Lightning Effect on Copper and Optical Fiber (FTTN) Access Networks of Telecommunication Systems

This paper presents a study of lightning effect on copper and optical fiber (fiber to the node) public switched telephone network (PSTN) in Sri Lanka. The study covers, firstly modeling and simulation of each PSTNs, secondly development and testing of PSTN laboratory prototypes and finally a case study for a selected real network. The individual PSTNs were modeled by PSCAD/EMTDC software and were verified in the laboratory by applying lightning surges to the developed prototypes. The study was extended by considering the effects of typical earth resistances in the PSTN and possible multiple reflections. In the case study, first the RLC parameters of the PSTN were estimated by field and laboratory measurements. Then, the PSTN was modeled and analyzed in PSCAD/EMTDC software. It was found that significant components of lightning surges may transfer to the consumer side unless proper surge protective devices are used. Further, this effect may be higher in the optical fiber network compared to the copper network. A significant effect may also appear at the subscriber side due to multiple reflections and the earth resistances characterized by high soil resistivity.


Introduction
Lightning is an electrostatic discharge occurring between electrically charged regions within clouds or between a cloud and the ground. As a result, lightning surges may propagate through service wires to the customer [1]. Thus the surges cause failures and damages to almost every electrical and electronic system that are exposed to thunderstorms directly or indirectly. The problem may become severe especially for telecommunication utilities that have exposed assets covering large areas [2].
As far as telecommunication system is concerned, lightning surges may propagate in three ways. The first type is called "direct lightning surge" i.e. due to high voltage surges generated by direct lightning discharges striking to the telecommunication cables or devices. Secondly, the voltage induced by electrostatic or electromagnetic coupling from a nearby lightning hit, is called "induced lightning surge". Thirdly when lightning strikes the ground, its potential rises and, as a result, ground potential of the appliances may become high. A lot of information has been collected on lightning effects on telecommunication systems during last few decades [3][4][5].
Lightning usually do not strike on telecommunication cables directly. However, if it does, the result could be melting of the cable and significant damages to the connected devices. Majority of lightning surges appearing on telecommunication cables occur due to indirect coupling [6][7][8].
Lightning surges can occur quite frequently especially in rainy seasons. Consequently, there have been growing concerns over the damages and disturbances caused by lightning at subscriber equipment, distribution points, cabinet and line cards of Multi-Service Access Nodes (MSAN) of telecommunication networks. Therefore a proper protection system should: (a) provide protection against indirect lightning strikes, (b) provide an effective earth termination network for discharge of lightning current and (c) prevent entering of conducting surges to equipment cabins [9][10][11].
2 This study basically addresses three important aspects of lightning on telecommunication systems: the instantaneous voltages created by the travelling wave due to indirect coupling mechanism, the grounding effect of the metallic cable sheath upon the resulting induced surge and the reflection and oscillation phenomena of transient voltage caused by the connecting cables. Finally a case study conducted with the aid of a practically verified simulation model in order to depict so called aspects is also presented.

Telecommunication networks
The telecommunication system belonging to Sri Lanka Telecom comprises with three main sections, namely: Core, Aggregation and Access Networks. This study focuses only on the access networks of the Public Switched Telephone Network (PSTN). There are two types of telecommunication networks namely (i) Copper wired network, and (ii) Optical fiber access network. The optical fiber access network is categorized into two groups as (i) Fiber To The Node (FTTN) and Fiber To The Home (FTTH), Sri Lanka Telecom presently use copper network, FTTN network and FTTH network.

Copper wired network
The copper wired access network comprises with different components such as main distribution frame (MDF), cabinet, distribution point (DP) and the subscriber end. Figure 1(a) shows a typical architecture of copper access network with all the entry points of the lightning surges into the system. Figure 1

Fiber To The Node Network (FTTN)
At present Sri Lanka Telecom"s communication access network is being converted to an optical fiber based network and the network edge will be extended closer to the subscriber. Main advantages of using optical fiber are less attenuation and high bandwidth. Fiber optic cables are normally composed of totally nonconductive materials. Optical fiber can only respond to light signals in the frequency range of THz whereas lightning surges" frequency spectrum covers in MHz. This leads to one of the most important advantages of this form of transmission, a fiber optic cable will not be affected by any electromagnetic interference such as lightning.
In optical fiber networks, MSANs are introduced and it extends the network edge closer to the subscriber. MSAN is a device typically installed in a telephone exchange (although sometimes in a roadside serving area interface cabinet) which connects customers" telephone lines to the aggregation network, to provide voice, data and media all from a single platform. MSAN unit consists of line cards, optical fiber cables, MDF, power unit etc. [12]. Gas discharge tube surge arrestors installed in tag blocks and earthing provide protection against over voltages.

Fiber to the Home (FTTH)
Currently Sri Lanka Telecom is expanding the FTTN access network to FTTH which provides multiple-services such as high definition channels on PEO TV, high speed internet access, DATA services, video on demand, Wi-Fi, VoIP etc. Fiber to the home (FTTH), also called fiber to the premises (FTTP), is the installation and use of optical fiber from a central point directly to individual buildings such as residences and apartment buildings. Since optical fiber is installed from the central office directly to the subscriber premises, it will not be affected by any electromagnetic interference such as lightning. However, this option is very expensive compared to other two networks.

Modelling and model verification
In this study two types of telecommunication networks: namely copper network and FTTN network were considered. As shown in Figure 3 first the individual components were modelled in PSCAD/EMTDC software. Voltages appearing at different points of the components were calculated by simulations and experimentally verified by measurements on individual components. Then models of individual components were combined to simulate the complete copper network and the FTTN network. Then laboratory prototypes of each network type were developed using hardware components which are used in Sri Lanka Telecom. The complete network models were verified by measurements on the prototypes.

Lightning surge
A surge corresponding to induced lightning surge specified in IEC 60950-1 was generated from HAEFELY PS1500 surge generator. The same surge waveform was modeled in PSCAD/EMTDC software as a double exponential function: Where α and β are constants related to rise time and time to half [13], which were 8µs and 45µs respectively in this case.
A resistive voltage divider consisting of two resistors having a ratio of 1000:1 (10 MΩ and 10kΩ) was used to reduce the output voltage to a low a level so that it can be monitored using an oscilloscope. Test setup used for the experiments is shown in Figure 4.

Figure 4-Block diagram of test setup
The comparison of normalized waveforms (modelled and observed) is shown in Figure 5.
It is very clear that modelled waveform is identical to the generated waveform.

Copper access network
First, each individual component (i.e. MDF, Cabinet, Discharger, DP and cables) of the copper network was modelled using PSCAD/EMTDC [14]. Suitable connection points as well as ground points were also included in the model. The transmission line model was used to model the twisted pair cables. The high frequency model for twisted pair cables consists of series resistors R, inductors L and parallel capacitors C and admittance G as shown in the Figure 6. The effect of G was neglected and the RLC components of a twisted pair cable were measured at the laboratory and verified with the theoretical values. Table 1 shows the calculated and measured line parameters for 0.4mm, 0.5mm and 0.65mm cables. The measured and calculated results were in similar range.

M-Measured values C-Calculated values
Where, R is the dc resistance, L is the inductance, and are capacitances between the twisted pair and between the shield and conductor.
After verifying each sub model (MDF, Cabinet, Discharger and DP) laboratory prototype was built-up by combining hardware components of each sub model. Grounding mesh of the high voltage laboratory at Faculty of Engineering, which has an earth resistance of 5Ω was used as the grounding of the prototype. Simulation model for the laboratory prototype was also developed by combining each sub model accordingly. In the simulation model, a 5Ω resister was used for grounding.
In order to verify the simulation model, the laboratory prototype was tested by applying 1kV lightning surge at the MDF. Voltages appearing at different locations were measured and compared with simulation results. Figure 7 shows a photograph of the laboratory prototype of copper access network.  As shown in the Table 2, experimental and simulation results resonably agree with each other. Slight diferences may be due to not accounting noise, multiple reflections and stray capacitance in the model. Figure 8 shows   Table 3 shows experimental and simulation results of optical fiber network when 1 kV surge is entered from MSAN unit.

Table 3 -Experimental and simulation results for verification of optical fiber network
Similar to copper access network, simulated results agree with experimental results thus verifying the model performance.

Comparison of surge propagation in copper and optical fiber networks
In order to compare the instantaneous surge voltages appearing at the subscriber premises, a laboratory experiment was conducted by using the developed laboratory prototypes. A lightning surge having peak value of 1 kV was applied at two locations, where practically lightning surges can enter in to the network. The locations are overhead lines (Case A) and MDF or MSAN unit (Case B).
In Table 4 peak voltages as a percentage at the subscriber premises and MDF or MSAN have been compared for two access networks.

Extended studies on copper network
The outcome of lightning surge entering the PSTN will not only be affected by the RLC components of the PSTN but also vary with the earth resistance characterized by soil resistivity as well as earth electrode configuration. This effect will be even severe due to multiple reflections at terminal loads i.e. mainly the shunt earth resistances. This work was extended by investigating the effect of earth resistance and multiple reflections by means of simulations. For this study, standard lightning impulse (1.2/50 us) voltage was applied.

Effect of earth resistance 4.1.1 Types of earth electrodes used in Sri Lanka Telecom
There are four main grounding points in PSTN: at MDF (central office), cabinet, distribution point and subscriber customer premises. GI pipes are used as grounding electrodes for the cabinet and DP whereas copper plated steel rods are used for the discharger at customer premises. Table 5 shows the details of the grounding electrodes used at each of these locations. When lightning surge current flow through a ground rod, the grounding system shows transient impedance characteristics depending on the frequency of current flowing into the grounding system. The lightning current gives a wideband frequency spectrum ranging from DC to a few MHz. The measurement of ground resistance with conventional low frequency measurements might not provide data indicative of the ground response to a lightning surge. Therefore it is necessary to evaluate the high frequency performance of a grounding system for protection against lightning surges [15].

Transmission line model for grounding rod
The distributed parameter circuit model for simulating the frequency-dependent impedance of grounding electrode systems is based on the transmission line theory and is assumed a single layer of soil with uniform resistivity [16]. For this study a single vertical grounding rod was considered. Equivalent circuit for a grounding rod is shown in Figure 9. For a vertically buried grounding rod, the ground resistance R is given by Tagg"s equation as follows. [17]

R = (ρ ln(4l/d))/2πl[Ω]
Here l, d and ρ represent length of the rod, diameter of the rod and soil resistivity respectively. The leakage conductance G, capacitance C per meter to the ground and inductance, L per meter of the ground rod can be calculated by dividing the lumped parameters by the length of a grounding rod l. Three types of soil conditions, which cover the whole range of soil resistivity, were selected in order to perform this simulation. Details of the selected soil are shown in Table 6.

Effect of multiple reflection
Objective of this study is to investigate the effect of the multiple reflections in PSTN and see whether it can create a potentially harmful situation to equipment and to the subscriber.

Reflection of lightning surge at a discontinuity point/a Junction/a terminal load
When a travelling wave (the lightning surge), enter a discontinuity point, or a junction or a terminal load, part of the surge may reflect while the other part transfer to the other side. These two components (reflected and transmitted) are determined by the reflection co-efficient and transmission coefficient of the system.
To analyze the multiple reflections, first characteristic impedances of each location of the telecommunication system was calculated and then transmission and reflection coefficients were calculated. The network can be simplified if the network nodes have considerable values for reflection coefficients.
So for other nodes, the incoming wave can be assumed to be completely transmitted.

Multiple reflections using Bewley"s method
For this analysis Bewley"s method, which was modeled using MATLAB software, was used. Consider the distribution point (point A) and the subscriber premises (point B) of a telecommunication network. Let α1 and α2 be transmission coefficients for a wave incident at A and B respectively and β1 and β2 be the corresponding reflection factors when surge enters between DP and discharger.
Characteristic impedances were calculated using the line parameters of the cables. Consider the following line diagram with a distribution point and a discharger. Z4 and Z5 are earth resistances at DP and discharger together with the characteristic impedance of surge arresters.  ….. (7) 9 In the Bewley"s lattice diagram, the junctions must be laid off at intervals equal to the time of transit of each section between junctions. (If all lines are overhead lines, then the velocity of propagation may be assumed to be the same and the junctions can be laid off proportional to the distance between them) [18, 19] Table 9 shows the relationship between voltage and time at DP and customer.
Where t0=L/2V, L is distance between A and B, and V is the velocity of surge propagation.
Voltage appearing at distribution point and discharger due to 1 kV lightning surge enters between DP and discharger is shown in Table  10.

Location
In order to analyse the effect of lightning on the real system of Sri Lanka Telecom, a route was selected inside the university premises. The route starts from the exchanger at Galaha Junction and terminates at the customer premises closer to the Wijewardana Hall at University of Peradeniya. The selected route is shown in Figure 11.

Earth resistances
As earth resistances at each location of the selected route were essential when modeling the network in PSCAD, a field investigation was carried out to measure the earth resistances of the network. At MDF earth resistance could not be measured due to inaccessibility. In this case a recently measured value from the central office was used. Earth resistance values of the selected network are given in the Table 11. 10  Figure 12 shows the single line diagram of the selected route inside the university premises.

Figure 12 -Single line diagram of the selected route
Here letters A-G show the nodes of the selected route. This selected telecommunication network consists of three types of cables having different diameters: 0.4mm, 0.5mm and 0.65mm. Line lengths between each node are shown in Table  12.

Simulation study
The real network described in section 5.1 was simulated using PSCAD. Different studies were carried out to analyze the performance of the existing surge protection during lightning surges. And instantaneous surge voltages at each location were compared with standard surge protection and existing surge protection.  Tables 14 and  15. It can be clearly seen from the Table 14 that without surge arresters a significant amount (i.e. 76%) of the surge voltage could be transferred to the customer equipment. If the surge arresters of the selected route functioned well then this voltage would be always less than the cut-off voltage of the surge arresters, in this case it is about 100 V. If both surge arresters functioned well and earth resistances were also within the acceptable limits then the voltage appearing at the subscriber end could be reduced further.
When a surge is applied in overhead lines similar variation could be observed. As far as the voltage wave forms at the MDF are concerned peak values were higher than 230 V which define the cutoff of the surge arresters. Consequently, the SPDs installed at the secondary side of the exchange would be operated ensuring the protection.

Effect of earth resistance
As discussed in section 4.1 the effect of earth resistances was analyzed for this real telecommunication system. In this study grounding resister used in earlier cases was replaced with the developed grounding model. Simulations were performed for three soil conditions cases: (a) Clay (b) Sandy clay and (c) Sand. Figure 13 shows instantaneous voltage waveforms at the subscriber premises for different soil conditions.
Waveforms with similar shapes were obtained at cabinet and DP. In order to analyze the results obtained from the simulation, the instantaneous peak voltage appeared at each location of the copper network for different soil conditions were tabulated when 1pu of lightning surge enters to the system. Results are shown in table 16.  12 performance can be enhanced by other means like treating soil or using parallel electrodes. Thus voltage appearing at subscriber end can bring to a somewhat lesser value than above.

Effect of multiple reflections
The telecommunication system described in section 5.1 was modeled in MATLAB, to analyze the effect of multiple reflections of selected access networks. Bewley"s method was used to obtain the surge waveform appeared at subscriber premises when the surge is entered between the distribution pole and the discharger. Effect of multiple reflections was analyzed for the worst case which can be happened in the copper wired access network. Surge waveforms at DP and discharger are shown in Figures 14(a) and 14(b). When 1 pu of lightning surge enters between DP and discharger, the surge voltage appered at subscriber end has reduced to 0.940 pu and peakvalue of surge voltage appeared at DP has increased to 1.100pu due to the effect of multiple reflections. Therefore this transmitted signal with higher surge voltage at DP due to the effect of multiple reflections can create a potentially harmful situation to the telecommunication system.

Conclusions
The characteristics of induced lightning surges in existing access networks of Sri Lanka Telecom vary according to the RLC components of the PSTN, earth resistance characterized by soil resistivity and earth electrode configuration and the multiple reflections characterized by cable lengths and surge impedances of the PSTN components. Induced voltage due to lightning would be at a lower level if the telecommunication network has proper surge protection. Instantaneous surge voltage at the subscriber side can be further reduced when the length of the line and number of sheath grounding increase. Results obtained from the laboratory experiments show that the effect of lightning is less significant in copper access network than the FTTN network. Moreover the voltage appearing at the subscriber end can be changed due to the soil condition. Therefore in such areas voltage appearing at subscriber end can bring to an acceptable value by treating soil or using parallel electrodes. It was found that the potentially harmful situations can be occurred in telecommunication systems due to the effect of multiple reflections.
In particular, the results obtained from the field investigation carried out inside the university premises show that the lightning surge can be completely transferred to the subscriber unless proper protective measures are taken. This analysis gives quantitative measurements of the voltages appearing at the end user for different cases discussed above.

Acknowledgement
The authors would like to express their gratitude to University of Peradeniya providing resources to complete this project. Special thank goes to Sri Lanka Telecom for providing equipment and for giving their valuable time in sharing their knowledge throughout the gathering of information relevant to this project. Finally, we offer our regards and blessing to all those who supported in any respect during the completion of the project.