Development of Test Method for Rodent Resistance of Optical Fiber ...

Development of Test Method for Rodent Resistance of Optical Fiber CableJi LeiBeijing CCS Optical Fiber CableCo., Ltd.. An Affiliate of CorningBeijing, China+86-10-67881199 ext: 8415 ·Lei.Ji@bjcorning.comXu RendongBeijing CCS Optical Fiber CableCo., Ltd.. An Affiliate of CorningBeijing, China+86-10-67881199 ext:8400·Rendong.Xu@bjcorning.comZhang LiBeijing CCS Optical Fiber CableCo., Ltd.. An Affiliate of CorningBeijing, China+86-10-67881199 ext: 8410 ·Li.Zhang@bjcorning.comAbstractDamage to communication cables by rodent gnawing leads tocostly maintenance expenditures and exposes thecommunication link to higher risk of service interruption.Several approaches have been used to enhance cable resistanceand extend the service life of the cables. The industry wouldbenefit from a statistically repeatable assessment of cableresistance to rodents. This paper introduces a testing procedurefor rodent resistance of metallic armored optical fiber cableagainst rodent damage. It is a repeated cycling of a rodentgnawing phase and an accelerated corrosion phase. The newmethod focuses on the remaining capability of the undamagedsections of cable after rodent gnawing. Experimental dataillustrate that the new procedure exhibits good repeatability andreproducibility with an acceptable accuracy. A ‘rodentresistance index’ of cables is proposed to reflect the overallresistance of cables against rodent induced damages.Keywords: Cable; Rodent resistance; Test method; Gnaw;Environmental corrosion1. IntroductionThe rodent population is large. In fact, 40 percent of all furcovered mammals are rodents. They have a long pair ofconstantly growing upper and lower incisor teeth which arerootless and covered with hard enamel. They brux and grindtheir teeth to a suitable length by gnawing on almosteverything. In the communications industry, rodents are aconcern because they gnaw on cables leading to costlymaintenance expenditures while exposing the communicationlink to a higher risk of service interruption.Research of cable resistance against rodent damages can betraced back to World War II when Bell Laboratories beganstudying the effects of gopher damage to buried cables [1] .Several approaches have been used to enhance cable resistanceand extend the service life of the cables. Special installation andplacement treatment approaches are available in the filed. Thispaper focuses on the enhancement of the cable itself. Theenhancements can be divided into physical protection andchemical repellents/pesticides. Chemical repellents/pesticidesdo not appear to be practical in the industry for several reasons:1. The effectiveness of repellents/pesticides is weakened by thebehavior of rodents to avoid getting any of the treated materialin their mouths while gnawing on the cable. 2.Repellents/pesticides may undergo dissolving into surroundings,oxidation from air, heat and light, and other chemical processesleading to decreasing effectiveness over time. 3. Leaching ofthese materials into the moist soils causes environmentalpollution issues. 4. The requisite properties of sheath material,e.g. strength, extrusion performance etc, may be altered afterblending with repellents/pesticide. 5. Repellents/pesticidesbring health hazard potential to factory employees andinstallation crews.In regards to physical cable protection, Howard believes at leastone species of rodent, gopher (Thomomys bottae), may have theability to penetrate all non-metallic and soft metallic kinds ofarmor [1] . Cables with multiple metallic armoring may be apromising solution option and was chosen to be the approachtested in our study.The industry would benefit from a statistically repeatableassessment of cable resistance to rodents. Such a test wouldallow cable products from different factories to be compared sothat improvements to the rodent resistance of the cable could bemade. Conventional test methods utilize rodent gnawing andsubjective-based rating system with either wild species orlaboratory rats [1, 2 ]. To better understand the mechanism ofhow gnawing initiates damage and to quantify the test results, avariety of studies have been conducted. These studies includethe following: using a gnathodynameter to study the bitingforce, gnawing frequency and failure modes of cables, thereforeexamining the material susceptibility to rodent damage [3] ;automated mechanical simulation of rodent behavior [4] ; andemployment of an alloy copy of rodent incisors [5] .2. Environmental CorrosionThe severity of damages by rodent gnawing is affected bycable design, installation method and geography. The lattertwo factors are closely related to species of rodent, theirpopulation density and environmental factors. Among thesefactors, the impact of environmental corrosion might havebeen underestimated in cable industry. Connolly and Cogeliapointed out that, although armored cables offer some degree ofprotection, corroded armor is less resistant to gopherpenetration [6] . The corrosion rate of carbon steel to fieldconditions in tropical/subtropical environments is up to0.958mm/year in Lagos, Lighthouse Beach, Nigeria. For urbanindustrial environments, it is 0.165 mm/year in Guiyang,China [7] . Exposed carbon steel tapes used in a conventionalcable, generally thinner than 0.2mm, would be perforated bycorrosion within several months or few years at best. Metalliccorrosion is electrochemical in nature, i.e. galvanic corrosion.Factors that affect the corrosion process are: rain, moisturecontent of the surrounding environment, pollutant content,mineral composition, grain size distribution, redox potential,electrical conductivity, pH, total acidity, chlorideconcentration, microbiological activity, etc. [7] . Cables in allinstallation environments undergo corrosions, whether in air,soil or water.International Wire & Cable Symposium 330 Proceedings of the 59th IWCS/IICIT

A hypothetical demonstration is presented in Figure 1 to showhow environmental corrosion causes deterioration of the steeltape which in turn increases the probability of one-bite cablepenetration. Assume the distribution of biting force for a givenspecies of rodent in installation environment and the strengthdeterioration of exposed metallic armor corresponding to aconstant environmental corrosion rate, for simplicity, arealready known. At a given time1, there is a correspondingmetallic armor strength2 , which has a capability to resist thebites not higher than the biting force 3 . Therefore thepossibility of one-bite penetration 4 at this time can beacquired by integrating possibilities of the upper sectionsalong the biting force distribution curve. With the sameprocedure, probability of one-bite penetration along timecoordinates could be obtained. The one-bite penetrationprobability curve shows the impact of environmental corrosionon an armoring layer. The risk of interruption increases overtime even though the initial strength might be higher than themaximum biting force. Furthermore, if an acceptanceprobability level was defined the service life related to thisprobability could be obtained.This probability curve presents the fact that because of the impactof environmental corrosion, one-bite penetration probability of anarmoring layer, thus the risk of interruption, increases over time,even though the initial strength might be higher than themaximum biting force. Moreover, theoretically, if an acceptanceprobability level was defined, the service life related to thisprobability could be obtained.3. Test Procedure and Configuration3.1 Test ProcedureThe proposed test procedure for cables with multiple metallic armors ismeant to reflect the damage process described above. It focuses moreon the remaining capability of the undamaged sections of cable afterrodent gnawing than the layers that have been damaged. The procedureis a repeated cycling of a rodent gnawing phase and a corrosion phase:1) 20 cages mounted with cable samples are exposed to rodent gnawingfor 10 days.2) Samples with exposed and partly damaged metallic armor bygnawing are put into a corrosive solution. Corrosive perforation timeshould be observed and recorded to quantify the test results. Theperforation time is a complicated quantity, which is dominated byphysical and chemical properties of metallic armor materials andaffected by damages from rodent gnawing and damages during themanufacturing process.3) Remove the corroded layers of cable samples.4) Repeat steps 1 through 3 until the innermost armor is corroded.3.2 Configuration3.2.1 Rodent. Laboratory rats are chosen in place of wild rodentspecies for better control of rodent characteristics. There are more than2000 species of rodent distributed around the world, and approximately216 species in China. Using different species of rodents would makeresults from multiple tests incomparable. In our study, Sprague Dawleyrats are used since they are considered to be slightly more aggressivethan other laboratory rats. They are 12 ~ 24 weeks old, 250g ~ 300g inweight, healthy with normal incisors, half male and half female. Each ofthem is put into separate cages and all of them are replaced with a newbatch every 5 days. No food but sufficient water is supplied for the first3 days. Normal food and water supply the last 2 days. Lightingconditions are controlled for 12-hour illumination and 12 hoursdarkness. As a general rule, configurations should be adjusted toconform to applicable laws and regulations when applying thisprocedure in different regions.Figure 1. A demonstration of environmental corrosioneffects in metallic armored cableBased on the analysis, the main failure mode of optical fiber cablesinitiated by rodent gnawing is a function of both gnawing andenvironmental corrosion. For a cable with multiple armoring layers,damage takes place from the outer part to the inner parts of the cable.The most damaging scenario would be the following: 1. Rodentgnawing destroys the outmost parts with non-metallic or softmetallic materials quickly. 2. The outermost hard metallic armoringdeteriorates by environmental corrosion and gnawing damagethereby exposing the inner parts of the cable with environmentalcorrosion being the dominating factor in this stage. 3. The formertwo stages take place repeatedly until the innermost part of cable,i.e. core of the cable, is exposed. 4. The final stage is that rodentsgnaws through the tube and destroys the optical fiber.Figure 2 . Schematic of cage with cable sampleinstalled3.2.2 Test Apparatus. Cages used in gnawing stage are made ofstainless steel wire grilles with a dimension of 350mm x 180mm x250mm. Cable samples are horizontally fitted onto the mountingboard separating the cage into two rooms with a length of 250mmand 100mm respectively. Rats are fed in the 250mm-long room.Some food is put into the other 100mm-long room as unreachableattractants. Cable samples are stably fixed with a 5mm~15mmseparation No displacement, slip or rotation is allowed during testperiod. Specific lengths of cable are exposed to the rat’s gnawingby a 70mm-wide window in the mounting board. The overlap partof outermost wrapped steel tape is placed away from the rats, asillustrated in figure 2.International Wire & Cable Symposium 331 Proceedings of the 59th IWCS/IICIT

3.2.3 Corrosive Solution. The corrosive solution used in thecorrosion phase is 0.3mol/L hydrochloric solution (HCl) withsodium chloride (NaCl ) added to the saturation point which leads toan accelerated corrosion process and a shorter test duration.4. Experiments4.1 Experiment 1. The first experiment was conducted in WestChina School of Public Health SiChuan University. It was astandard configuration trial (group A) with a replication (group B),40 cages in total. 120 pieces of cable samples ( 3 pieces per cage)were taken from a cable shorter than 20m with corrugated,electrolytically chrome coated low carbon steel tape as outer andinner armors. The structures and other cable characteristics can beconsidered constant.Due to the put-up procedure, a solution with saturated sodiumchloride (NaCl) was made first, and it was used to adjust thehydrochloric solution to the pH of 0.3mol/L. The chloride ionconcentration of the corrosive solution used was slightly lowerthan aforementioned solution, but remained constant in thisexperiment.4.1.1 Gnawing Phase. The percentage of samples with exposedsteel tape for two groups during the gnawing phase exhibits similarevolution tendency, as illustrated in Figure 3. Chi-square statisticaltesting verified that there are no sufficient reasons to refuse thehypothesis that data of two different groups come from the samepopulation of a random variable. Since the outer and inner jacketshave different thickness and different diameters, the exposed sampleamounts are significantly different for the inner and outer armors.This reflects the variation of cable structures. Whether the steel tapearmors have been damaged in different levels has not been readilyinvestigated at this point. Further tests will reveal the differences.Sample amount /pcs6050403020100Group A: Inner armorGroup B: Inner armorGroup A: Outer armorGroup B: Outer armor0 1 2 3 4 5 6 7 8 9 10Time /daysFigure 3.The evolution of exposed sample amountAfter the gnawing phase, it can be observed that damage levelvaries from no obvious damage to fully exposed steel tape, asillustrated in Figure 4. The rating variances combined with resultsthat are not quantified determine the inherent instability and poorrepeatability of conventional subjective-based rating system.Figure 4. Damage level variation during gnawingSignificant rusting was also observed on several exposed steel tapesbecause of rat urine, saliva or other excreta further emphasizing theimpact of corrosion.4.1.2 Corrosion Phase. Cable samples with exposed steel tapewere dipped into the corrosive solution in separate containers. Forcomparison, 14 samples for outer armor and 15 samples for innerarmor, with part of the jacket on each sample manually ripped offto expose a ~30mm x 5mm window of corresponding steel tapewere corroded with the same procedure as the rodent exposedsamples. Corrosive solution was replaced at the same time everydaywhen inspections were made. The corrosion process was continueduntil all steel tapes corroded to visible perforation (with a hole about1mm 2 in size).Figure 5 presents the results of perforation time. Severalconclusions can be directly drawn from it: 1. Perforation times ofGroup A and Group B exhibit similar distributions. 2. Rodentgnawing causes damage on the steel tape, leading to significantperforation time reduction when compared to manually ripped-offsamples that had minor initial defects. 3. The severity of damagefrom sample to sample is variable since the perforation of gnawedsamples shows a widened distribution. 4. The inner tape damageis a bimodal distribution. The causes lie in that the chromecoating of steel tapes of some samples had been completelydamaged.Percentage /%Percentage /%40 Group AGroup B30 Manual rip-off201004 5 6 7 8 9 10 11 12 13 14 15 1680604020Outer ArmorInner Armor04 5 6 7 8 9 10 11 12 13 14 15 16Perforation Time /daysFigure 5. Corrosive perforation time of experiment 1The average perforation times for Group A are outer armor: 8.69days; inner armor: 7.17 days. For Group B the times are outerarmor: 8.86 days; inner armor: 7.32 days. For the manually exposedsteel tape reference group the times are outer armor: 14.27 days;inner armor: 12.6 days. There is a small deviation between groups[~2%, Deviation=abs(D 1 -D 2 )/min(D 1 ,D 2 )] and a significantreduction in corrosion time compared to the reference samples. Inaddition, even though the Sprague Dawley rats can not completelypenetrate the armor, they have the ability to damage it.Several statistical tests, including 1-Way ANOVA and N-WayANOVA, were performed on the data and consistent results wereobtained. The analysis illustrates that there is a significant differencebetween two armors, no significant difference between two groupsand no significant effect of interaction. Analysis also shows nosignificant difference between male and female rats. The result ofN-Way ANOVA is included in Table 1 where X1 is a factor whichcomes from differences between outer armor and inner armor, X2 isInternational Wire & Cable Symposium 332 Proceedings of the 59th IWCS/IICIT

a factor that comes from differences between two groups, X1*X2 isthe interaction of those two factors.Table 1. Results of N-Way ANOVASource Sum Sq. d.f. Mean F Prob>FSq.X1 89.401 1 89.401 46.774 1.5004E-10X2 0.966 1 0.966 0.505 0.478X1*X2 0.0047 1 0.0047 0.0024 0.961Error 313.46 164 1.911Total 404.29 167Note that less than 10 rats were typically usually used in previousliteratures. In this experiment, 20 cages were used with rat changesevery 5 days. A convention is that results of any zoologicalexperiment may exhibit repeatability and are considered acceptablewhen more than 15 animals are used.A way to determine the number of cages is to view the datadifferently. Pick out cages 1; 1~2; 1~3; …; 1~40 in sequence andcalculate the average perforation time to generate 40 virtual testresults of different populations of cages. To put this into a moregeneral sense, randomly pick out 1 cage, 2cages, …, 40 cages andcalculate the average perforation time. A virtual evolutionsequence varying on populations of cages is obtained. The resultsare demonstrated in Figure 6, where the average time of all 40cages is treated as the theoretical mean.Absolute deviation ofaverage perforation time /%Absolute deviation ofaverage perforation time /%25201510252015105Randomly selected (5 sets)Prearranged order00 5 10 15 20 25 30 35 405Cage population /pcsa) Outer armoringRandomly selected (5 sets)Prearranged order00 5 10 15 20 25 30 35 40Cage population /pcsb) Inner armoringFigure 6. The dependence of the deviation of averageperforation time on the population of cagesWhile the population of cages increases, the fluctuation at thebeginning tends to become more stable and convergence. A deviationon a level of less than 5% is achieved when the population of cagesreaches 20. The amount of 20 appears to be a reasonable choice togain acceptable repeatability.4.2 Experiment 2. The purpose of the second experiment is toverify if consistent results could be reproduced by different people atdifferent locations with same test procedure. Two experimental trialswere conducted separately by different staff in both China School ofPublic Health SiChuan University (located in Chendu, Sichuanprovince, China;) and Capital Medical University (located in Beijing,China). These two locations represent Southern China and NorthernChina respectively. The stainless steel outer armor of cable sampleswere tested in this experiment. The amount of samples with exposedmetallic armor after gnawing is 16 for both two institutes. Theaverage perforation times are 45.31 days and 44.31 days respectively.The deviation is ~2.26%. The distribution of perforation time ispresented in Figure 7. The two institutes produced consistent testresults indicating good reproducibility of the procedure.Percentage /%40302010SiChuan UniversityCapital Medical University040 41 42 43 44 45 46 47 48 49 50 51Perforation Time /daysFigure 7. Corrosive perforation time of experiment 25. Rodent Resistance Index of CableBased on data acquired in the test, a ‘rodent resistance index’ ofcable, R Cable , is proposed here. R Cable is a weighted sum of averageperforation time of test cycles and meant to reflect the overallperformance of the cable resistance against rodent initiated damages.N( i )R = ∑ lao T + c(1)Cable i i i ii=1Where, T i is the average perforation time for the ith metallic armor.i indicates the layers of metallic armors, from outmost to innermost,or the sequence number of corrosion phase if some soft metallicarmor was directly penetrated by rodent gnawing. N is the totalamount of metallic armors or corrosion phases. l i is the weightcorresponding to the layers of armors to reflect the fact that partiallydamaged armor can still provide some protection to inner structureof the cable; therefore, an inner armor persists longer than if it is theoutmost armor. o i is the overlap coefficient. If an armor islongitudinally wrapped around the cable, the overlap part might beweakened depending on the quality of that part. c i is a compensationtime; because some jacket layers of cables are so thick that theprolonged gnawing duration is necessary to get a sufficient amountof exposed samples. l i , o i and c i need further study to be determined.a i is the wrapped metallic wire coefficient if an armor is constructedby several helically wrapped metallic wires. This coefficient iscalculated by comparing the effective cross-sectional area of wiresto corresponding metallic tube. a i is to be calculated by followingformula, while other parameters are defined in Figure 7.S m⋅πabai= =(2)S ' ( ) 2 2π ⎡ r+ a −r⎤⎣ ⎦Where, m is the count of metallic wires.International Wire & Cable Symposium 333 Proceedings of the 59th IWCS/IICIT

Figure 8. The cutaway view of a helically wrappedmetallic armorFor example, the rodent resistance index of the cable samples used inthe first experiment can be calculated with the test data (see Table 2). Inthe calculation, it is specified that l i =1, o=1 i , c i =0, a i =1, therefore theindex is actually the sum of average perforation times of outer armorand inner armor. As expected, the index numbers for each cable samplein replicated tests are consistent with a deviation of ~2%.Table 2. Calculation of rodent resistance indexSubsetAverage perforation time/daysOuter armorInner armorRodent resistance indexGroup A 8.69 7.17 15.86Group B 8.86 7.32 16.18Deviation* 1.96% 2.09% 2.02%* Deviation=abs(D 1 -D 2 )/min(D 1 ,D 2 ).6. ConclusionsThis paper investigates the mechanism and failure mode of damage tocables initiated by rodent gnawing while bringing out the impact ofenvironmental corrosion in the process. A test procedure with arepeated cycling of a rodent gnawing phase and a corrosion phase, aswell as a rodent resistance index constructed by quantified perforationtime, is proposed. Study of the corrosion phase gives insight into thesubtle damage of steel tape caused by rodent gnawing which cannot beeasily measured by inspection. Experimental data gathered in twodifferent institutes illustrate that the new procedure exhibits goodrepeatability and reproducibility with an acceptable accuracy.7. AcknowledgmentsSpecial thanks to Professor Wang ZhenYue from Hua Fei Institute,Beijing University of Posts & Telecommunications and ProfessorJiang XueLong from Kunming Institute of Zoology, ChineseAcademy of Sciences for their valuable advices on rodent gnawingexperimental designing.Special thanks to Professor Cheng WeiBo and Wang Ye fromWest China School of Public Health SiChuan University andProfessor Huang HaiXia from Capital Medical University for theconduction of experiments.8. References[1] Craig A. Remey, and Geraldine R.McCann. 1997. Evaluatingcable resistance to pocket gopher damage- A review. 13thGreat Plains Wildlife Damage Control Workshop Proceedings.Year 1997, pp.107-110[2] Yonghua Zhong, Shen Wang, Mingzhu Li, Changshun Wu,September 7-11, 2008, Studying on gnawing testing method ofrodent-proof wire and cable, Proceedings of 2008 InternationalSymposium on Electrical Insulating Materials, Yokkaichi, Mie,Japan, pp.334-337[3] Cogelia, N. J., G. K. LaVoie, and J. F. Glahn. 1976. Rodentbiting pressure and chewing action and their effects on wireand cable sheath. Int. Wire and Cable Symp. 25, pp.117-124[4] D. E. Moss, S. B. McMaster, E. Castafieda, and R. L. Johnson,An automated method for studying stereotyped gnawing,Psychopharmacology 69, 1980, pp.267- 269[5] Shixin Yang, Bin Xiao, Wenge Xiao, Yanan Tan, A study onthe evaluation method for optical fiber cable’s rodentresistance, Proceedings of the 58th IWCS, 2009[6] Connolly, R. A. and N. J. Cogelia. 1970. The gopher andburied cable. The Bell Lab. Rec. 48(4) , pp.98-103[7] Acidification in Tropical Countries, Chapter 6 The Effect ofAcidification on Corrosion of Structures and Cultural Property,John Wiley & Sons Ltd (Import); 1 edition (August 4, 1988),ISBN-10: 0471918709, ISBN-13: 978-04719187079. Pictures of AuthorsJi Lei works in the R&D department ofBeijing CCS Optical Fiber Cable Co., Ltd.He received his PhD Degree in PhysicalElectronics from Tianjin University in2006 and joined Beijing CCS in the sameyear.Rendong Xu, Technology Director ofBeijing CCS Optical Fiber Cable Co., Ltd.He leads R&D, engineering and technicalsupport for customers. He graduated fromWuhan University of Technology. Beforejoining Beijing Corning, he worked forLucent Technologies Beijing Optical FiberCable for about seven years in variousroles such as manufacturing, quality, testand R&D.Zhang Li, Research & Development(R&D) Manager of Beijing CCS OpticalFiber Cable Co., Ltd. She graduated fromXi 'an Jiaotong University in 1993 and hasmore than fifteen years technicalexperience in optical fiber and cableindustry.International Wire & Cable Symposium 334 Proceedings of the 59th IWCS/IICIT