Failure Report Circuit owner: Entergy (Gulf States) Segment: 00001301-2 Work Order: Goslin Feeders (PO 10174636) Termination 1: Gosling RD Termination 2: Woodlands DR/Gosling RD Failure type: Cable Termination Splice Other buried elbow near termination on bend Failure location & description live-front An Elastimold 35PCJ1P [14-10] splice failed on the B phase of the Goslin Feeder circuit at the second pit near the corner of Gosling and Woodlands. The splice joined a short segment of new EPR cable and a treated PE cable. The splice failed on the side adjacent to the EPR cable. The Elastimold splice body, a short section of EPR cable, a short section of treated PE cable, and the compression connector/injection adaptor assembly (hereafter the assembly ) were sent to Kent. Injected: May 12, 2008 Failed: May 23, 2008 at 5:30 A.M. Elapsed: 11 days Summary Figure 1 shows the sample components excluding the splice body as they were received. The cable was severed at the interface between the new EPR cable and the compression connector. Figure 2 shows a sectioned view of the splice body interior as it was received. The Elastimold splice body included a molding flaw at the interface between the EPDM insulation and the outer semiconducting shield. This flaw is visible in Figure 2 and enlarged in Figure 3. This flaw may have caused reliability issues had it remained in service, but its presence was in no way related to the failure. The inside of the splice body, enlarged in Figure 4, demonstrates extensive tracking along the entire 360 circumference of the interface between the EPR cable insulation and the EPDM insulation of the splice body. Within the splice body, the EPR experienced very high temperatures over a prolonged period of time, as the EPR is thinner and asymmetrical. This deformation was not present outside of the splice body. The asymmetry and the thinning of the insulation are illustrated in Figure 5, where the sample was cut at approximately 6 inch intervals to reveal the cable cross section. Figure 6 was prepared by cutting the Copyright 2008, Novinium, Inc. All rights reserved. Page 1 of 5
compression connector/injection adaptor assembly and a portion of the connected PE cable with an abrasive water jet to preserve internal features. Compression connector/injection adaptor assembly PE Cable EPR Cable Sample as received Figure 1. Internal components as received. Cable and assembly were severed. Index marks (each one inch apart) were added to the EPR cable starting with zero at the right cable end. The EPR cable was approximately 22.5 long. Molding flaw (see Fig. 3) 360 tracking (see Fig. 4) Figure 2. Splice body cut along an axis to allow inspection of the internal surfaces. Molding flaw increases local electrical stress and voids within flaw may create partial discharges Figure 3. Molding flaw unrelated to the failure of the splice may have created future reliability issues. The failure sample handling was not performed according to NRI 93. For example, solvents where used to clean portions of the sample prior to the sample being shipped to the Novinium laboratory. Improper sample handling may reduce the reliability of failure analysis. NRI-93 was first promulgated on May 25 th, two days after the failure and hence the injection team had not yet been exposed to the failure sample handling instructions. Copyright 2008, Novinium, Inc. All rights reserved. Page 2 of 5
Figure 4. Close-up of 360 tracking shows uniform and extensive damage. 240 mils 350 mils 5.5 11.5 17.5 22.5 Figure 5. EPR cable in cross section approximately every 6 inches. From left to right, the cross sections are at 5.5, 11.5, 17.7 and 22.5 index marks respectively. Only the cable within the splice body was deformed indicating that the fault was the cause of the deformation. This deformation could only occur with prolonged temperatures well above 130 C. 0.6 0.1 0.2 2.7 Figure 6. An abrasive water jet was used to section the compression connector/injection adaptor assembly and the attached PE cable. Gaps in the assembly indicate improper craft work. Figure 6 is annotated with three gaps and the PE insulation cutback in red. The first 0.6 gap delineates the volume between the EPR strands and the water stop in the compression connector. There should be no Copyright 2008, Novinium, Inc. All rights reserved. Page 3 of 5
more than a 0.1 gap here. Installation instructions require that the strands touch the water stop. At least 0.5 of excess length was added to the assembly length by this craft error. This error would have been obvious if the installation template were used. The second 0.1 gap is the volume between the PE cable conductor strands and the compression connector water stop. This gap is within normal installation parameters and is created by differential elongation upon application of the swage. The third gap is the 0.2 between the injection adaptor shoulder stop and the compression connector. There should not be a gap here. In fact, when properly installed some aluminum from the compression connector extrudes past the shoulder. This error would have been obvious if the installation template were used. Finally, the cutback length of 2.7 is 0.2 greater than the required 2.5 and is the root cause of the aforementioned 0.2 gap. These errors together lengthened the assembly by about 0.7. When assembled correctly there is 0.61 of clearance within the conductive insert on each side or a total of 1.22. At least 0.7 of that clearance was eliminated by improper assembly. A properly centered splice would have 0.52 of total clearance or 0.26 on either side. This is an unacceptably small margin even if the splice body were perfectly centered. The splice body was almost certainly not centered. The pattern of internal tracking around the entire inside circumference of the splice body suggests uniform encroachment by the compression connector past the conductive insert and into the insulation interface. The installer did not use a template and did not provide adequate permanent marks to center the splice. As seen in Figure 7, there may be some residual marking tape on one side of the splice, but it is not applied perpendicular to the axis. There is no evidence that marking tape was applied to the other side of the splice. Figure 7. Splice before removal Marking tape? Though probably not directly related to the fault, Figure 1 illustrates that the insulation shield cutbacks were incorrect on both cable ends. The installer clearly did not use the template for these cutbacks as such errors are obvious with even a cursory inspection. The individual (Mr. Gabe Durant) who installed this splice did not provide the following permanent marks, which are required by the Novinium NRIs: Copyright 2008, Novinium, Inc. All rights reserved. Page 4 of 5
1. There was no permanent tape or mark to demarcate a centered splice. See NRI 32, step 2 and 20. 2. There was no permanent tape or mark to demarcate the cleaned area of the insulation shield where a molded splice is parked. See NRI 32, step 2 and 12. 3. There was no permanent mark on the injection adaptor to indicate who performed the craftwork. See NRI 32, step 3. 4. There was no positioning line drawn on the insulation. See NRI 30, step 18 and NRI 32, step 4. Corrective Novinium has taken 1 the following case specific and general corrective actions to reduce the probability of a reoccurrence. Case Specific Corrective Actions Mr. Gabe Durant of Finway, Inc. has been decertified and will not be allowed to work for Novinium until he is recertified. Novinium recommends 3M splices and terminations. General Corrective Actions Asymmetrical splice templates (i.e. a splice with a single injection adaptor) are being redrafted to remove the unnecessary and excess gap between the insulation on the non-injected side and the compression connector. This change provides for a greater margin of craft error on assembly length and centering. For an Elastimold 35PCJ1P [14-10], the total margin is increased by about 0.25. Proper failure handling instructions (NRI 93) are being pushed to all field injection personnel. Durable metallic craftsman tags are being distributed and will be attached to all compression connectors to document workmanship responsibility. Lead Investigator: Glen J. Bertini Date: August 9, 2008 1 In some cases the corrective actions require some developmental work and may not be fully implemented by the report date. Copyright 2008, Novinium, Inc. All rights reserved. Page 5 of 5