A Low Cost Distributed Solar DC Nanogrid: Designand Deployment with Remote Monitoring Unit

Rustam Kumar, Ranjan Kumar Behera, Senior Member IEEEDepartment of Electrical EngineeringIndian Institiute of Technology Patna

Patna, IndiaEmail: rustamkumar22@gmail.com, rkb@iitp.ac.in

Abstract—In this paper, a low cost islanded dc microgridarchitecture is proposed for rural electrification in developingcountry. The proposed architecture is superior to existing archi-tecture in comparison to 1) High distribution efficiency (due todistributed generation and distributed storage) 2) Provide powerto communal load (without large central power generation) 3)Less extensive control and reliable power management scheme(because of distributed control using dc-bus signaling) 4) Low costonline real-time monitoring scheme. The architecture is composedof several nanogrid, which have self-generation, storage and bi-directional power flow capability inside the system. The detailedanalysis of individual nanogrid is carried out in this paper. Usingmathematical modeling of the nanogrid, a less extensive controlwith reliable power management scheme is implemented alongwith low-cost monitoring scheme. To validate the compatibilityof the schemes with the microgrid architecture, a scale downprototype is developed in the laboratory and experimental resultsare discussed.

Index Terms—DC nanogrid, distributed generation, rural elec-trification, solar PV.

I. INTRODUCTION

Most of the rural areas in India are disconnected from theutility grid and even if the grid is connected, there is frequentpower cut, power fluctuation and voltage sag due to powermismatch between generation and demand. To cater the powerdemand in rural areas, a photovoltaic (PV) power generation isthe most attractive and reliable source [1],[2]. Recent researchproved that renewable energy fueled microgrids can be moreefficient, reliable and cost-effective [3]. Further, the efficiencyand reliability of the microgrid can be achieved using dcdistribution system [4]. Due to its inherent advantages suchas the management challenges of frequency, phase, reactivepower as well as the synchronization and harmonics in acmicrogrids do not exist in dc based microgrids [5]. Hence, thedc microgrid is beneficial for the small-scale commercial andresidential applications.

Due to aforementioned advantages of a dc microgrid, re-searcher has proposed different architectures for rural elec-trification in the literature 1) Central generation with centralstorage architecture (CGCSA) [6]-[9], 2) Central generationwith distributed storage architecture (CGDSA) [10],[11]. The

Authors acknowledge the support provided by the Ministry of HumanResource Development and Ministry of Power, Govt. of India under IMPRINTProject. Sanction letter no.: F. No. 1-18/2015-TS-TS.I.

Fig. 1: Islanded microgrid architecture.

central positioning architecture is typically beneficial for theperspective of control, where generation, distribution, andreliability of the system can be easily monitored. However,this results in higher losses in generation and distribution alongwith rigidity in future expansion [12]. More critically in theterm of capital cost, the system requires higher deploymentcost because of central generation of PV panel. One realapplication project Mero Goa Power stand out based on thisarchitecture in the India, where a centralized generation anddistribution system is implemented in a village to provideessential dc power up to 5 W (24 V) to each house at the limitof 0.2 amp-enough for two led and a mobile charging point.Due to the limited power supply, this architecture unlikelyto elevate villagers socio-economical condition [13]. If thisscheme is implemented at high power (50W) then the lossesassociated with the system will be higher, therefore, thisscheme is unlivable. In this [10],[11] a CGDSA based systemis proposed. Due to distributed storage, its result in higherefficiency than CGCSA but this system suffers from highdistribution losses and its require high capacity PV generationwhich results in higher installation cost. Several other studieshave been also done in the literature [14]-[16] in perspective ofadvancement in control but these increases the complexity andcost of the system which is unsuitable for rural electrification.978-1-5386-6159-8/18/$31.00 © 2018 IEEE

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India

(a) Circuit diagram of the nanogrid. (b) Time-Average modeling diagram of the nanogrid

Fig. 2: The nanogrid diagram

Considering the limitation of 1) Low distribution efficiency2) Unable to provide power to communal load 3) Complexcontrol 4) Inaccessibility of the system. A distributed genera-tion and distributed storage architecture (DGDSA) microgridis discussed with a distributed control and real-time monitoringscheme [17]. The architecture consists of several nanogrid witha communal load illustrated in Fig. 1.

This paper primarily discussed modeling of converters, con-verters control, power management and a monitoring schemeof the individual nanogrid system. The converter control andpower management scheme works based on dc-bus signalingmethod [18], which reduces the extra communication linkamong converters and makes the system more reliable. Theproposed monitoring scheme uses a low-speed communicationprotocol. To validate the proposed schemes, a scale downprototype of a dc nanogrid with less extensive controllers, aredesigned in the laboratory and tested by imitating differentenvironmental and loading conditions.

II. DESCRIPTION OF THE MICROGRID

The microgrid architecture consists several distributednanogrids and communal loads, which are connected to a com-mon dc-bus, where individual dc nanogrid is composed usingrenewable source, battery energy system and local load shownin Fig. 2(a), where renewable generation (PV) is connectedto the dc-bus using a dc-dc Converter1. The dc-dc converteris employed using boost converter with a suitable controllerusing Pulse Width Modulation (PWM) technique is to extractmaximum power from the solar panel at intermittent solarirradiance. A Battery Energy Storage (BES) is also integratedto the dc-bus using a bidirectional buck-boost converter, whichis charged and discharged. The charging and discharging ofthe BES is controlled by their controller. A local load isconnected to the dc-bus using a buck converter to supply powerat low rated voltage by the way medium rated voltage load orcommunal load can be also directly connected to the dc-bus,if the load voltage rating matches with the dc-bus voltage.

A. Mathematical Modeling

In this section, a mathematical modeling of the nanogridhas been developed. For the simplicity, an averaged method isused for developing the average dynamic model and neglected

the switching frequency. An average model of the nanogridis illustrated in Fig. 2(b), where d1, d2, d3, and d4 arecorresponding duty cycle of the converters. To derive themathematical modeling, KVL and KCL are applied to eachconverter model. For boost converter, the dynamic equationof the converter can be written as (1-2). Similarly, for bidi-rectional converter in boost operation, buck operation can bewritten as (3-4), (5-6) respectively, where d2 and d3 are dutyratio in respective modes and finally the dynamic equation forbuck converter, dc bus is derived given in (7-8), 9.

For Boost Converter

dILp

dt= −ILpRp

Lp− (1− d1)vd

Lp+

vpLp

(1)

dvpdt

=IpCp− ILp

Cp(2)

For Bi-directional ConverterIn Boost Mode:

dILB

dt= −ILBRb

Lb− (1− d2)vd

Lb+

vbLb

(3)

dvbdt

=IbCb− ILB

Cb(4)

In Buck Mode:dILb

dt= −ILbRb

Lb− vb

Lb+

d3vdLb

(5)

dvbdt

=ILb

Cb+

IbCb

(6)

For Buck ConverterdILL

dt= −ILL

RL

LL− vL

LL+

d4vdLL

(7)

dvLdt

=ILL

CL− vL

RLoadCL(8)

After applying KCL at dc-bus; where, Cd=C1+C2+C3;

dvddt

=(1− d2)ILB

Cd+

(1− d1)ILp

Cd− d3ILb

Cd

−d4ILL

Cd− vd

RdCd

(9)

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India

(a) Boost converter controller. (b) Bidirectional converter controller.(c) Buck converter controller.

Fig. 3: Controller diagram of the nanogrid.

B. Controller Design

A controller is designed for each dc-dc converter such thatthe system will operate in stable region. To control the boostconverter, a controller is designed using Proportional Integral(PI) shown in Fig. 3(a). The Gv1, Gv2 are PI controllers tunedat considering sufficient gain and phase margin. This controllerhas two modes of Constant Voltage (CV) and Maximum PowerPoint Tracking (MPPT) operation, which is selected basedon the power management scheme. In CV control mode, dc-bus voltage (Vd) is regulated at a constant voltage. For this,a reference signal (V ∗

d ) is given to the subtractor block andcompared with Vd and the generated error signal is given to thecontroller block (Gv1), which generates duty ratio for Conv1.Similarly, in MPPT mode, PV voltage (Vp) is compared withPV MPP voltage (Vmpp) and the generated error signal is givento the controller (Gv2), which further generates duty cycle forthe converter.

A complete controller diagram of the bidirectional buck-boost converter is shown in Fig. 3(b), where two modes ofcharging and discharging control have been used. The chargingcontrol becomes active during battery charging mode, whiledischarging control active during discharging of the battery. Inthe charging mode, battery reference current (I∗b ) is generatedby, the comparing battery fully charged voltage (VbH ) andactual battery voltage (Vb). The error between this is givento controller block. That further generates a reference currentin charging mode. In discharging mode, battery referencecurrent is obtained by comparing dc-bus voltage between thereference bus voltage and given to the associated controller.That generates battery reference current in discharging mode.A current limiter is used in both modes to prevent excessivedrawn or injected current into the battery. And finally, thegenerated reference currents is selected using selection block.The selected reference current control the battery current. Forthis, an inner current control block is implemented, where thereference current is compared with battery current (Ib) andgiven to the controller block (Gi), which further generatesduty signal for the converter.

For the buck converter, a closed-loop voltage control methodis used to maintain a constant voltage at the load. Thecontroller diagram is shown in Fig. 3(c), where load referencevoltage (V ∗

L ) is compared with load voltage (VL) and givento the controller, which furthers generates duty cycle for thebuck converter.

C. Power Management Scheme

A power management scheme is implemented based ondc-bus signaling to manage power inside the system underdifferent environmental and loading condition. The operationof the system is divided based on PV module power generationand it is categorized into three different modes of operations(CV, MPPT +Battery charging, MPPT + Discharging). WhenPV generated power is more than the required local powerthen PV panel operates in CV mode. Means if the powergeneration in the system is more than the required loadpower and the battery is fully charged. In this condition,the mismatched power cannot be compensated by the battery.Therefore, the generated PV power is reduced by changing theboost converter controller mode of operation to CV mode. Inthis mode, the boost converter maintains a constant 1.05Vdn

and deliver power to the dc-bus.In other hand, if the generated power is more than required

local power and battery is not fully charged then the mis-matched power can be compensated by charging the battery.In this mode of operation, the PV operates at MPP and batteryis being charged by regulating a constant Vdn, which is lesserthan the above constant 1.05Vdn. However, if the generatedpower is less than the required power then the extra amountof power is supplied using battery by maintaining a constantdc-bus voltage 0.95Vdn. These three modes of operation areidentified using dc-bus voltages signaling method shown inFig. 4, where the system mode of operation is detected bymeasuring the dc-bus voltage. The detailed mode of operationis given below, where the nominal dc-bus voltage is consideredas Vdn=40 V.

Fig. 4: Operation modes based on dc-bus voltage.

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India

Fig. 5: Proposed low-speed communication scheme.

Mode I [1.075Vdn − 1.025Vdn]: In this mode, if the batteryis fully charged or in light load condition. The generated powercannot be consumed by connected load or battery. Due to thispower mismatch in the system, its causes dc-bus voltage rise.To balance the powers in the system, the generated power isreduced by changing the boost converter operation mode fromMPPT to CV and regulated at V ∗

d =42 V. The bidirectionalconverter remains disabled and buck converter operates inclosed loop voltage control.

Mode II [1.025Vdn − 0.975Vdn]: In this mode of operation,the generated power is more than the demand power andbattery is not full charged then the extra amount of poweris stored in the battery to supply the energy when powergeneration is less. For this, the boost converter operatesin MPPT mode and the bidirectional converter operates incharging mode by maintaining at dc-bus voltage V ∗

d = 40 V.Mode III [0.975Vdn − 0.95Vdn]: Due to the absence

of power generation or less generated power than the localdemand power, the connected load is being supplied usinga bidirectional converter. In this mode of operation bidirec-tional converter regulates the dc-bus voltage by dischargingthe battery and boost converter operates in MPPT mode ordisabled if the generated power is less than the threshold powerby maintaining dc-bus voltage V ∗

d =38 V. The buck converteroperates in closed loop voltage control and maintains loadvoltage VL=12V.

III. MONITORING SCHEME

In this section, an implemented monitoring scheme isdiscussed. System parameters (voltage, current) are sensedthrough sensors at the terminal of the each converter and sentto a cloud server to monitor these parameters. To send thedata at a fast rate and less package drop, it requires highbandwidth, which may not be suitable for the low-cost system.To achieve the same feature using low bandwidth, a schemeis proposed shown in Fig. 5. In this scheme, an algorithmis implemented at each transmitter and receiver sides. Thetransmitter side algorithm only generates an encoded message

if there is sufficient change (ε) in the parameter and sendthe message to a receiver. This may compromise in accuracyof the parameter value but its result in data size reduction,which needs to be transmitted through the communicationchannel. Although the parameter value will be less precisestill it can be used where the requirement of the parametervalue is not much sensitive. The proposed scheme is used fora nanogrid monitoring application. The nanogrid consist ofthree converters at the common dc-bus and these convertersinput, output parameters are important and need to be sent tothe cloud for monitoring real-time status of the system. Forthis, each converter requires a transmitter along with a suitablealgorithm to implement the scheme.

In converter 1, there are two sets of the parameter, wherethe bus parameters usually remain constant for time beingand changes when there is abrupt change in the other systemparameters, however, solar parameter changes indeterminately,so considering into account of this event. Converter1 algorithmgenerates encoded data (|D|A|T |) using a set of changingparameter and its embed zero instead of another set of pa-rameter, which implies that the set of the parameter is notchanging abruptly. Similarly, for converter 2, battery State OfCharge (SOC) changes with the significant amount of time andbattery parameter (ib) can change random, its mostly dependon the burden of the system, therefore the converter2 algorithmsends data (|D|A|), if it sees εs change in the SOC. However,this parameter can be predicted at receiver terminal based onbattery current but it requires extra computational effort, so toreduce the effort, the algorithm detects εs change in SOC andthen generates a set of data. Similarly, in the converter 3, Analgorithm detects a sudden change in the load parameter andgenerates a data (|D|A|). These data are decoded at receiverend using a suitable algorithm and further send it to a cloudserver, where it can be remotely monitored.

IV. EXPERIMENTAL RESULTS AND DISCUSSION

Experimental verification of the proposed system is obtainedthrough a prototype experimental setup in the laboratory as

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India

Fig. 6: Online monitoring portal.

Fig. 7: Hardware prototype.

shown in Fig. 7. The system parameters are provided in TableI. The controller and encoding scheme of each dc-dc convert-ers are implemented using a digital signal processor (DSP)TMS320F240 board and decoding algorithm is implementedon Raspberry Pi. A serial communication protocol is used toestablish a connection between transmitter and receiver. UsingHyper Text Transfer Protocol (HTTP), the data are sent to theweb server. To monitor these data an online monitoring portalis developed as shown in Fig. 6.

Finally, the prototype is implemented in one of the villagesin Araria district, Bihar and tested on the site. To observethe system redundancy, a house with essential loads suchas Led bulbs (25 watt), Mobile charger (15 watt), Fan (15watt), is connected to the system through buck converter andexperimental results are taken as shown in Fig. 8. During theoperation, boost converter is operating in MPPT mode canbe seen from PV output current (ip) in Fig. 8(a). To observeeach converter dynamic performance under different loadingcondition, their inductor voltage, inductor current, output volt-age, and output current are captured. During this experiment,solar is feeding power to the common dc-bus and the dc-bus is feeding power to led bulbs through buck converter.Because of lower power generation than the demand, the

(a) Experimental Results of boost converter i) Inductor voltage (vLp),ii) Inductor current (iLp), iii) dc-bus voltage (vd), iv) PV current (ip)[vLp: 10 V/div, iLp: 1 A/div, vd: 20 V/div, ip: 0.5 A/div, Time: 5 s/div].

(b) Experimental Results of bidirectional converter i) Inductor voltage(vLb), ii) Inductor current (iLb), iii) Battery voltage (vb), iv) Batterycurrent (ib) [ vLb: 20 V/div, iLb: 1 A/div, vb: 10 V/div, ib: 1 A/div,Time: 5 s/div ].

(c) Experimental Results of buck converter i) Inductor voltage (vLL ),ii) Inductor current (iLL ), iii) Load voltage (vL), iv) Load current (iL)[ vLL : 20 V/div, iLL : 1 A/div, vL: 10 V/div, iL: 2 A/div, Time: 5s/div].

Fig. 8: Experimental results during prototype testing

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India

TABLE I: Prototype system parameters

Parameters ValuesPV Input Voltage (vp) 14 − 24 V

Boost Converter Inductor (Lp) 2.6 mHBoost Converter Inductor Resistance (Rp) 0.8 Ω

Boost Converter Input Capacitor (Cp) 1000 µFBoost Converter Output Capacitor (C1) 480 µFBoost Converter Switching Freq (fswp ) 20 kHz

Battery voltage (vb) 12 VNominal DC BUS Voltage (vd) 40 V

Bidirectional Converter Inductor (Lb) 0.727 mHBidirectional Converter Inductor resistance (Rb) 0.2 Ω

Bidirectional Converter Input Capacitor (Cb) 1000 µFBidirectional Converter Output Capacitor (C2) 480 µF

Bidirectional Converter Switching Freq (fswb,B ) 20 kHz

Load Voltage (vo) 12 VBuck Converter Inductor (LL) 1 mH

Buck Converter Inductor Resistance (RL) 0.3 ΩBuck Converter Input Capacitor (C3) 480 µF

Buck Converter Output Capacitor (CL) 1000 µFBuck Converter Switching Freq (fswL ) 20 kHz

Power Ratting (Psys) 150 watt

grid is operating in mode III and the mismatched power isdelivered using battery while maintaining a constant dc-busvoltage 37.25 V. The battery current (ib), dc-bus voltage (vd)can be seen in Fig. 8(b), Fig. 8(a) respectively. Next, the ledbulb is being switched OFF means no load condition. The gridstarts operating in mode II and the generated surplus power isobserved in the battery. Again the led bulbs are connected tothe grid and the grid starts operating in mode III and finally thesystem is burdened with a dc fan. The complete load profilecan be seen during the experiment in Fig. 8(c).

V. CONCLUSION

This paper discusses a DGDS architecture along withdistributed control technique, power management, and low-cost monitoring scheme for rural electrification. The resultshows that the distributed control technique and the powermanagement scheme is compatible with the architecture. Inaddition, it can enhance the distribution efficiency throughmodular interconnection. Moreover, the proposed architectureis scalable along with that it has some extra key advantages1) High distribution efficiency 2) Provide power to communalload 3) Less extensive control 4) Low-cost real-time moni-toring scheme. Therefore, the architecture is suitable for ruralelectrification.

REFERENCES

[1] M. E. Baran and N. R. Mahajan, ”DC distribution for industrialsystems: opportunities and challenges,” in IEEE Transactions on IndustryApplications, vol. 39, no. 6, pp. 1596-1601, Nov.-Dec. 2003.

[2] S.-Y. Lu, L. Wang, T.-M. Lo, and A. V. Prokhorov, ”Integration of windpower and wave power generation systems using a DC microgrid,” IEEETrans. Ind. Appl., vol. 51, no. 4, pp. 27532761, Jul./Aug. 2015.

[3] D. Kumar, F. Zare and A. Ghosh, ”DC Microgrid Technology: SystemArchitectures, AC Grid Interfaces, Grounding Schemes, Power Quality,Communication Networks, Applications, and Standardizations Aspects,”in IEEE Access, vol. 5, pp. 12230-12256, 2017.

[4] T. Dragievi, X. Lu, J. C. Vasquez and J. M. Guerrero, ”DC Mi-crogridsPart II: A Review of Power Architectures, Applications, andStandardization Issues,” in IEEE Transactions on Power Electronics, vol.31, no. 5, pp. 3528-3549, May 2016.

[5] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, ”Overviewof control and grid synchronization for distributed power generationsystems,” IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398-1409,Oct. 2006.

[6] S. Balijepalli, V. S. K. M. Khaparde, and S. A. Dobariya, ”Deploymentof microgrids in India,” in Proc. IEEE Power Energy Soc. Gen. Meet.,2010, pp. 17.

[7] P. Loomba, S. Asgotraa, and R. Podmore, ”DC solar microgridsAsuccessful technology for rural sustainable development,” in Proc. 2016Int. Conf. IEEE PES PowerAfrica, 2016, pp. 204208.

[8] ] D. Palit, G. K. Sarangi, and P. Krithika, ”Energising rural India usingdistributed generation: The case of solar mini-grids in Chhattisgarh State,India,” in Mini-Grids for Rural Electrification of Developing Countries,New York, NY, USA: Springer, 2014, pp. 313342.

[9] D. Palit and G. K. Sarangi, ”Renewable energy based mini-grids forenhancing electricity access: Experiences and lessons from India,” inProc. Int. Conf. Utility Exhib. Green Energy Sustain. Develop., Mar.2014, pp. 18.

[10] P. A. Madduri, J. Rosa, S. R. Sanders, E. A. Brewer, and M. Podolsky,”Design and verification of smart and scalableDCmicrogrids for emerg-ing regions,” in Proc. 2013 Int. Conf. IEEE Energy Convers. Congr.Expo., 2013, pp. 7379.

[11] P. A. Madduri, J. Poon, J. Rosa, M. Podolsky, E. Brewer, and S. R.Sanders, ”Scalable DC microgrids for rural electrification in emergingregions,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 4, no. 4, pp.11951205, Dec. 2016.

[12] M. Nasir, N. A. Zaffar, and H. A. Khan, ”Analysis on central anddistributed architectures of solar powered DC microgrids,” in Proc. 2016Int. Conf. Clemson Univ. Power Syst., 2016, pp. 16.

[13] F. Pearce, ”Worlds poor need grid power, not just solar panels. Energyand Fuels 2980,” New Scientist, 2014.

[14] J. Chi,W. Peng, X. Jianfang, T. Yi, and C. Fook Hoong, ”Implementationof hierarchical control in DC microgrids,” IEEE Trans. Ind. Electron.,vol. 61, no. 8, pp. 40324042, Aug. 2014.

[15] T.Vandoorn, J. DeKooning, B. Meersman, and L.Vandevelde, ”Reviewofprimary control strategies for islanded microgrids with power-electronicinterfaces,” Renew. Sustain. Energy Rev., vol. 19, pp. 613628, 2013.

[16] T. Dragicevic, J. M. Guerrero, J. C. Vasquez, and D. Skrlec, ”Supervi-sory control of an adaptive-droop regulated DC microgrid with batterymanagement capability,” IEEE Trans. Power Electron., vol. 29, no. 2,pp. 695706, Feb. 2014.

[17] K. Sun, L. Zhang, Y. Xing and J. M. Guerrero, ”A Distributed ControlStrategy Based on DC Bus Signaling for Modular Photovoltaic Gener-ation Systems With Battery Energy Storage,” in IEEE Transactions onPower Electronics, vol. 26, no. 10, pp. 3032-3045, Oct. 2011.

[18] L. Zhang, T. Wu, Y. Xing, K. Sun and J. M. Gurrero, ”Power controlof DC microgrid using DC bus signaling,” 2011 Twenty-Sixth AnnualIEEE Applied Power Electronics Conference and Exposition (APEC),Fort Worth, TX, 2011, pp. 1926-1932.

Proceedings of the National Power Systems Conference (NPSC) - 2018, December 14-16, NIT Tiruchirappalli, India