IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO.

3, JULY 2020 1657

Techno-Economical Model Based Optimal Sizing

of PV-Battery Systems for Microgrids
Soumya Bandyopadhyay , Student Member, IEEE, Gautham Ram Chandra Mouli , Member, IEEE,
Zian Qin , Member, IEEE, Laura Ramirez Elizondo, Member, IEEE, and Pavol Bauer , Senior Member, IEEE

Abstract—Microgrid with integrated photo-voltaics (PV) and Additionally, energy efficiency of the Microgrid can be im-
battery storage system (BSS) is a promising technology for future proved by choosing direct current (dc) distribution instead of
residential applications. Optimally sizing the PV system and BSS conventional ac based distribution [6]–[9].
can maximise self-sufficiency, grid relief, and at the same time
can be cost-effective by exploiting tariff incentives. To that end, Several publications have proposed optimal sizing of PV-
this paper presents a comprehensive optimisation model for the battery systems by maximising the economic value created
sizing of PV, battery, and grid converter for a microgrid sys- by using battery and PV system focusing on improving self-
tem considering multiple objectives like energy autonomy, power consumption or energy autonomy [10], [11]. Other studies have
autonomy, payback period, and capital costs. The proposed ap- focused on optimal sizing of only the battery with the goal of
proach involves developing a holistic techno-economic microgrid
model based on variables like PV system power, azimuth angle, maximising peak-shaving [12]–[14]. A MILP based optimisa-
battery size, converter ratings, capital investment and electricity tion model for system sizing for grid-connected and off-grid
tariffs. The proposed method is applied to determine the optimum microgrids is presented in [15]–[17].
capacity of a PV system and BSS for two case residential load However, there are several limitations in current literature re-
profiles in the Netherlands and Texas, US to investigate the effect
garding the optimal sizing of PV-battery systems for microgrids.
of meteorological conditions on the relative size of PV and battery.
Based on the optimisation results, thumb rules for optimal system First, most studies focus on a single objective like increasing
sizing are derived to facilitate microgrid design engineers during self-consumption, or reduce demand peaks or maximise eco-
the initial design phase. nomic benefits [13], [15], [16]. Due to the mutually conflicting
Index Terms—Batteries, microgrids, optimal sizing, particle
nature of the targets, the single objective optimisation approach
swarm optimisation, renewable energy, techno-economical is unable to provide valuable insights regarding the trade-offs be-
analysis. tween these objectives. Additionally, the inherent simplification
associated with formulating a complex multi-objective problem
into a weighted single objective problem fails to capture under-
I. INTRODUCTION lying trends. Second, in PV system modelling, most studies use
NCREASING energy consumption of buildings (both res- solar irradiation as the only input [18], occasionally combined
I idential and commercial) has led to 40% of total energy
consumption in developed countries [1]. The rise of energy
with temperature [15]. Some studies utilise more accurate PV
models which take into account the tilt and azimuth angle
demand in buildings will continue in the near future be- of the panel orientations. However, during the design process
cause of population growth, urbanisation, increasing penetration they select fixed values of tilt and azimuth angle [14], [19]
of electric vehicles (EVs) [2], and electrification of house- for maximum solar generation. This approach may not result
hold heating [3], [4]. To solve these problems, “Microgrid” in optimal PV and storage sizes as they do not consider the
based future electrical power systems have been proposed. A degree of temporal match of the PV profile and the load profile.
microgrid is a low voltage (LV) power network containing Third, many studies do not take into account the effect of load
distributed energy sources such as photovoltaic (PV) arrays, profiles and power management strategy on battery lifetime [11],
micro-wind turbines, fuel cell and energy storage devices [5]. [13], [17]. Fourth, the impact of incentives given by real-life
electricity tariffs on the optimal sizing is not investigated in the
literature [14], [15].
Manuscript received March 28, 2019; revised July 1, 2019 and July 26, 2019; The aim of this paper is thus to develop a multi-objective opti-
accepted August 10, 2019. Date of publication August 19, 2019; date of current misation (MOO) framework to solve the system sizing problem
version June 19, 2020. This work is part of the research programme P 13-
21 with project number A, which is financed by the Netherlands Organisation for a grid-connected residential microgrid system to incorporate
for Scientific Research (NWO). Paper no. TSTE-00350-2019. (Corresponding multiple mutually opposing objectives while taking into account
author: Soumya Bandyopadhyay.) the effect of battery degradation, incentives like feed-in tariffs,
The authors are with the Department of Electrical Sustainable Energy,
DCE&S group, TU Delft, Mekelweg 4, 2628 CD, Delft, The Nether- and PV system orientation. The developed MOO framework
lands (e-mail: [email protected]; [email protected]; is applied to optimally size the PV-battery-converter system in
[email protected]; [email protected]; [email protected]). two residential load profiles in Cabauw, Netherlands (NL), and
Color versions of one or more of the figures in this article are available online
at http://ieeexplore.ieee.org. in Austin, United States (US). The main contributions of this
Digital Object Identifier 10.1109/TSTE.2019.2936129 paper compared to previous works are:

1949-3029 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1658 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

1) Develop a multi-objective optimisation framework to size

PV system, grid converter, and battery storage capacity
resulting in Pareto fronts of trade-offs between multiple
objectives like lifetime capital cost, self-sufficiency, power
autonomy and simple payback period.
2) Study the effect of solar meteorological potential on op-
timal PV and battery sizing by comparing results on two
different geographical locations: Cabauw, NL and Austin,
US.
3) Investigate the effect of electricity pricing tariffs and feed- Fig. 1. (a) Annual energy yield of a 5kW PV system in: Cabauw, Netherlands,
in tariffs on optimal sizing of PV and battery system. (b) Annual energy yield of a 5 kW PV system in: Austin, Texas.
4) Draw guidelines for selecting the optimal azimuth angle
for a residential PV system.
5) Derive sizing equations and thumb rules to optimally 1 + cos θm
size PV-battery-converter systems for microgrids based SmDHI = S DHI (2)
2
on solar potential and specific load profile.
Sm = SmDNI + SmDHI (3)
The paper is structured in five parts. In Section II, the techno-
economical model of the microgrid is presented. Section III where SmDHI , SmDNI are the components of DHI and DNI which are
develops the multi-objective optimization (MOO) framework incident on the panel. The above equations show that the solar
and optimises the sizing of PV-battery based microgrid for energy generation by a panel can be controlled by changing the
two residential load profiles in NL and US. The results of module azimuth (Am ) and the tilt angle (θm ).
the multi-objective optimisation are presented and analysed in To improve the PV model accuracy, the effect of the ambient
Section IV to derive insights into optimal design and thumb temperature on solar power generation is also taken into con-
rules for optimal system sizing. Finally, general conclusions are sideration. The E20-327 PV module is rated for 327 W at the
summarised based on the results. ambient temperature of 25◦ . For other ambient temperatures, the
PV array output power Psolar (t) at a certain time instant can be
computed using [20]:
II. TECHNO-ECONOMICAL MODEL OF A MICROGRID
Sm (TNOCT − 20)
The techno-economical model of PV-battery based micro- Tcell = Ta + (4)
grids is presented in this section. The technical model comprises 800
of the PV system model, the battery lifetime model, and the Np Pr Sm [1 − γ(Tcell − 25)]
Psolar (t) = (5)
power management strategy. The economic model presents the 1000
methodology to compute the lifetime cost which consists of Fig. 1 shows the annual energy yield for a 5kWp PV array for
the capital cost and the operational cost. Finally, the figure of different azimuth angle and module tilt is estimated for the case
merits (FoMs) of a grid-connected PV-battery based residential of Netherlands (NL) and Texas (TX) based on equations (1)–(5).
microgrid are highlighted based on the different metrics pro- In case of Netherlands, the maximum annual yield is 5800 kWh
duced by the two models. obtained for south-facing panels with Am = 185◦ , θm = 28◦ .
The technical model is presented in three sub-sections. First, The maximum energy yield of 7830 kWh for Austin, Texas is
the power and energy output of a rooftop PV system is modelled obtained for panels with Am = 175◦ , θm = 18◦ .
considering azimuth and tilt angle of the PV panels. Second, To elaborate the effect of module azimuth orientation, Fig. 2
the power management strategy to control battery bank power is shows the power output profile of a 5 kW PV system during a
presented. Finally, the battery lifetime methodology is discussed summer day in the Netherlands with different azimuth (Am )
briefly. angles at an optimal tilt angle of θm = 28◦ . By changing
1) PV System Modeling: To estimate the energy and power the azimuth, the time of the day when maximum PV power
generated by the rooftop PV array, an accurate PV system model is available can be controlled at the cost of lower energy yield.
is built in this paper. Based on meteorological data of Nether- To investigate the effect of module azimuth on optimal storage
lands and Texas, US, Global Horizontal Irradiance (S GHI ), size sizing, the azimuth angle (Am ) is considered as a design
Diffuse Horizontal Irradiance (S DHI ), Direct Normal Irradi- variable in the optimisation framework.
ance (S DNI ) and ambient temperature (Ta ) are obtained. The 2) Power Management Strategy: The goal of the power man-
PV array is modelled in MATLAB using Sun power E20-327 agement strategy is to determine the charging/discharging power
modules rated at 327 W. of the battery Pbess (t) and grid power Pgrid (t) at a certain time in-
At a certain sun position, the irradiance on a panel with a spe- stant based on the load power Pload (t) and PV power generation
cific orientation (Am , θm ) can be computed using the geometric Ppv (t) at that particular instant. Different power management al-
models and the isotropic sky diffused model: gorithms lead to different solutions to the optimal storage and PV
sizing problem. In this study, the state-based power management
SmDNI = S DNI [sin θm cos as cos(Am − As ) + cos θm sin as ] (1) algorithm approach is considered due to their simplicity, low

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
BANDYOPADHYAY et al.: TECHNO-ECONOMICAL MODEL BASED OPTIMAL SIZING OF PV-BATTERY SYSTEMS FOR MICROGRIDS 1659

TABLE I
OVERVIEW OF ECONOMIC PARAMETERS

convention is used in this analysis for residential applications.

Therefore, the SoH of the battery becomes:
 
ξ
SoH = 1 − .100% (6)
Q0
where ξ is the total capacity fade of the battery during operation
and Q0 is the nominal capacity of the battery. It can be seen
that when 20% of the nominal battery capacity has faded the
Fig. 2. Power generated by 5 kW PV system for a summer day (Day 165 of
year 2017) in Netherlands for different azimuth angles = 0◦ , 100◦ , 180◦ , and
SoH of the battery becomes 80% based on (6), and thus the
260◦ with constant tilt angle of 28◦ . battery system has reached the end-of-life (EOL) condition.
The main goal of lifetime modelling process is to estimate the
lifetime as a function of the operating conditions associated with
the application: Lbatt = f (SoC, DoD, Ctot ), where Ctot is the
total Ah processed by the battery in that load cycle. In this paper,
a detailed empirical Li-ion battery lifetime model developed
in [22] is used.
This concludes the details of the technical model of the
grid-connected microgrid. The economic modelling approach
is presented in the next section.
An economic model of the microgrid is required to quantify
the benefits of installing a PV and battery system in grid-
connected microgrids. The economic benefit comes in form of
savings in the electricity bill due to: (a) using PV energy and
battery stored energy for houshold loads thereby reducing grid
dependency, and (b) selling unused PV power to the grid.
During the system lifetime, the total cost of the system can be
Fig. 3. Power profiles (1 minute resolution) of different sources and overall
load along with battery SoC profile on a summer day (day = 240) in Netherlands divided in two parts: (a) capital cost, and (b) operational costs
with battery as primary source algorithm. The simulated house has installed PV which include cost of electricity. The cost of battery replace-
power is 5 kW, a lithium ion battery of 10 kWh capacity with Crate of 1 and a ments and maintenance is considered within the capital costs.
5 kW front-end converter.
Thus, the total capital cost (κtotal ) of a PV-battery integrated
grid-connected microgrid can be formulated as the following:

computational requirement and ease of real-life implementation. κtotal = κbatt + κpv + κgrid-conv
In this algorithm, the battery is used as the primary source all = πbatt Cbatt (nreplace + 1) + πpv Ppv,r + πconv Pgrid-conv,r
the time. The utility grid is used only on two scenarios: (1) when (7)
the battery has reached the minimum allowable SoC, and there
is not enough solar generation to provide power to the load, where, κBatt , κpv , and κgrid-conv are the capital cost associated
and (2) when the battery has reached the maximum allowable with the battery storage system, PV system, and the front end
SoC, and the excess solar power is fed into the grid. Fig. 3 grid converter. πbatt , πpv , and πconv are unit-price for the battery
shows the power profiles of the grid, battery, PV and load along system, the PV system, and the grid interfacing converter. nreplace
with the SoC of the battery on a summer day in NL using the denotes the number of battery replacements needed during the
algorithm. total system lifetime. Table I shows the values of the constants
3) Battery Lifetime Modelling: In literature, battery ageing used in the economic model along with their sources. Next, the
is characterised and quantified by the term state of health (SoH). electricity tariffs associated with NL and US are discussed in
For residential based grid storage application, capacity fad- detail.
ing (permanent capacity loss) is used as the primary indicator 4) NL Tariff: The electricity tariff in the Netherlands chosen
for SoH of the battery. By convention, in EV batteries, the for this study is based on Eneco residential rates [25]. Eneco
end-of-life (EOL) condition is reached when the battery capacity tariff is based on time-of-use (TOU) prices with peak and off-
has dropped to 80% of its nominal capacity [21]. The same peak rates as shown in Table II.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1660 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

TABLE II Austin Energy credits solar customers for the solar energy pro-
NETHERLANDS ENECO ELECTRICITY TARIFFS
duced by their on-site solar energy system. However, it must be
noted that the regulatory charge and the power supply adjustment
charges discourages consumers to be dependent on the grid in
terms of both drawing and feeding in power. In conclusion,
multiple objectives like self-consumption or energy autonomy,
peak shaving or power autonomy need to be considered while
TABLE III
AUSTIN ENERGY ELECTRICITY TARIFF STRUCTURE optimally sizing PV and battery system for lowering electricity
bill in the Austin Energy tariff structure.
Figure of merits (FoMs) of Microgrids: A detailed description
of the modelling methodology of the technical and economic
aspects of the PV-battery system integrated microgrid is pre-
sented in the previous section. However, to optimally size the PV
system, battery storage and the converters, certain performance
metrics or figure of merits (FoMs) need to be defined to evaluate
and differentiate between designs objectively. To that end, the
following four FoMs are introduced.
Energy autonomy factor (α): The grid energy autonomy factor
is a metric to measure self-sufficiency or energy independence
of the microgrid design. It is calculated as:
Eload − Egrid,buy
Based on the data provided, the total electricity bill (κelec ) can γ= × 100 (%) (13)
Eload
be computed based on the following:
Power autonomy (ρ): Power autonomy factor is a metric to
κelec = κpeak + κoff-peak (8) quantify the power independence of the microgrid from the
κpeak = Enet,p cp if Enet,p > 0 utility grid. It is computed as the following:
 N

= Enet,p cfed if Enet,p < 0 (9) 1  |Pgrid,i |
ρ= 1− × 100 (%) (14)
κoff-peak = Enet,off coff if Enet,off > 0 N i=1 |Pload,i |

= Enet,off cfed if Enet,off < 0 (10) where N depends on the resolution of the power profiles used.
In this study, 1 min resolution is chosen.
where Enet,i is the net energy exchanged between the grid and Therefore, to compute the power autonomy factor for an
the house for the period i={p,off}: annual load profile N is = 24 × 60 × 365. As the energy au-
Enet,i = Egrid-drawn,i − Egrid-fed,i (11) tonomy factor (α) measures the energy independence of the
microgrid, the power autonomy factor (ρ) measures the power
5) Texas Tariff: The electricity tariff system for Texas consid- independence which includes both drawn and fed power. For
ered in this study is based on Austin energy residential rates [26]. example, the Texas electricity bill as shown in equation (12)
Austin Energy has a five-tier rate structure that incentivises incentivises customers to regulate their peak by charging the
customers on lowering their electric usage resulting in lower regulatory and the power supply adjustment costs. However, in
bills. Details of the electricity bill are presented in Table III. Netherlands traiff structure the customer is encouraged to be
Based on the rate structure, the annual electricity bill of a energy independent and not necessarily power independent.
household in Austin, Texas is computed using the following Lifetime capital cost (κtotal ): The capital cost of the entire
equation: system is an economic metric to quantify the total lifetime
κelec = cfixed + Etotal crc + Esum cpsa,s + Ewin cpsa,w · · · cost of the system which includes the initial investment cost,
maintenance, and the replacement costs during the system life-
5
 time [27]. Detailed modeling of the capital cost is already shown
+ Etier,i ctier,i − Efed cfed (12)
in equation (7).
i=1
Simple payback period (TPB ): The simple payback period is a
where Etotal is the total energy exchanged with the grid (drawn metric to measure the economic viability of the PV-battery based
and fed), Esum and Ewin are the energy exchanged with the grid system [23], [28]. It is defined as the number of years needed to
during the summer and the winter months. Etier,i for i = 1–5 are pay back the capital cost with the savings related to electricity
obtained from the net drawn energy from the grid. Efed is the bill (Rsavings ):
net energy fed into the grid.
κtotal κtotal
The tariff equation (12) shows that self-consumption by con- TPB (year) = = (15)
sumers is encouraged as the electricity tariff increases sharply Rsavings κelec,o − κelec,pv-batt
between tiers. Additionally, consumers are encouraged to be where κelec,o and κelec,pv-batt are the annual electricity bills with-
producers as well with the Value of Solar (VOS) tariff which out and with integrated pv-battery system in the house.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
BANDYOPADHYAY et al.: TECHNO-ECONOMICAL MODEL BASED OPTIMAL SIZING OF PV-BATTERY SYSTEMS FOR MICROGRIDS 1661

TABLE IV next part, the multi-objective optimisation algorithm and sizing

OPTIMIZATION VARIABLES AND THEIR RANGE
methodology is discussed in detail.

B. Multi-Objective Optimisation Algorithm and Methodology

Particle swarm optimisation (PSO) algorithm is used for opti-
mising the proposed PV-battery sizing problem. PSO is an evo-
lutionary gradient-free algorithm inspired by the movement of
birds or insects in a swarm which potentially requires fewer func-
tion calls [31]. However, PSO is a single-objective optimisation
algorithm. To make it suitable for multi-objective optimisation
problems, the concept of Pareto dominance is combined [32] to
The choice of performance metrics is motivated to facilitate generate non-dominated solutions or Pareto-optimal solutions,
both the end-users and the distribution system operators (DSOs). which results in Pareto fronts. A repository is used to store the
Economic metrics like lifetime system cost and the simple Pareto-optimal solutions which are updated at the end of each
payback period are useful to end-users to analyse the cost- iteration. Fig. 4 presents the multi-objective optimisation routine
effectiveness of the solutions. On the other hand, fundamental in a flowchart depicting the inputs, outputs, and system analysis
technical metrics like energy autonomy and power autonomy of the PV-battery optimal sizing problem.
can be utilised by DSOs to design tariff schemes to incentivise Thirty particles or designs per iteration and two-hundred
users to achieve their higher system-level goals like reducing the iterations are evaluated to generate a stable Pareto front between
operating costs, delay expensive grid upgrades [12], and solving two conflicting objectives. MATLAB is used to compute all the
network congestion [29], [30]. analytical equations presented in the previous section to model
This concludes the techno-economical modelling of the PV- the performance parameters or figure of merits (FoMs) of the
battery based microgrid. Based on the model metrics, four FoMs microgrid design. The time required to evaluate a complete
are identified which will be used for the optimisation problem design varies between 15 s–30 s.1 In this paper, an approach
formulation. based on placing particles on the border of the search space
using a combination of variable clipping and reflecting [31].
The optimisation results for two case studies (Netherlands and
III. MULTI-OBJECTIVE OPTIMISATION FRAMEWORK
US) are discussed in the next section.
A multi-objective optimisation framework is developed in
this section to optimally size the PV system, the grid converter, IV. OPTIMISATION RESULT ANALYSIS
and the battery of a microgrid. It utilises the techno-economical
The MOO framework developed in the previous section is
model developed in the previous section. Initially, the optimisa-
utilised to optimally size the PV-battery-grid converter system
tion targets and variables are described, followed by a discussion
in both the Netherlands and the US residential case studies. Fig. 5
on the system analysis flowchart. In the final part of this study,
shows the real-life load profiles used in this paper. The data for
the developed multi-objective optimisation framework is utilised
the NL residential load profile is obtained from a Dutch DSO
to optimally size the PV-battery for a microgrid in different
company, and the US profile is obtained from Pecanstreet online
operational scenarios. The results obtained are presented and
database [33]. Table V presents the details regarding the selected
analysed in detail in the next section.
household load profiles.
It must be noted that the PV-battery sizing results obtained
A. Optimisation Targets and Variables from the MOO framework for the two case studies are particular
Based on the FoMs introduced in the modelling section, the solutions for the particular combination of the load profile,
targets of the optimisation are: solar irradiance profile, battery technology, power management
1) Maximize energy autonomy factor (α) strategy, and electricity tariff structure. Altering any of the above
2) Maximize power autonomy factor (ρ) aspects will lead to different conclusions on the PV-battery sizing
3) Minimize lifetime capital cost (κtotal ) problem.
4) Minimize simple payback period (TPB ) The optimisation returns a 4-D Pareto optimal front. To aid
The objectives mentioned above are selected strategically to visualisation and insight into the results, a detailed analysis is
ensure that the optimisation progresses towards designs with conducted in three steps. First, higher level results are analysed
acceptable economic and technical performances. using sub-fronts of two targets. At a second level, complete
Table IV presents the optimisation variables and their range. fronts are shown to aid explanation of underlying trends. Finally,
The optimisation variables are mainly categorised into three two optimal designs are selected and analysed to verify the
groups: battery variables, PV system variables and the grid con- efficacy of the optimisation process.
verter variables. Crate is chosen as an independent optimisation Fig. 6 shows the side views of the 4D Pareto optimal front
variable which decides the rated power of the battery Pbatt,r based which highlight the trade-offs between lifetime system cost,
on the battery capacity Ecap . To ensure feasible designs, specific
constraints are put on the optimisation solution space. In the 1 Intel Xeon CPU E5-1620 v2 @ 3.70 GHz, 16 GB RAM.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1662 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

Fig. 4. Flowchart of the proposed main multi-objective optimisation routine for PV-battery sizing for microgrids. The routine calculates and identifies the
Pareto-optimal designs as a combination of the optimisation design variables, which include the PV system parameters, battery management parameter, and grid
converter ratings. The system analysis for individual designs evaluates the optimisation targets based on the swarm algorithm-generated designs and feeds it back
to the optimisation routine to update the Pareto-optimal design repository.

TABLE V
LOAD PROFILE DETAILS

A. Pareto Front Analysis

α − κtotal Pareto front: In the case of Texas load profile, it
Fig. 5. (a) NL house daily load profile for four days of the year with maximum is evident from the fronts shown in Fig. 6, that full energy
load Pload,max,nl = 7 kW, (b) Texas, US house daily load profile for four days of autonomy is achievable after significant capital investment.
the year with maximum load Pload,max,us = 9.8 kW.
However, in the case of the NL residential profile, it is difficult to
achieve full energy autonomy, which asymptotes around 70%.
energy autonomy, power autonomy, and payback period. Indi- It can be explained due to the low solar potential combined
vidual 2D Pareto fronts are discussed in the following: with the temporal mismatch between the yearly solar generation

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
BANDYOPADHYAY et al.: TECHNO-ECONOMICAL MODEL BASED OPTIMAL SIZING OF PV-BATTERY SYSTEMS FOR MICROGRIDS 1663

Fig. 6. Results of multi-objective optimisation for PV-battery sizing with NL and US houseload profile: (a) α − κtotal : Pareto fronts of trade off between energy
autonomy factor and lifetime system cost, (b) ρ − κtotal : Pareto fronts of trade off between power autonomy and lifetime system cost, and (c) TPB − κtotal :
Pareto-fronts of simple payback period and power autonomy factor.

Fig. 7. Effect of PV module azimuth angle on: (a) energy autonomy for different PV power for Cabauw, NL, (b) energy autonomy for different PV power for
Texas, US, (c) correlation coefficient between annual pv generation profile and load profile for the case of NL and US.

profile (Fig. 10a) and daily energy usage profile (Fig. 10b) of Austin load profile leading to higher the savings potential for the
the Netherlands. Therefore, to achieve full energy autonomy in NL case is much higher and therefore leads to lower payback
case of NL load profiles, battery-based storage is not sufficient period, and (b) the NL electricity tariff heavily incentivises the
and seasonal storage is required. end-user to be energy autonomous by installing PV with battery
ρ − κtotal Pareto front: Fig. 6b shows that full power autonomy storage compared to the Austin tariff. This result underlines the
is hard to achieve in both NL and US case studies with maximum importance of incorporating the cost of electricity model in the
possible values of 50% and 70% respectively. Additionally, it is optimisation problem.
evident that power autonomy can be negative in certain cases.
Negative power autonomy is due to feeding more power into the
grid compared to the actual in-house load power demand which B. PV-battery System Design and Sizing Trends
leads to even lower electric bills. Underlying trends regarding the battery system design, PV
TPB − α Pareto front: The trade-off between simple payback system design and sizing are analysed in this section. Fig. 8
period and energy autonomy is presented in Fig. 6c. Previously presents the maximum allowable depth of discharge (DoD)
it is shown that the US-based PV-battery system performs much of the Li-ion battery system of the optimised designs. The
better than the NL system in terms of both power autonomy maximum allowable DoD of the Li-ion battery system increases
and energy autonomy for the same capital investment. However, with the increase of the battery capacity for both NL and US.
the NL based PV-battery system performs significantly better in Based on that, it is concluded that for optimal performance of a
the metric of simple payback period (TPB ) for certain energy Li-ion based microgrid, one should select a proper DoD range
autonomy. It can be explained due to the following reasons: (a) depending on the size of the battery. For these particular case
the annual electricity bill for the NL load profile without any studies, a smaller capacity battery (≤5 kWh) has an optimal
PV and battery is 1630 $ compared to 840 $ in case of the US maximum DoD of 40%–70%. The optimal maximum DoD

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1664 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

Fig. 8. Maximum allowable DoD vs. the battery capacity of the optimal Pareto
system designs.

is 65%–90% for battery capacities ranging from 10 kWh to

20 kWh. 100% DoD of the battery can be utilised for battery
sizes bigger than 20 kWh when the capacity fading due to daily
cycles is insignificant compared to the nominal battery capacity Fig. 9. Results of multi-objective optimisation for PV-battery sizing with NL
for this particular application. and US houseload profile: (a) variation of energy autonomy factor with storage
hours (Sh ) or ratio of battery capacity to rated pv power in case of NL, (b)
In Section II-1, the potential of changing the azimuth angle variation of annual cost of electricity with Sh in case of NL and (c) variation of
Am to improve the temporal match between the pv generation energy autonomy factor with Sh in case of US, (d) variation of annual cost of
profile and the houseload profile was presented. To that end, electricity with Sh in case of US.
Az was considered to be a design variable in the optimisation
framework (Table IV). Fig. 7a and 7b presents the range of
Am for the optimal designs for the NL and US case studies case study. Fig. 10a and Fig. 10b show that there is high load
respectively. In case of the NL designs the optimal range is demand and low solar generation during winter coupled with
between 180◦ –220◦ whereas for the case of Austin, US the comparatively lower load demand and higher solar generation
optimal range is 240◦ –300◦ . It is interesting to highlight that during the summer. However, in the US case study, the seasonal
the optimal designs have significantly different Am from the load demand variations match the seasonal PV profile variations
maximum energy generation Am,max which is 185◦ for NL and leading to high values of correlation coefficient γ.
175◦ for the US case (Fig. 1). To explain this result, a new Finally, it is important to derive design thumb rules to select
metric based on the Pearson correlation coefficient is defined PV system size and battery capacity to achieve optimal per-
to quantify the temporal match between the PV power profile formance for a certain capital investment. In this considered
and load profile: PV-battery system for grid-connected microgrids there are four
N    unknown variables: Ppv,r , Pbatt,r , Pgrid,r , and Ecap . For the relative
1  Ppv,i − μpv
Pload,i − μload sizing of PV power and battery capacity, a metric called storage
γ(Ppv , Pload ) =
N − 1 i=1 σpv σload hours (Sh ) is defined as following:
(16)
where Pj , μj , and σj are the annual power profile, mean and Ecap
Sh = (17)
the standard deviation of the power profile for the jth system Ppv,r
with j = {pv, load}. The correlation co-efficient γ represents
the degree of temporal match between the PV profile and the Fig. 9a shows the effect of storage hours on the energy
load profile. γ ranges from 1 (complete temporal match) to - autonomy of the PV-battery based microgrid for the NL case
1 (complete temporal mismatch). Fig. 7c shows the variation study. It can be seen that the optimal range of Sh is 2 to 4 which
of correlation coefficient γ with the choice of azimuth angle for also results in the minimum annual cost of electricity as shown
both the NL and the US case studies. It is evident from Fig. 7, that in Fig. 9b. Applying similar analysis to the case of US as shown
the optimal choice of azimuth angle Am for pv-battery system in Fig. 9c and 9d, the optimal range of Sh turns out to be 4 to 6.
design is the one which results in maximum temporal match Since Texas, US has higher solar potential (≈35%) compared to
between the pv profile and the load profile. Additionally, the Cabauw, NL, it is intuitive that a bigger sized battery is needed to
correlation coefficient γ for the case of NL load profile and PV properly harness the excess solar potential. This ratio can be used
profile is almost 10 times lower compared to the case of the as a thumb rule for sizing PV and battery system for a microgrid
US profile. It is expected since there is significant mismatch in Netherlands and US, Texas to ensure high performance for a
between the annual solar profile and load profile for the NL fixed capital cost.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
BANDYOPADHYAY et al.: TECHNO-ECONOMICAL MODEL BASED OPTIMAL SIZING OF PV-BATTERY SYSTEMS FOR MICROGRIDS 1665

TABLE VI
SYSTEM SIZING THUMB RULES

equations derived from analyzing the data obtained from the

optimisation procedure.
Fig. 10. (a) Annual daily energy usage profile for the NL with a daily average
of 18 kWh and US load profile with a daily average of 17 kWh, (b) comparison of
solar energy production throughout the year in Cabauw, Netherlands and Austin, C. Analysis of Selected Optimal Designs
Texas for a 5 kW PV system designed for maximum annual solar output.
Higher level design sizing trends are obtained in the previous
section by analysing the optimisation results. To highlight the
efficacy of the sizing optimisation procedure, we have selected
two Pareto optimal microgrid designs for more in-depth analysis.
The designs are chosen with the same total lifetime cost of
13,000 $ to ensure a fair comparison.
Table VII presents the important design and performance
metrics of the two selected design cases. In terms of pv-battery
system sizing, the NL microgrid has a higher power PV system
and lower battery capacity compared to the US counterpart.
Due to higher solar potential, the US microgrid needs a slightly
bigger battery to store the excess solar energy. Both the mi-
Fig. 11. Variation of energy autonomy factor with (a) sum of all the source crogrid designs under-utilise the battery capacity by selecting a
converter ratings Psum,total , (b) ratio of the sum of PV-battery converter ratings maximum allowable DoD of 63% (NL) and 70% (US) to extend
to the grid converter rating Pratio . the lithium-ion battery lifetime.
In performance factors like energy autonomy and power
autonomy, the NL microgrid lags behind the US microgrid.
To optimally size the different converters associated with the Still, the NL microgrid performs much better economically with
grid, PV, and the storage, two additional metrics or equations simple payback period almost half that of the US microgrid
are introduced: design. This is mainly due to two reasons. First, the annual cost
of electricity for the US house is significantly lower than the
Psum,total = Pgrid,r + Ppv,r + Pbatt,r (18) NL house (see Table V) for almost similar energy consump-
(Ppv,r + Pbatt,r ) tion, thereby limiting the savings margin. Second, the Dutch
Pratio = (19) electricity provider incentivises high energy autonomy with
Pgrid,r
net-metering based tariff system, whereas the US tariff system
where Psum,total represents the sum of all the source/storage incentivises high power autonomy, which is a comparatively
converter ratings, and Pratio is the ratio of the sum of PV-battery more difficult metric to achieve. Fig. 12 presents the annual grid
converters to the grid converter rating. Fig. 11 presents the effect power profiles of the two selected case studies. It is evident
of the two aforementioned metrics on the energy autonomy of the from the profile (Fig. 12a) that the NL house microgrid interacts
microgrid optimal designs for both the case studies. In case of the heavily with the utility grid throughout the year, which results
NL house, a Psum,total of 15 kW or above will result in high energy in weak power autonomy of 15%. During winter, when the solar
autonomy. Similarly a Psum,total of 20 kW or above is required for generation is low, and the load requirement is high, it depends
high energy autonomy for the Austin, US house. It is interesting solely on the grid to supply the load. However, during the
to point out that the maximum load power in case of NL and US summer days when the load is comparatively lighter, and solar
are 7 kW and 9.8 kW respectively which are approximately half power is higher, the microgrid consistently dumps the excess
of the thresholds required for high energy autonomy. Similarly, solar energy to the utility grid. Therefore, the NL microgrid is
from analysing Fig. 11b, the optimal converter ratio factor Pratio highly energy-positive during the winter days and highly energy-
lies somewhere around 1–1.5 for both the case studies of NL negative during the summer days, which eventually balances
and US. As for sizing guidelines, one can select a certain rated out due to the electricity tariff’s sole dependence on energy
power for the PV system depending on the avaliable area for PV autonomy rather than power autonomy. This leads to low simple
installation and use equations (17), (18), (19) as thumb rules to payback period for the NL microgrid. However, unlike the NL
optimally size the battery capacity, and the converter sizes for microgrid, the US microgrid has weaker interaction with the
a grid connected microgrid. Table VI summarizes all the sizing grid as the Texas electricity tariff incentivises power autonomy,

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1666 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

TABLE VII
RESULTS OF SELECTED OPTIMISED MICROGRID DESIGNS

2) Solar potential of a location has a significant impact on the

relative sizing of the battery capacity relative to the rated
Ecap
PV power. The optimal value of storage hour (Sh = Ppv,r )
for Cabauw is between 2 to 4, whereas for Austin it is
between 4 to 6.
3) Optimal value of the azimuth angle (Am ) for the PV system
is found to be the one which results in a maximum temporal
match between the annual PV and load profile.
4) Thumb rules for optimal system sizing are derived to size
the battery power rating, battery capacity, PV power rating,
Fig. 12. Annual profile of the power exchange of microgrids with the utility and the grid converter rating in equations in Table VI for
grid (Pgrid ): (a) NL house microgrid, (b) US house microgrid. grid-connected microgrid application.
In conclusion, the presented multi-objective optimisation pro-
cess provides a platform to optimally size PV-battery systems
which leads to less dumping of solar power in the grid. There- during the initial design process taking into account a multitude
fore, it is economically more viable to install PV and battery of design variables and multiple objectives. For future work, this
system for the NL case study compared to the US case study study can be extended to compare different battery technologies
although the solar potential of Texas, US is much higher than to select the most economical design. Additionally, the effect
Cabauw, NL. of intelligent power management algorithms with forecasting
capability on the system sizing problem can be investigated with
this framework.
V. CONCLUSION
This paper presents a multi-objective optimisation (MOO)
ACKNOWLEDGMENT
procedure to size the PV-battery-converter system for micro-
grid applications. To validate the advantages of the proposed The authors would like to thank Netherlands Organisation
method, the MOO framework is applied to optimally size for Scientific Research (NWO) for providing funding for this
PV-lithium-ion (Li-ion) battery-converter system for two resi- research.
dential case studies in Cabauw, Netherlands and Austin, US. An
in-depth microgrid model considering battery degradation, PV
REFERENCES
design variables, and real-life electricity tariffs are coupled with
the MOO framework with the goal of drawing design guidelines [1] L. Pérez-Lombard, J. Ortiz, and C. Pout, “A review on buildings energy
consumption information,” Energy Buildings, 2008.
for optimal system sizing. Some essential observations obtained [2] A. Dubey and S. Santoso, “Electric vehicle charging on residential distri-
from a detailed analysis of the optimisation results are presented bution systems: Impacts and mitigations,” IEEE Access, vol. 3, 2015.
below: [3] E. Veldman, M. Gibescu, H. Slootweg, and W. L. Kling, “Impact of
electrification of residential heating on loading of distribution networks,”
1) The local electricity tariffs in Cabauw, NL heavily incen- in PowerTech, 2011 IEEE Trondheim, pp. 1–7, 2011.
tivises energy autonomy with significant feed-in tariffs [4] A. Peacock and M. Newborough, “Controlling micro-CHP systems to
leading to low payback periods for an initial investment. modulate electrical load profiles,” Energy, 2007.
[5] D. Kumar, F. Zare, and A. Ghosh, “DC microgrid technology: System
However, Austin electricity tariffs incentivise power au- architectures, AC grid interfaces, grounding schemes, power quality, com-
tonomy by power-based tariffs, which lead to higher pay- munication networks, applications, and standardizations aspects,” IEEE
back periods for an initial investment. Access, vol. 5, pp. 12230–12256, 2017.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
BANDYOPADHYAY et al.: TECHNO-ECONOMICAL MODEL BASED OPTIMAL SIZING OF PV-BATTERY SYSTEMS FOR MICROGRIDS 1667

[6] V. Vossos, K. Garbesi, and H. Shen, “Energy savings from direct-dc in US [30] J. H. Yoon, R. Baldick, and A. Novoselac, “Dynamic demand response
residential buildings,” Energy Buildings, 2014. controller based on real-time retail price for residential buildings,” IEEE
[7] T.-F. Wu, Y.-K. Chen, G.-R. Yu, and Y. Chang, “Design and development Trans. Smart Grid, vol. 5, no. 1, pp. 121–129, Jan. 2014.
of dc-distributed system with grid connection for residential applications,” [31] Y. Del Valle, G. K. Venayagamoorthy, S. Mohagheghi, J.-C. Hernandez,
in Proc. IEEE 8th Int. Conf. Power Electron. ECCE Asia (ICPE & ECCE), and R. G. Harley, “Particle swarm optimization: Basic concepts, variants
IEEE, 2011, pp. 235–241. and applications in power systems,” IEEE Trans. Evol. Comput., vol. 12,
[8] U. Boeke and M. Wendt, “DC power grids for buildings,” in Proc. IEEE no. 2, pp. 171–195, Mar. 2008.
1st Int. Conf. DC Microgrids (ICDCM), IEEE, 2015, pp. 210–214. [32] C. A. C. Coello et al., Evolutionary Algorithms for Solving Multi-objective
[9] E. Rodriguez-Diaz, J. C. Vasquez, and J. M. Guerrero, “Potential energy Problems. Springer, 2007.
savings by using direct current for residential applications: A Danish [33] P. Street, “Pecan street online database,” Pecan Street: Austin, TX, USA,
household study case,” in Proc. IEEE 2nd Int. Conf. DC Microgrids 2016.
(ICDCM), IEEE, 2017, pp. 547–552.
[10] T. Beck, H. Kondziella, G. Huard, and T. Bruckner, “Assessing the influ-
ence of the temporal resolution of electrical load and PV generation profiles
on self-consumption and sizing of PV-battery systems,” Appl. Energy,
vol. 173, pp. 331–342, 2016.
[11] E. Tervo, K. Agbim, F. DeAngelis, J. Hernandez, H. K. Kim, and Soumya Bandyopadhyay received the B.Tech. de-
A. Odukomaiya, “An economic analysis of residential photovoltaic sys- gree (Hons.) in electrical and electronics engineer-
tems with lithium ion battery storage in the United States,” Renewable ing from Jadavpur University, Kolkata, India, in
Sustain. Energy Rev., vol. 94, pp. 1057–1066, 2018. 2011 and the M.Sc. degree in electrical engineer-
[12] J. Tant, F. Geth, D. Six, P. Tant, and J. Driesen, “Multiobjective battery ing in 2015 from the Delft University of Technol-
storage to improve PV integration in residential distribution grids,” IEEE ogy, Delft, The Netherlands. Since 2016, he has
Trans. Sustain. Energy, vol. 4, no. 1, pp. 182–191, 2013. been working toward the PhD degree in of key
[13] W. L. Schram, I. Lampropoulos, and W. G. van Sark, “Photovoltaic power electronics in low voltage dc distribution
systems coupled with batteries that are optimally sized for household systems. His research interests include multi-port
self-consumption: Assessment of peak shaving potential,” Appl. Energy, dc-dc converter design for renewable sources and
vol. 223, pp. 69–81, 2018. storages, smart charging of electric vehicles, and
[14] J. Moshövel et al., “Analysis of the maximal possible grid relief from wireless power transfer.
PV-peak-power impacts by using storage systems for increased self-
consumption,” Appl. Energy, 2015.
[15] R. Atia and N. Yamada, “Sizing and analysis of renewable energy and
battery systems in residential microgrids,” IEEE Trans. Smart Grid, vol. 7,
no. 3, pp. 1204–1213, 2016. Gautham Ram Chandra Mouli received the bach-
[16] T. Dragičević, H. Pandžić, D. Škrlec, I. Kuzle, J. M. Guerrero, and elor’s degree in electrical engineering from the Na-
D. S. Kirschen, “Capacity optimization of renewable energy sources and tional Institute of Technology Trichy, Tiruchirappalli,
battery storage in an autonomous telecommunication facility,” IEEE Trans. India, in 2011, and the master’s degree in electrical
Sustain. Energy, 2014. engineering from the the Delft University of Tech-
[17] S. Kahrobaee, S. Asgarpoor, and W. Qiao, “Optimum sizing of distributed nology, Delft, The Netherlands, in 2013, and the
generation and storage capacity in smart households,” IEEE Trans. Smart Ph.D. degree from the Delft University, Delft, The
Grid, vol. 4, no. 4, pp. 1791–1801, 2013. Netherlands, in 2018, for the development of a solar
[18] M. Bortolini, M. Gamberi, and A. Graziani, “Technical and economic de- powered V2G electric vehicle charger compatible
sign of photovoltaic and battery energy storage system,” Energy Convers. with CHAdeMO, CCS/COMBO and designed smart
Manag., vol. 86, pp. 81–92, 2014. charging algorithms. He is currently an Assistant
[19] L. Xu, X. Ruan, C. Mao, B. Zhang, and Y. Luo, “An improved optimal Professor with the Department of Electrical Sustainable Energy, Delft University
sizing method for wind-solar-battery hybrid power system,” IEEE Trans. of Technology, Delft, The Netherlands. The project was in collaboration with
Sustain. Energy, vol. 4, no. 3, 2013. PRE Power Developers, ABB, and UT Austin. His current research focuses on
[20] G. C. Mouli, P. Bauer, and M. Zeman, “System design for a solar powered electric vehicles, EV charging, PV systems, power electronics, and demand-side
electric vehicle charging station for workplaces,” Appl. Energy, vol. 168, management. He was awarded the Most Significant Innovation in electric
2016. vehicles at the IDTechEx Show 2018 and the Best Tech Idea of 2018 by KIJK.
[21] M. O. Ramoni and H.-C. Zhang, “End-of-life (eol) issues and options for From 2017 to 2019, he was a Postdoctoral Researcher with TU Delft, working
electric vehicle batteries,” Clean Technol. Environmental Policy, vol. 15, on Flexgrid, Trolley 2.0, and Orchestrating Smart Charging in mass Deployment
no. 6, pp. 881–891, 2013. project.
[22] L. Lam and P. Bauer, “Practical capacity fading model for Li-ion battery
cells in electric vehicles,” IEEE Trans. Power Electron., vol. 28, no. 12,
pp. 5910–5918, Dec. 2013.
[23] C. N. Truong, M. Naumann, R. C. Karl, M. Müller, A. Jossen, and
H. C. Hesse, “Economics of residential photovoltaic battery systems in Zian Qin (M’15) received the B.E. degree in au-
germany: The case of teslas powerwall,” Batteries, 2016. tomation from Beihang University, Beijing, China,
[24] D. Akinyele, J. Belikov, and Y. Levron, “Battery storage technologies in 2009, the M.E. degree in control science and en-
for electrical applications: Impact in stand-alone photovoltaic systems,” gineering from the Beijing Institute of Technology,
Energies, vol. 10, no. 11, p. 1760, 2017. Beijing, China, in 2012, and the Ph.D. degree in
[25] Eneco, “Residential electricity tariffs netherlands,” 2017. [Online]. Avail- power electronics from Aalborg University, Aalborg,
able: https://www.eneco.nl/zonnepanelen/saldering/ Denmark, in 2015. He is currently an Assistant Pro-
[26] A. Energy, “Residential electricity tariffs texas,” [Online]. Available: fessor with the Delft University of Technology, Delft,
https://austinenergy.com/ae/residential/rates Netherlands. He was a Visiting Scientist with the
[27] B. Zakeri and S. Syri, “Electrical energy storage systems: A compara- Institute for Power Generation and Storage Systems
tive life cycle cost analysis,” Renewable Sustain. Energy Rev., vol. 42, (PGS), Aachen University, Aachen, Germany, from
pp. 569–596, 2015. April to July 2014. From August 2015 to June 2017, he was a Postdoctoral
[28] J. Li, Z. Wu, S. Zhou, H. Fu, and X.-P. Zhang, “Aggregator service for Research Fellow with Aalborg University.
pv and battery energy storage systems of residential building,” CSEE He is a member of IEEE Power Electronics Society, IEEE Industrial Elec-
J. Power Energy Syst., vol. 1, no. 4, pp. 3–11, 2015. tronics Society, and IEEE Industry Application Society. His current research
[29] F. Shen, S. Huang, Q. Wu, S. Repo, Y. Xu, and J. Østergaard, “Comprehen- interests include application of wide bandgap devices, high frequency power
sive congestion management for distribution networks based on dynamic electronics, modelling, control and stability of grid-tied power electronics, en-
tariff, reconfiguration and re-profiling product,” IEEE Trans. Smart Grid, ergy storage and battery management systems, renewable energy technologies,
vol. 10, no. 5, pp. 4795–4805, Sep. 2019. DC technologies.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.
1668 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 11, NO. 3, JULY 2020

Laura Ramirez Elizondo (M’06) was born in San Pavol Bauer (SM’07) received the master’s degree
Jos, Costa Rica. She received the bachelor’s degree in electrical engineering from the Technical Univer-
in electrical engineering from the Universidad de sity of Kosice, Kosice, Slovakia, in 1985, and the
Costa Rica in 2003 and the M.Sc. degree (Hons.) in Ph.D. degree from the Delft University of Technol-
electrical power systems from the Delft University of ogy, Delft, The Netherlands, in 1995.
Technology, The Netherlands, in 2007. From Septem- He is currently a Full Professor with the Depart-
ber 2007 to December 2011, she worked on her Ph.D. ment of Electrical Sustainable Energy, Delft Univer-
project. sity of Technology, and the Head of DC Systems,
She is currently an Assistant Professor with the DC Energy Conversion and Storage Group. From 2002
Systems, Energy Conversion & Storage group. She to 2003, he was working partially with KEMA (DNV
is involved in the projects STW Perspektief P13-21: GL, Arnhem) on different projects related to power
Smart Energy Management and Services in Buildings and Grids (SES-BE), KI electronics applications in power systems. He authored or coauthored more
Switch 2 Smart Grids: Flexible and future power links for smart grids (FLINK), than 95 journal and more than 300 conference papers in his field (with H
and NWO URSES project: Gaming beyond the Copper Plate. Her research factor Google scholar 35, Web of science 23). He is an author or coauthor of
interests include integration of renewable sources, direct current networks, eight books, holds four international patents, and organized several tutorials at
microgrids, optimization, and power control. In 2013, she was awarded with the international conferences. He has worked on many projects for industry
the Erasmus Energy Science Award. concerning wind and wave energy, power electronic applications for power
systems such as Smarttrafo; HVDC systems, projects for smart cities such as
photovoltaic (PV) charging of electric vehicles, PV and storage integration,
contactless charging; and he participated in several Leonardo da Vinci and
H2020 EU projects as a Project Partner (ELINA, INETELE, E-Pragmatic) and
a Coordinator (PEMCWebLab.com-Edipe, SustEner, Eranet DCMICRO).
Dr. Bauer is the Former Chairman of Benelux IEEE Joint Industry Applica-
tions Society, Power Electronics and Power Engineering Society Chapter, the
Chairman of the Power Electronics and Motion Control Council, a member of
the Executive Committee of European Power Electronics Association, and also
a member of the International Steering Committee at numerous conferences. He
was a recipient of the title Professor from the President of Czech Republic at the
Brno University of Technology, Brno, Czech Republic, in 2008, and the Delft
University of Technology, in 2016.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:44:26 UTC from IEEE Xplore. Restrictions apply.