The Scottish Government

« Previous | Contents | Next »

Listen

Appendix D - Response on Draft Report by Weidlinger/Faber Maunsell

Faber Maunsell/Weidlinger

RESPONSE TO AUDIT REPORT DATED DEC 2005

Item 1

The audit states that the NCHRP 534 Guidelines require that a cable band be removed whenever (numerous) broken wires or stage 4 are present. The more detailed requirements for the first and second internal inspections give clearer requirements for cable band removal:

1. There is no statement requiring removals during the first inspection, which is intended to be an exploratory inspection providing information upon which future inspections will be based. Only six locations are recommended; eight locations were in the original project statement for the Forth because the cable was already 40 years old, rather than 30.
2. In the second inspection, at least two cable bands are to be removed "to facilitate inspection to the center of the cable and under the bands".
3. On the Forth Road Bridge, the cable band spacing is (about) 20 meters, which is more than adequate to inspect to the center of the cable and bands need only be removed if "numerous" broken wires or Stage 4 corrosion to a depth of more than 3 wires are present.
4. Clearly, cable band removal will be required during the second inspection because of the depth of Stage 4. The term "numerous" is not defined, but seven broken wires is not considered numerous. On the Ben Franklin Bridge, one cable band was removed during the first inspection (at the age of about 75 years) because internal broken wires at locations where the wires crossed were found, and with a cable band spacing of only 6 meters, deep wedging was impossible. Another was removed when over 120 broken wires were found in a single panel during the subsequent oiling operation.
5. The decision to remove a cable band is thus flexible and, especially during a first inspection, at the discretion of the engineer when unusual conditions are found. During the first inspection, it is rare that provision for removal is made, and the resulting delay makes it impractical to include removal into the work unless the need is very clear. This was not the case on the Forth, and cable bands were not removed.

Item 2

The audit states that it is not common practice to take samples only from the outer 11 rings of wires. While this is true, the taking of deeper samples is usually done in inspections later than the first, when some knowledge of conditions is available. On the Forth, the first panel that was opened was East 100S-100N, which was only later found to be the worst panel opened (it is, of course, not known to be the worst on the bridge; only if all are opened can one say this with any certainty). Because it was the first, and 8 broken wires were found, the inspection team focused primarily on Stage 4 wires for sampling. The first strength estimates were made using the data from the Bronx-Whitestone Bridge, with 64% of Stage 4 cracked, because tests were not yet available; this resulted in a low cable strength. Realizing that the cracked wires were the key to the cable strength, wires from the outer few rings, which usually contain a higher percentage of cracked wires, were removed to obtain an early estimate of the fraction of Stage 4 wires that are cracked (of the seven Stage 4 wires tested, none were found with cracks. This was surprising, because where there are broken wires there must also be cracked wires. In other panels, the emphasis continued to be on the outer wires). In this first panel, the additional wedge line was driven to investigate further the large amount of Stage 4 in the top of the cable and one sample was removed from the eleventh ring (there are Stage 4 wires to a depth of 14 rings at this wedge line and the last 3 were not as corroded as the eleventh). By the time that the first 6 panels had been inspected, it was intended that some deeper samples be taken; none were found.

The sampling of outer wires in the first panel was also the result of:

1. The crew had not done wire splicing on the actual cable before, and some training was necessary. This is best done on outer wires. Inner wires are difficult to reach for the purpose of tightening the ferrules, and often a wire would be spliced with zero or small stress.
2. It had been decided that only one sample would be removed from each wedge line to minimize the introduction of too many ferrules, which introduce voids that can collect water inside the cable. There would be enough ferrules from splicing broken wires.
3. Since no broken wires were found in the interior of the first panel opened (except for one adjacent to a wood inclusion in the cable, an anomaly), it was expected that fewer cracked wires would be present in the interior (this has also been found on several other cables).
4. It was expected that a reduced sample (compared with the sample recommended in the Guidelines) would be taken because this was a first internal inspection, and it was desired to learn as much as possible about cracked wires that are prevalent in the outer rings.

The distribution of cracked wires found in the samples is as follows:

The fraction of samples cracked is a maximum at the outer ring and is smaller in the next three rings. A larger sample would probably show the fraction dropping as the depth increases. The 72 broken wires found in all 10 panels are distributed vs. depth as shown in the following chart.

This chart does not include the three wires found broken at depth which are assumed to be anomalies.

Item 3

The cable strengths quoted are like comparing apples and bananas. Several different assumptions have been made in computing the current cable strength. These are listed below, along with the resulting current estimated loss in cable strength for each.

The percent loss in cable strength in all these cases is based on the initial cable strength using the Normal distribution for Stage 1 & 2 wires, which have been found in other investigations to have the same properties as new wires. The Normal distribution results in an initial strength that is 1.2% lower than the nominal design strength, while the Weibull distribution results in an initial strength that is 1.6% lower than that. The first of these differences may be the result of simplifications used in the NCHRP Brittle-Wire Model; the second is the result of the use of the Weibull distribution for these wires. The Weibull distribution has a bias towards lower strength wires that is appropriate for a degraded population, and, for the same mean and standard deviation of tensile strength will result in a lower strength than the Normal distribution.

The minimum tensile strengths, based on a probable maximum fraction of cracked wires, are calculated only to estimate the error caused by the small number of samples and the error is considered to be covered by the safety factor. The probable maximum fraction was continuously estimated as the data from the laboratory was received (testing was by batches on wires from several panels) to determine whether or not it was stabilizing (it was).

The 6.7% strength loss should thus be compared with a strength loss by NCHRP of 12.7%. The difference between the nominal and "minimum" cable strengths is 2.7% at the most; this is accommodated by the "safety factor", though it should be considered in reliability calculations.

The assumption of cracks occurring only in the outer 6 rings is made to assess an upper bound on the cable strength. It is based on the observation that broken wires (other than three anomalies) are found only in the outer 5 rings. This assumption is applied to the entire cable despite no broken or cracked wires being found in the top portion of the cable, to account that more rings of cracked wires may be present in the bottom portion of the cable. The calculation of cable strength using 3 zones is made to further check this assumption, resulting in strength 1.3% greater. This same 3 zone assumption is also used to estimate future cable strengths to avoid excessive conservatism that may occur from the assumption of linear progression of corrosion "category" with time.

Item 4

It is agreed that when different assumptions are made for the rate of deterioration for each category, the future strength will be different. Tests made to date indicate that a uniform rate (linear) through Stage 4 is reasonable. There are no test results to date for higher categories. Extending the calculations out fifteen years is excessive when the cable is only 40 years old; five years (about 10% of the age) is as far into the future that can be reasonably estimated. At this age, the strength loss would vary from 9.8% to 7.6% by these calculations, barely different from the 6.7% estimated at the current time. Even a linear rate of strength loss assumption from the completion of the cable would result in 6.7 x 45 / 40 = 7.5%; a linear rate from the onset of Stage 3 (estimated to be about 19 years age, in 1983) would result in 6.7 x (21+5) / 21 = 8.3% five years hence. This same linear rate would result in a strength loss of 11.5% in 15 years (2019). Assumption 2 with half the rate of deterioration after Stage 4 is reached is about the most that could be considered reasonable.

It is noted that the values for T that are used seem to be applied to all categories of wires (not likely) and that the strength loss is taken as a percentage of the strength in 1983, since the strength given for 1983 is smaller than the initial strength; it should be taken from the initial strength of 345144 kN, resulting in 17.2% in 2019 (admittedly a quibble, but the difference increases with increasing T. For T = 2, the loss becomes 15.4%, and for T = 5, it becomes 14.3%).

Future planning should be based on conservative assumptions, and waiting for 5 or 10 years to start remedial action, which could be assumed reasonable if T = 5 is assumed, is not prudent.

The numbers are, of course, useful in providing an indication of the effects of deterioration rate, and how inaccurate estimating 40% of the age of the cable into the future can be.

Without further effort to estimate the effect of changing the rate only after Stage 4 is reached, we would not guess how your tables would change. The effect on the table for assumption 2, of course, would be minor

Item 5

The outer ring of wires represents 3.2% of all the wires in the cable. The refinement in the total number of wires in each stage of corrosion in the cable would be affected by perhaps 0.5% if each wire were assigned a stage in accounting for the wires. If, for example all the wires were Stage 4 and Stage 3 was seen at the wedge, the inspectors should assign Stage 4 at the wedge lines, or they should move the wedge over by two wires. When these wires are half Stage 3 and half Stage 4 between the wedge lines all around the cable and all are recorded as Stage 3, the error in counting could be about 1.7%. This last scenario is extremely unlikely, and there has to be some reliance on the inspector to use judgment. Quite honestly, with 97% of the wires being inside the cable and counted by half octants, there was to be no compelling reason to make a separate calculation for the outer ring. The statement, however is correct, and perhaps should be clarified in the Guidelines if the opportunity presents itself.

To try to inspect the entire second ring of wires would be very time consuming, except for broken wires, which almost always make themselves known. For broken wires, which are often more prevalent in the outer two rings, interior wires and exterior wires are counted separately.

Item 6

The paragraph is, for the most part, correct. It would be most desirable to have a series of repeat inspections to calibrate the methods, and it is assumed that eventually, as repeat inspections are made as recommended, this will occur. It will take many years, because of the recommended inspection intervals. It should be assumed that when a five year inspection interval is recommended, that other steps to remediate the conditions inside the cable will be taken, and calibration in this case will not be effective.

Item 7

As discussed in Item 3 above, the strength losses to be compared should be 8.0% strength loss when cracks are assumed to occur only in the outer 6 rings and 12.7% by the NCHRP method. When refined further to analyzing the cable in 3 zones with a different fraction cracked in each zone, the strength loss is 6.7%; this analysis breaks the test data into smaller units, resulting in larger probable errors.

Item 8

We agree with this statement. However, with the current data and lack of knowledge of deterioration rates beyond Stage 4, we are not comfortable with any assumptions other than those made.

« Previous | Contents | Next »