Predictive Control Strategy for DC Microgrid

Integrated EV Charging Stations in Oslo

K. Raghavendra Naik Mohan lal Kolhe M. Krishna Siva Prasad
AAU Energy Department Faculty of Engineering and Science Asst. Prof at Computer Science and
Aalborg University & Asst. prof in NIT University of Agder Engineering Department
Jamshedpur Kristiansand, Norway SRM university, AP.
Aalborg, Denmark [email protected] Guntur, Andhra Pradesh
[email protected] [email protected]
2024 IEEE 4th International Conference on Sustainable Energy and Future Electric Transportation (SEFET) | 979-8-3503-8399-7/24/$31.00 ©2024 IEEE | DOI: 10.1109/SEFET61574.2024.10718036

Juan C. Vasquez
Gibran David Agundis Center for Research on Microgrids Josep M. Guerrero
Center for Research on Microgrids AAU Energy Department Center for Research on Microgrids
AAU Energy Department Aalborg University AAU Energy Department Aalborg
Aalborg University Aalborg, Denmark University
Aalborg, Denmark [email protected] Aalborg, Denmark
[email protected] [email protected]

Abstract Norway holds highest per capita Electric vehicle The researchers in [3] have employed a second-life
(EV) shares globally. To meet the recharge demand from such battery storage units to mitigate peak loads caused by EV
large EVs fleet, fast charging infrastructure is essential. But, charging at ultra-fast charging stations. While the research
the fast charging stations are unable to fulfill the charging in [4] reveals that peak load of Germany power grid with
request of EV owners due to restricted grid power supply. EVCS is reduced by 10% through the integrated hydrogen
Peak shaving has partially addressed this issue with a solar
storage-wind-solar PV infrastructure.
PV-Battery energy storage system (BESS) integrated
Microgrid (MG) configuration. But, maintaining the reliability
and stability of Microgrid (MG) against intermittent EV In [5, 6], an optimal power flow management is
arrivals during peak load of the grid is a critical objective. To proposed for EV charging station supplied by a grid-
address this issue a Micro hydro generator-Solar PV-BESS integrated solar PV-Battery energy storage (BESS) DC
integrated DC MG with iterative predictive control (ITPC) microgrid. Reinforcement learning with Markov Decision
algorithm is proposed in this paper. With the prediction of EV Process is utilized to solve the optimal power flow problem
load arrivals on iteration basis, the proposed algorithm with a reduction in peak load as the objective. Similarly, [7]
operates the Micro hydro generator in such a way that EV load addresses the same objective using a Genetic Algorithm,
is supplied under the peak load condition with optimal
dependency on BESS. The performance of the proposed
alleviating peak loading on the grid through a solar PV-
strategy evaluated against the real time EV load data of Oslo as Wind-BESS microgrid system.
a case study. The proposed strategy achieves 9.3% improved
fall in DC-link voltage and 67.735% reduced depth of The Danish researchers in [8, 9], investigated the cost
discharge of BESS. benefits of deploying BESS in the grid connected DC Fast
charging stations. Also, experimental study in [8, 9] reveals
Keywords Electric Vehicle arrival rate, Norway EV that with 437kWhr BES, the peak load on grid due to 5*150
charging station, Peak shaving, Micro hydro generator, kW FCS is reduced by 30%.
Predictive control of DC Microgrid, DOD In [10], peak loads resulting from EV charging at FCSs
I. INTRODUCTION are reduced by strategically charging vehicles based on their
charging requirements, utilizing a solar PV-diesel generator
The global shift towards zero-emission sustainable and BESS as the energy supply system.
transportation marks a significant stride in combating CO2
emissions, air quality issues, and oil dependence. In line The research methodologies outlined in [3-10] primarily
with zero emission policy and with a vision to improve oil concentrate on peak shaving through solar PV and BESS
exports, Norway is implementing incentive-driven economic solutions. However, in real-time scenarios, peak shaving
policies to increase the presence of electric vehicles (EVs) concepts can potentially generate marginal profits for
on its roads. As per the report of [1] for 2022, Norway has Electric Vehicle Charging Stations (EVCSs). Additionally,
approximately 70,000 EVs, with 87% of new car sales being these methodologies do not address the impact of
fully electric. However, according to [2], it is projected that intermittent EV arrivals during peak load conditions on the
on linear scale, energy consumption of transportation sector Microgrid stability and reliability.
consumes 23.8% extra energy from 2012 to 2030. While the research presented in [11-13] has considered
the effect of transient loading on DC MG. But these
The rise in EV adoption highlights the need for a strong strategies have not experimented on real time EV load data.
Fast charging (FC) infrastructure to accommodate the Apparently, the researchers have not addressed the impact
growing number of vehicles and the expected rise in energy of intermittent arrivals of EVs at EVCS during peak load
consumption. But the intermittent arrivals of EVs at (electric condition on DC MG. An iterative predictive control
vehicle charging station) EVCS have detrimental impact on algorithm is designed to reduce the detrimental impact of
peak electricity demand of distribution grid. A critical intermittent EV arrivals during peak condition. To supply
review of research methodologies which addressed this the peak load of EVCS, a Micro hydro generator (MHG)-
objective are reviewed in this section. solar PV-BESS integrated DC MG is proposed. The

979-8-3503-8399-7/24/$31.00 ©2024 IEEE

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.
 proposed strategy considered the real time EV load data of The detailed modeling parameters of solar PV are given in
Oslo, Norway given in [14] and its performance is evaluated [13].
on this data.
C. Modeling of BESS
II. MODELING OF DC MG
The instantaneous power output of BESS to compensate
In this section, mathematical modeling of Microgrid the untracked EV load is given as
constitutes to demonstrate the proposed objectives of the
proposed work are discussed. The primary component of the
PBT t I BT t VOC I BT (t ) Rint (6)
DC MG structure shown in Fig. 1 is MHG. MHG comprise,
Semi-Kaplan Turbine, PI governor control loop, rectifier
and DC-DC converter as shown in Fig.1. The PI driven Internal resistance of BESS, Input current of
control loop in Fig. 2 sets the mechanical power to be BESS, Open circuit voltage which is
delivered by hydro turbine. With reference to this power, The state of charge of BESS is
MHG generates AC power output. To supply DC output the
EVCS load, a uncontrolled AC-DC rectifier with supportive (7)
DC-DC converter for regulating the power level at certain
voltage is deployed as shown in Fig. 2. For a given capacity, BESS compensates the unbalanced EV
A. Modeling of MHG load in such a way that using the cascaded
The modeling equations regarding power conversion of control loop as shown in |Fig. 2. The cascade loop equations
hydro generator are given as follows to with abiding the rule is given as

(8)
(1)
(2)
Where, , , ,
The mechanical response time of MHG can be expressed as
(3)
Mechanical power developed by hydraulic system
(Watt)
Ideal and real gate opening of servo mechanism
Torque developed by PMSG (Nm)
Height of drafts tube (m)
Water velocity (m/sec)
Fig. 1 Microgrid topology proposed for Oslo case study
3
Flow rate of the water (m /sec),
Area of draft tube (m2)
The electrical output power of DC-DC converter is given as
(4)

Output voltage and current of uncontrolled

converter,
The detailed modeling of MHG is given in [13].
B. Modeling of solar PV
The modeling of solar PV which drives the MPPT in
maximum power point mode is given in (5). In this work, Fig. 2 Control configuration for the proposed MG
solar PV never operates in the voltage control regardless of topology
the weather condition.
The instantaneous duty cycle of BESS controller is set based
As the temperature of Oslo is cold, the solar PV output on the following equation
cannot deliver the maximum power relative to its installed (9)
capacity. But for the given irradiance value, it operates in
the MPP mode with the following power output
Tuning the duty cycle using the current control law
ensures shown in Fig. 2.
(5)

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.
 III. PROPOSED CONTROL STRATEGY the grid support and loss of EV load are the two solutions to
The prosed control strategy destined to regulate the EV constitute the voltage regulation within the limits. The same
load through the active operation of MHG against the is shown in the Fig. 3. The distributive mathematical
transient arrival rates of EVs at EVCs. The active power function based on which ITPA predicts the EV load arrivals
contribution of MHG to regulate EV load is evaluated at the EVCS on iteration basis is given below.
through DC link voltage control. The preciseness of the
n 1 PEVj 2 n 2
1
PEVn 1 PEVj PEVj
j 1` n 1 j
n 2 j n 1` j 1` j

2
1 j 1
1 PEVj
n 3. j n 1
`n 3 j
PEVn 1 PEVn 5 ax1 PEVn 2 ax 2 PEVn 4
n 4 j
PEVn 1 ax3 PEVn 2 ax 4 PEVn 3 ax5 PEVn 4 PEVn 5
n 5 j

(10)
Where

EV load demand before steady state

no. of EV load dynamics before steady state,
Iteration number of ITPC
Fig. 3 Control framework of proposed ITPC Algorithm
The distributive constants of ITPA for 6 EV load dynamics
are given as
prediction control against the EV load is evaluated through
BESS power output control. In this section, the control ax1 2
framework to achieve the aforementioned objective of EV ax2 2
load regulation is going to be discussed. a x3 4
ax4 4
The control paradigm shown in Fig. 2 tracks the EV a x5 4
arrivals at EVCS periodically for
possible uneven transient load condition. The senses the
The dispatchable power output of MHG supervised through
uneven situation through monitoring BESS power output
ITPC algorithm is demonstrated below
limit PBT t 0.8 5PBTmax . This indicates the generation-
load balance. The control algorithm is designed to operate PEVn PEVn 1
the MHG without intervene of grid operator to balance the PMHGKref j n 1
PPV (t )
EVCS load. As the EV arrivals are transient in nature, a PEVn 1 (t ) CPAdJ j
prediction algorithm uses the prediction model defined in j 1
the (10) to predict these load fluctuations. The constraints in (11)
the algorithm are shown in Fig 3. If the prediction
over/under fit the actual EVCS load, the BEES may get The modeling of APFC to incorporate adjustable power
overcharged or under charged. To alleviate this situation, an flow is given as
adjustable power flow control (APFC) is designed. Its role is
demonstrated in Fig. 2. The transfer function which drives 1
L Q 1 q1 | PBT max |
APFC is given in (12). PADJ L 1
Q 1 sTADJ L 0
(1 q1 ) 0.55 | PBT max |

The excessive arrival rates of the EVs due to range

anxiety, low ambient temperature of the BESS, insufficient (12)
EV charging may cause the unbalanced hydro power ; in charging mode or
generation-EV load. This situation is identified by 85% discharging mode.
exceeded BEES power as shown in Fig. 3. To regulate , Else
Q= No. of power stamped steps for APFC,

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.
 Response time of each adjusted power flow If occurs and arrival rates of the
Constraints based on which ITPC senses the transient EV less the defined load limit then the following equation
arrivals of EVs at EVCS are given as follows holds true

(13)

If & and arrival The EV charging reveals that around 3:00 AM 6:00
rates of the EV exceed the defined load limit then the AM, the charging status low and after 6AM a gradual
following condition holds true increment in the load curve can be observed in each month.
Keen observation of Fig. 4 reveals that for 14-18hrs, the
(14) charging events reach the peak value and stand stable for
couple of hours and then drop in the evening hours. But,
IV. RESULTS AND INTERPRETATION minority demands around 12AM-1AM are the customers
In this section, the results regarding the proposed strategy who really want to charge their EVs at low EV price.
against the randomized behavior of the EV load are
demonstrated. Prior to the demonstration of the evaluation In this work, the peak demand in Fig. 4 for 10min-24min
of the proposed methodology, it is important to analyze the for the coldest month is considered and the ITP control
dynamic characteristic EV load. Charging data recorded for a
Oslo as case study is presented in Table I. intermittent transient load.
The EV charging load curves shown in Figure 4 presents
average EV charging record for each day of respective
month.

Fig. 4 EV charging load curve of Oslo, Norway

Fig. 5 (a) EV load dynamics with cumulative charging events for case 1, (b) Solar PV dynamics, (c) MHG output power, (d)
BESS dynamic response, (e) BESS output current, (f) DC- link voltage

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.
 Fig. 6 (a) EV load dynamics with cumulative charging events for case 2, (b) Solar PV dynamics, (c) MHG output power, (d)
BESS dynamic response, (e) BESS output current, (f) DC- link voltage

The presented research work in this paper captures two power demand of [0.3 0.58 0.75 0.9 1] p.u. at [0.5 2 3.2 4.2
windows out of peak load dynamics of EVCS as shown in 4.9]min, the ITPA let the MHG deliver [0.38 0.8 0.95] p.u.
Fig. 4. The load curve in Fig. 4 represents the average load with 2.5 min response time for iterations K=2, 3,4
at a charging site. But, the windows captured during the respectively. The observation of the same can be seen in
peak load at multiple charging sites are shown in Fig. 4. The Fig. 5(c).
same windows are simulated and the results for the same are In Fig. 5 (d) & Fig. 6 (c) it can be observed that due to
shown in Fig. 5 and Fig. 6. In these two windows the arrival fast acting response characteristics of MHG during transient
rates of EVs are highly instinctive. Moreover, the nature of EV load, the battery has never over discharge throughout
the charging events is highly stochastic in nature. the transient period. With lack of prediction, the application
of control strategies proposed in [15-17] for case 1 and case
Table I. The charging site data characteristics 2 has loaded the BESS to maximum discharging power. The
wave forms without control strategies refer to strategies
The peak load of the charging site =290 kW proposed in [15-17] and the outcomes with these are
No. of charging events=30722 represented with red in Fig. 5 & Fig. 6. In Fig. 5(f) & Fig.
The peak load simulated=165kW 6(e) it can be observed that as the ITPA has balanced the
Average charging power hydro power generation and power demand of EV arrivals,
consumption=10kwhr/EV owner the stand stable through the dynamics. Whereas, the
Charging events observation period= 19 min existing control strategies fail to do maintain the stable DC
link voltage due the aforementioned incompetent feature.
Charging period simulated during the peak To showcase the potential observation with and without
load= 10-24 min applying the prediction strategy, a relative comparison is
demonstrated in Table II.
The EV load dynamics in peak load condition for 24 The formulation of evaluating parameters are given as
minutes interval is shown in Fig.5 (a). As it can be observed follows
in Fig. 5(a) that the load is fluctuating with [0.1 0.2 0.295
0.38 0.45 0.6 0.74 0.88 0.69 0.5 0.7] p.u. magnitude at [4.7 Average Energy Disch arg ing by BESS during peak load
5.5 7 9.4 11.7 121. 14 15.9 17.5 18.8] minutes respectively. DOD
Battery capacity in kWhr
At t=11.7min, the sudden fall in p.u. cause the
(15)
abnormal situation with battery storage exceeds 85% limit.
Utilization of MHG Energy
This triggered the ITPA control algorithm supervise the
MHG based on the load fluctuations before 11.7min. Based Average Energy Dispached during peak load condition
on the p.u, the ITPA as Time period MHG capacity in kWh
response time as shown in Fig. 5 (c). In the similar manner, (16)
for case 2, against the intermittent EV load arrivals with

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.
 https://www.nve.no/energy-consumption-and-efficiency/
(accessed Nov. 08, 2018).
% Averageloading of BESS [3] -in electric vehicle
Average disching current during peak load
100 no. 3, pp. 193 201, 2018.
duration of peak Maximum discharging current [4] D. Luca de Tena and T.
(17) a future renewable energy-based power system in europe with a
Table II. Qualitative comparison between with and without vol. 42, no. 8, pp. 2670 2685, 2018.
ITPA [5] Q. Chen and K. Folly, "Impact of Renewable Energy and
Dynamic Pricing on Electric Vehicles Charging," 2022 30th
Southern African Universities Power Engineering Conference
Parameter Without With % (SAUPEC), Durban, South Africa, 2022, pp. 1-6
Strategy Strategy Improvement [6] S. Paudyal and S. Dahal, "Impact of Plug-in Hybrid Electric
MHG 0.49410 0.729 23.49 Vehicles and their optimal deployment in Smart Grids," AUPEC
2011, Brisbane, QLD, Australia, 2011, pp. 1-6.
utilization [7]
(p.u.)
DOD of 92.295 24.56 67.735 storage systems. Int. J. Electr. Power Energy Syst. 105, 46 58
BESS (%) (2019)
[8] Gjelaj1: M. Gjelaj, C. Træholt, S. Hashemi and P. B. Andersen,
% Average 92.295 24.56 67.735 "Optimal design of DC fast-charging stations for EVs in low
loading of voltage grids," 2017 IEEE Transportation Electrification
BESS Conference and Expo (ITEC), Chicago, IL, USA, 2017, pp. 684-
689
% Average 9.3 0 9.3
[9] Gjelaj2: Gjelaj, M., et al.: Multifunctional applications of
Fall in
4873
% Fall in DC link voltage (2019)
[10] Mehrjerdi, H., Hemmati, R.: Electric vehicle charging station
Average volatge during peak load
1 100 diesel generation incorporating comfort of drivers. J. Energy
Nominal volatge
Storage. 26, 100924 (2019)
(18) [11] K Raghavendra Naik, Bhooshan Rajpathak, Arghya Mitra,
Mohan Lal Kolhe,"Assessment of energy management
The Qualitative comparison reveals that with 23.049%, technique for achieving the sustainable voltage level during grid
outage of hydro generator interfaced DC Micro-Grid," in
67.735% and 9.3% improvement in utilization of Micro Sustainable Energy Technologies and Assessments, pp. 1-18,
hydro generation, DOD and fall in DC link voltage, the vol. 46, April. 2021. Doi:10.1016/j.seta.2021.101231.
proposed ITPA has outperformed the existing control [12] K Raghavendra Naik, Bhooshan Rajpathak, Arghya Mitra,
strategies to control the peak load demand. Consequently, Mohan Lal Kolhe,"Adaptive energy management strategy for
sustainable voltage control of PV-hydro-battery integrated DC
enhanced battery life and stable voltage control are achieved microgrid," in Journal of Cleaner Production, pp. 1-18, vol.315,
despite the intermittent EV charging during peak load June. 2021. Doi:10.1016/j.jclepro.2021.128102.
condition. [13] K Raghavendra Naik, Bhooshan Rajpathak, Arghya Mitra, C.
Sadanala, Mohan Lal Kolhe,"Power management scheme of DC
V. CONCLUSION micro-grid integrated with photovoltaic - Battery - Micro hydro
power plant," in Journal of Power Sources, pp. 1-16, vol.525,
The proposed Iterative predictive control algorithm in this Feb. 2022. Doi:10.1016/j.jpowsour.2022.230988.
research work has achieved the objective of supplying the [14]
charging demand of EV charging station with intermittent Peak shaving through a battery energy storage A case study
-80.
EV arrival rates during peak load scenario. The assessment [15] S. Kewat, B. Singh and I. Hussain, "Power management in PV-
of presented modeling results against the real time EV load battery-hydro based standalone microgrid," in IET Renewable
curves of Oslo reveals that the proposed strategy as Power Generation, vol. 12, no. 4, pp. 391-398, 19 3 2018.
outperformed the existing control strategies with 23.049%, [16] R. Sharma, S. Kewat and B. Singh, "Robust 3IMPL Control
Algorithm for Power Management of SyRG/PV/BES-Based
67.735% and 9.3% improvement in utilization of Micro Distributed Islanded Microgrid," in IEEE Transactions on
hydro generation, depth of discharge (DOD) and fall in DC Industrial Electronics, vol. 66, no. 10, pp. 7765-7777, Oct. 2019.
link voltage respectively. The low DOD achieved in this [17] R. Sharma, S. Kewat and B. Singh, "SyRG-PV-BES-Based
research work guarantee enhanced life cycles of battery Standalone Microgrid Using Approximate Multipliers Based
Adaptive Control Algorithm," in IEEE Transactions on Industry
storage. As futuristic direction, the presented research work Applications, vol. 56, no. 3, pp. 2913-2924, May-June 2020.
can further be extended to study the economic feasibility of
hydro generation centered DC MG for supplying the FCS in Appendix:
Oslo. Also, the researchers are destined to present the RNN Table III. Parameters considered for the modeling
Parameter Capacity Base
learning based prediction and price control strategy to Peak load on EV 290 kW 290 kW
alleviate the peak loading on grid due to FCSs. charging station
Micro hydro 100 kW- case 2 100 kW- case 2
generator 150 kW- case 2 150 kW- case 2
REFERENCES Dynamic EV load 100 kW- case 2 100 kW- case 2
165kW- case 2 165kW- case 2
[1] Vox, "Electric Vehicles" https://www.vox.com/electric-vehicles.
Battery capacity 30 kW 30 kW
(accessed June 2, 2022).
[2] The Norwegian Water Resources and Energy Directorate,

Authorized licensed use limited to: GOVERNMENT COLLEGE OF ENGINEERING. Downloaded on November 18,2024 at 06:29:39 UTC from IEEE Xplore. Restrictions apply.