Year
3
Year
4
E [GF]
E [PV]
E [WT]
E [ST]
figure 8. Incremental investment with and without interaction.
0.3
0.25
0.2
0.15
0.1
0.05
0
ST (Year 4) Model 1
ST (Year 4) Model 2
0
2
4
6
8
Year 1
9.255
0
4.895214286
0
0
15.73020833
0
0
Model 1
Year 3
0
9.039292654
0
0.076904827
Year 4
10
12
MW
14
16
18
0
1.769791667
4.359785714
0
Year 2
GF
PV
WT
ST
0
2
4
6
8
Year 1
9.255
0
4.895214286
0
0
15.73020833
0
0
Model 2
Year 3
0
9.039292654
0
0.262601719
Year 4
10
12
MW
14
16
18
0
1.769791667
4.359785714
0
Year 2
GF
PV
WT
ST
58 ieee power & energy magazine july/august 2013
and operation. The microgrid cost savings function is calcu-
lated from a model that optimizes day-ahead planning and
the same-day operation of the microgrid under a host of sto-
chastic variables. This functional form is fed into a stochastic
long-term investment model, which decides when to invest in
microgrid components and expansions. The investment model
captures long-term market and nancing volatilities, such as
the investment costs of PV and storage, natural gas prices, the
availability of investment funds, and the correlation between
peak electricity prices and natural gas prices.
The analysis is performed on the basis of cash ow, reect-
ing actual outows and inows of monetary values. It requires
proper identication of the costs and benets resulting from
the investment, including any marginal values introduced to
the system by the investment. The analysis includes the sunk
cost incurred by a new investment as well as its opportunity
cost (the benet forgone if the investment is undertaken). An
opportunity cost is also incurred if the asset or resource can be
used in some alternative way and with some positive return.
Cash ow at the end of the planning horizon plus the value
of beyond-horizon cash ows at the end of the horizon is the
investment criterion to be maximized within the investment
optimization model.
We look at incremental investment decisions over a spe-
cic time horizon to evaluate microgrid investments. Deci-
sions regarding how much (if any) capacity of each type of
resource, i.e., gas-red generation (GF), PV, wind turbine
(WT), and electricity storage (ST), should be purchased at
the beginning of each one-year time period.
Figures 6 and 7 show investment strategies using deter-
ministic and stochastic investment models. In this example,
uncertainty exists in future PV capital costs. Therefore, a
stochastic model would suggest a strategy that is more dis-
tributed over the horizon. This could be viewed as a hedging
mechanism against the future uncertainty.
Results may not match expectations if interactions between
assets exist in the actual microgrid but operation and invest-
ment models ignore them. For example, PV and ST have inter-
action effects on the cost savings of the microgrid. Depending
on the value that it generates, the interaction between PV and
ST may make investment in these assets more or less attractive.
The incremental investment decisions about various resources
are shown in Figure 8. The interaction between PV and ST
forces the investment to allocate more capacity to these assets
in the fourth year in comparison with the same case without
such an interaction. PV dominates the investment because of its
higher contributions to the savings generated by the microgrid.
Allowing for interactions between the two assets permits the
use of storage not only for arbitrage but also coupled with
PV production. Therefore, at some point in time (here, in the
fourth year), the value of storage exceeds its costs and thus
becomes more attractive as an investment. This observation
tends to verify our hypothesis, and it necessitates the use of an
appropriate model in cases where such interactions exist.
Future Work
Participation in capacity markets and ancillary services mar-
kets are attractive revenue streams for microgrids. Inclusion
of ancillary market commitments in day-ahead and intraday
operations is a well-understood problem; the mathematics is
very similar to that used for the co-optimization that indepen-
dent system operator (ISO) market operations practice when
scheduling grid resources today. As with ISO-level market
operations, incorporating signicant storage in the formulation
and obtaining co-optimized solutions are challenges. Incorpo-
rating ancillary participation into investment decisions is more
complicated, however, as bidding strategies come into play. In
the examples shown above, the microgrid is a simple “price
taker” in the market that optimizes its resources once market
prices are known. But to participate in the ancillary markets,
the microgrid operator must make informed decisions about
what ancillaries and what energy to offer the markets as a bid-
der. This complicates the decision process and the investment
decisions required to enable that participation.
There is also interest from very large facility operators in
co-optimizing energy operations across multiple microgrids.
This is an area being intensively investigated at Rutgers.
For Further Reading
C. S. Park and G . P. Sharp-Bette, Advanced Engineering
Economics. New Yo r k : Wiley, 1990.
H. M. Wei nga r tne r, Mathematical Programming and the
Analysis of Capital Budgeting Problems. Englewood Cliffs,
NJ: Princeton-Hall, 1963.
W. El-Khattam, K. Bhattacharya, Y. Hegazy, and M. M. A.
Salama, “Optimal investment planning for distributed gen-
eration in a competitive electricity market,” IEEE Trans.
Power Syst., vol. 19, no. 3, pp. 1674 –1684, Aug. 2004.
H. Asano, W. Ariki, and S. Bando, “Value of investment in
a microgrid under uncertainty in the fuel price,” in Proc. IEEE
Power and Energy Society General Meeting, 2010, pp. 1–5.
S. Bruno and C. Sagastizabal, “Optimization of real asset
portfolio using a coherent risk measure: Application to oil
and energy industries,” in Proc. Int. Conf. Engineering Opti-
mization, Rio de Janeiro, Brazil, 2008.
Biographies
Farna z Fa rzan is with DNV KEMA Inc., Chalfont,
Pennsylvania.
Sudipta Lahiri is with DNV KEMA Inc., Chalfont,
Pennsylvania.
Michael Kleinberg is with DNV KEMA Inc., Chalfont,
Pennsylvania.
Kaveh Gharieh is with Rutgers University, New Bruns-
wick, New Jersey.
Farbod Farz an is with Rutgers University, New Bruns-
wick, New Jersey.
Mohsen Jafari with Rutgers University, New Bruns-
wick, New Jersey. p&e
