Selectio of Suitable Features of Lightig Rods for Secure Lightig Protectio Hariza Che Mat Haris, Nor Farahaida Abdul Rahma, Nur Ashida Salim Faculty of Electrical Egieerig, Uiversiti Tekologi MARA 40450 Shah Alam, Selagor, MALAYSIA farahaida@salam.uitm.edu.my Abstract Lightig rods play a very importat role i lightig protectio system. I this work, a study of the effect of usig with differet tip surfaces ad diameters is coducted. Hece, suitable features of ca be idetified. All experimetal results were aalyzed usig descriptive ad iferetial aalyses. From this research, blut tip surface with smaller diameters were prove to be the best lightig rod for safer lightig protectio. Idex Terms Voltage Impulse; High Voltage Protectio; Aalysis Of Variace (ANOVA) I. INTRODUCTION Lightig is a electrical discharge betwee cloud ad earth; it comprises oe or more impulses that carry very high curret. Accordig to [1], the Earth is gettig hit by lightig more tha 8 millio times per day. Thus, for the purpose of protectio, a lightig rod should be istalled at the top of high buildigs or structures exposed to lightig. This techique has bee i existece sice 250 years ago, whe Bejami Frakli iveted the lightig rod [2]. I a lightig protectio system, a copper (or lightig) rod with earth coectio is the oly compoet exists i the protectio system. It comes with a variety of desig such as hollow, solid, poited, rouded, flat strips ad others. Nevertheless, regardless of all desigs, each lightig rod should be able to coduct electricity efficietly. Hece, whe a lightig bolt strikes a buildig, the lightig rod will directly coduct it to the groud, prevetig electrocuted ad fire. The effectiveess of usig differet shape of has become a edless debate amog researchers. It is because there are o proper stadards for the shape of. Therefore, may studies have bee coducted primarily associated with the tip surface of [3]. I additio, usual failure of to protect earby objects has also ispired by may researchers to come up with ew shapes of [4]. I 1990s, a experimet coducted at the Lagmuir Laboratory had prove that a lightig rod with moderate blut tip was a best receptor for lightig strike. Hece, may researchers believed that a lightig rod with poited tip, elevated ad grouded is uable to discharge lightig effectively. However, i 2013, Karl Berggre had prove that a lightig rod with sharp tip is a better lightig strike receptor [5]. Nevertheless, some people still believe that a blut tip is better tha a shape tip, eve amog people i the lightig protectio busiess [6]. Therefore, this work focuses o studyig the relatioship betwee the lightig impulse voltage ad the tip surface ad diameter of. Hece, suitable features of for better protectio system ca be selected. To achieve these objectives, laboratory experimet ad descriptive ad iferetial aalyses were coducted. I this work, both sharp ad blut tips of were used. The diameter of each lightig rod was varied while the distace betwee the rod ad the lightig source was kept costat [7]. A high voltage impulse (of lightig) was geerated withi a rage of 50 kv to 70 kv. II. METHODOLOGY Figure 1 shows the methodology of the research work. It starts with the experimetal setup to geerate a lightig impulse ad applies it to a lightig rod. To achieve the research objective, a lightig impulse voltage test was coducted. The test was selected sice it is able to geerate lightig impulse voltages i accordace to IEC 60060-1 stadard [8]. The impulse was geerated withi a rage of 50 kv to 70 kv. The lightig impulse voltage test system comprises a impulse voltage geerator, voltage divider, choppig sphere gap ad overvoltage correctio. Additioally, it also has a digital MIAS trasiet recorder to record high voltage geeratio ad breakdow activity. Hece, accurate measuremets of the voltage ad evaluatio of its values ca be achieved [9]. Experimetal setup ad schematic diagram of the test system are show i Figure 2 ad Figure 3, respectively. The geerated lightig impulse is applied o sharp or blut tip with differet diameters. I order to justify all experimetal results, the gap betwee the lightig impulse ad each lightig rod is kept costat at 3cm. Meawhile, the diameters of both types of are NOMENCLATURE i j l Y i.. Y. j. Y Y ij. Y ijl Idexes levels of factor A; i =1,,a Idexes levels of factor B; j =1,,b Idexes plot (for each factor combiatio); l = 1,, Number of trials Mea of the i th factor level of factor A Mea of the j th factor level of factor B Overall mea of all observatios Mea of observatios at the i th level of factor A ad the j th level of factor B Observatios at the i th level of factor A, the j th level of factor B ad th level of factor combiatio e-issn: 2289-8131 Vol. 10 No. 1-9 43
Joural of Telecommuicatio, Electroic ad Computer Egieerig 9.5 mm, 10 mm, 13 mm ad 16 mm. The experimet was coducted o various combiatios of lightig source surfaces ad lightig rod s tips ad diameters. All of the combiatios are preseted i Table 1. Table 1 List of Combiatio of Lightig Source Surfaces ad Lightig Rod s Tips ad Diameters Lightig source surface Flat Flat Sharp Sharp Lightig rod Tip Diameter (mm) Blut 9.5, 10, 13, 16 Sharp 9.5, 10, 13, 16 Blut 9.5, 10, 13, 16 Sharp 9.5, 10, 13, 16 Oce the test was completed, all the experimetal results were assessed i several types of aalyses ad tests. I this work, descriptive ad iferetial aalyses were coducted. Accordig to [10], a descriptive aalysis ca be defied as the umbers that sum up the data with the idea of describig what happeed i the sample. The aalysis was able to evaluate samples from oe study to aother. Moreover, it ca also assist researchers to detect sample characteristics that may ifluece their coclusios. For this work, the choppig time of the lightig impulse was measured for five times i every coditio. idetify factors that ca result variatios beyod the iheret experimetal variatio. Additioally, F-test was performed to determie whether the overall variace of all samples is sigificatly greater tha the iheret experimetal variace. If the F-test gives a sigificat result, it ca be cocluded that there is a sigificat differece betwee the mea values of all samples [11]. I order to obtai a valid ANOVA result, all samples must comply with these followig assumptios: 1. The depedet variable is measured at the iterval or ratio level, ad it is a cotiuous data. 2. The idepedet variable must cosist of two or more categories, ad it is i idepedet groups. 3. There is o associatio betwee the observatios i each group or betwee the groups. 4. There are o sigificat outliers. 5. The depedet variable is approximately ormally distributed for each category of the idepedet variable. 6. The variaces are homogeous. Figure 2: Experimetal setup for lightig impulse voltage test Figure 3: Schematic diagram for geeratig lightig impulse voltage Sice this study has two mai effect variables with differet levels, the F-test falls ito two-factor Completely Radomized Desig (CRD) factorial experimet. All parameters of ANOVA table format are preseted i Table 2. Table 2 Aalysis of Variace Table for Two-factor CRD Factorial Figure 1: Flowchart of the research work Next, a iferetial aalysis was coducted to verify the fidig of the descriptive aalysis. It was coducted usig statistical method kow as aalysis of variace (ANOVA) o Miitab software. ANOVA was used to determie possible differeces i the mea values of more tha 2 data samples [11]. It ivolves variace aalysis of overall data to Source of variatio Degree of Freedom (DF) Sum of Square (SS) Mea Square (MS) First factor A DF A A = a 1 SSA MSA = SSA/DFA Secod factor DFB = b 1 SSB MSB = B SSB/DFB Iteractio DFAB = SSAB MSAB = AB (a 1)(b 1) SSAB/ DFAB Error DFE = SSE = ab( 1) SSE/DFE Total DFT = SST ab 1 F MSA/ MSB/ MSAB/ 44 e-issn: 2289-8131 Vol. 10 No. 1-9
Selectio of Suitable Features of Lightig Rods for Secure Lightig Protectio From the table, SS ca be calculated usig the followig equatios: SSA = b a i=1 (Y i.. Y ) 2 (1) SSB = a b (Y. j. Y ) 2 j=1 (2) SSAB = a b (Y ij. Y i.. Y. j. Y ) 2 i=1 j=1 (3) SSE = a b (Y ijl Y ij. ) 2 i=1 j=1 l=1 (4) SST = a b (Y ijl Y ) 2 i=1 j=1 l=1 (5) Results of ANOVA were validated usig two tests, amely as the assumptio of ormality of variace ad the assumptio of homogeeity of variace. The assumptio of ormality of variace was coducted to determie the ormality of a set of data. To satisfy this assumptio, the variace distributio must exhibit a symmetrical bell-shaped (ormal) curve. For this study, the variats (or residuals) of the model was decided to be ormally distributed depedig o the majority of the observatio values o the ormal probability plot. O the other had, the assumptio of homogeeity of variace was coducted to idetify the correlatio betwee all variaces [12]. Accordig to this assumptio, a satisfied model should have a structure with less variace. I this case, the variaces should be urelated to ay other variables, icludig the predicted respose. For this study, the residuals of the model are expected to be scattered i the plot of residuals agaist fitted value. Moreover, this plot should ot reveal ay obvious patter. III. RESULTS AND DISCUSSIONS As metioed, the objective of this work is to determie suitable features of for better protectio system. I the experimetal work, there are two maipulative variables: tip surface ad diameter of copper. To verify the effectiveess of each lightig rod to chop the lightig impulse, all waveforms resulted from the experimet were compared to a full lightig impulse voltage waveform. The respose of this experimet is choppig time T c. Figure 4 shows a waveform of full lightig impulse voltage geerated by the lightig impulse voltage test. Accordig to the stadard, a full lightig impulse voltage should raise its peak value less tha a few microsecods ad falls, appreciably slower tha the ormal value, ultimately back to zero [13]. Sice the waveform i Figure 4 iherits the same characteristics stated i the stadard, the productio of lightig impulse usig the lightig voltage test has bee achieved successfully. The peak voltage V p of the waveform is 51.03 kv. The poits correspodig to the 30% ad 90% of V p is the frot time T1 which is 1.13 µs. Next, 50% of V p is called the tail time T2; it is 48.97 µs. Pictures of the geeratio of lightig impulse voltage are show i Figure 5. Each lightig impulse voltage was applied o a blut tip lightig rod as show i Figure 5(a) or a sharp tip lightig rod as show i Figure 5(b). Additioally, Figure 6 presets the waveforms of lightig impulse voltage for all coditios show i Figure 5. Figure 4: Full lightig impulse voltage Accordig to Figure 6, we ca see that all the waveforms exhibit differet characteristics as compared to the stadard waveform of full lightig impulse voltage show i Figure 4. From the startig poit, we ca observe that the raise of voltage of all the waveforms is ot as fast as the full lightig impulse voltage waveform. However, after V p, the voltage value of all the waveforms drops drastically. These waveforms are kow as lightig impulse voltages chopped o the frot. The poit where voltage starts to decrease is called T c. From the result, we ca otice that the use of with differet tips has resulted differet T c ad peak voltage. By referrig to Figure 6, we ca observe that the use of the blut tip lightig rod has resulted higher V p tha the sharp tip lightig rod. It may happe because the blut tip has a bigger surface as compared to the sharp tip. Hece, it receives higher voltage tha the sharp tip. Other tha that, because of T c, all the waveforms do ot have T2. Nevertheless, based o the results, we ca cofirm that all types of are capable to chop the lightig impulse. Hece, they serve as lightig protectio devices. As discussed i the methodology, the experimet usig the same cofiguratio as show i Table 1 was repeated for five times. The, the average T c ad V p for all cofiguratios were plotted agaist the diameter of each lightig rod. Figure 7 ad Figure 8 depict 2 graphs: diameter versus T c ad diameter versus V p. (a) Figure 5: Lightig impulses applied o (a) blut ad (b) sharp tips of (b) e-issn: 2289-8131 Vol. 10 No. 1-9 45
Joural of Telecommuicatio, Electroic ad Computer Egieerig By referrig to Figure 7, we ca see that the use of blut tip has resulted the lowest T c tha usig sharp tip. O average, the use of blut tip has recorded 85.6% lower T c tha usig sharp tip lightig rods. Nevertheless, by cosiderig V p i Figure 8, we ca ote that the use of blut tip has resulted higher V p tha usig sharp tip. O average, the use of blut tip has recorded 16.3% higher V p tha usig sharp tip. This percetage is cosidered low as compared to the differet percetage of T c. Based o all coditios, we ca say that lower T c ca be obtaied whe V p is higher tha 60 kv. Moreover, we ca see that lightig rods with smaller diameter exhibit lower T c tha lightig rods with bigger diameter. I overall, the use of the 9.5 mm diameter of blut tip lightig rod has recorded the lowest T c tha other diameters. I order to verify the descriptive aalysis, a CRD factorial experimet was coducted. Table 3 tabulates all ANOVA outputs. Referrig to the ANOVA outputs, the two mai effect factors have a sigificat differet towards respose variable (time). Accordig to both tables, the tip surface show high statistically sigificat differet towards time; F(1,32) = 9758.75, p < 0.001. This coditio also appear for the rod diameter mai effect; F(3,32) = 2622.16, p < 0.001. Moreover, the fidigs have revealed that the iteractio betwee tip surface ad rod diameter also show high statistically sigificat differet towards time; F(3,32) = 2586.91, p < 0.001. Additioally, the statistic R 2 was used to represet the percetage of variatio i a respose variable (time) that is explaied by its relatioship with oe or more predictor variables. I the table, the adjusted R 2 percetages idicate that 99.85% of the variatio is explaied by the tip surface ad the rod diameter. (b) Figure 6: Waveforms of lightig impulses applied o (a) blut ad (b) sharp tips of Table 3. ANOVA Table Factor Type Level Value Tip surface Fixed 2 Blut ad sharp Rod diameter (D) Fixed 4 D9.5, D10, D13 ad D16 Aalysis of variace for time, usig adjusted SS Source of variatio DF Adj SS Adj MS F Sigificat level, p Tip surface 1 9.5326 9.5326 9758.75 0.000 Rod diameter 3 7.6842 2.5614 2622.16 0.000 Tip surface* 3 7.5809 2.5270 2586.91 0.000 Diameter Error 32 0.0313 0.0010 Total 39 24.8289 R 2 (adj) = 99.85% I order to verify all the ANOVA results, the assumptio of residual ormality distributed ad also the assumptio of homogeeity of the residual were performed. (a) Figure 7: Relatioship betwee average T c ad the tip ad diameter of 46 e-issn: 2289-8131 Vol. 10 No. 1-9
Selectio of Suitable Features of Lightig Rods for Secure Lightig Protectio IV. CONCLUSION Figure 8: Relatioship betwee average V p ad the tip ad diameter of Accordig to the ormal probability plot i Figure 9, we ca cofirm that the residuals of the model are ormally distributed. It is because the majority of the observatio values (small red dots) lie o the straight lies. Meawhile, the residuals agaist fitted value plot as show i Figure 10 were used to verify the assumptio of homogeeity of the residual. Based o the figure, the residuals of the model do ot show ay serious o-costat residuals patter; the plot exhibits radom patter. Sice the importat assumptios of the ANOVA aalysis have bee met, it ca be cofirmed that the fidigs from the ANOVA results are reliable ad valid. Subsequetly, the results of the descriptive aalysis are also verified. Percet Residual 99 95 90 80 70 60 50 40 30 20 10 5 1-0.075 0.050 0.025 0.000-0.025-0.050 Normal Probability Plot (respose is Time) -0.025 0.000 Residual 0.025 Figure 9: Normal probability plot Versus Fits (respose is Time) 0.050 I this study, we have foud that with differet tips ad diameters exhibit differet T c ad V p, despite of differet lightig source surface. Accordig to the result, the blut tip exhibit lower T c tha the sharp tip. However, i order to achieve low T c, V p should be higher tha 60 kv. Furthermore, based o ANOVA results, we ca determie that the tip surface factor ad the rod diameter factor have high statistically sigificat differet towards respose variable (time). Based o the descriptive ad ANOVA aalyses, a lightig rod with a blut tip ad 9.5 mm diameter is the most suitable for lightig protectio system. Hece, the objectives of the study are achieved. ACKNOWLEDGMENT The authors would like to thak the Research Maagemet Istitute (RMI), Uiversiti Tekologi MARA, Malaysia ad the Miistry of Higher Educatio (MOHE), Malaysia through research grat 600-RMI/RAGS 5/3 (192/2014) for the fiacial support of this research. REFERENCES [1] C. A. Pfortmueller, Y. Yiku, M. Haberker, E. Wuest, H. Zimmerma, A. K. Exadaktylos, Ijuries, Sequelae ad Treatmet of Lightig Iduced Ijuries : 10 Years of Experiece at a Swiss Trauma Ceter, Emergecy Medicie Iteratioal, 2012. [2] A. A. Hossam-Eldi ad M. I. Houssi, Study of the Effect of Dissipatio Poits o the Lightig Protectio, Iteratioal Joural of Scietific & Techology Research, vol. 1, issue 2, 2012. [3] F. D. Alessadro, O The Optimum Rod Geometry for Practical Lightig Protectio Systems, Joural of Electrostatics, 2007. [4] A. Farouk ad M. Rizk, Modelig of Lightig Exposure of Sharp ad Blut Rods, IEEE Trasactios o Power Delivery, Vol.25, No.4, 2010. [5] K. K. Berggre, How Do You Make ad Istall A Lightig Rod?, Ask A Egieer, 2013. [6] G. D. Aulich, C. B. Moore, W. Riso, The Power of the Poit : Bejami Frakli, the Lightig Rod ad Two Miscoceptios That Have Plagued Us to This Day, AGU Fall Meetig Abstracts, 2006. [7] S. F. Sazali, A Study o the Effects of Strikig Distace for Capturig Lightig Leaders, Fial Year Project Thesis, 2014. [8] IEC 60060-1 : High Voltage Test Techiques-Part 1 : Geeral Defiitios ad Test Requiremets, 2010. [9] HIGHVOLT Prüftechik Dresde GmbH, High Voltage Module Test System, 2012. [10] J. Pallat, SPSS Survival Maual, 4 th ed, Alle & Uwi Book Publishers, 2010. [11] G. Currell ad A. Dowma, Essetial Mathematics ad Statistics for Sciece, 2 d ed, Wiley-Blackwell, 2009. [12] [12] D. C. Motgomery, Desig ad Aalysis of Experimets, 7 th ed, Joh Wiley & Sos, 2009. [13] K. Scho, High Impulse Voltage ad Curret Measuremet Techiques Fudametals Measurig Istrumets Measurig Methods, Spriger, 2013. -0.050 1.0 1.5 2.0 2.5 Fitted Value 3.0 3.5 Figure 10: Plot of residuals agaist fitted value e-issn: 2289-8131 Vol. 10 No. 1-9 47