LEAD-ACID BATTERY STATE DETECTION FOR AUTOMOTIVE ELECTRICAL ENERGY MANAGEMENT PhD thesis Ga´bor Ba´ra´ny Supervisor: Dr. P´eter G´asp´ar BUDAPEST UNIVERSITY OF TECHNOLOGY AND ECONOMICS FACULTY OF TRANSPORTATION ENGINEERING AND VEHICLE ENGINEERING March 18, 2015 Alul´ırott, B´ar´any G´abor kijelentem, hogy ezt a doktori ´ertekez´est magam k´esz´ıtettem ´es abban csak a megadott forr´asokat haszn´altam fel. Minden olyan r´eszt, amelyet sz´o szerint, vagy azonos tartalomban, de ´atfogalmazva m´as forr´asb´ol ´atvettem, egy´ertelmu˝en, a forr´as megad´as´aval megjel¨oltem. A dolgozat b´ır´alatai ´es a v´ed´esr˝ol k´eszu¨lt jegyz˝ok¨onyv a k´es˝obbiekben a Budapesti Mu˝szaki ´es Gazdas´agtudom´anyi Egyetem d´ek´ani hivatal´aban lesz el´erhet˝o. Budapest, March 18, 2015 .............................. Ba´ra´ny Ga´bor Magyar nyelvu˝ o¨sszefoglalo´ Az ´olomsavas akkumul´atorok k¨ozponti helyet t¨oltenek be a modern g´epj´armu˝vek elektro- mos energia ell´at´o h´al´ozat´aban, f˝ok´epp az aut´oipari felhaszn´al´as szempontj´ab´ol vonz´o tu- lajdons´agaik ´es kedvez˝o ´aruknak k¨osz¨onhet˝oen. Az egyre szigorod´o t¨orv´enyi szab´alyoz´asok arrak´esztetik azaut´ogy´art´okat,hogyfejlettu¨zemanyag´esk´arosanyagkibocs´ajt´ast cs¨okkent˝o technol´ogi´akat vezessenek be, mint pl. a Start/Stop, f´ekenergia visszat´apl´al´as ´es a vitorl´az´as funkcionalit´as. Egy energia menedzsment rendszer keru¨lt bevezet´esre hogy biztos´ıtsa az aut´o elektromos rendszer´enek megb´ızhat´osa´g´at, meghosszabb´ıtsa az akkumul´ator ´elettartam´at ´es t´amogassa a szigorod´o CO kibocs´at´ast korl´atoz´o t¨orv´enyi hat´ar´ert´ekek betart´as´at mind 2 hagyom´anyos mind hibrid j´armu˝vekben. Az elektromos energia menedzsment rendszer pon- tos inform´aci´okat ig´enyel az akkumul´ator ´allapot´ar´ol. Ezt ´altal´aban egy olyan elektromos akkumul´atorszenzor biztos´ıtja, amirendelkezik ´allapotdetekt´al´as´espredikt´ıvfunkci´okkalis. Az akkumul´ator t¨olt¨otts´egi ´allapot´anak (SOC)´es ¨oregedetts´egi ´allapot´anak (SOH)´eszlel´esre alapvet˝o ahhoz, hogy az akkumul´ator bet¨olthesse kulcsszerep´et a funkcionalit´as ´es biztons´ag garant´al´as´anak ´erdek´eben. Agyakorlatbanazakkumul´atortulajdons´agaitk¨ozvetlenu¨lnemlehetm´ernik¨olts´eghat´ekony m´odon, ez´ert ez az akkumul´ator feszu¨lts´eg´enek, ´aram´anak ´es h˝om´ers´eklet´enek m´er´es´eb˝ol keru¨lkik¨ovetkeztet´esre. Ezent´ezisazakkumul´ator´allapotfelismer´esk´etalapvet˝okih´ıv´as´aval, azSOCu´jrakalibr´al´as´aval´esazakkumul´atorbels˝oellen´all´as´anakmeghat´aroz´as´avalfoglalkozik. A rendszeres, le´all´ıtott j´armu˝ben t¨ort´en˝o SOC u´jrakalibr´al´as alapvet˝o fontoss´agu´ az ´aramintegr´al´asb´ol sz´armaz´o SOC sodr´od´as korrig´al´as´ahoz. Szu¨ks´eges hozz´a a nyugalmi feszu¨lts´eg gyors´es megb´ızhat´o meghat´aroz´asa. Egy nyugalmi feszu¨lts´egre optimaliz´alt akku- mul´ator modell keru¨l megfogalmaz´asra, amihez k´et ku¨l¨onb¨oz˝o megfigyel˝o keru¨l javasol´asra: egy t¨obbsz¨or¨os modellalapu´ adapt´ıv becsl˝o (MMAE)´esegy Unscented K´alm´anszu˝r˝o(UKF) alapu´ megfigyel˝o. Mindk´et megk¨ozel´ıt´es k´epesanyugalmifeszu¨lts´eg robusztusmegbecsl´es´ere alacsony sz´am´ıt´asi ig´eny mellett. A megfigyel˝ok teljes´ıtm´enye numerikusan sz´amszeru˝s´ıt´esre keru¨l val´os aut´oban fell´ep˝o k¨oru¨lm´enyeket figyelembe vev˝o szimul´aci´ok ´altal. Az ´olom akkumul´ator bels˝o ellen´all´asa kulcs szerepet j´atszik az aut´o ind´ıthat´os´aga ´es az elektromos ´aram ell´at´o h´al´ozat stabilit´as´anak szempontj´ab´ol. Ebben a dolgozatban a de-facto bels˝o ellen´all´as defin´ıci´ok keru¨lnek bemutat´asra ´es ¨osszehasonl´ıt´asra a frekven- cia tartom´anyban, egy nagy frekvenci´as akkumul´ator modell ´es elektrok´emiai impedancia spektroszk´opi´as m´er´esek seg´ıts´eg´evel. A t´ezis f˝o hozz´aadott ´ert´eke a t¨ortszeru˝ ellen´all´as fo- galm´anak bevezet´ese a differenci´alis ellen´all´as fogalm´anak ´altal´anos´ıt´asak´ent. Ezt k¨ovetve egy matematikai keretrendszer ´es egy alacsony sz´am´ıt´asi kapacit´ast ig´enyl˝o becsl˝o algorit- mus koncepci´oja keru¨l levezet´esre. Tov´abb´a egy´eb bels˝o ellen´all´as sz´am´ıt´asi elj´ar´asok, mint a differenci´alis elj´ar´as, line´aris regresszi´os illeszt´es, ARX rekurz´ıv identifik´aci´os elj´ar´as ´es a szu˝r˝obankos RMS elj´ar´asok keru¨lnek ¨osszefoglal´asra ´es teljes´ıtm´eny ´ert´ekel´esre val´os aut´os m´er´esek felhaszn´al´as´aval. Legv´egu¨l egy k¨ovezketet´es keru¨l levon´asra az algoritmusok gyako- rlati alkalmazhat´os´ag´ar´ol. Summary Lead-acid batteries play a major role in modern vehicles’ power net due to their attractive properties for automotive application and favorable price. The ever stricter legal regula- tions motivate carmakers to develop advanced fuel consumption and emission reduction technologies, like Start/Stop, recuperation and coasting. An energy management system is introduced to assure the reliability of the vehicle’s electric system, prolong the battery lifetime and support the fulfillment of more stringent CO emission legal regulations in both 2 conventional and hybrid vehicles. The electrical energy management system requires a pre- cise knowledge about the battery status. This is usually provided by a battery sensor with state detection and predictive functions. The detection of State-of-Charge (SOC) and State- of-Health (SOH) is essential to help the battery fulfill its role as a key element for vehicle functionality and safety. In the praxis the battery properties cannot be measured directly in a cost effective way, they are deducted from the measurement of battery voltage, current and temperature. In this thesis two fundamental challenges of the automotive battery state detection, the SOC recalibration and the internal resistance determination are addressed. A periodic SOC recalibration during vehicle key-off is essential to correct the SOC drift due to the current integration during vehicle active mode. It requires a fast and reliable determination of the quiescent voltage. A quiescent voltage battery model is formalized and two different observers are proposed: a Multiple Model Adaptive Estimator (MMAE) and an Unscented Kalman Filter (UKF). Both approaches are capable of robust quiescent voltage estimation in a computationally inexpensive way. The performance of the observers is numerically quantified by simulations within application relevant boundary conditions. The internal resistance of lead-acid batteries plays a key-role in vehicle startability and power-net stability. In this thesis the de-facto internal resistance definitions are described and compared in the frequency domain using a high frequency battery model and electro- chemical impedance spectroscopy measurements. The main contribution is the introduction of the fractional resistance as a generalization of the differential resistance. Following this definition a mathematical framework and the concept of a computationally inexpensive es- timation algorithm is derived. Furthermore other internal resistance calculation methods like the differential method, linear regression fit, ARX recursive identification method and the RMS with filter bank method are summarized and benchmarked based on real vehicle measurements. Finally a conclusion is drawn on the applicability of the algorithms. Acknowledgement The author wishes to thank Mr. Ju¨rgen Motz, Dr. Martin H. K¨onigsman, Mr. Clemens Schmucker and Dr. P´eter G´asp´ar, my supervisor at the Hungarian Academy of Sciences in the Institute for Computer Science and Control. Without their support, in various forms, the challenges encountered during the process of writing this thesis might never have been solved. Thanks also to my parents, who provided the item of greatest worth - opportunity. And last but not least, I’m thankful to Csilla, my fianc´ee for her patience and continuos support over the last three years. Budapest, March 18, 2015 Ga´bor Ba´ra´ny v Contents 1 Introduction 1 1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Structure of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Lead-acid batteries 5 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.1 Main reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2.2 Oxygen reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.2.3 Hydrogen reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.3 Double layer effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4 Porous electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.5 Electrode utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.6 Mass transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.7 Current density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.8 Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.9 Inductance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.10 Aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.11 Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3 Energy Management 31 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.2 Generator management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.3 Load management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 3.4 Advanced energy management functions . . . . . . . . . . . . . . . . . . . . 39 3.4.1 Start/Stop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.4.2 Start/Stop with coasting . . . . . . . . . . . . . . . . . . . . . . . . . 41 vi 3.4.3 Boost-recuperation system . . . . . . . . . . . . . . . . . . . . . . . . 42 3.5 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4 Model-based State-of-Charge recalibration 46 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.2 Quiescent voltage battery model . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.2.1 Electrochemical model . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.2.2 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.2.3 State-space model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.3 Observer design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.3.1 Multiple Model Adaptive Estimator . . . . . . . . . . . . . . . . . . . 55 4.3.2 Unscented Kalman filter . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.4 Validation of the observers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.4.1 Synthetic data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.4.2 Battery measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5 Internal resistance 65 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 5.2 Factors of internal resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 5.3 Impedance measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.4 High frequency model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 5.4.1 Randles circuit with capacitor . . . . . . . . . . . . . . . . . . . . . . 73 5.4.2 Randles circuit with constant phase element . . . . . . . . . . . . . . 75 5.4.3 Data fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 5.5 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.6 Estimation algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 5.6.1 Differential method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 5.6.2 Linear regression fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 5.6.3 ARX recursive identification with forgetting factor . . . . . . . . . . . 88 5.6.4 RMS with filter bank . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 5.6.5 Fractional resistance method . . . . . . . . . . . . . . . . . . . . . . . 91 5.6.6 Spectral analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 5.7 Benchmarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 5.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 vii 6 Summary of theses 108 viii List of Tables 2.1 Electrochemical reactions in a lead-acid battery . . . . . . . . . . . . . . . . 10 2.2 Symbols of the Butler-Volmer equation . . . . . . . . . . . . . . . . . . . . . 11 2.3 Volume to charge constant of the main reaction . . . . . . . . . . . . . . . . 19 2.4 Dependency of active surface area from electrode utilization . . . . . . . . . 22 2.5 BCI standard for SOC estimation of a 12V flooded lead acid car battery . . 29 3.1 CO legal limit targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2 3.2 Typical current consumption of the electrical loads . . . . . . . . . . . . . . 37 3.3 Summary of hybrid car types and functions . . . . . . . . . . . . . . . . . . 45 5.1 EIS test cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 5.2 Model with capacitor: Parameters of Exide 70Ah wet battery at 25◦C . . . . 78 5.3 Model with CPE: Parameters of Exide 70Ah wet battery at 25◦C . . . . . . 79 5.4 Comparison of internal resistance determination algorithms . . . . . . . . . . 99 5.5 Comparison of internal resistance calculation methods . . . . . . . . . . . . . 100 ix
Description: