Inter national J our nal of P o wer Electr onics and Dri v e Systems (IJPEDS) V ol. 9, No. 1, March 2018, pp. 357 364 ISSN: 2088-8694 357       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     Corr elation between Battery V oltage under Loaded Condition and Estimated State of Char ge at V alv e-Regulated Lead Acid Battery on Dischar ge Condition using Open Cir cuit V oltage Method Ahmad Qurthobi , Anggita Bayu Krisna P amb udi , Dudi Darmawan , and Reza F auzi Iskandar Department of Engineering Ph ysics, School of Electrical Engineering, T elk om Uni v ersity , Indonesia Article Inf o Article history: Recei v ed No v ember 24, 2017 Re vised January 23, 2018 Accepted February 18, 2018 K eyw ord: Battery State of Char ge Open Circuit V oltage ABSTRA CT One of the common methods that de v eloped to predict sta te of char ge is open circuit v oltage (OCV) method. The probl em which commonly occurs is to find the correc- tion parameter between open circuit v oltage and loaded v oltage of the battery . In this research, correlation between state of char ge measurement at loaded condition of a P anasonic LC-V A1212N A1, which is a v alv e-re gulated lead acid (VRLA) battery , and open circuit v oltage had been analyzed. Based on the results of research, cor - relation between battery’ s measured v oltage under loaded condition and open circuit v oltage could be approached by tw o linearization area. It caus ed by K v s v alues tend to increase when measured v oltage under loaded condition V M < 11 : 64 v olt. Ho w- e v er , K v v alues w ould be relati v ely sta ble for e v ery V M 11 : 64 v olts. Therefore, estimated state of char ge v alue, in respect to loaded battery v oltage, w ould increase slo wer on V M < 11 : 64 v olts and f aster on other range. Copyright c 2018 Insitute of Advanced Engineeering and Science . All rights r eserved. Corresponding A uthor: Ahmad Qurthobi Department of Engineering Ph ysics, School of Electrical Engineering, T elk om Uni v ersity Jalan T elek omunikasi No 1 T erusan Buah Batu +62-813-20262968 qurthobi@telk omuni v ersity .ac.id 1. INTR ODUCTION Lead acid battery is a type of dry elements. Its commonly used as ener gy backup storage for r emote communities, telecommunication systems, or electric v ehicles[1]-[4]. The problems which commonly occur on lead acid type batteries are o v erchar ge and o v erdischar ge. Both condition create temperature changes and ph ysical deformation on the battery which af fect on its performances. Therefore, it needs to analyzed battery’ s state of char ge (SoC)[5]-[10]. State of char ge (SoC) is a ratio between a v ailable and maximum electrical char ge that stored in t he battery[2],[11]. SoC is a non-dimentional unit and it could be e xpressed in percent or v alue between 0 and 1. SoC is an important parameter which not only to a v oid harmful condit ion, such as o v erchar ge and o v erdis- char ge, b ut also to e xpand the battery lifetime. One of the methods used to estimate state-of-char ge is open circuit v oltage (OCV). Statistically , correlation between state of char ge and open circuit v oltage could be repre- sented as an linear equation. Ho we v er , as loads attached to the battery , the loading ef fect w ould be occured. It w ould create g ap between measured v oltage and open c ircuit v oltage[12]. Therefore, correlation between open circuit v oltage and loaded battery v oltage which culminated on their correlation with state of char ge should be determined. J ournal Homepage: http://iaescor e .com/journals/inde x.php/IJPEDS       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     DOI:  10.11591/ijpeds.v9n1.pp357-364 Evaluation Warning : The document was created with Spire.PDF for Python.
358 ISSN: 2088-8694 Figure 1. An e xample of typical correlation between state of char ge and open circuit v oltage[13] In pre vious research, correlation between both v ariables in a Y uasa SWL2500 could be written as a constant v alue for se v eral e xamination point[1]. Therefore, in this research, correlation between loaded battery v oltage on dischar ge condition and open circuit v oltage had been analyzed on a P anasonic LC-V A1212N A1, which is a v alv e-re gulated lead acid (VRLA) battery . This research observ es the changes of rest period param- eter on the battery based on its loaded v oltage at se v eral points which leads to produce a preci se estimation of state of char ge. 2. RESEARCH METHOD 2.1. State of Char ge and Open Cir cuit V oltage As mentioned in section 1., state of char ge or SoC is a non-dimentional unit which represents ratio between a v ailable and maximum char ge stored in a battery . Hence, SoC is proportional to char ge residues inside a battery and it could be represented as equation (1), where q t and q max represent, current and maximum char ges v alues inside battery , respecti v ely . SOC % = q t q max 100 (1) On the real condition, neither zero point of a battery equal to zero char ge ( q 0 6 = 0 ) nor its maximum v alue equal to peak char ge ( q max = q peak ). The minimum and maximum char ges v alues inside the battery are frequently mentioned by the manuf acturers o n battery’ s datasheet to k eep its performances. Hence, equation (1) should be re written as equation (2), where q min represents minimum char ges v alue. SOC % = q t q min q max q min 100 (2) There are some methods de v eloped for SoC estimation. One of those esti mation methods is open circuit v oltage (OCV)[1]. OCV method has been de v eloped based on an assumption that a battery is equal to a capacitor and the amount of char ges ( q ) inside a capacitor is linearly proportional to its output v oltage ( V t )(see equation (3)). V t = q t C (3) Statistically , if it is assumed that typical correlation between SoC and OCV is sho wn by figure 1, it could be approached by a linear formula such as equation (4)[13], where V min , 1 , and 2 are, respecti v ely , battery’ s minimum v oltage (recommended by manuf acturer), a multiplier constant, and a correction f actor . SOC % = 1 ( V t V min ) + 2 (4) IJPEDS V ol. 9, No. 1, March 2018: 357 364 Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS ISSN: 2088-8694 359 Figure 2. Lead acid equi v alent netw ork for both dischar ge and char ge[12] Figure 3. T ypical v oltage and current profile for a constant current dischar ge[12] 2.2. Battery Modelling and Rest P eriod Equation (3) is only v alid in specific conditions, such as when all of its loads are disconnected from the battery . Ho we v er , remo ving loads from the battery for v oltages e xamination creates a non-real time mea- surement. Barsali and Ceraolo(2002) define a lead acid battery’ s dynamic model to in v estig ate its dischar ging dynamic beha vior[12]. Lead acid equi v alent netw ork for char ge and dischar ge are sho wn by figure (2). Further - more, the y also represent typical v oltage and current profile for constant current dischar ge as sho wn on figure 3. Figure 3 sho ws that there is e xist a transition time required by the battery to sho w it open cir cuit v oltage v alue after loads are remo v ed. This transition time kno wn as rest period [1],[2],[12]. Hence, OCV ( V t oc ) of a battery could be written as equation (5)[1], where V M is measured v oltage when loads a re attached to the battery and K v is a parameter deri v ed from V t oc V M after the battery is rested. V t oc = V M K v (5) 2.3. Obser ving K v As mentioned in subsection 2.2., K v is a parameter deri v ed from dif ferences between open circuit v oltage and loaded v oltage. It v alues could be v aries on e v ery point. Therefore, it needs to observ e K v v alues at se v eral point to determine it correlation with battery’ s v oltage. T o determine the v alue of K v at some specified point, a testing had beed done with this follo wing steps: 1. As pre-conditioning process, a VRLA battery (P anasonic LC-V A1212N A1) has being char ged until it reach its maximum v oltage v alue recommended by its manuf acturer , in this case is 13 v olt[14](Figure 4(a)). 2. Connect the battery with a 10 w att load and measure its terminal v oltage v(Figure 4(b)). Corr elation between Loaded Battery V olta g e and Estimated SoC at VRLA Battery ... (Qurthobi) Evaluation Warning : The document was created with Spire.PDF for Python.
360 ISSN: 2088-8694 Figure 4. Battery char ging and dischar ging process Figure 5. The changes of battery v oltage measurement for 150 minutes 3. As soon when terminal v oltage reach first point, remo v e load from battery and l ea v e it rest for 150 minutes. 4. Monitor the change in battery v oltage v alues, starting from the load released until the rest period ends. 5. Repeat step (2) to (4) for another monitoring point. 3. RESUL T AND AN AL YSIS Figure 5 sho ws the change of e xperimental bat tery’ s v oltage v alue when it rested for 150 minutes. This figure sho ws that loading ef fects which occurs on the battery results the measured v oltage at loaded con- dition is lo wer than its open circuit v oltage. The v alues are v aries in range 0.22 - 0.74 v olt. Correlation between measured v oltage under loaded condition V M and open circuit v oltage V oc after rest period on the battery are sho wn in figure 6. It informs the gradient of correlation curv e of V M and V oc for V M 11 : 64 v olt is relati v ely constant with @ V oc @ V M 1 with small zero of fset addition. Ho we v er , for IJPEDS V ol. 9, No. 1, March 2018: 357 364 Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS ISSN: 2088-8694 361 Figure 6. Correlation between measured v oltage V M under loaded condition and open circuit v oltage V oc after 150 minutes of rest period Figure 7. Correlation between measured v oltage V M under loaded condition and correction parameter K v Figure 8. Measured v oltage under loaded condition and estimated open circuit v oltage in respect to time Corr elation between Loaded Battery V olta g e and Estimated SoC at VRLA Battery ... (Qurthobi) Evaluation Warning : The document was created with Spire.PDF for Python.
362 ISSN: 2088-8694 Figure 9. Correlation between V M and SOC p (%) V M < 11 : 64 , correlation between both v ariabl e relati v ely fluctuati v e and could be approached by equation (6). V oc = 0 : 50 V M + 6 : 02 (6) This condition yeilds correlation between K v and V M had ne g ati v e gradient for e v ery V M < 11 : 64 v olts. In the other hand, it is relati v ely stable with a v erage K v = 0 : 23 v olt for V M 11 : 64 v olts (see figure 7). Hence, e v olution o v K v could be de vided into tw o linearization condition and could be written as equation(7). K v = 0 : 49 V M + 5 : 90; for V M < 11 : 64 0 : 23; for others (7) Ho we v er , the result sho ws dif ferent outcome with another research with another VRLA battery , where K v v alues are almost constant for e v ery inspection points(80%, 60%, 40%, and 20% of SoC) on Y uasa SWL2500 battery with 240 minutes of rest period[1]. Based on equation (5) and (7), correlation between measured loaded v oltage and estimated open ci rcuit v oltage v alue in respect to time could be dra wn as figure 8. As results of implementation of K v v alues, the g ap between both curv es w ould increase as soon as measured v oltage under loaded condition is less than 11.64 v olts. Therefore, correlation between state of char ge and V M could be sho wn as figure 9 and could be written as equation (8). SoC % = 20 : 44 V M 184 : 64; for V M < 11 : 64 40 V M + 410 : 8; for others (8) As v alidation process, testing on subsection 2.3. had been repeated with addition of ne w e xamination point at V M = 12 : 24 v olts. Figure 10 sho ws correlation between estimated open circuit v oltage V toc p and actual v oltage V real . This figure sho ws correlation between both v ariables could be approached with a linear formula and could be written as equation (9). It also informs the a v erage absolute error between V toc p and V real is 0.02 v olt. V real = 1 : 02 V toc p + 0 : 28 (9) 4. CONCLUSION Based on the results of research, correlation between battery’ s measured v oltage under loaded con- dition and open circuit v oltage could be approached by tw o linearization area. It caused by K v s v alues tend to increase when measured v oltage V M < 11 : 64 v olt and correlation between both v ariables could be written as 0 : 49 V M + 5 : 90 v olt. Ho we v er , K v v alues w ould be relati v ely stable on 0.23 v olt for e v ery V M 11 : 64 v olts. Therefore, in respect to loaded battery v oltage, estimated state of char ge v alue w ould increase slo wer on V M < 11 : 64 v olts and f aster on other range. It sho ws dif ferent outcome with another research with another VRLA battery , where K v v alues are almost constant for four inspection points(80%, 60%, 40%, and 20% of SoC) on Y uasa SWL2500 battery with 240 minutes of rest period. IJPEDS V ol. 9, No. 1, March 2018: 357 364 Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS ISSN: 2088-8694 363 Figure 10. Correlation between V toc p and V real REFERENCES [1] A. Mariani, T . Stockle y , K. Thanapalan, and J. W illiams, Simple and Ef fecti v e OCV Prediction Mech- anism for VRLA Battery Systems, in The 3rd International Conference on Mechanical Engineering and Mechatronics, pp. 140(1)140(10), Aug. 2014. [2] S. Mischie and L. T oma, Beha vior of the Lead Acid Battery after the Rest Period, WSEAS TRANSA C- TIONS on PO WER SYSTEMS, v ol. 3, pp. 111117, Mar . 2008. [3] K. Belmokhtar , H. Ibrahim, Z. Fger , and M. Ghandour , Char ge Equalization Systems for Serial V alv e Re gulated Lead-Acid (VRLA) Connected Batteries in Hybrid Po wer Systems Applications, in International Rene w able Ener gy Storage Conference, no. 99, pp. 277284, Mar . 2016. [4] A. Selmani, A. Ed-Dahhak, M. Outanoute, A. Lachhab, M. Guerbaoui, and B. Bouchikhi, Performance Ev aluation of Modelling and Simulation of Lead Acid Batteries for Photo v oltaic Applications, International Journal of Po wer Electronics and Dri v e System, v ol. 7, pp. 472480, June 2016. [5] C. Unterri eder , C. Zhang, M. Lunglmayr , R. Prie w asser , S. Marsili, and M. Huemer , Battery state-of-char ge estimation using approximate least squares, Journal of Po wer Sources, pp. 274286, 2015. [6] S. Xie, R. Xiong, Y . Zhang, and H. He, The Estimation of State of Char ge for Po wer Battery P acks used in Hybrid Electric V ehicle, in The 8th International Conference on Applied Ener gy , pp. 2678 2683, 2017. [7] V . Surendar , V . Mohakumar , S. Anand, Prasanna, and D. V adana, Estimation of State of Char ge of a Lead Acid Battery Using Support V ector Re gressi on, in SMAR T GRID T echnologies,, pp. 264 270, Aug. 2015. [8] Y . W ang, C. Zhang, and Z. Chen, State-of-char ge estim ation of lithium-ion batteries based on multiple filters method, in The 7th International Conference on Applied Ener gy , v ol. 25, pp. 2635 2640, Else vier , 2015. [9] X. T ang, B. Liu, and F . Gao, State of char ge estimation of LiFePO4 battery based on a g ain-classifier observ er , in The 8th International Conference on Applied Ener gy , v ol. 105, pp. 2071 2076, 2017. [10] S. Prasad and D. V . K umar , Hybrid fuzzy char ged system search algorithm based state e stimation in distrib ution netw orks, Engineering Science and T echnology , an International Journal, v ol. 20, pp. 922933, 2017. [11] T . W u, L. Liu, Q. Xiao, Q. Cao, and X. W ang, Research on SOC Estimation Based on Second-order RC Model, TELK OMNIKA, v ol. 10, pp. 16671672, No v . 2012. [12] S. Barsali and M. Ceraolo, Dynamical Models of Lead-Acid Batteries: Implementation Issues, IEEE TRANSA CTIONS ON ENERGY CONVERSION, v ol. 17, pp. 1623, Mar . 2002. [13] W . Y . Chang, The State of Char ge Estimating Methods for Battery: A Re vie w , ISRN Applied Mathemat- ics, v ol. 2013, no. 953792, pp. 17, 2013. [14] P anasonic, LC-V A1212. P anasonic Storage Battery , Co. Ltd. Corr elation between Loaded Battery V olta g e and Estimated SoC at VRLA Battery ... (Qurthobi) Evaluation Warning : The document was created with Spire.PDF for Python.
364 ISSN: 2088-8694 BIOGRAPHIES OF A UTHORS Ahmad Qurthobi is a lecturer and researcher at Department of Engineeering Ph ysics, School of Electrical Engineeering, T elk om Uni v ersity (2013-present). He obtained Bachelor De gree in T elecommunication Engineering from STT T elk om (Indonesia) in 2007 and Master De gree in in- strumentation and Control from Institut T eknologi Bandung (Indonesia) at 2011. His researches are in fields of electric po wer , instrumentation, control system, rene w able ener gy , and po wer electronics. He is af filiated with PII as member . F or further info on his homepage: http://qurthobi.staf f.telk omuni v ersity .ac.id/ Anggita Bayu Kr esna P amb udi is an engineer of N ational Electric Compan y of Indonesia. He obtained Bachelor De gree in Engineering Ph ysics from T elk om Uni v ersity (Indonesia) at 2016. His researches are in fields of instrumentation, and measurement. Beside his acti vity as engineer , he also runs a start up design compan y called Dodlezig. F or further info on his homepage: http://www .anggitabay .com/ Dudi Darmawan is a lecturer and researcher at Department of Engineeering Ph ysi cs, School of Electrical Engineeering, T elk om Uni v ersity (1999-present). He obtained Bachelor De gree in Ph ysics in 1998, Master De gree in instrumentation and Control at 2004, and Doc toral De gree in Engineeering Ph ysics at 2015 from Institut T eknologi Bandung(Indonesia). His researches are in fields of non-destructi v e testing, instrumentation, tomograph y , and electro- magnetics. F or further info on his homepage: http://dudiddw .staf f.telk omuni v ersity .ac.id/ Reza F auzi Iskandar is a lecturer and researcher at Department of Engineeering Ph ysics, School of Electrical Engineeering, T elk om Uni v ersity (2012-present). He obtained Bachelor De gree in Ph ysics Education from Uni v ersitas Pendidikan Indonesia at 2003, and Master De gree in instru- mentation and Control from Institut T eknologi Bandung (Indonesia) at 2011. His researches are in fields of instrumentation, control system, po wer electronics, and rene w able ener gy . F or further info on his homepage: http://rezaf auzii.staf f.telk omuni v ersity .ac.id/ IJPEDS V ol. 9, No. 1, March 2018: 357 364 Evaluation Warning : The document was created with Spire.PDF for Python.