Experimental Study of the Response of Efficient Lighting Technologies to Complex Voltage Fluctuations

Izaskun Azcarate, Jose Julio Gutiérrez, Andoni Lazkano, Puri Saiz, Koldo Redondo, Luís Alberto Leturiondo, “Experimental study of the response of efficient lighting technologies to complex voltage fluctuations”, International Journal of Electrical Power and Energy Systems, Volume 63, 2014, Pages 499-506, ISSN 0142-0615, https://doi.org/10.1016/j.ijepes.2014.06.039 (https://www.sciencedirect.com/science/article/pii/S0142061514003883) Abstract: The replacement of incandescent lamps with more energy-efficient lighting technologies has a direct influence on the way flicker is measured. The International Electrotechnical Commission (IEC) established in the 61000-4-15 standard the functional specifications of a flickermeter, taking a standard incandescent lamp’s response to voltage fluctuations as the reference. During the past ten years, different works have studied the sensitivity of modern lamps to analytical voltage fluctuations of low complexity. From these studies, the most widespread conclusion is that modern lamps are less sensitive to flicker than are incandescent lamps. Based on these results, international standardization organizations are currently studying two different possibilities for updating the flicker assessment procedure: adjusting the IEC flickermeter according to a new less sensitive reference lamp, or increasing the established compatibility levels for voltage fluctuations. This work presents for the first time a sensitivity analysis of a set of modern lamps subjected to real voltage signals that are more complex than analytical voltage fluctuations. The obtained results lead to the following conclusions: not all efficient lamps have a lower sensitivity to fluctuations than do incandescent lamps; the response of some lamps depends on the complexity of the input voltage fluctuation; and the response of some lamps in real scenarios, i.e., more complex voltage fluctuations, does not correlate with their response to simple voltage fluctuations. Keywords: Efficient Lighting, Voltage Fluctuations, Flicker, Power Quality, Complexity. Experimental Study of the Response of Efficient Lighting Technologies to Complex Voltage Fluctuations. I. Azcarate∗,a, J.J. Gutierreza, A. Lazkanoa, P. Saiza, K. Redondoa, L.A. Leturiondoa Affiliations and addresses: aCommunications Engineering Department. University of the Basque Country UPV/EHU. Alameda Urquijo S/N 48013 Bilbao, Spain Corresponding author: ∗ Izaskun Azcarate email: [email protected] Tel. : +34946018209 Fax. : +34946014259 Abstract1 The replacement of incandescent lamps with more energy-efficient lighting technologies2 has a direct influence on the way flicker is measured. The International Electrotechnical3 Commission (IEC) established in the 61000-4-15 standard the functional specifications of4 a flickermeter, taking a standard incandescent lamp’s response to voltage fluctuations as5 the reference. During the past ten years, different works have studied the sensitivity of6 modern lamps to analytical voltage fluctuations of low complexity. From these studies,7 1 the most widespread conclusion is that modern lamps are less sensitive to flicker than are8 incandescent lamps. Based on these results, international standardization organizations are9 currently studying two different possibilities for updating the flicker assessment procedure:10 adjusting the IEC flickermeter according to a new less sensitive reference lamp, or increasing11 the established compatibility levels for voltage fluctuations. This work presents for the first12 time a sensitivity analysis of a set of modern lamps subjected to real voltage signals that13 are more complex than analytical voltage fluctuations. The obtained results lead to the14 following conclusions: not all efficient lamps have a lower sensitivity to fluctuations than15 do incandescent lamps; the response of some lamps depends on the complexity of the input16 voltage fluctuation; and the response of some lamps in real scenarios, i.e., more complex17 voltage fluctuations, does not correlate with their response to simple voltage fluctuations.18 Keywords19 Efficient Lighting, Voltage Fluctuations, Flicker, Power Quality, Complexity.20 2 1. Introduction21 International regulations have prompted the mass replacement of incandescent lamps22 with energy-efficient lighting technologies. There is a rising concern over the relationship23 between the large-scale introduction of such lamps and the Power Quality [1–3]. In this24 sense, the way flicker is measured now takes special relevance [4]. Flicker is defined as25 the irritation suffered by humans when exposed to illuminance fluctuations produced by26 changes in the supply voltage. The standard IEC 61000-4-15 [5], whose predecessor was IEC27 868 [6], establishes the functional specifications for a flickermeter—a device that objectively28 quantifies the level of irritation by using the short-term (10 min) and long-term (2 hours)29 flicker severity, Pst and Plt, respectively. The specifications and reference values, detailed in30 the standard from its first publication in 1986, were based on the leading lighting technology31 at that time: the 60 W incandescent lamp. Compatibility levels for voltage fluctuations were32 specified based on the assumption that a value of Pst = 1 would lead to complaints from at33 least 50% of the people exposed to the light fluctuation produced by an incandescent lamp.34 The gain curve is traditionally used to characterize the sensitivity of lamps subjected to35 sinusoidal voltage fluctuations. This method calculates the relationship between the relative36 amplitudes of the illuminance and voltage fluctuations for a given fluctuation frequency.37 Modern lamps’ gain curves are generally lower than those of incandescent lamps, so new38 lighting technologies are considered significantly less sensitive to voltage fluctuations [7, 8].39 Taking into account the progressive replacement of incandescent lamps and assuming the40 lower sensitivity of energy-efficient lighting technologies to voltage fluctuations, international41 organizations for the standardization and improvement of electric power systems are cur-42 rently studying alternative flicker measurement protocols adapted to current technologies:43 for instance, adjusting the IEC flickermeter according to a new, modern reference lamp, or44 increasing the established compatibility levels for voltage fluctuations [9].45 The specifications of the IEC flickermeter use the gain curve of an incandescent lamp46 as a linear model, combined with the human eye’s response to light fluctuations. The IEC47 flickermeter’s behavior is thus always linear, irrespective of the complexity of voltage fluc-48 3 tuations. However, it has not been demonstrated that this mirrors the behavior of modern49 lighting technologies [10, 11]. The current work analyzes the behavior of modern lamps with50 complex voltage supplies, such as rectangular analytical fluctuations and real voltage sig-51 nals. The main objective was to study whether their gain curves, which were obtained from52 simple sinusoidal fluctuations, were valid for characterizing their response to more complex53 fluctuations, in which case the assumed insensitivity would be convincingly demonstrated54 and the aforementioned proposals would go ahead. However, our work revealed that not all55 modern lamps show a consistent relationship between their gain curve and their behavior56 in real scenarios. Moreover, the present work clearly challenges the assumed insensitivity of57 new lighting technologies. The results point to uncertainty in the outcome of increasing the58 flicker compatibility levels, and difficulties in adapting the IEC flickermeter to a less-sensitive59 reference lamp.60 2. Experimental Setup61 This section describes the set of lamps under test (LUTs) and the system used to supply62 the lamps and to record the illuminance signals.63 2.1. Set of LUTs64 The main characteristics of the selected LUTs are given in Table 1. We selected a set65 of commercially available lamps from different manufacturers and using different lighting66 technologies, including halogen, linear fluorescent (LFL), compact fluorescent (CFL; using67 electronic or electromagnetic ballast), and light-emitting diode (LED) lamps. The study68 also included a CFL dimmable lamp because this feature involves additional brightness-69 control methods that also affect the flicker [12]. The LUTs had different energy efficiency70 ratings; these were based on European Union energy labeling, which uses classes from the71 most efficient (A) to the least efficient (G).72 2.2. Voltage Generation and Illuminance Recording73 Fig. 1 depicts the experimental setup used to generate the supply voltage for the LUTs74 and to record their illuminance signals. The setup could generate real or analytical voltage75 4 signals. The real voltage signals were previously recorded at a sampling rate of 6400 Hz by an76 acquisition system based on an analog-to-digital (A/D) converter (National Instruments (NI)77 USB-6281) with 18-bit resolution. The digitized analytical or real signal was converted into78 an analog signal by a digital-to-analog (D/A) converter (NI USB-6211) at a rate of 6400 Hz.79 The analog signal was amplified to 230 V by a 7500 Krohn-Hite Amplifier (75 W, from DC80 to 1 MHz) and a 120/230 V transformer. The output of the transformer was supplied to the81 LUT, which was enclosed in a white box together with the light sensor, and was connected82 to a luxmeter (E4-X Hagner Digital Luxmeter). This provided the illuminance signal, l(t),83 which was digitized by another A/D converter (NI USB-6211) at a rate of 6400 Hz with84 16-bit resolution, and then finally stored.85 3. Response to Analytical Voltage Fluctuations86 The responses of lamps to voltage fluctuations have traditionally been studied by means87 of their gain curves [8, 10, 13, 14]. A lamp is more sensitive to voltage fluctuations when88 its gain curve is higher. Each data point on this curve represents the gain factor for a given89 frequency of the voltage fluctuation, fm. This parameter assesses the relationship between90 the relative amplitudes of the illuminance and voltage fluctuations, ∆L Land ∆V V, respectively:91 G(fm) = ∆L/L ∆V/V .(1) The supply voltage of the LUT, when subjected to analytical voltage fluctuations, can92 be expressed as follows:93 u(t) = √2A1 + gm(t)cos(ωot),(2) where ωo= 2πforepresents the frequency of the mains supply, Arepresents its Root94 Main Square (RMS) value, and gm(t) represents the fluctuation.95 5 For a sinusoidal voltage fluctuation, gm(t) consists of a single frequency, fm, of relative96 amplitude ∆V V. In this case, the illuminance fluctuation also consists of a single frequency, fm,97 of relative amplitude ∆L L. Fig. 2a depicts the gain curve of each LUT, GLUT, normalized to98 the values corresponding to the reference incandescent lamp, GI1. Each LUT was subjected99 to 32 sinusoidal fluctuations with fmvalues ranging from 1 to 32 Hz, with ∆V V= 1%,100 according to (2). The results reveal that the sensitivity of the halogen lamp (H1) is quite101 close to that of I1 over the whole frequency range, while the rest of the LUTs show reduced102 sensitivity (relative to I1) over a wide frequency range, from approximately 3 to 25 Hz.103 3.1. Sensitivity analysis for rectangular fluctuations104 For rectangular voltage fluctuations, which are characterized by an unlimited bandwidth,105 it is not possible to calculate ∆L Lfor some of the LUTs. Fig. 3 depicts the waveforms of106 the illuminance fluctuations for the I1 and F1 lamps for a rectangular voltage fluctuation107 with fm= 8 Hz and ∆V V= 1.5%. In the case of I1 (Fig. 3a), the illuminance envelope still108 consists of a predominant frequency fluctuation that allows the calculation of the relative109 amplitude, ∆L. However, in the case of F1 (Fig. 3b), the complexity of the waveform does110 not allow the direct identification of ∆Lor the accurate calculation of its gain factor.111 An alternative method for studying the responses of the lamps to complex voltage fluc-112 tuations should consider all the frequency components of the illuminance fluctuation that113 affect the flicker [15]. Thus, it would consider the real characteristics of the LUT and provide114 a closer approximation to the real irritation it produces. Assessment of the flicker severity115 by means of an illuminance flickermeter meets these requirements.116 The IEC 61000-4-15 standard defines the design specifications for a flickermeter based117 on the voltage supply, according to a physiological model of the lamp–eye–brain chain.118 Some of the five blocks of the model are based on the performance characteristics of the119 60 W incandescent lamp. An illuminance flickermeter would be based on the illuminance120 rather than the voltage signal, but should otherwise be based on the IEC flickermeter, with121 modification of the blocks in the IEC model that use the incandescent lamp’s response to122 voltage fluctuations as a reference. An illuminance flickermeter only includes four blocks123 6 (Fig. 4) because the quadratic demodulation (Block 2 of the IEC standard) is eliminated.124 Moreover, the weighting filter in Block 3 of the IEC standard, corresponding to the lamp–eye125 response, is modified so that the frequency characteristics of the human eye are only used126 in Block B. The current work used a highly accurate implementation of an illuminance127 flickermeter that was described in detail and validated in [16].128 The acquisition system depicted in Fig. 1 registered the illuminance of each LUT sub-129 jected to rectangular voltage fluctuations, generated according to (2). Each LUT was sub-130 jected to 32 fluctuations with fmvalues from 1 to 32 Hz, corresponding to ∆V Vvalues that131 produced one unit of flicker severity for an incandescent lamp, i.e., Pst,I1 = 1. The recorded132 illuminance signals were processed by the illuminance flickermeter, and the corresponding133 flicker severity values, Pst,LUT, were obtained.134 Fig. 2b depicts these Pst,LUT values. The results reveal three different behaviors. First,135 the C1 and L1 lamps exhibit low sensitivities, clearly below that of the incandescent lamp,136 with results comparable to their corresponding gain curves generated based on sinusoidal137 voltage fluctuations (Fig. 2a). Second, the sensitivity of lamp H1 is close to, or even higher138 than, that of the incandescent lamp, with results comparable to the gain curve. Third,139 the rest of the LUTs show different responses to sinusoidal and rectangular fluctuations.140 The response of C2 to rectangular fluctuations is higher than the gain curve for the whole141 frequency range and shows more sensitivity than I1 from 18 Hz onward. The response of142 F1 to rectangular fluctuations is similar to the gain curve up to 25 Hz; from this frequency143 onward, the sensitivity of F1 is quite close to that of the incandescent lamp. The response144 of C3 is also quite similar to the gain curve up to 25 Hz, but from this frequency onward it145 shows greater sensitivity than I1.146 The analysis showed that the behavior of some LUTs was different when supplied with147 rectangular versus sinusoidal voltage fluctuations. The higher complexity of the rectangular148 fluctuations produced flickering behavior in some LUTs that differed from their expected149 responses according to their gain curves. Hence, it is necessary to analyze the responses of150 the LUTs when they are supplied with real signals, i.e., with fluctuations that are presumably151 more complex because of their lack of repetitive characteristics.152 7 4. Behavior in Real Scenarios153 We analyzed the response of each LUT to voltage signals registered at four different154 locations of the Low Voltage (LV) network (230 V/50 Hz) in the north of Spain. The155 locations were selected based on features such as the population, type of disturbing loads,156 and level of flicker severity.157 4.1. Description of the sites158 Fig. 5 depicts the evolution of the Pst and Plt values for the real voltage signals at each159 site over approximately one week, as well as their 99th percentiles, assessed by means of160 the IEC flickermeter [5]. Additionally, the figures include the percentage of time, TTH, for161 which both parameters exceed the irritability threshold (Pst = 1, Plt = 1). The Plt values162 were calculated using a time interval of two hours, i.e., 12 short time intervals. A Plt value163 was calculated as the cubic average of 12 consecutive Pst values. Each new Plt value was164 obtained with the 11 most recent Pst values used in the calculation of the previous Plt value165 and the next new Pst value. Following this procedure, one Plt value for each Pst value was166 obtained. A brief description of each site is detailed next.167 •Site a (Sa): A metropolitan area of 900,000 inhabitants with relevant industrial activity,168 including steel mills working with arc furnaces. This site did not present excessive169 flicker severity levels: Pst,99 = 1.21 and Plt,99 = 0.93, being close to the irritability170 threshold, with only 3.8% of the Pst values exceeding this limit and all the Plt values171 being below it (Fig. 5a).172 •Site b (Sb): A tourist destination of 100,000 inhabitants (residents and visitors) during173 the holiday period. This site is located far from big industrial loads. In this case, the174 flicker severity values remained above the irritability threshold: Pst,99 = 1.74 and175 Plt,99 = 1.49. The flicker severity values exceeded the limit 67% of the time for Pst and176 86.6% of the time for Plt (Fig. 5b).177 •Site c (Sc): A small town of 15,000 inhabitants located in a steel industry area, pre-178 dominantly with installations equipped with arc furnaces. The duty cycle of these179 8 protocol for new lamps.349 Acknowledgments350 This work received financial support from the Government of Basque Country through351 the PhD studentship BFI-2012-315 and from the University of the Basque Country UPV/EHU352 through the project UFI11/16 and the PhD studentship PIF2011/169.353 The authors would like to thank LightLabs department of Philips Lighting (Eindhoven)354 for providing us with lamp samples and the necessary information regarding the working355 principles of these lamps.356 References357 [1] M. L. Schenne, A. K. Ziarani, T. H. Ortmeyer, A novel adaptive flicker measurement technique, Int. J.358 Electr. Power. Energy Syst. 33 (10) (2011) 1686–1694. doi:10.1016/j.ijepes.2011.08.008.359 [2] S. Uddin, H. Shareef, A. Mohamed, Power quality performance of energy-efficient low-wattage LED360 lamps, Measurement 46 (10) (2013) 3783–3795. doi:10.1016/j.measurement.2013.07.022.361 [3] A. Gil-de Castro, S. K. R¨onnberg, M. H. J. Bollen, A. Moreno-Mu˜noz, Study on harmonic emission of362 domestic equipment combined with different types of lighting, Int. J. Electr. Power. Energy Syst. 55363 (2014) 116–127. doi:10.1016/j.ijepes.2013.09.001.364 [4] A. Hooshyar, E. El-Saadany, Development of a flickermeter to measure non-incandescent lamps flicker,365 IEEE Trans. Power Deliv. 28 (4) (2013) 2103–2115. doi:10.1109/TPWRD.2013.2267153.366 [5] IEC-61000-4-15ed2.0, Electromagnetic Compatibility (EMC) Part 4: Testing and measurement tech-367 niques - Section 15: Flickermeter functional and design specifications (2010).368 [6] IEC-868, Flickermeter. Funtional and design specifications (1986).369 [7] M. Halpin et al., Review of flicker objectives for HV, MV and LV systems, CIGRE/CIRED WG C4.108370 (2009).371 [8] R. Cai et al., Flicker responses of different lamp types, IET Gener. Transm. Dis. 3 (9) (2009) 816–824.372 [9] Terms of reference, Tech. rep., Review of LV and MV Compatibility Levels for Voltage Fluctuations373 CIGRE WG C4.111 (2010).374 [10] I. Azcarate et al., Sensitivity to flicker of dimmable and non-dimmable lamps, in: Proc. of I2MTC375 Conference, 2012, pp. 344 –347.376 [11] R. Cai, Flicker interaction studies and flickermeter improvement, Ph.D. thesis, Faculty of Electrical377 Engineering, Eindhoven University of Technology (2009).378 15 [12] J. Smith, J. Speakes, M. Rashid, An overview of the modern light dimmer: design, operation, and379 application, in: Proc. of the 37th NAPS Conference, 2005, pp. 299 – 303.380 [13] K. Chmielowiec, Flicker effect of different types of light sources, in: Proc. of 11th EPQU Conference,381 IEEE, 2011, pp. 1–6. doi:10.1109/EPQU.2011.6128852.382 [14] J. Drapela, P. Toman, Interharmonic - Flicker curves of lamps and compatibility level for interharmonic383 voltages, in: Power Tech, Lausanne, IEEE, 2007, pp. 1552–1557.384 [15] I. Azcarate, J. Gutierrez, P. Saiz, A. Lazkano, L. Leturiondo, K. Redondo, Flicker characteristics385 of efficient lighting assessed by the IEC flickermeter, Electr. Power Syst. Res. 107 (2014) 21–27.386 doi:10.1016/j.epsr.2013.09.005.387 [16] I. Azcarate et al., Type testing of a highly accurate illuminance flickermeter, in: Proc. of 15th ICHQP388 Conference, 2012, pp. 897 –903.389 [17] S. Kay, J. Marple, S.L., Spectrum analysis–A modern perspective, Proceedings of the IEEE 69 (11)390 (1981) 1380–1419. doi:10.1109/PROC.1981.12184.391 [18] I. Rezek, S. Roberts, Stochastic complexity measures for physiological signal analysis, IEEE Trans.392 Biomed. Eng 45 (9) (1998) 1186–1191. doi:10.1109/10.709563.393 [19] C.E. Shannon, A mathematical theory of communication, Bell Syst. Tech. J. 27 (1948) 379–423 and394 623–656.395 [20] L. Frater, N. Watson, Light flicker sensitivity of high efficiency compact fluorescent lamps, in: Proc. of396 AUPEC Conference, IEEE, 2007, pp. 1–6. doi:10.1109/AUPEC.2007.4548067.397 [21] B. Heffernan, L. Frater, N. Watson, LED replacement for fluorescent tube lighting, in: Proc. of AUPEC398 Conference, IEEE, 2007, pp. 1–6. doi:10.1109/AUPEC.2007.4548064.399 [22] K. Taekhyun et al., LED lamp flicker caused by interharmonics, in: Proc. of IMTC Conference, IEEE,400 2008, pp. 1920–1925. doi:10.1109/IMTC.2008.4547361.401 [23] D. Arlt, M. Stark, C. Eberlein, Examples of international flicker requirements in high voltage networks402 and real world measurements, in: Proc. of 9th EPQU Conference, 2007, pp. 1 –4.403 [24] J. Ruiz et al., A review of flicker severity assessment by the IEC flickermeter, IEEE Trans. Instrum.404 Meas. 59 (8) (2010) 2037–2047.405 [25] McKinsey & Company Inc., Lighting the way: Perspectives on the global lighting market, http://406 img.ledsmagazine.com/pdf/LightingtheWay.pdf, [Online] (2011).407 [26] J. Slezingr et al., A new simplified model of compact fluorescent lamps in the scenario of smart grids,408 in: Proc. of the 15th ICHQP Conference, 2012, pp. 835 –841.409 16 Figure Captions410 Figure 1 Diagram of the voltage generation and illuminance recording411 processes.412 Figure 2 LUT sensitivity curves. (a) Gain curves for sinusoidal voltage413 fluctuations and (b) Pst values for rectangular voltage fluctua-414 tions.415 Figure 3 Waveform of the illuminance fluctuation for a rectangular volt-416 age fluctuation. (a) I1 lamp and (b) F1 lamp.417 Figure 4 Functional diagram of the illuminance flickermeter.418 Figure 5 Time evolution of Pst and Plt values for the real voltage signals419 at each site; (a) Sa, (b) Sb, (c) Scand (d) Sd.420 Figure 6 Real voltage signal at Scsite. (a) Temporal evolution of the421 voltage fluctuation and (b) power spectral density by Welch422 estimator.423 Figure 7 Plt values for each LUT at the selected sites. (a,b,c,d) Time424 evolution and (e,f,g,h) box-plots, corresponding to Sa, Sb, Sc 425 and Sd.426 Table Titles427 Table 1 Set of lamps under test (LUTs).428 Table 2 Spectral Entropy of the analyzed signals.429 Table 3 TS: percentage of Plt,LUT values exceeding the unit and simulta-430 neously the 90% of the Plt,I1 values.431 17 Voltage generation Illuminance recording Analytical signal generator Real Signals Record. Reprod. D/A conv. NI USB-6211 Krohn-Hite amplifier 120/230V transformer u(t) Hagner Luxmeter Light sensor A/D conv. NI USB-6211 l(t) Data storing Figure 1 Fluctuation Frequency (Hz) (b) Pst, LUT Fluctuation Frequency (Hz) (a) GLUT GI1 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0.2 0.6 1 1.8 0.2 0.6 1 1.8 Figure 2 Lux Time (s) (a) ∆L 0 0.10.20.30.40.5 1450 1480 L 1510 1540 Lux Time (s) (b) 0 0.10.20.30.40.5 1445 1455 L 1475 1485 Figure 3 l(t) BLOCK A INPUT ILLUMINANCE ADAPTER BLOCK B 0.05 35 10 DEMODUL. AND EYE-RESPONSE FILTERS BLOCK C SQUARING MULTIPLIER + SLIDING MEAN FILTER Pinst BLOCK D STATISTICAL EVALUATION Pst Figure 4 Plt Pst Flicker Severity Time (h) (a) TTH,Plt =0.0% TTH,Pst =3.8% Plt,99 =0.93 Pst,99 =1.21 Thurs.Wed.Tues.Mon. Sun. Sat. Fri. 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 0 0.5 1 1.5 2 2.5 Plt Pst Flicker Severity Time (h) (b) TTH,Plt =86.6% TTH,Pst =67.0% Plt,99 =1.49 Pst,99 =1.74 Sat.Fri.Thurs.Wed. Tues. Mon. Sun. 17h 01h 09h 17h 01h 09h 17h 01h 09h 17h 01h 09h 17h 01h 09h 17h 01h 09h 17h 0 0.5 1 1.5 2 2.5 Plt Pst Flicker Severity Time (h) (c) TTH,Plt =68.6% TTH,Pst =50.5% Plt,99 =1.63 Pst,99 =2.06 Mon.Sun.Sat.Fri. Thurs. Wed. Tues. 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 0 0.5 1 1.5 2 2.5 Plt Pst Flicker Severity Time (h) (d) TTH,Plt =95.3% TTH,Pst =77.2% Plt,99 =1.92 Pst,99 =2.29 Fri.Thurs.Wed. Tues. Mon. Sun. 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 00h 08h 0 0.5 1 1.5 2 2.5 Figure 5 PSD (dB/Hz) Frequency (Hz) (b) Time (min) (a) RMS (V) 50 150 250 350 13579 −100 −50 0 227 228 229 Figure 6 Time (h) (a) Plt 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 0 0.3 0.6 1 Plt LUT (e) L1 F1C3C2C1H1I1 0 0.3 0.6 1 Time (h) (b) Plt 17h 01h 09h 17h 01h 09h 17h 01h 09h 17h 0 0.5 1 1.5 Plt LUT (f) L1 F1C3C2C1H1I1 0 0.5 1 1.5 Time (h) (c) Plt 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 0 0.7 1.4 2.1 Plt LUT (g) L1 F1C3C2C1H1I1 0 0.7 1.4 2.1 Time (h) (d) Plt 16h 00h 08h 16h 00h 08h 16h 00h 08h 16h 0 0.7 1.4 2.1 Plt LUT (h) L1 F1C3C2C1H1I1 0 0.7 1.4 2.1 Figure 7