Review of the PEA Method for Space Charge Measurements on HVDC Cables and Mini-Cables

energies

Review Review of the PEA Method for Space Charge Measurements on HVDC Cables and Mini-Cables

Giuseppe Rizzo * , Pietro Romano * , Antonino Imburgia * and Guido Ala *

L.E.PR.E. HV Laboratory, Dipartimento di Ingegneria-Università di Palermo, Viale delle Scienze, Ediﬁcio 9, 90100 Palermo, Italy * Correspondence: [email protected] (G.R.); [email protected] (P.R.); [email protected] (A.I.); [email protected] (G.A.)

Received: 25 July 2019; Accepted: 8 September 2019; Published: 12 September 2019

Abstract: This review takes into account articles and standards published in recent years concerning the application of the Pulsed Electro Acoustic (PEA) method for space charge measurement on High Voltage Direct Current (HVDC) cables and mini-cables. Since the 80s, the PEA method has been implemented for space charge measurements on flat specimens in order to investigate space charge phenomena and to evaluate the ageing of dielectrics. In recent years, this technique has been adapted to cylindrical geometry. Several studies and experiments have been carried out on the use of the PEA method for full size cables and HVDC cable models. The experiments have been conducted using different arrangements of the measurement setup and focusing attention on different aspects of space charge phenomena. In this work, the importance of space charge measurement is highlighted and the state-of-the-art PEA method application to full size cables and mini-cables is described. The main aim of this paper is to offer a complete and current review of this technique. In addition, limits on the use of PEA method are examined and main possible directions of research are proposed in order to improve the applicability, reliability, and replicability of this method.

Keywords: space charge; pulsed electroacoustic technique (PEA); HVDC cables

1. Introduction During the last decades, the interest on DC transmissions has grown. This is due to the need of increasing the transnational exchange of electrical power, strictly related to the increase of the availability of energy produced by renewable randomly distributed sources, installed offshore and/or at long distances from the users. For several reasons, High Voltage Direct Current (HVDC) lines are the best solution for long distances and the unique solution for undersea applications [1]. The use of extruded cables for HVDC transmission has shown the onset of premature aging due to space charge accumulation phenomena in the bulk of the dielectric and near the electrodes. In recent years, many research groups have developed different techniques useful for detecting space charges accumulation in dielectric materials. Most of these techniques can be grouped as acoustic and thermal methods [2]. Thanks to a relatively simple measurement system and a good spatial resolution, the Pulsed Electro Acoustic (PEA) method is currently the most reliable way to detect space charge accumulation into the dielectric material of HVDC cables [3]. Since 1985, the year in which the group of Professor Tatsuo Takada demonstrated the possibility of measuring space charge accumulation with a non-destructive method, the scientific community has made important progress in the use of the PEA method [4]. The presence of space charge in the dielectric is a typical phenomenon occurring in HVDC cables. This is due to the DC high voltage supply of the cable line, which determines the application of an electric field for a long time in the dielectric region: this enables space charge accumulation phenomena

Energies 2019, 12, 3512; doi:10.3390/en12183512 www.mdpi.com/journal/energies Energies 2019, 12, 3512 2 of 23 to be established, as will be discussed in the next sections. Since the time constants of most space charge phenomena have a minimum order of magnitude equal to several dozen seconds, their manifestation in HVAC application is negligible. Many researchers have used the PEA method to carry out space charge measurements on flat samples of dielectric materials and more recently on HVDC mini and full size cables. Much progress has been made in the knowledge of performance and applicability of the PEA method. In addition, this technique has shown the occurrence of new phenomena observed for the first time in recent years and it has been used to evaluate the ageing of cables. However, further improvements to the PEA method are needed to obtain a universally accepted measurement protocol and to achieve the possibility of measuring on-site the aging of HVDC cables during their lifetime. To date, a protocol for the space charge measurements on cables has been proposed by IEEE Std 1732TM-2017 entitled “IEEE Recommended Practice for Space Charge Measurements on High Voltage Direct-Current Extruded Cables for Rated Voltages up to 550 kV” [5]. This standard recalls the contents reported in [6] and assumes that PEA method is suitable for its use for long term pre-qualification test and type test of HVDC extruded cables.

2. Space Charge Injection, Formation, and Accumulation Phenomena in HVDC Cables Space charge formation and accumulation is one of the most important causes of faults in HVDC cable transmission lines. This phenomenon is essentially due to the following reasons [1]: due to their dependence on the temperature, a spatial gradient both of conductivity and permittivity occurs in loaded HVDC cables. This leads to the establishment of space charge in the bulk of dielectric; in presence of high electric field discontinuity, an injection of charges from electrodes occurs; these phenomena are amplified by the presence of imperfections at the interfaces between different materials; thermal ionization leads to the formation of space charge inside the insulation; the charge carriers injected by the electrodes can remain in proximity of these ones, or can migrate leading to a hetero charge accumulation in the proximity of the opposite electrode or remain in the bulk of the insulation in traps due to the inhomogeneity of the material itself. In Figure1, the typical stratigraphic distribution of materials in an HVDC cable for land application is shown while, in Figure2, a mini-cable is shown. The latter is typically devoid of the outer layers of the external semiconductor in order to easily perform space charge measurements. This is commonly consideredEnergies 2019, a12 good, x FOR solution PEER REVIEW for the qualification of semicon-dielectric-semicon compounds [7,8].3 of 23

Figure 1. Typical on High Voltage Direct Current (HVDC) full size cable. Figure 1. Typical on High Voltage Direct Current (HVDC) full size cable.

Figure 2. Typical HVDC mini-cable.

A high thermal gradient leads to a high conductivity gradient in the insulation until it causes a reversal of the radial electric field profile. For this reason, during recent years, new insulating materials have been developed in order to achieve high thermal conductivity and less dependence of the dielectric properties on temperature [9]. As shown in Figure 3, with typical operating temperature gradients for the HVDC cables, the radial component of the electric field at the external semiconductor layer can reach values close to the initial ones at the internal semiconductor layer. The profiles shown in Figure 3 can be found through a semi-empirical expression useful to calculate the electric field across the dielectric EDC of an HVDC cable under thermal and electrical stress, as is reported in Equation (2) [1]:

Energies 2019, 12, x FOR PEER REVIEW 3 of 23

Energies 2019, 12, 3512 3 of 23 Figure 1. Typical on High Voltage Direct Current (HVDC) full size cable.

Figure 2. Typical HVDC mini-cable. Figure 2. Typical HVDC mini-cable. It has been demonstrated that, in the presence of a radial thermal gradient, the pattern of the electricA high field thermal over the gradient insulation leads layer to ofa high HVDC conductivity cables is strongly gradient influenced in the insulation by the current until it field causes over a thereversal insulation of the layer radial [1]. electric field profile. For this reason, during recent years, new insulating materialsCombining have been the currentdeveloped continuity in order equation, to achieve the high Ohm’s thermal law forconductivity current density and less and dependence the Gauss law of forthe spatialdielectric charge, properties it has on been temperature shown that [9]. if there As sh isown a permittivity in Figure 3, and with conductivity typical operating gradient, temperature there is a spacegradients charge for density the HVDC in accordance cables, withthe theradial following component Equation: of the electric field at the external semiconductor layer can reach values close to the initial ones at the internal semiconductor layer. ε ε The profiles shown in Figure 3 can σbeE found0 r through= ρ, a semi-empirical expression useful (1)to ∇ σ calculate the electric field across the dielectric EDC of an HVDC cable under thermal and electrical wherestress, σasis is the reported electric in conductivity, Equation (2)E [1]:is the electric field, ε0 is the vacuum permittivity, εr is the relative permittivity, and ρ is the free charge volume density. Within the operating temperature range, the conductivity of existing insulating materials can vary up to three orders of magnitude, therefore, the distribution of electric field is heavily influenced by thermal gradient [1]. The thermal profile inside the dielectric of a cable typically follows a logarithmic law and depends on the conductive losses, the external surface heat exchange conditions, the material properties, and cable’s geometry. A high thermal gradient leads to a high conductivity gradient in the insulation until it causes a reversal of the radial electric field profile. For this reason, during recent years, new insulating materials have been developed in order to achieve high thermal conductivity and less dependence of the dielectric properties on temperature [9]. As shown in Figure3, with typical operating temperature gradients for the HVDC cables, the radial component of the electric field at the external semiconductor layer can reach values close to the initial ones at the internal semiconductor layer. The profiles shown in Figure3 can be found through a semi-empirical expression useful to calculate the electric field across the dielectric EDC of an HVDC cable under thermal and electrical stress, as is reported in Equation (2) [1]:

δ 1 δU0(r/r0) − EDC(r) = , (2) r [1 (r /r )δ] 0 − i 0 Energies 2019, 12, x FOR PEER REVIEW 4 of 23

𝛿𝑈 𝑟/𝑟 𝐸 𝑟 = , (2) 𝑟1−𝑟/𝑟 Energies 2019, 12, 3512 4 of 23 where: 𝑎∆𝑇 𝑏𝑈 + where: 𝑙𝑛𝑟 /𝑟 𝑟 −𝑟 𝛿= a∆T + bU 0 , (3) ln(r /r ) 𝑏𝑈(r r ) δ = 1+0 i 0− i , (3) bU 1 + 𝑟 −𝑟0 (r0 ri) and ΔT is the temperature difference between the inner− and the outer radius of the dielectric, a and b and ∆T is the temperature difference between the inner and the outer radius of the dielectric, a and b are, are, respectively, the temperature coefficient and the electrical stress coefficient of electrical respectively, the temperature coefficient and the electrical stress coefficient of electrical conductivity, r0 conductivity, r0 and ri are, respectively, the outer and the inner radius of the cable insulation, and U0 and r are, respectively, the outer and the inner radius of the cable insulation, and U is the voltage is thei voltage across the dielectric. 0 across the dielectric.

Figure 3. Example of no-load and full load electric field distribution in HVDC cable—figure reproduced fromFigure [9]. 3. Example of no-load and full load electric field distribution in HVDC cable—figure reproduced from [9]. Equation (2) can be used to evaluate the electric field in a DC cable under thermal and electrical stressEquation without considering (2) can be used other to space evaluate charge the formation electric field and in accumulation a DC cable under phenomena, thermal strictly and electrical related tostress the interface without quality, considering the material other properties,space charge and formation the presence and of defectsaccumulation or traps. phenomena, In real applications, strictly otherrelated phenomena to the interface lead toquality, the establishment the material ofproperti space chargees, and inthe the presence dielectric of suchdefects as theor traps. injection In real of chargeapplications, carriers other by the phenomena electrodes orlead their to formingthe establishmen by meanst of space thermal charge ionization in the e dielectricffects as well such as as their the trappinginjection inof thecharge bulk carriers due to theby the presence electrodes of defects, or their discontinuities, forming by means additives, of thermal etc. [1 ].ionization effects as wellThe as typical their trapping crosslinking in the by-products bulk due to of the XLPE presen (cross-linkedce of defects, polyethylene) discontinuities, insulated additives, cables etc. are[1]. one ofThe the typical main types crosslinking of defects by-products that enhance of theXLPE accumulation (cross-linked of spacepolyethylene) charge. The insulated degassing cables of theare insulationone of the materialmain types is aof good defects solution that enhance to reduce the the accumulation presence of of crosslinking space charge. by-products The degassing [10]. of Trap the depthsinsulation can material be estimated is a good by applying solution theto reduce PEA method the presence for space of crosslinking charge measurements by-products in [10]. different Trap scenarios:depths can equilibrium, be estimated charging, by applying and dischargingthe PEA method [11]. Thefor sumspace of charge the aforementioned measurements phenomena in different leadsscenarios: to a not equilibrium, negligible charging, modification and of dischargin the radialg electric [11]. The field sum distribution of the aforementioned inside the insulation phenomena of a HVDCleads to cable, a not withnegligible respect modification to the ideal of one. the Theseradial phenomenaelectric field playdistribution an important inside rolethe insulation during some of a transientHVDC cable, scenarios with that respect occur to during the ideal the lifetimeone. Thes ofe an phenomena HVDC cable, play such an asimportant fast polarity role inversion during some and transienttransient overscenarios voltages. that occur during the lifetime of an HVDC cable, such as fast polarity inversion and Attransient the beginning over voltages. of a fast polarity inversion transient, the inverted electric field profile due to the thermalAt gradientthe beginning and to of the a otherfast polarity above-mentioned inversion phenomena,transient, the is inverted added to electric the new field conditions profile related due to tothe the thermal polarity gradient inversion. and to Under the other certain above-mentio conditionsned depending phenomena, on thermal is added gradient, to the new age level,conditions and characteristicsrelated to the ofpolarity the cable, inversion. the maximum Under certain local electric conditions stress depending could significantly on thermal exceed gradient, the rated age valuelevel, leadingand characteristics to the ageing of ofthe the cable, cable. the This, maximum in the presencelocal electric of defects, stress could such significantly as crosslinking exceed by-products, the rated couldvalue leadleading to partialto the dischargeageing of the phenomena cable. This, and in to the the presence aging of of the defects, cable’s such insulation, as crosslinking causing by- its breakdownproducts, could [12]. lead to partial discharge phenomena and to the aging of the cable’s insulation, causing its breakdown [12].

Energies 2019, 12, 3512 5 of 23

3. Principles of the Pulsed Electro Acoustic (PEA) Method—Application on Flat Specimens The Pulsed Electro Acoustic (PEA) method is currently the most used technique for the space charge accumulation measurements into dielectric materials. As already underlined, the widespread use of the PEA method in detecting space charge phenomena in HVDC cables and mini-cables has led to the development of international standard and protocols for its implementation in Prequalification (PQ) tests, Type Tests (TT) [5,6], and ageing programs [13]. This section briefly explains the principles of the PEA method through the description of its application on flat specimens. The PEA method is physically based on the measurement of the acoustic waves generated by the vibrations of charge carriers, laying into the dielectric material, by means of a superimposed pulsed voltage applied between the electrodes. In Figure4, the path of the pressure waves generated by the vibration of the charge carriers within the dielectric is illustrated. After having reached one of the specimen interfaces, the pressure waves propagate toward a metallic grounded electrode and are converted into an electric signal by a piezoelectric membrane. Downstream of the latter, an absorbent material avoids the reflection of the pressure waves. In order to minimize the pressure waves reflection before their detection with the piezoelectric sensor, the adherence of the contact surfaces between the crossed materials should be optimized. In Figure5, a schematic of the measurement setup for flat specimens is shown. On the left-hand side, the high voltage aluminum electrode is connected with the output of the DC voltage source VDC and with the pulse generator Vp(t). A bias resistor RDC and a decoupling capacitor CC are inserted between the high voltage electrode and the VDC source and the pulse generator, respectively, in order to correctly superimpose the pulses on the DC voltage. A backing material is interposed between the high voltage electrode and the dielectric sample in order to optimize the interface from the point of view of the electrical and acoustic characteristics. On the right-hand side of the sample, a grounded aluminum electrode is crossed by the pressure waves generated by the vibration of charges and is linked to the piezoelectric transducer. The latter is linked to another aluminum electrode by means of a backing material. The electric signal produced by the interaction of the pressure waves with the piezoelectric membrane is measured between the grounded and the extreme right-hand side electrodes, and it is amplified in order to allow the measurement by the oscilloscope. The value of RDC is typically much larger than the impedance of the sample capacity at the frequencies of the pulse generator. A semiconductor is used as backing material due to its acoustic impedance value which is similar to that of the dielectric material. A thin polivinylidene difluoride (PVDF) film is used widespread as a piezoelectric transducer when the measurements are carried out at environmental temperature [14]. If higher temperatures are reached, a crystal sensor of Lithium Niobate is more reliable due to its stability in the presence of temperature changes [15]. The thickness of the piezoelectric membrane plays an important role for the goodness of the measurement since the wave crossing time through the film is proportional to the spatial resolution of the results [16]. On the other hand, a thin film generates a small signal amplitude. A compromise must be therefore reached between the need for a satisfactory spatial resolution and a sufficient signal amplitude output from the piezoelectric. Energies 2019, 12, x FOR PEER REVIEW 6 of 23

𝑏/𝑢 𝜂 = , (5) 𝑑/𝑢 where b and up are the piezoelectric film thickness (with an order of magnitude of about 50–500 μm) and sound speed within 2600 m/s for PVDF film, respectively. Nonetheless, the ratio b/up has to be much smaller than ΔTp in order to accurately transform the measured pulse by the means of signal deconvolution. As mentioned above, the presence of material interfaces along the path of the pressure waves leads to reflections depending on the change of the acoustic impedances. Multiple reflections of the generated waves cause the distortion of the measurement output pattern. In particular, with reference to Figure 5, the high value of the sound velocity in the aluminum grounded electrode (about equal to 6400 m/s) could lead to the superimposing of the doubly reflected waves generated at the closer sample interface with the direct wave started at the farthest one [18]. In order to avoid the above- mentioned waveform distortion, the thicknesses of the layers of different materials of the PEA cell have to be calibrated, taking into account the multiple reflections. Nevertheless, since the acoustic impedances depend on the temperature and on the mechanical stress of the materials (a clamping Energiesforce generates2019, 12, 3512 a variation of density), the status of the sample has to be well known before6 ofthe 23 beginning of the PEA measurements.

Figure 4. Pressure wave propagation into a ﬂatflat specimenspecimen of dielectric material—ﬁgurematerial—figure reproduced Energies 2019, 12, x FOR PEER REVIEW 7 of 23 from [[17].17].

Again with reference to Figure 5, the voltage signal existing between the grounded and the extreme right hand side electrodes, after the conversion of a pressure wave pulse, has an order of magnitude of some millivolts. The signal is then amplified and sent to the oscilloscope with a time resolution sufficiently smaller than the measurement resolution in time domain.

Figure 5. MeasurementMeasurement setup scheme for flat ﬂat spec specimens—ﬁgureimens—figure reproduced from [[18].18].

TheIn order spatial to resolutionobtain a reliable of the PEAspace setup charge may pattern be defined at the as output the minimum of the PEA distance setup, z overthe outgoing which a volumetricsignal of the charge amplifier, density measured variation by the can oscilloscope, be measured. has The to be spatial deconvoluted resolution and is relatedcalibrated to the [17]. pulse The widthsignal ∆atT pthe, to amplifier the sample is thicknessaffected dby, anddistortion to the soundbecause velocity of the in acoustic the sample, reflection about and 2000 due m/s into polyethylene,characteristic nature as follows of the [16 detection,17]: circuit. A deconvolution technique is therefore applied in order ∆Tp to obtain the waveform with two peaks shownηsa =in Figure. 6 [19]. Since the aforementioned distortion (4) can be considered as a systematic error, the decond/volutionusa can be based on the signal obtained by applyingUsually, a small the ηelectricsa value field is in at the the range sample. between In this 0.02 condition, and 0.05. it can The be spatial assumed resolution that the of charges the PVDF lay filmexclusively is calculated at the by electrode the time necessaryinterfaces for while the pressureno charge wave is toin crossthe bulk the piezoelectricof the dielectric. membrane With withthis theassumption, time necessary the surface to cross charge the sample density [ 16at]: the electrode σ0 can be calculated as: 𝑉 b/up 𝜎 =𝜀𝜀 , (6) ηPVDF = 𝑑 , (5) d/usa 𝜎 𝜌 = , (7) where b and up are the piezoelectric film thickness𝜏 (with𝑢 an order of magnitude of about 50–500 µm) and sound speed within 2600 m/s for PVDF film, respectively. Nonetheless, the ratio b/up has to where VDC is the applied voltage, τS is the sampling time due to the resolution of the oscilloscope, and usa is the sound speed into the sample. The space charge at the electrode surface is detected as a volumetric density 𝜌 due to the non-zero spatial resolution of the oscilloscope equal to τS usa. The measurement output in the above described conditions involves two peaks with different heights due to the partial absorption of the pulse generated at the farthest electrode. The higher peak measured at the oscilloscope can be compared with the calculated value of 𝜌 in order to evaluate the transfer function. The deconvolution process in PEA measurements, schematized in Figure 6, is extensively described in [19].

Figure 6. Example of deconvolution.

Energies 2019, 12, x FOR PEER REVIEW 7 of 23 Energies 2019, 12, 3512 7 of 23

be much smaller than ∆Tp in order to accurately transform the measured pulse by the means of signal deconvolution. As mentioned above, the presence of material interfaces along the path of the pressure waves leads to reflections depending on the change of the acoustic impedances. Multiple reflections of the generated waves cause the distortion of the measurement output pattern. In particular, with reference to Figure5, the high value of the sound velocity in the aluminum grounded electrode (about equal to 6400 m/s) could lead to the superimposing of the doubly reflected waves generated at the closer sample interface with the direct wave started at the farthest one [18]. In order to avoid the above-mentioned waveform distortion, the thicknesses of the layers of different materials of the PEA cell have to be calibrated, taking into account the multiple reflections. Nevertheless, since the acoustic impedances depend on the temperature and on the mechanical stress of the materials (a clamping force generates a variationFigure of density), 5. Measurement the status setup of thescheme sample for flat has spec toimens—figure be well known reproduced before the from beginning [18]. of the PEA measurements. In order to obtain a reliable space charge pattern at the output of the PEA setup, the outgoing Again with reference to Figure5, the voltage signal existing between the grounded and the signal of the amplifier, measured by the oscilloscope, has to be deconvoluted and calibrated [17]. The extreme right hand side electrodes, after the conversion of a pressure wave pulse, has an order of signal at the amplifier is affected by distortion because of the acoustic reflection and due to magnitude of some millivolts. The signal is then amplified and sent to the oscilloscope with a time characteristic nature of the detection circuit. A deconvolution technique is therefore applied in order resolution sufficiently smaller than the measurement resolution in time domain. to obtain the waveform with two peaks shown in Figure 6 [19]. Since the aforementioned distortion In order to obtain a reliable space charge pattern at the output of the PEA setup, the outgoing signal can be considered as a systematic error, the deconvolution can be based on the signal obtained by of the amplifier, measured by the oscilloscope, has to be deconvoluted and calibrated [17]. The signal applying a small electric field at the sample. In this condition, it can be assumed that the charges lay at the amplifier is affected by distortion because of the acoustic reflection and due to characteristic exclusively at the electrode interfaces while no charge is in the bulk of the dielectric. With this nature of the detection circuit. A deconvolution technique is therefore applied in order to obtain assumption, the surface charge density at the electrode σ0 can be calculated as: the waveform with two peaks shown in Figure6[ 19]. Since the aforementioned distortion can be considered as a systematic error, the deconvolution can𝑉 be based on the signal obtained by applying a 𝜎 =𝜀𝜀 , (6) small electric field at the sample. In this condition, it can𝑑 be assumed that the charges lay exclusively 𝜎 at the electrode interfaces while no charge is𝜌 in= the bulk, of the dielectric. With this assumption, the 𝜏 𝑢 (7) surface charge density at the electrode σ0 can be calculated as: where VDC is the applied voltage, τS is the sampling time due to the resolution of the oscilloscope, and VDC usa is the sound speed into the sample. Theσ0 space= ε0ε rcharge, at the electrode surface is detected as (6) a d volumetric density 𝜌 due to the non-zero spatial resolution of the oscilloscope equal to τS usa. σ0 The measurement output in the above describedρ0 = conditions, involves two peaks with different (7) heights due to the partial absorption of the pulse geτSneratedusa at the farthest electrode. The higher peak wheremeasuredVDC atis the the oscilloscope applied voltage, can beτS coismpared the sampling with the time calculated due to the value resolution of 𝜌 in of order the oscilloscope, to evaluate theand transferusa is the function. sound speed The deconvolution into the sample. process The space in PEA charge measurements, at the electrode schematized surface is in detected Figure 6, as is a volumetricextensively densitydescribedρ0 indue [19]. to the non-zero spatial resolution of the oscilloscope equal to τS usa.

Figure 6. Example of deconvolution.

Energies 2019, 12, 3512 8 of 23

The measurement output in the above described conditions involves two peaks with different heights due to the partial absorption of the pulse generated at the farthest electrode. The higher peak measured at the oscilloscope can be compared with the calculated value of ρ0 in order to evaluate theEnergies transfer 2019, 12 function., x FOR PEER The REVIEW deconvolution process in PEA measurements, schematized in Figure8 6of, is23 extensively described in [19]. OnceOnce thethe outputoutput deconvoluteddeconvoluted signal signal is is obtained obtained in in millivolts, millivolts, a furthera further operation operation is necessaryis necessary in orderin order to calculateto calculate the valuethe value of volume of volume density density charge. charge. By assuming By assuming that the that pulsed the pulsed voltage voltage is square is shaped,square shaped, the following the following expression expression can be used can inbe orderused toin convertorder to the convert output the deconvoluted output deconvoluted signal in spacesignal charge in space [20 charge]: [20]: vdecon(t) = Kρ, (8) 𝑣𝑡 =𝐾𝜌, (8) where K is calculated on the basis of a comparison with the signal obtained from a known value of where K is calculated on the basis of a comparison with the signal obtained from a known value of space charge [20]. space charge [20]. The above described principles are the basis for using the PEA method on flat specimens. The above described principles are the basis for using the PEA method on flat specimens. The The measurement setup shown in Figure5 can be used for single or multiple dielectric layers [ 2,19]. measurement setup shown in Figure 5 can be used for single or multiple dielectric layers [2,19]. In In the case of space charge measurements on multilayer flat specimens, particular attention must the case of space charge measurements on multilayer flat specimens, particular attention must be be paid to take into account the different permittivities and acoustic properties of the layers [21,22]. paid to take into account the different permittivities and acoustic properties of the layers [21,22]. The The pressure distribution at the piezoelectric membrane depends on the acoustic impedances (Z) of pressure distribution at the piezoelectric membrane depends on the acoustic impedances (Z) of the the crossed materials, defined as the product between the acoustic velocity and the mass density. crossed materials, defined as the product between the acoustic velocity and the mass density. When When the acoustic wave passes from a material with acoustic impedance Z1 to a material with acoustic the acoustic wave passes from a material with acoustic impedance Z1 to a material with acoustic impedance Z2, the ratio of the transmitted pressure amplitude T can be calculated as follows: impedance Z2, the ratio of the transmitted pressure amplitude T can be calculated as follows: 2Z T = 2𝑍2 . (9) 𝑇=Z + Z . (9) 𝑍1 +𝑍2 OnOn thethe otherother hand,hand, thethe ratioratio ofof thethe reflectedreflected pressurepressure amplitudeamplitude isis calculablecalculableby by the the following: following:

𝑍 −𝑍 𝑅=Z1 Z2 . (10) R = 𝑍 −+𝑍 . (10) Z1+ Z2 From the above Equations, it can be deduced that high differences in acoustic impedances of the From the above Equations, it can be deduced that high differences in acoustic impedances of the crossed layers lead to high reflection ratios. Multiple reflections can distort the measurements as crossed layers lead to high reflection ratios. Multiple reflections can distort the measurements as shown shown in Figure 7 [23]. In addition, for the aforementioned reasons, it is necessary to optimize the in Figure7[ 23]. In addition, for the aforementioned reasons, it is necessary to optimize the continuity continuity between the piezoelectric sensor and the sample in order to avoid the presence of gas. The between the piezoelectric sensor and the sample in order to avoid the presence of gas. The acoustic acoustic impedance of the latter is several orders of magnitude lower than that of the solid materials, impedance of the latter is several orders of magnitude lower than that of the solid materials, therefore, therefore, high values of reflection factors could be reached shielding a portion of the sample. high values of reflection factors could be reached shielding a portion of the sample.

Figure 7. Acoustic wave multiple reflections within a multilayer flat specimen—figure reproduced Figure 7. Acoustic wave multiple reflections within a multilayer flat specimen—figure reproduced from [22]. from [22].

4. Measurements of Space Charge on Cables and Mini-Cables with the PEA Method In 1990, K. Fukunaga and their group developed for the first time the application of the PEA method to cables [24]. The space charge measurement by the PEA method in cables and mini-cables is a reproduction of the above described principles on cylindrical geometry. With reference to Figure 5, the upper and lower electrodes are substituted by the inner and outer semicon layers. The pulsed voltage generating the pressure waves is therefore applied between the cable conductor and an external flat or rounded

Energies 2019, 12, 3512 9 of 23

4. Measurements of Space Charge on Cables and Mini-Cables with the PEA Method Energies 2019, 12, x FOR PEER REVIEW 9 of 23 In 1990, K. Fukunaga and their group developed for the first time the application of the PEA methodelectrode to cables adjacent [24 to]. the outer semicon layer. With reference to Figure 8, the space charge distribution canThe be spacemeasured charge in measurementa portion of cable by the having PEA method sizes depending in cables and on mini-cablesthe extensions is a of reproduction piezoelectric ofmembrane the above describedand the contact principles surface on cylindrical between the geometry. cable outer With and reference the bottom to Figure electrode.5, the upper Unlike and the lowerapplication electrodes of PEA are substituted method to flat by thespecimens, inner and in outercable semiconsystems, layers.one of the The two pulsed electrodes voltage is generating not directly theaccessible. pressure wavesFor this is reason, therefore various applied measurement between the setups cable conductorhave been anddeveloped an external in the flat last or decades rounded for electrodethe application adjacent of to the the PEA outer method semicon on layer. cables With and reference mini-cables. to Figure 8, the space charge distribution can beIn measured all setups in described a portion below, of cable the havingpulse generator sizes depending output is onindirectly the extensions connected of to piezoelectric electrodes of membranethe measurement and the area. contact This surface leads to between obtain a the deform cableed outer and delayed and the waveform bottom electrode. at the interfaces Unlike of the the applicationdielectric ofportion PEA methodin which to the flat space specimens, charge in has cable to systems,be measured. one of The the twomain electrodes purpose isof notthe directlydifferent accessible.approaches For is this to find reason, an optimized various measurement solution to setupsapply the have pulsed been developedelectric field in to the the last measured decades forarea thebased application on the cable of the thickness PEA method and onlength. cables and mini-cables.

Figure 8. Space charge measurement on a cable portion by the Pulsed Electro Acoustic (PEA) method. Figure 8. Space charge measurement on a cable portion by the Pulsed Electro Acoustic (PEA) method. In all setups described below, the pulse generator output is indirectly connected to electrodes of 4.1. SETUP 1—Pulsed Voltage Applied to the Cable Conductor the measurement area. This leads to obtain a deformed and delayed waveform at the interfaces of the dielectricThe portion first PEA in whichmeasurement the space setup charge described has to in be this measured. review and The here main called purpose Setup of the1 is diillustratedfferent approachesin Figure 9. is It to was find proposed an optimized for the solution first time to by apply K. Fukunaga the pulsed et electrical. [24] and field derives to the measureddirectly from area the basedconfiguration on the cable adopted thickness in the and case length. of flat specimens shown in Figure 5. In this approach, the HVDC supply and the pulsed voltage are both applied to the cable conductor. In Figure 10 a schematic of 4.1.the SETUP Setup 1—Pulsed1 is shown Voltage where Applied a bias resistor to the Cable Rb is Conductor interposed between the HVDC source and the cable conductorThe first while PEA measurementa capacitor C0setup decouples described the latter in this from review the andpulse here generator called Setup output. 1 is The illustrated sizes ofin the Figurecable9 are. It taken was proposed into account for thethrough first timethe impedance by K. Fukunaga Zcab and et the al. capacitance [ 24] and derives Ccab. In directly order to from maximize the configurationthe amplitude adopted of Vsa, the in theimpedance case of flat of C specimensx has to be shownmuch higher in Figure than5. that In this of C approach,0. The schematic the HVDC shown supplyin Figure and 10 the is pulsedvalid if the voltage cable are specimen both applied is short to enough the cable to be conductor. treated as In a lumped Figure 10 capacitance a schematic within of the frequency region of the pulsed voltage [25]. For long cables it has to be taken into account that the Setup 1 is shown where a bias resistor Rb is interposed between the HVDC source and the cable the pulses are reflected at the terminals due to the impedance mismatch. For this reason, the Setup 1 conductor while a capacitor C0 decouples the latter from the pulse generator output. The sizes of the is inapplicable to cables with length greater than 0.5 m. In the first version of Setup 1, a round PEA cable are taken into account through the impedance Zcab and the capacitance Ccab. In order to maximize cell electrode was faced to a sector of the outer semicon layer of the cable specimen [26,27]. In this the amplitude of Vsa, the impedance of Cx has to be much higher than that of C0. The schematic shown inway, Figure the 10 curvatures is valid if theof the cable ground specimen electrode is short and enough cable external to be treated surface as a must lumped be the capacitance same, therefore within a thecertain frequency PEA regioncell is ofusable the pulsed just for voltage a certain [25]. cabl Fore long size. cables Moreover, it has tothe be impossibility taken into account of obtaining that the in pulsespractice are reflecteda perfect atcontact the terminals of the facing due to surfaces the impedance leads to mismatch. the presence For thisof air reason, gaps theresponsible Setup 1 is of inapplicablereflection of to a cables part of with the lengthpressure greater waves. than In 0.5order m. Into theovercome first version the aforementioned of Setup 1, a round disadvantages, PEA cell electrodethe configuration was faced toof athe sector Setup of the1 illustrated outer semicon in Figure layer 8 of was the proposed cable specimen by M. [ 26Fu,27 et]. al. In in this 2000 way, [28]. the In this case, the absence of air gaps in the contact surface between the semicon layer and the ground electrode is guaranteed by a clamping force applied to the cable.

Energies 2019, 12, 3512 10 of 23 curvatures of the ground electrode and cable external surface must be the same, therefore a certain PEA cell is usable just for a certain cable size. Moreover, the impossibility of obtaining in practice a perfect contact of the facing surfaces leads to the presence of air gaps responsible of reﬂection of a part of the pressure waves. In order to overcome the aforementioned disadvantages, the conﬁguration of the Setup 1 illustrated in Figure8 was proposed by M. Fu et al. in 2000 [ 28]. In this case, the absence of Energiesair gaps 2019 in, the12, x contact FOR PEER surface REVIEW between the semicon layer and the ground electrode is guaranteed10 of by 23 a clampingEnergies 2019 force, 12, x appliedFOR PEER to REVIEW the cable. 10 of 23

Figure 9. Setup 1: pulsed voltage applied to the cable conductor. Figure 9.9. Setup 1: pulsed voltage applied to the cable conductor.

Figure 10. Schematic of Setup 1. Figure 10. Schematic of Setup 1. 4.2. SETUP 2—Pulsed Voltage AppliedFigure to the 10. Outer Schematic Semicon of LayerSetup at1. the Sides of the Measurement Point 4.2. SETUP 2—Pulsed Voltage Applied to the Outer Semicon Layer at the Sides of the Measurement Point In Figure 11 the PEA measurement configuration here named Setup 2 is shown. This measurement configuration,4.2. SETUPIn Figure 2—Pulsed extensively11 the Voltage PEA described Appliedmeasurement into the [29 ],Outer involvesconfiguration Semicon in the Layer applicationhere at thenamed Sides of ofSetup the the pulsed Measurement 2 is voltageshown. Point to This two measurementsymmetricIn Figure conductive configuration, 11 the tapes PEA woundextensiv measurement aroundely described theconfiguration outer in semicon[29], involves here layer named in at the sidesSetupapplication of 2the is measurementofshown. the pulsed This voltageregion.measurement to The two pulsed configuration,symmetric voltage conductive isextensiv transmitted tapesely described wound to the cablearound in [29], conductor theinvolves outer by semiconin meansthe application layer of the at dielectricthe of sides the pulsed of layer, the measurementtherefore,voltage to the two decoupling region.symmetric The capacitor conductive pulsed neededvoltage tapes in iswound Setup transm 1 around isitted not necessary tothe the outer cable in semicon this conductor configuration. layer byat the means The sides semicon of of thethe dielectriclayermeasurement is partially layer, region. therefore, stripped The aspulsedthe shown decoupling voltage in Figure iscapaci transm 11 torsoitted thatneeded theto the totalin cableSetup capacitance conductor1 is not between necessaryby means the in lateralof thisthe configuration.conductivedielectric layer, tapes The andtherefore, semicon the conductor the layer decoupling is twicepartially thecapaci capacitancestrippedtor needed as at shown the in measurementSetup in Figure 1 is not point.11 necessaryso Withthat referencethe in total this capacitancetoconfiguration. the diagram between The illustrated semicon the lateral in Figurelayer conductive is 12 partially, the pulsedtapes stripped and voltage the asconductor is shown applied inis at twiceFigure the cable the 11 capacitance conductorso that the byat total the measurementmeanscapacitance of two between equivalentpoint. Withthe capacitanceslateral reference conductive toC 1the. The diagramtapes regions and illustrated between the conductor the in lateral Figure is twice tapes 12, the and capacitancepulsed the measurement voltage at the is appliedpointmeasurement can at bethe modelled cablepoint. conductor With as areference mainly by the capacitive tomeans the diagramof two impedance equivalent illustratedZg. capacitances Thein Figure Setup 212, C can1 . theThe be pulsed regions considered voltage between as is a thesemi-destructiveapplied lateral at tapesthe cable and method conductor the measurement because by athe mechanical means point ofcan action two be equivalentmodelled on the outer ascapacitances a semicon mainly layercapacitive C1. The is necessary. regions impedance between For thisZg. Thereason,the lateralSetup the 2tapes Setupcan be and 2considered is the mainly measurement as applicable a semi-destructive point on mini-cables.can be method modelled Nevertheless, because as a mainly a mechanical the capacitive partial action stripping impedance on the ofouter theZg. semiconexternalThe Setup semiconductivelayer 2 can is be necessary. considered layer inducesFor as a thissemi-destructive a floatingreason, voltagethe methodSetup on the2 because surfaceis mainly a adjacent mechanical applicable the grounded action on mini-cables.on electrodethe outer Nevertheless,ofsemicon the PEA layer cell the whichis partialnecessary. is no stripping longer For uniformly ofthis the reason, external grounded. the semiconductive Setup For this2 is reason, mainly layer thisinduces applicable measurement a floating on mini-cables.setup voltage is noton theapplicableNevertheless, surface foradjacent longthe partial cables the grounded stripping [17]. electrodeof the external of the semiconductive PEA cell which layeris no induceslonger uniformly a floating grounded.voltage on Forthe thissurface reason, adjacent this measurementthe grounded setupelectrode is not of applicable the PEA cell for which long cables is no longer[17]. uniformly grounded. 4.3.For SETUPthis reason, 3—Pulsed this measurement Voltage Applied setup to the is Outernot applicable Semicon Layerfor long at the cables Measurement [17]. Point In Figure 13, the measurement configuration here called Setup 3 is shown, where the pulsed voltage is applied directly to the outer layer of the measuring region by the means of a conductive film coiled around the cable. In the first version proposed by N. Hozumi et al. [25], a PET film was interposed between the aforementioned conductive layer and the aluminum block of the PEA cell in order to decouple the measurement devices from the high voltage. Compared to the other setups, in this case, the pressure waves have to cross two additional layers before reaching the piezoelectric

Figure 11. Setup 2: pulsed voltage applied to the partially stripped outer semiconductor. Figure 11. Setup 2: pulsed voltage applied to the partially stripped outer semiconductor.

Energies 2019, 12, x FOR PEER REVIEW 10 of 23

Figure 9. Setup 1: pulsed voltage applied to the cable conductor.

Figure 10. Schematic of Setup 1.

4.2. SETUP 2—Pulsed Voltage Applied to the Outer Semicon Layer at the Sides of the Measurement Point In Figure 11 the PEA measurement configuration here named Setup 2 is shown. This measurement configuration, extensively described in [29], involves in the application of the pulsed voltage to two symmetric conductive tapes wound around the outer semicon layer at the sides of the measurement region. The pulsed voltage is transmitted to the cable conductor by means of the dielectric layer, therefore, the decoupling capacitor needed in Setup 1 is not necessary in this configuration. The semicon layer is partially stripped as shown in Figure 11 so that the total capacitance between the lateral conductive tapes and the conductor is twice the capacitance at the measurement point. With reference to the diagram illustrated in Figure 12, the pulsed voltage is applied at the cable conductor by the means of two equivalent capacitances C1. The regions between the lateral tapes and the measurement point can be modelled as a mainly capacitive impedance Zg. TheEnergies Setup2019 2, 12 can, 3512 be considered as a semi-destructive method because a mechanical action on the 11outer of 23 semicon layer is necessary. For this reason, the Setup 2 is mainly applicable on mini-cables. Nevertheless, the partial stripping of the external semiconductive layer induces a floating voltage on membrane. An improved version of Setup 3, proposed by N. Hozumi et al. in 1994, consists of the the surface adjacent the grounded electrode of the PEA cell which is no longer uniformly grounded. application of the pulsed voltage to the PEA cell electrode and in the decoupling of the ampliﬁer from For this reason, this measurement setup is not applicable for long cables [17]. the oscilloscope by means of an optical ﬁber [30].

Energies 2019, 12, x FOR PEER REVIEW 11 of 23 Figure 11. Setup 2: pulsed voltage applied to the partially stripped outer semiconductor. Figure 11. Setup 2: pulsed voltage applied to the partially stripped outer semiconductor.

Energies 2019, 12, x FOR PEER REVIEW 12 of 23

charge phenomena are not predominant. Indeed, the degassing effect of load cycles leads to a reduction of crosslinking by-products and consequently the tendency to accumulate space charge. An alternative to the above procedure is the implementation of the PEA measurements on the primary cable sample near one of the terminals in order to enable the removal of the region altered by the exposure of the outer semicon. As it can be deduced from the above described measurement setups, several parameters have a relevant influence on the results of a space charge measurement in a cable carried out with the PEA method. For the sake of simplicity, the following is a brief list of the most relevant aspects to consider when preparing a PEA measurement setup on cables or mini-cables in order to perform replicable and reliableFigure 12.measures.Schematic of Setup 2. Figure 12. Schematic of Setup 2.

4.3. SETUP 3—Pulsed Voltage Applied to the Outer Semicon Layer at the Measurement Point In Figure 13, the measurement configuration here called Setup 3 is shown, where the pulsed voltage is applied directly to the outer layer of the measuring region by the means of a conductive film coiled around the cable. In the first version proposed by N. Hozumi et al. [25], a PET film was interposed between the aforementioned conductive layer and the aluminum block of the PEA cell in order to decouple the measurement devices from the high voltage. Compared to the other setups, in this case, the pressure waves have to cross two additional layers before reaching the piezoelectric membrane. An improved version of Setup 3, proposed by N. Hozumi et al. in 1994, consists of the application of the pulsed voltage to the PEA cell electrode and in the decoupling of the amplifier from the oscilloscope by means of an optical fiber [30]. The use of this method requires the insulation of the PEA cell and of the amplifier from the ground. On the measurement region sides, beyond a stretch of cable with the outer semicon layer Figure 13. Setup 3: pulsed voltage applied to the PEA cell’s electrode. exposed, the remaining external surface is grounded by the means of a conductive tape. With referenceThe use to Figures of this method13 and 14, requires the impedance the insulation Zg is,of in thethis PEA case, cell mainly and ofresistive. the amplifier Since the from pulsed the ground.voltage is On directly the measurement applied to the region measurement sides, beyond region, a stretch this setup of cable is less with influenced the outer by semicon reflection layer and exposed,attenuation the remainingof the HV externalsignal than surface the isother grounded methods by thedescribed means ofabove a conductive [25]. This tape. makes With this reference method toapplicableFigures 13 for and long, 14, full the size impedance cables asZ proposedg is, in this by case, N. Hozumi mainly et resistive. al. in 1998 Since [31]. the In this pulsed configuration, voltage is directlythe PVC applied jacket toand the the measurement screen are removed region, this near setup the measurement is less influenced point, by reflectionpositioned and at the attenuation center of ofthe the exposed HV signal semicon than theregion. other The methods pulsed describedvoltage is aboveapplied [25 between]. This makesthe PEA this cell method electrode applicable and the forwire long, screen full of size the cables cable. as The proposed amplitude by N.of Hozumithe pulsed et voltage al. in 1998 is proportional [31]. In this configuration,to the ratio between the PVC the jacketimpedance and the of screen the capacitance are removed at the near measurement the measurement point point,and the positioned characteristic at the impedance center of the of the exposed cable. semiconThis makes region. this The method pulsed also voltage applicable is applied to thick between insulations the PEA because cell electrode the amplitude and the of wire the screen pressure of waves generated at the farthest electrode is proportional to the amplitude of the voltage pulses. The properties of the E/O and O/E converters, particularly in terms of sampling rate, has to be taken into account in order to avoid cutting high frequency contributions. For this purpose, an Figure 14. Schematic of Setup 3. improved version of Setup 3 is proposed in [16] and it consists of the direct connection between the amplifier and the oscilloscope. The latter is therefore insulated from the ground and connected to a Electrical stress application. Different levels of electric field are related to different space charge PC by means of optical fiber. formation and accumulation phenomena, therefore, several electrical stresses should be tested in the Thanks to its applicability to long and thick specimens, Setup 3 is mentioned by the standard [5] range of the typical working conditions (10–30 kV/mm). which recommends carrying out PEA measurements before and after PQ (Pre-Qualification) and TT Thermal conditions. The presence of a thermal gradient leads to the establishment of space (Type Test) on full size cables. As specified by this protocol, the PEA cell has to be adjacent to the charge within the dielectric due to the dependence on the temperature of the material conductivity. outer semi conductive layer, therefore the removal of the external shield is necessary. Since a different Nevertheless, electric and acoustic impedances of the sample and of the measurement setup can vary degassing could occur to the cable section with or without the external coating during the load cycle, with the temperature. In light of the above, the environmental and sample temperature should be two different procedures could be carried out. monitored in order to carry out replicable measurements. In addition, the temperature at the The first of them consists of PEA measurements performed before and after PQ test or TT on a piezoelectric film should be compared with its tolerability. The output signals generated with secondary virgin cable sample identical to that subjected to the load cycles. In this case, the secondary piezoelectric films in PVDF are stable up to 70 °C whilst, at 100 °C, the intensity decreases to halve sample is subjected to the same load cycles as the primary one. This procedure can be used if the the value at 27 °C. Piezoelectric films in LiNb03 give a stable signal until 100 °C [15]. In order to carry humidity of the laboratory is carefully controlled and the effect of crosslinking by-products on space out measurements at high temperatures, a short-circuited HVDC sample cable is passed inside the

primary of a transformer and becomes its secondary. The primary induces a current in the conductor, which then heats it up [32]. An example of the aforementioned setup is shown in Figure 15. In this example, a current of 32 A passes through a mini-cable with an external radius of 3.05 mm to reach

Energies 2019, 12, x FOR PEER REVIEW 12 of 23

Energies 2019, 12, 3512 12 of 23 the cable. The amplitude of the pulsed voltage is proportional to the ratio between the impedance of the capacitance at the measurement point and the characteristic impedance of the cable. This makes this method also applicable to thick insulations because the amplitude of the pressure waves generated at the farthest electrodeFigure is 13. proportional Setup 3: pulsed to the voltage amplitude applied of to the the voltagePEA cell’s pulses. electrode.

Figure 14. Schematic of Setup 3. Figure 14. Schematic of Setup 3. The properties of the E/O and O/E converters, particularly in terms of sampling rate, has to Electrical stress application. Different levels of electric field are related to different space charge be taken into account in order to avoid cutting high frequency contributions. For this purpose, an formation and accumulation phenomena, therefore, several electrical stresses should be tested in the improved version of Setup 3 is proposed in [16] and it consists of the direct connection between the range of the typical working conditions (10–30 kV/mm). amplifier and the oscilloscope. The latter is therefore insulated from the ground and connected to a PC Thermal conditions. The presence of a thermal gradient leads to the establishment of space by means of optical fiber. charge within the dielectric due to the dependence on the temperature of the material conductivity. Thanks to its applicability to long and thick specimens, Setup 3 is mentioned by the standard [5] Nevertheless, electric and acoustic impedances of the sample and of the measurement setup can vary which recommends carrying out PEA measurements before and after PQ (Pre-Qualification) and TT with the temperature. In light of the above, the environmental and sample temperature should be (Type Test) on full size cables. As specified by this protocol, the PEA cell has to be adjacent to the monitored in order to carry out replicable measurements. In addition, the temperature at the outer semi conductive layer, therefore the removal of the external shield is necessary. Since a different piezoelectric film should be compared with its tolerability. The output signals generated with degassing could occur to the cable section with or without the external coating during the load cycle, piezoelectric films in PVDF are stable up to 70 °C whilst, at 100 °C, the intensity decreases to halve two different procedures could be carried out. the value at 27 °C. Piezoelectric films in LiNb03 give a stable signal until 100 °C [15]. In order to carry The first of them consists of PEA measurements performed before and after PQ test or TT on a out measurements at high temperatures, a short-circuited HVDC sample cable is passed inside the secondary virgin cable sample identical to that subjected to the load cycles. In this case, the secondary primary of a transformer and becomes its secondary. The primary induces a current in the conductor, sample is subjected to the same load cycles as the primary one. This procedure can be used if the which then heats it up [32]. An example of the aforementioned setup is shown in Figure 15. In this humidity of the laboratory is carefully controlled and the effect of crosslinking by-products on space example, a current of 32 A passes through a mini-cable with an external radius of 3.05 mm to reach charge phenomena are not predominant. Indeed, the degassing effect of load cycles leads to a reduction of crosslinking by-products and consequently the tendency to accumulate space charge. An alternative to the above procedure is the implementation of the PEA measurements on the primary cable sample near one of the terminals in order to enable the removal of the region altered by the exposure of the outer semicon. As it can be deduced from the above described measurement setups, several parameters have a relevant influence on the results of a space charge measurement in a cable carried out with the PEA method. For the sake of simplicity, the following is a brief list of the most relevant aspects to consider when preparing a PEA measurement setup on cables or mini-cables in order to perform replicable and reliable measures. Electrical stress application. Different levels of electric field are related to different space charge formation and accumulation phenomena, therefore, several electrical stresses should be tested in the range of the typical working conditions (10–30 kV/mm). Thermal conditions. The presence of a thermal gradient leads to the establishment of space charge within the dielectric due to the dependence on the temperature of the material conductivity. Nevertheless, electric and acoustic impedances of the sample and of the measurement setup can vary with the temperature. In light of the above, the environmental and sample temperature should be monitored in order to carry out replicable measurements. In addition, the temperature at the piezoelectric film should be compared with its tolerability. The output signals generated with Energies 2019, 12, 3512 13 of 23

piezoelectric ﬁlms in PVDF are stable up to 70 ◦C whilst, at 100 ◦C, the intensity decreases to halve the value at 27 ◦C. Piezoelectric ﬁlms in LiNb03 give a stable signal until 100 ◦C[15]. In order to carry out measurements at high temperatures, a short-circuited HVDC sample cable is passed inside the primary of a transformer and becomes its secondary. The primary induces a current in the conductor, Energies 2019, 12, x FOR PEER REVIEW 13 of 23 which then heats it up [32]. An example of the aforementioned setup is shown in Figure 15. In this example, a current of 32 A passes through a mini-cable with an external radius of 3.05 mm to reach an outer temperature of 60 °C and an inner temperature of 70 °C [33]. The induced current has to be an outer temperature of 60 C and an inner temperature of 70 C[33]. The induced current has to be calibrated in order to generate◦ the necessary thermal power ◦density inside the cable conductor to calibrated in order to generate the necessary thermal power density inside the cable conductor to reach reach the desired thermal gradient over the insulation. the desired thermal gradient over the insulation.

Figure 15.15. ExampleExample ofof PEA PEA setup setup for for measurements measurements at highat high temperatures—figure temperatures—figure reproduced reproduced from from [33]. [33]. Spatial resolution. As specified in Section2, the spatial resolution of a PEA measurement system dependsSpatial on resolution. the relation As between specified the in widthSection of 2, the the voltage spatial resolution pulse at the of measurementa PEA measurement point and system the thicknessdepends on of thethe componentsrelation between crossed the by width the acoustic of the voltage wave generated pulse at the in the measurement sample. Compared point and to the originalthickness signal, of the thecomponents width of crossed the voltage by the pulse acoustic at the wave measurement generated pointin the couldsample. be Compared enlarged and to the/or deformedoriginal signal, by the the aforementioned width of the voltage reflection pulse phenomena. at the measurement Nonetheless, point other could parameters be enlarged can haveand/or a relevantdeformed role by inthe the aforementioned definition of the reflection spatial resolution phenomen sucha. Nonetheless, as the sampling other rate parameters of the oscilloscope can have or a ofrelevant the E/O role converter, in the definition the bandwidth of the spatial of the amplifier,resolution and such the asdeconvolution the sampling rate process. of the oscilloscope or 3 of theSensitivity. E/O converter, The minimumthe bandwidth amount of the of spaceamplifier, charge and expressed the deconvolution in C/m which process. is detectable with a PEASensitivity. system essentially The minimum depends amount on the of thicknessspace charge of the expressed sample andin C/m the3 noisewhich level. is detectable The latter with can a bePEA reduced system averaging essentially over depends time theon the system thickness output. of Thethe sample typically and reached the noise sensitivity level. The has latter an order can be of 3 magnitudereduced averaging of 1–10 Cover/m . time the system output. The typically reached sensitivity has an order of magnitudeCable clamping.of 1–10 C/m The3. cable has to adhere perfectly to the PEA cell electrode in order to avoid the presenceCable clamping. of air at The the cable interface, has to therefore, adhere perfectly a clamping to the force PEA shouldcell electrode be applied in order to to the avoid sample. the Thepresence establishment of air at the of mechanical interface, therefore, stresses, especiallya clamping into force the cable,should induces be applied a variation to the ofsample. the sound The velocityestablishment within. of In mechanical light of the above,stresses, the especially clamping forceinto the has tocable, be carefully induces monitored a variation in orderof the to sound carry outvelocity replicable within. measurements. In light of the above, the clamping force has to be carefully monitored in order to carryPulse out replicable application. measurements. As specified above, the choice of the measurement setup for a certain sample shouldPulse be doneapplication. in order As to specified apply an above, appropriate the choice pulse ofat the the measurement measurement setup point. for For a certain this purpose, sample absorptions,should be done reflections, in order and to apply delays an of appropriate the electric pulsepulse shouldat the measurement be taken into account.point. For The this impedance purpose, mismatchabsorptions, at reflections, the interface and between delays theof the pulse electric generator pulse should output be coaxial taken cableinto account. and the The sample impedance lead to mismatch at the interface between the pulse generator output coaxial cable and the sample lead to the partial reflection of the pulsed voltage. An attenuator is typically connected to the coaxial cable output terminal as a matched load, in parallel with the cable that is effectively equivalent to a capacitor. This connection makes it practically difficult to match the impedance of the coaxial cable at high frequencies. In addition, the switching frequency of the pulse generator is typically between 100 and 200 Hz, therefore the switch remains open for 5–10 ms after sending out each pulse. During this time, the pulse generator is an open circuit for the reflected pulses that oscillate forwards and

Energies 2019, 12, x FOR PEER REVIEW 14 of 23

backwards along the coaxial cable with a decreasing amplitude, as shown in Figure 16. On the other hand, the distortion of the effective pulse output from a perfect square pulse has to be taken into account. An example of effective shape of the pulse generator output signal is illustrated in Figure 17.

Energies 2019, 12, 3512 14 of 23 the partial reflection of the pulsed voltage. An attenuator is typically connected to the coaxial cable output terminal as a matched load, in parallel with the cable that is effectively equivalent to a capacitor. This connection makes it practically difficult to match the impedance of the coaxial cable at high frequencies.Energies 2019, 12 In, x addition,FOR PEER REVIEW the switching frequency of the pulse generator is typically between 10014 of and 23 200 Hz, therefore the switch remains open for 5–10 ms after sending out each pulse. During this time, thebackwards pulse generator along the is coaxial an open cable circuit with for a thedecreasing reflected amplitude, pulses that as oscillate shown in forwards Figure 16. and On backwards the other alonghand, thethe coaxialdistortion cable of withthe effective a decreasing pulse amplitude, output from as showna perfect in Figuresquare 16pulse. On has the to other be taken hand, into the distortionaccount. An of theexample effective of effective pulse output shape from of the a perfectpulse generator square pulse output has signal to be taken is illustrated into account. in Figure An example17. of effective shape of the pulse generator output signal is illustrated in Figure 17.

Figure 16. Effect of the pulse generator output delay line—figure reproduced from [14].

The extent of this phenomenon depends on the length of the cable, impedances mismatch, and circuital schematic. For these reasons, some researchers have carried out sensibility analysis varying the pulse generator’s output cable length in order to minimize the disturbances related to these phenomena that reach the measurement region. Since the pulse propagation speed is close to light speed while the pressure waves are propagated across the measurement region and the PEA cell with the sound velocity within the crossed materials, the length of the pulse generator’s output cable should be optimized in function of the sample geometry. It has been found that a pulse generator’s output cable a few hundred meters long should be suitable for PEA measurements on full size cables [14].

Figure 16. Eﬀect of the pulse generator output delay line—ﬁgure reproduced from [14]. Figure 16. Effect of the pulse generator output delay line—figure reproduced from [14].

Figure 17. Example of shape of a real pulse generator output—ﬁgure is reproduced from [34]. Figure 17. Example of shape of a real pulse generator output—figure is reproduced from [34]. The extent of this phenomenon depends on the length of the cable, impedances mismatch, and circuital schematic. For these reasons, some researchers have carried out sensibility analysis varying the pulse generator’s output cable length in order to minimize the disturbances related to these phenomena that reach the measurement region. Since the pulse propagation speed is close to light speed while the pressure waves are propagated across the measurement region and the PEA cell with the sound velocity within the crossed materials, the length of the pulse generator’s output cable should

Figure 17. Example of shape of a real pulse generator output—figure is reproduced from [34].

Energies 2019, 12, 3512 15 of 23

Energiesbe optimized 2019, 12, x in FOR function PEER REVIEW of the sample geometry. It has been found that a pulse generator’s15 output of 23 cable a few hundred meters long should be suitable for PEA measurements on full size cables [14]. InIn addition, addition, it it is is advisable advisable that that the the cable cable subjecte subjectedd to measurement is closed as a ring in order to avoid complete reflection reflection of the pulse at the air-terminalair-terminal interface. Nonetheless, it it is inevitable that thethe cable cable coupled coupled to to the the capacitor constituted constituted by by it itss coupling coupling with with the the PEA PEA cell cell acts acts as as a a resonator disturbing the pulsed voltage across the insulation [[35].35]. Bottom PEA cell electrode. Mainly Mainly two two different different types types of of bottom electrodes electrodes have have been been used used by by researchers:researchers: flat flat and and round round electrodes. electrodes. The The use use of of a a flat flat bottom electrode involves the establishment of a contact surface with the cable depending on thethe clamping force. An An illustration illustration of of the stress distribution in in a section of a clamped cable is shownshown in FigureFigure 1818.. As th thee temperature rises, the polymeric materialmaterial of of the the insulation insulation subjected subjected to a mechanical to a mechanical stress could stress undergo could plastic undergo deformation plastic deformationas well as the as softening well as the of thesoftening material of atthe high material temperature at high couldtemperature lead to could an intolerable lead to an amplification intolerable amplificationof the contact areaof the between contact the area cable between and the the electrode. cable and Nevertheless, the electrode. an Nevertheless, increasing deformation an increasing with deformationthe temperature with leads the temperature to a progressive leads reduction to a progre andssive displacement reduction and of the displacement measured radiusof the measured as well as radiusto a loss as of well time-spatial as to a loss correspondence. of time-spatial Indeed,correspondence. permanent Indeed, deformation permanent prevents deformation the replicability prevents of thethe replicability measurements. of the For measurements. these reasons, For particularly these reason carefuls, particularly attention has careful to be attention paid to the has behavior to be paid of tothe the cable behavior under of mechanical the cable under stress dependingmechanical on stress the temperaturedepending on to the be reached. temperature For example, to be reached. a thermal For example,cycle before a thermal the beginning cycle before of a test the couldbeginning stabilize of a thetest material could stabilize [36]. the material [36].

Figure 18. Example of distribution of Von Mises equivalent stress under the application of a clamping Figure 18. Example of distribution of Von Mises equivalent stress under the application of a clamping force—figure reproduced from [37]. force—figure reproduced from [37]. A larger contact surface with the cable with a weaker clamping force can be obtained thanks to the useA larger of a round contact bottom surface electrode. with the This cable solution with a is we thereforeaker clamping suitable force for measurements can be obtained carried thanks out to at thehigh use temperatures. of a round bottom Nonetheless, electrode. if the This contact solution surface is therefore is relatively suitable large, for pressure measurements waves propagated carried out in ata non-radialhigh temperatures. direction could Nonetheless, reach the if piezoelectric the contac filmt surface increasing is relatively the noise large, level andpressure reducing waves the propagatedspatial resolution in a non-radial [37]. In addition, direction as could mentioned reach above,the piezoelectric a round bottom film increasing electrode can the onlynoise be level used and for reducinga certain cablethe spatial diameter resolution and particularly [37]. In addition, careful as attention mentioned has above, to be paid a round to the bottom imperfect electrode cohesion can onlybetween be used the contact for a certain surfaces. cable diameter and particularly careful attention has to be paid to the imperfect cohesion between the contact surfaces. 5. Boundary Conditions for PEA Measurements 5. Boundary Conditions for PEA Measurements After appropriate calibration and deconvolution of the output signals of PEA measurement setup, obtainedAfter resultsappropriate are typically calibration expressed and deconvolution as charge density of the vs. output radius signal and electrics of PEA field measurement vs. radius. setup,The application obtained ofresults the PEA are methodtypicallyshows expressed several as phenomenacharge density occurring vs. radius by varyingand electric the following field vs. radius.boundary The conditions: application of the PEA method shows several phenomena occurring by varying the following boundary conditions: Electric stress value; • Electric stress value; Temperature; • Temperature; • Thermal radial gradient; • Fast and long-term measurement; • HVDC constant application or voltage transient (over voltage, polarity inversion, etc.).

Energies 2019, 12, 3512 16 of 23

Energies 2019, 12, x FOR PEER REVIEW 16 of 23 Thermal radial gradient; • Fast and long-term measurement; • Electric stress value. As mentioned above, a low applied electric field to the sample leads to the HVDC constant application or voltage transient (over voltage, polarity inversion, etc.). •accumulation of charges at electrodes but not into the bulk of the dielectric. The PEA output results obtainedElectric at low stress electric value. field As mentioned are used for above, calibratio a lown appliedand as reference electric field for the to thedeconvolution sample leads process. to the accumulationAt high electric of field, charges as describ at electrodesed above, but several not into phenomena the bulk of of the inje dielectric.ction, forming, The PEA and output accumulation results obtainedof space atcharge low electriccould occur. field are A phenomenon used for calibration of formation and as of reference hetero charges for the deconvolutionnear the electrodes process. has Atbeen high recently electric observed field, as described at low electric above, field several suddenly phenomena after ofthe injection, application forming, of a DC and source accumulation [38,39].of In spaceFigure charge 18 the could behavior occur. of Aa “fast phenomenon charge pulse” of formation is graphically of hetero illustrated. charges nearThe charge the electrodes carriers’ has mobility been recentlyobserved observed in this atphenomenon low electric field is 4–5 suddenly orders afterof ma thegnitude application higher of athan DC sourcethe one [38 occurring,39]. In Figure during 18 theconventional behavior of conduction a “fast charge processes pulse” is[40]. graphically The occurri illustrated.ng of fast The charge charge carriers’pulses at mobility design observedvalues of inapplied this phenomenon electric field is can 4–5 lead orders to the of magnitudeaging of the higher insulation than material the one occurring also if no duringovervoltage conventional transient conductionhappens during processes the [cable40]. The lifetime. occurring In order of fast to charge observe pulses this at phenomenon design values with of applied a sufficient electric spatial field canresolution, lead to thea small aging width of the of insulation voltage pulses material (<10 also ns) ifis no required overvoltage to carry transient out the happens PEA measurements during the cable[41]. lifetime. In order to observe this phenomenon with a sufficient spatial resolution, a small width of voltageTemperature. pulses (< 10It ns)has is been required demonstrated to carry out that the PEAthe temperature measurements has [41 an]. important role in the injection,Temperature. transport, It has and been recombination demonstrated of that charges the temperature in the insulation has animportant of a HVDC role cable. in the injection,A higher transport,temperature and particularly recombination enhances of charges the injection in the insulation process from of a the HVDC inner cable. electrode, A higher and temperatureit leads to an particularlyincreasing mobility enhances of the charge injection carriers process [42]. fromThe temperature the inner electrode, at the electrodes and it leads of a to cable an increasing has to be mobilitycarefully of taken charge into carriers account [42 for]. The the temperaturereplicability of at thePEA electrodes measurements. of a cable has to be carefully taken into accountThermal for radial the replicability gradient. As of described PEA measurements. in Section 1, the establishment of thermal gradient across the radiusThermal of radialthe dielectric gradient. of HVDC As described cable leads in Section to an1 inversion, the establishment of the electric of thermal field as gradientshown in across Figure the3. Since radius this of thebehavior dielectric can of be HVDC calculated cable and leads accu to anrately inversion predicted, of the the electric PEA field measurement as shown inresults Figure can3. Sincebe divided this behavior into a contribution can be calculated due to and thermal accurately gradient predicted, and another the PEA due measurementto other phenomena. results can be divided“Instantaneous” into a contribution and duelong-term to thermal measurement. gradient and The another output due pattern to other of an phenomena. “instantaneous” PEA measurement“Instantaneous” is the result and long-term of time averaging, measurement. deconvolution, The output and pattern calibration of an of “instantaneous” the amplifier output PEA measurementsignal. In Figure is the 19 resultan example of time of averaging,instantaneous deconvolution, space charge and measurements calibration ofis illustrated. the amplifier As outputmentioned signal. in Section In Figure 1, most19 an of example the space of charge instantaneous injection, space formation, charge and measurements accumulation is phenomena illustrated. Ashave mentioned characteristic in Section time1 ,with most an of order the space of magnit chargeude injection, between formation, a few minutes and accumulation and few hours. phenomena In order haveto carry characteristic out the assessment time with of an slow order space of magnitude charge phenomena, between a fewa long-term minutes analysis and few is hours. indispensable. In order toThe carry Figure out the20 shows assessment an example of slow of space long-term charge meas phenomena,urement graphic a long-term representation analysis is obtained indispensable. by the collection of several “instantaneous” measurements. In this way, placing the time on the abscissa and The Figure 20 shows an example of long-term measurement graphic representation obtained by the the radius on the ordinate, thanks to the use of a chromatic scale for the charge density, long-term collection of several “instantaneous” measurements. In this way, placing the time on the abscissa and transient phenomena are illustrated. the radius on the ordinate, thanks to the use of a chromatic scale for the charge density, long-term transient phenomena are illustrated.

Figure 19. Example of fast charge pulses observed detecting the PEA measurements output each Figure 19. Example of fast charge pulses observed detecting the PEA measurements output each 50 50 ms—ﬁgure reproduced from [38]. ms—figure reproduced from [38].

Energies 2019, 12, x FOR PEER REVIEW 17 of 23

Energies 2019, 12, 3512 17 of 23 Energies 2019, 12, x FOR PEER REVIEW 17 of 23

Figure 20. Example of migration of negative charges at −30 kV (a) and −55 kW (b) during charging (left side) and discharging (right side)—figure reproduced from [33]. Figure 20. Example of migration of negative charges at 30 kV (a) and 55 kW (b) during charging Figure 20. Example of migration of negative charges at− −30 kV (a) and −−55 kW (b) during charging (leftHVDC(left side)side) constant andand dischargingdischarging application (right(right side)—ﬁgure side)—figureor voltage reproduced reproduced transient. from from Long [ 33[33].]. term measurements show the establishment of several phenomena such as the migration of charges from an electrode to another HVDC constant application or voltage transient. Long term measurements show the establishment or theHVDC accumulation constant of applicationcharges near orphysical voltage or chemicaltransient. defects. Long Onterm the measurements other hand, space show charge the of several phenomena such as the migration of charges from an electrode to another or the accumulation accumulationestablishment is of particularly several phenomena dangerous such during as the fast mi voltagegration transient of charges such from as the an polarityelectrode inversion to another or of charges near physical or chemical defects. On the other hand, space charge accumulation is overvoltageor the accumulation transient of due charges to a fault.near physicalIndeed, orthe chemical experimental defects. evaluation On the other of the hand, evolution space chargeof the particularly dangerous during fast voltage transient such as the polarity inversion or overvoltage electricaccumulation field distribution is particularly into dangerous the dielectric during of a fast HVDC voltage cable transient during sucha fast as voltage the polarity transient inversion appears or transient due to a fault. Indeed, the experimental evaluation of the evolution of the electric ﬁeld particularlyovervoltage interesting.transient due In toFigure a fault. 21, Indeed,a graphical the illustrationexperimental shows evaluation the results of the of evolutiona simulation of the of distribution into the dielectric of a HVDC cable during a fast voltage transient appears particularly polarityelectric fieldinversion distribution occurring into thein adielectric cable in of the a HVDC presence cable of during a thermal a fast gradient. voltage transient The simulation appears interesting. In Figure 21, a graphical illustration shows the results of a simulation of polarity inversion demonstratesparticularly interesting. that the most In Figuresevere condition21, a graphical is reached illustration immediately shows after the resultsthe polarity of a simulationreversal near of occurring in a cable in the presence of a thermal gradient. The simulation demonstrates that the most thepolarity inner inversionelectrode [43,44].occurring in a cable in the presence of a thermal gradient. The simulation severe condition is reached immediately after the polarity reversal near the inner electrode [43,44]. demonstrates that the most severe condition is reached immediately after the polarity reversal near the inner electrode [43,44].

Figure 21. Simulation results of polarity inversion +/ 15 kV at temperature gradient of 10 C—ﬁgure − ◦ reproducedFigure 21. Simulation from [44]. results of polarity inversion +/−15 kV at temperature gradient of 10 °C—figure reproduced from [44]. 6. LimitsFigure of 21. the Simulation PEA Method results of polarity inversion +/−15 kV at temperature gradient of 10 °C—figure 6. Limits of the PEA Method Despitereproduced the from growing [44]. demand of extruded HVDC cables, and the recent publication of an internationalDespite the standard growing [5], demand there is stillof extruded no universally HVDC accepted cables, protocoland the forrecent the publication execution of of PEA an 6. Limits of the PEA Method measurementsinternational standard [13]. As mentioned[5], there is above still no and univ experimentedersally accepted by many protocol researchers, for the several execution parameters of PEA havemeasurementsDespite an important the [13]. growing role As mentioned in the demand quality above of of extruded PEAand experime measurements. HVDCnted cables, by In many particular, and researchers, the recent the cable several publication length, parameters the of cell an arrangement,haveinternational an important thestandard equivalent role [5],in the capacitances,there quality is still of noPEA and univ impedancesmeasersallyurements. accepted of the In measurement particular, protocol for the setup the cable execution (see length, Figures theof9 –PEA cell14) shouldarrangement,measurements be optimized the [13]. equivalent As for mentioned a certain capacitances, cable above [45]. and and For experime example,impedancesnted a special of by the many apparatus measurement researchers, with setup low several pass (see ﬁltersparameters Figures at the 9– oscilloscope14)have should an important be input, optimized high role voltage infor the a certain quality pulse cable generator of PEA [45]. meas andFor example,urements. an appropriate a Inspecial particular, thickness appara the oftus the cablewith PEA lowlength, cell pass electrode the filters cell shouldatarrangement, the oscilloscope be implemented the equivalent input, for high measurements capacitances, voltage pulse onand genera thick impe cabletordances and insulations ofan the appropriate measurement [46]. Indeed, thickness setup the compatibility of (see the Figures PEA cell of9– aelectrode14) commercial should should be PEAoptimized be apparatus implemented for a certain should for cable be measurements carefully [45]. Fo evaluatedr example, on thick ona special the cable basis appara insulations of thetus characteristics with [46]. low Indeed, pass offilters the at the oscilloscope input, high voltage pulse generator and an appropriate thickness of the PEA cell electrode should be implemented for measurements on thick cable insulations [46]. Indeed, the

Energies 2019, 12, x FOR PEER REVIEW 18 of 23

compatibility of a commercial PEA apparatus should be carefully evaluated on the basis of the Energiescharacteristics2019, 12, 3512 of the specific cable on which space charge measurements should be done. In18 addition, of 23 a physical model of the measurement region of the cable and of the PEA cell should be implemented in order to correctly evaluate the superposition of reflected signals on the direct ones that reach the specific cable on which space charge measurements should be done. In addition, a physical model of piezoelectric film [47]. the measurement region of the cable and of the PEA cell should be implemented in order to correctly To date, the PEA method is not considered a non-destructive measurement on laying cables evaluate the superposition of reflected signals on the direct ones that reach the piezoelectric film [47]. since a partial stripping of the outer layers is necessary. To date, the PEA method is not considered a non-destructive measurement on laying cables since a partial7. Comparison stripping ofbetween the outer the layers PEA isand necessary. Thermal Step Method (TSM) 7. ComparisonIn addition between to the the PEA, PEA the and Thermal Thermal Step Step Method Method (TSM) (TSM) is a recognized method for space charge measurements on full sized cables also mentioned by the standard IEEE Std 1732TM-2017. In Figure In addition to the PEA, the Thermal Step Method (TSM) is a recognized method for space charge 22 a schematic of the TSM method as described by the aforementioned standard is shown. This measurements on full sized cables also mentioned by the standard IEEE Std 1732TM-2017. In Figure 22 method was invented by Tourielle in 1987 [48] and it relies on the application of a thermal step across a schematic of the TSM method as described by the aforementioned standard is shown. This method the insulation in order to generate a temporary displacement of the charges within. This influences was invented by Tourielle in 1987 [48] and it relies on the application of a thermal step across the the amount of charges at the interfaces generating a current with a typical order of magnitude of pA insulation in order to generate a temporary displacement of the charges within. This influences if the dielectric is short-circuited. The generated current signal is then amplified, deconvoluted, and the amount of charges at the interfaces generating a current with a typical order of magnitude of calibrated in order to obtain the relation between the charge density and the radial coordinate. The pA if the dielectric is short-circuited. The generated current signal is then amplified, deconvoluted, thermal step can be generated heating the conductor or cooling the outer surface. In the first case, the and calibrated in order to obtain the relation between the charge density and the radial coordinate. short-circuited cable works as the secondary of a transformer and it is crossed by a current of several The thermal step can be generated heating the conductor or cooling the outer surface. In the first kA. Alternatively, the cooling of the outer surface is carried out by means of a cylindrical thermal case, the short-circuited cable works as the secondary of a transformer and it is crossed by a current diffuser crossed by a cold fluid and adjacent to the cable surface. While the inner heating method of several kA. Alternatively, the cooling of the outer surface is carried out by means of a cylindrical provides information about the whole cable, the outer cooling is used to get space charge thermal diffuser crossed by a cold fluid and adjacent to the cable surface. While the inner heating measurements in a small portion of the sample (20–40 cm) [6]. The outer cooling method is the one method provides information about the whole cable, the outer cooling is used to get space charge recommended by the above cited standard. measurements in a small portion of the sample (20–40 cm) [6]. The outer cooling method is the one recommended by the above cited standard.

Figure 22. Thermal Step Method (TSM) for cables, (a) conditioning phase, (b) measurement phase—figureFigure 22. reproducedThermal Step from Method [6]. (TSM) for cables, (a) conditioning phase, (b) measurement phase— figure reproduced from [6]. The current amplifier cannot be in contact with high voltage as well as the conduction and polarizationThe currentcurrent could amplifier mask cannot the TS signal.be in contact For these with reasons, high thevoltage protocol as well described as the in conduction [6] proposes and a “doublepolarization capacitor” current configuration could mask whereby the TS a compensationsignal. For these cable reas identicalons, tothe that protocol under measurementdescribed in [6] is insertedproposes into a the“double setup. capacitor” During the configuration first step of the whereb proposedy a compensation TSM measurement cable procedure,identical to both that theunder cable conductors are connected to the HV source whilst the outer semiconductor layer is grounded. After an appropriate conditioning time, the HVDC generator is disconnected, the thermal step is Energies 2019, 12, 3512 19 of 23 applied to the cable under measurement, and the current amplifier is connected to the compensation cable. In other words, the two cables are in parallel during the conditioning phase and in series during the measurement phase. Unlike the PEA, the TSM provides information about space charge averaged among a cable segment and it is suitable for thick cables. The TSM measurement setup is typically bulkier than the PEA one due to the necessity to use a compensation cable. Moreover, with respect to the latter, the former requires a significantly greater measuring time due to the necessity to also perform a conditioning phase. Like the PEA, the TSM is also a partially destructive method [2].

8. Possible Direction of Research on the Use of the PEA Method for HVDC Cables Recently, the demand of HVDC extruded cables is rapidly increasing especially thanks to the growing preference for the use of underground links and the increasing amount of energy produced by off-shore wind farms. In this context, the development of an online, embedded, and automatic PEA device would enable an optimized use of the cable system by the point of view of both electrical and thermal stresses [49]. A first step in this direction has been taken with the implementation of an automatic PEA setup developed by Jongmin Kan et al. at the Honam University, Gwangju Korea [50]. The PEA method can be applied to oil-paper insulation, although further aspects such as the environmental conditions and the acoustic wave recovery need to be addressed to acquire more accurate results [51,52]. An improvement of the existing commercial PEA cells is desirable in order to allow their adoptability in different conditions. A more versatile PEA apparatus would be more easily used in the PQ tests and in the TT of different cables. In order to enable the use of the PEA method for online applications, the measurement device should be integrated within the cable structure. A new embedded PEA measurement setup should be developed for this purpose. Finally, the simultaneous measurements of space charge and partial discharge on samples could improve the knowledge of the phenomena responsible for ageing in HVDC cables [53]. Some first experiments were carried out on flat specimens by means of a PEA cell and a wireless sensor system for partial discharges detection. The development of this technique and its application on cables could provide more information about the correlation between these phenomena.

9. Conclusions The extruded HVDC cables’ demand has grown in recent years due to their increasing use for new underground and undersea lines as well as the replacement of existing Mass Impregnated HVDC cables. The initial installation of extruded HVDC cables manufactured with the same insulating materials used for HVAC application has highlighted the occurrence of premature ageing of cables due to the injection, formation, and accumulation of space charge within the insulation layer. In order to investigate these phenomena, several non-destructive techniques for space charge measurement within insulating materials have been developed by researchers. Hitherto, the most widespread used technique is the Pulsed Electro Acoustic (PEA) method. This method is based on the detection, by means of a piezoelectric membrane, of the pressure waves generated from the interaction of a pulsed voltage with charges inside the lattice of the dielectric. During the last decades, the PEA method has been implemented for space charge measurements on cables and mini-cables with satisfactory results in terms of reliability and replicability. To date, the PEA technique is used for space charge phenomena characterized by a time scale ranging from fraction of a second to several hours. Different research groups in the world have developed different measurement setups in order to correct apply the voltage pulse to the insulation in the measurement region. The optimal choice of setup and of electrical parameters depends on the cable sizes. In 2017, a published standard [5] included the use of the PEA and TSM methods in the context of PQ tests and TT provided for HVDC extruded cables. Nevertheless, further improvements are needed Energies 2019, 12, 3512 20 of 23 to obtain a PEA measurement procedure universally accepted and adaptable to different types of cable. Indeed, commercial PEA cells currently available on the market are optimized for a limited combination of measurement parameters as above specified. The development of a more versatile PEA device is therefore desirable in order to spread the use of this measurement method in the industrial field. In addition, more studies should be carried out in order to evaluate the behavior of the cable’s insulation during fast transients such as polarity inversion and overvoltage. In this way, the use of the HVDC grid could be optimized taking into account the damage occurred to the cable due to the frequency and the extent of such transient phenomena [9]. In conclusion, the development of an “online” PEA system could lead to an optimization of the use of the cables by the point of view of electric and thermal stresses.

Author Contributions: Conceptualization, G.R. and P.R.; validation, A.I., P.R. and G.A.; formal analysis, G.R., A.I. and P.R.; investigation, G.R. and A.I.; writing—original draft preparation, G.R.; writing—review and editing, P.R. and G.R.; supervision, G.A. Funding: This research received no external funding. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Mazzanti, G.; Marzinotto, M. Extruded Cables for High-Voltage Direct-Current Transmission—Advances in Research and Development; IEEE Press Series on Power Engineering; John Wiley & Sons: Hoboken, NJ, USA, 2013. 2. Imburgia, A.; Miceli, R.; Sanseverino, E.R.; Romano, P.; Viola, F. Review of space charge measurement systems: Acoustic, thermal and optical methods. IEEE Trans. Dielectr. Electr. Insul. 2016, 23, 3126–3142. [CrossRef] 3. Hussaini, H.; Adam, A.A.; Susan, A.A. Review of Space-charge Measurement using Pulsed Electro-Acoustic Method, Advantages and Limitations. Int. J. Eng. Res. Appl. 2015, 5, 90–95. 4. Maeno, T.; Kushibe, H.; Takada, T.; Cooke, C.M. Pulsed electroacoustic method for the measurement of volume charges in E-beam irradiated PMMA. In Proceedings of the Conference on Electrical Insulation & Dielectric Phenomena, Amherst, NY, USA, 20–24 October 1985. 5. IEEE Recommended Practice for Space Charge Measurements on High Voltage Direct-Current Extruded Cables for Rated Voltages up to 550 kV. IEEE Std 1732™-2017. Available online: https://ieeexplore.ieee.org/ servlet/opac?punumber=7949156 (accessed on 11 September 2019). 6. Mazzanti, G.; Chen, G.; Fothergill, J.C.; Hozumi, N.; Li, J.; Marzinotto, M.; Mauseth, F.; Morshuis, P.; Reed, C.; Tzimas, A.; et al. A Protocol for Space Charge Measurements in Full-size HVDC Extruded Cables. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 21–34. [CrossRef] 7. Tsekmes, I.A.; van der Born, D.; Morshuis, P.H.F. Space Charge Accumulation in Polymeric DC Mini-Cables. In Proceedings of the 2013 IEEE International Conference on Solid Dielectrics, Bologna, Italy, 30 June–4 July 2013. 8. Ghorbani, H.; Abbasi, A.; Jeroense, M.; Gustafsson, A.; Saltzer, M. Electrical Characterization of Extruded DC Cable Insulation—The Challenge of Scaling. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 1465–1475. [CrossRef] 9. Chen, G.; Hao, M.; Xu, Z.; Vaughan, A.; Cao, J.; Wang, H. Review of High Voltage Direct Current Cables. CSEE J. Power Energy Syst. 2015, 1, 9–21. [CrossRef] 10. Li, S.; Lan, L.; Wu, J.; Yin, Y. Investigation of Space Charge Behavior of HVDC XLPE Cables Using PEA Method. In Proceedings of the 7th International Symposium on Electrical Insulating Materials, Niigata City, Japan, 1–5 June 2014. 11. Van der Born, D.; Tsekmes, I.A.; Morshuis, P.H.F.; Smit, J.J.; Person, T.J.; Sutton, S.J. Evaluation of Apparent Trap-Controlled Mobility and Trap Depth in Polymeric HVDC Mini-Cables. In Proceedings of the 2013 IEEE International Conference on Solid Dielectrics, Bologna, Italy, 30 June–4 July 2013. 12. Zhan, Y.; Chen, G.; Hao, M.; Pu, L.; Zhao, X.; Duan, W.; Sun, H.; Ju, Z.; Wang, N. Thermal Transient Eﬀects on Space Charge Dynamics in HVDC Extruded Cable Insulation. In Proceedings of the 12th IEEE International Conference on the Properties and Applications of Dielectric Materials, Xi’an, China, 20–24 May 2018. Energies 2019, 12, 3512 21 of 23

13. Tzimas, A.; Boyer, L.; Mirebeau, P.; Dodd, S.; Castellon, J.; Notingher, P. Qualitative analysis of PEA and TSM techniques on a 200kV extruded cable during a VSC ageing program. In Proceedings of the 2016 IEEE International Conference on Dielectrics, Montpellier, France, 3–7 July 2016. 14. Zheng, H. Application of PEA Technique to Space Charge Measurement in Cylindrical Geometry HV Cable Systems. Ph.D. Thesis, Department of Engineering, University of Leicester, Leicester, UK, 2015. 15. Kitajima, H.; Tanaka, Y.; Takada, T. Measurement of space charge distribution at high temperature using the pulsed electro-acoustic (PEA) method. In Proceedings of the 1996 Seventh International Conference on Dielectric Materials, Measurements and Applications, Bath, UK, 23–26 September 1996. 16. Wang, X.; Hao, J.Q.; Xiong, J.Z.; Chen, C.; Wu, K.; Fu, M.L.; Chen, G. Comparison and Analysis of Three Pulse Injection Methods in the Pulsed Electroacoustic Technique Used for Long Cables. IEEE Electr. Insul. Mag. 2018, 34, 17–31. [CrossRef] 17. IEC/TS 62758, Calibration of Space Charge Measuring Equipment Based on the Pulsed Electroacoustic (PEA) Measurement Principle. 2012. Available online: https://webstore.iec.ch/preview/info_iec62758%7Bed1.0% 7Den.pdf (accessed on 11 September 2019). 18. Ala, G.; Caruso, M.; Cecconi, V.; Ganci, S.; Imburgia, A.; Miceli, R.; Romano, P.; Viola, F. Review of Acoustic Methods for Space Charge Measurement. In Proceedings of the 2015 AEIT International Annual Conference (AEIT), Naples, Italy, 14–16 October 2015. 19. Imburgia, A. Modelling of Pulsed Electro Acoustic Method for Space Charge Detection on Single and Multilayer Dielectrics. Ph.D. Thesis, DEIM, University of Palermo, Palermo, Italy, 2018. 20. Jeroense, M.J.P. Charges and Discharges in HVDC Cables. Ph.D. Thesis, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University, Delft, The Netherlands, 1997. 21. Bodega, R.; Morshuis, P.H.F.; Smit, J.J. Space Charge Signal Interpretation in a Multi-layer Dielectric Tested by Means of the PEA Method. In Proceedings of the International Conference on Solid Dielectrics, Toulouse, France, 5–9 July 2004. 22. Bodega, R.; Morshuis, P.H.F.; Smit, J.J. Space charge measurements on multi-dielectrics by means of the pulsed electroacoustic method. IEEE Trans. Dielectr. Electr. Insul. 2006, 13, 272–281. [CrossRef] 23. Andrade, M.D.; Candela, R.; de Rai, L.; Bononi, S.F.; Imburgia, A.; Sanseverino, E.R.; Romano, P.; Viola, F. Interpretation of PEA Output Signal in a Multilayer Specimen. In Proceedings of the 2018 IEEE Conference on Electrical Insulation and Dielectric Phenomena, Cancun, Mexico, 21–24 October 2018. 24. Fukunaga, K.; Miyata, H.; Sugimori, M.; Takada, T. Measurement of Charge Distribution in the Insulation of Cables using Pulsed Electroacoustic Method. Ieej Trans. Fundam. Mater. 1990, 110, 647–648. [CrossRef] 25. Hozumi, N.; Okamoto, T.; Imajo, T. Space charge distribution measurement in a long size XLPE cable using the Pulsed Electroacoustic Method. In Proceedings of the Conference Record of the 1992 IEEE International Symposium on Electrical Insulation, Baltimore, MD, USA, 7–10 June 1992. 26. Yasuda, M.; Ito, M.; Takada, T. Measurement of charge distribution in coaxial cable using the Pulsed Electroacoustic Method. Jpn. J. Appl. Phys. 1991, 30, 71. [CrossRef] 27. Liu, R.; Takada, T.; Takasu, N. Pulsed electro-acoustic method for measurement of space charhe distribution in power cables under both DC and AC electric field. Jpn. J. Appl. Phys. 1993, 26, 986. 28. Fu, M.; Chen, G.; Davies, A.E.; Tanaka, Y.; Takada, T. A Modified PEA Space Charge Measuring System for Power Cables. In Proceedings of the 6th International Conference on Properties and Applications of Dielectric Materials, Xi’an, China, 21–26 June 2000. 29. Perego, G.; Peruzzotti, F.; Montanari, G.C.; Palmieri, F.; de Robertis, G. Investigating the effect of additives for high-voltage polymeric dc cables through space charge measurements on cables and films. In Proceedings of the 2001 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, Kitchener, ON, Canada, 14–17 October 2001. 30. Hozumi, N.; Suzuki, H.; Okamoto, T.; Watanabe, K.; Watanabe, A. Direct Observation of Time-dependent Space Charge Profiles in XLPE Cable under High Electric Fields. IEEE Trans. Dielectr. Electr. Insul. 1994, 1, 1068–1076. [CrossRef] 31. Hozumi, N.; Takeda, T.; Suzuki, H.; Okamoto, T. Space Charge Behavior in XLPE Cable Insulation under 0.2-1.2 MV/cm dc Fields. IEEE Trans. Dielectr. Electr. Insul. 1998, 5, 82–90. [CrossRef] 32. Raja, A.R.A.; Vissouvanadin, B.; Vu, T.T.N.; Teyssedre, G.; Sinisuka, N.I. Space Charge Measurement on XLPE Cable for HVDC Transmission using PEA Method. In Proceedings of the 4th International Conference on Electrical Engineering and Informatics (ICEEI 2013), Bangi, Malaysia, 24–25 June 2013. Energies 2019, 12, 3512 22 of 23

33. Vu, T.T.N.; Teyssedre, G.; Vissouvanadin, B.; Steven, J.Y.; Laurent, C. Transient space charge phenomena in HVDC model cables. In Proceedings of the9th International Conference on Insulated Power Cables (JiCable), Versailles, France, 21–25 June 2015. 34. Lee, J.-H.; Kim, C.-H.; Park, Y.-H.; Hwangb, S.; Kim, Y.H.; Kim, J.-T.; Morshuis, P.; Hozumi, N. A Development of Space Charge Measurement Device for Model Cable. In Proceedings of the 2016 International Conference on Condition Monitoring and Diagnosis, Xi’an, China, 25–28 September 2016. 35. Hozumi, N.; Hori, M. Measuring Space-Charge in HVDC Cables. Cigré Science Engineering. 2016. Available online: https://www.ee.co.za/wp-content/uploads/2017/02/Energize-TD-page35-38-Feb17.pdf (accessed on 11 September 2019). 36. Zheng, H.; Dodd, S.J.; Dissado, L.A.; Fothergill, J.C.; Allais, A.; Kebbabi, L. Feasibility of using a flat bottom electrode for PEA space charge measurement on mini-cables under high temperatures. In Proceedings of the 2013 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, Shenzhen, China, 20–23 October 2013. 37. Zheng, H.; Dodd, S.J.; Dissado, L.A.; Fothergill, J.C.; Allais, A.; Kebbabi, L. Influence of cable clamping force on PEA measurement on HVDV mini-cables. In Proceedings of the 2013 IEEE International Conference on Solid Dielectrics, Bologna, Italy, 30 June–4 July 2013. 38. Montanari, G.C.; Fabiani, D.; Dissado, L.A. Fast Charge Pulses: The Evidence and its Interpretation. In Proceedings of the 2013 IEEE International Conference on Solid Dielectrics, Bologna, Italy, 30 June–4 July 2013. 39. Fabiani, D.; Montanari, G.C.; Dissado, L.A. Measuring a Possible HVDC Insulation Killer: Fast Charge Pulses. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 45–51. [CrossRef] 40. Montanari, G.C.; Dissado, L.A.; Serra, S. The Hidden Threat to HVDC Polymeric Insulation at Design Field: Solitonic Conduction. IEEE Electr. Insul. Mag. 2014, 30, 39–50. [CrossRef] 41. Zhu, Z.; Yang, L.; Guo, S.; Li, D.; Zhang, Y. Comparison Between PEA and PWP Method of Space Charge Measurement in Model Cable. In Proceedings of the 2015 IEEE 11th International Conference on the Properties and Applications of Dielectric Materials, Sydney, Australia, 19–22 July 2015. 42. Zimmerling, I.M.; Tsekmes, I.A.; Morshuis, P.H.F.; Smit, J.J.; Person, T.J.; Geussens, T. Space charge analysis of modified and unmodified XLPE model-cables under different electric fields and temperatures. In Proceedings of the 2015 IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), Ann Arbor, MI, USA, 18–21 October 2015. 43. Adi, N.; Vu, T.T.N.; Teyssèdre, G.; Baudoin, F.; Sinisuka, N. DC Model Cable under Polarity Inversion and Thermal Gradient: Build-Up of Design-Related Space Charge. Technologies 2017, 5, 46. [CrossRef] 44. Adi, N.; Teyssedre, G.; Vu, T.T.N.; Sinisuka, N.I. DC Field Distribution in XLPE-Insulated DC Model Cable with Polarity Inversion and Thermal Gradient. In Proceedings of the 2016 IEEE International Conference on High Voltage Engineering and Application (ICHVE), Chengdu, China, 19–22 September 2016. 45. Albertini, M.; Bononi, S.F.; Giannini, S.; Troia, I.; Sica, G.; Mazzanti, G.; Pini, D. Issues in Space Charge Measurments with the PEA Technique in HVDC Cables: Applicative Case Study. In Proceedings of the 2018 IEEE CEIDP Conference on Electrical Insulation and Dielectric Phenomena, Cancun, Mexico, 21–24 October 2018. 46. Fechner, T.; Zhou, Y.; Lei, X.; Zhou, M. Important Design Aspects of Pulsed Electroacoustic Measurement Systems for Thick Full-Size Cable Specimen. In Proceedings of the 2018 International Conference on Power System Technology (POWERCON), Guangzhou, China, 6–8 November 2018. 47. Imburgia, A.; Romano, P.; Sanseverino, E.R.; de Rai, L.; Bononi, S.F.; Troia, I. Pulsed Electro-Acoustic Method for specimens and cables employed in HVDC systems: Some feasibility considerations. In Proceedings of the 2018 AEIT International Annual Conference, Bari, Italy, 3–5 October 2018. 48. Toureille, A.; Reboul, J.P. The thermal-step-technique applied to the study of charge decay in polyethylene thermoelectrets. In Proceedings of the 6th International Symposium on Electrets (ISE 6) Proceedings, Oxford, UK, 1–3 September 1988; pp. 23–27. 49. Morshuis, P. A Short History of High-Voltage Direct Current Cables and Space Charge Measurements at Delft University of Technology. Sens. Mater. 2017, 29, 1099–1108. 50. Kang, J.; Hwangbo, S.; Min, W.; Jung, H.; Park, J. Automatic Measurement System for Space Charge Distribution in HVDC Model Cable. In Proceedings of the 8th International Symposium on Electrical Insulating Materials, Toyohashi City, Japan, 12–15 September 2017. Energies 2019, 12, 3512 23 of 23

51. Tang, C.; Huang, B.; Hao, M.; Xu, Z.; Hao, J.; Chen, G. Progress of Space Charge Research on Oil-Paper Insulation Using Pulsed Electroacoustic Techniques. Energies 2016, 9, 53. [CrossRef] 52. Chen, Q.; Zhang, J.; Chi, M.; Tan, P.; Sun, W. Eﬀect of Temperature on Space Charge Distribution of Oil–Paper Insulation under DC and Polarity Reversal Voltage. Energies 2018, 11, 2271. [CrossRef] 53. Imburgia, A.; Romano, P.; Viola, F.; Madonia, A.; Candela, R.; Troia, I. Space charges and partial discharges simultaneous measurements under DC stress. In Proceedings of the 2016 IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), Toronto, ON, Canada, 16–19 October 2016.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).