European pipeline performance improving, spill study shows

Dec. 7, 1998
Recent analysis of 25 years of spill statistics for Western European pipelines suggests that a pipeline system poses no greater risk because it is aging. That's the major conclusion of a report prepared by Concawe's D. Lyon and released earlier this year based on data compiled and reported annually 1971-1995. Concawe ("Conservation of Clean Air & Water, Europe") is the oil companies' European organization for environmental and health protection.
Warren R. True
Pipeline/Gas Processing Editor
Recent analysis of 25 years of spill statistics for Western European pipelines suggests that a pipeline system poses no greater risk because it is aging.

That's the major conclusion of a report prepared by Concawe's D. Lyon and released earlier this year based on data compiled and reported annually 1971-1995. Concawe ("Conservation of Clean Air & Water, Europe") is the oil companies' European organization for environmental and health protection.

In the report, the organization says its goals were to "record the reported-on pipeline system development over time, quantify environmental performances, and reveal trends in causes of spills."

The pipeline system covered in Concawe's review has grown from 12,800 km (approximately 7,950 miles) carrying 310 million cu m (approximately 2.05 billion bbl) in 1971 to 30,600 km transporting 634 million cu m in 1995. Over the same period, 63 pipelines with a total length of 2,650 km have been permanently shut down.

In 1971, furthermore, 70% of the Western European system was 10 years old or younger. By 1995, only 8% was 10 years old or less and 30% was more than 35 years old.

During 1971-1995, there were 341 reported spills, mostly very small, and averaged fewer than 14/year. Less than 5% were responsible for 50% of the gross volume spilled. Moreover, the frequency of spills improved over the 25-year period, from 1.2 spills/1,000 km of pipeline to 0.4 spills/1,000 km.

Pipelines carrying hot oils, such as fuel oils, suffered relatively severely from external corrosion because of design and construction problems, says the study. Many were shut down or switched to cold service. The majority of pipelines carry unheated petroleum products and crude oil.

The two most important causes of spills, the study concluded, are third-party accidents and mechanical failure, with corrosion placing a distant third and operational and natural hazards making minor contributions. Significant gains have been made in reducing third-party incidents.

But the report says that the progress made reducing mechanical failures appeared to have reversed somewhat 1992-1997. Such failures mainly occur in fittings rather than in the pipe itself.

If it continues, the increase may prove to be a function of more fittings as pipeline design complexity increases, says Concawe. Or, it could be related to the age of the fittings.

Definition; inventory

The basic definition of pipelines to be reported in the Concawe inventory has remained unchanged since 1971:
  • In Western Europe, that is, OECD Europe as originally defined, consisting of 19 countries excluding Turkey. (OECD = Organization for Economic Cooperation and Development)
  • For transporting crude oil or petroleum products
  • Of 2 km or more long and in the public domain
  • Over cross-country, including short estuary crossings, but excluding other subsea pipeline systems as well as tanker loading and unloading facilities.
The pipeline inventory it monitors, says Concawe, has changed over the years. This evolution has been the result of physical changes to the system, changes to country status (specifically the inclusion of the former East Germany), the pipelines actually reported-on as distinct from the total in existence, and ownership (originally only commercial companies being included).

Concawe says at least 99.5% of all the pipelines meeting its definition are currently reported on.

Moreover, the organization set a minimum spill size at 1 cu m for reporting except when there are exceptional serious safety and/or environmental consequences to be reported for a spill of less than 1 cu m.

Concawe believes its spill-size reporting criterion to be more rigorous than that commonly used in reports for other regions. In the U.S., for example, published spill performances use a criterion of 100 bbl, the organization says in its report but cites no source.

Following German reunification, the pipelines in the former East Germany were added to the database since 1991. From 1971 to 1987, only pipelines owned by oil companies were included, but from 1988, noncommercially owned pipeline systems have been covered also, says the organization.

In 1990, Concawe began to gather data from eastern European countries, but they were not included in the report under discussion here. Initial findings indicate that performance in certain respects is not directly comparable with Western European data.

How many companies and nonindustry bodies reported data to Concawe in 1971 was unrecorded, but in 1980 approximately 70 companies participated in the annual Concawe survey. The group says that number has remained relatively constant with several new companies coming in and others closing down or merging.

For 1995 (the final year of the 25-year period covered in the report), 69 companies reported results. Affiliates and other operating entities of certain large companies are counted individually in these numbers.

Currently, there are some 250 pipelines with physical details recorded in the inventory, reported in terms of some 540 discrete sections.

Concawe gathers the annual throughput and traffic, spill data, and intelligent-pig inspection activity from questionnaires sent to pipeline operating companies early in the year following the year of report.

When the Concawe pipeline inventory was assembled in 1971, the pipeline system already evinced a wide age distribution, says the organization. The oldest pipelines were already in the 26-30 year bracket, although they represented only a tiny fraction of the inventory. The system was comparatively new with some 9,000 km out of 13,000 km (70%) being 10 years old or younger.

The inventory has grown with new pipelines being commissioned and reporting additions of existing pipelines either individually or as entire pipeline systems. Noncommercial pipelines (9,700 km) were added in 1988 onward and East German pipelines (1,700 km) in 1991.

Overall, Concawe's figures show that the system has been aging progressively.

By 1995, only some 2,000 km (8%) out of 31,000 km were 10 years old or younger and some 9,000 km (30%) were older than 35 years. To assess whether age is significant in spill performance, Concawe uses the inventory of the pipeline lengths in each age bracket (1-5 years, 6-10 years, so forth) each year to calculate the relative frequency of the reported spills per unit of length in each age bracket over the 25-year period.

Only the potentially age-related spill causes are included in the analysis. If the occurrence of spills were increasing with increasing age, the plot of spill frequencies/year/1,000 km of pipeline in the same age bracket would increase in successively older age brackets. These data are discussed in detail later.

As yet, says the report, no evidence shows that aging (up to 45 years old at least) is affecting environmental security.

Spill frequency, volumes

The Concawe report says that the reduction over time in the spill frequency per unit length of pipeline represents significant progress on pipeline-spill performance over the years. Fig. 1 [62,219 bytes] shows this overall frequency trend, broken down into major cause categories.

As stated, the frequency of spill has been reduced to a third of what it was at the start of the 1970s: from about 1.2/year/1,000 km to about 0.4.

Data on gross and net spill volumes also indicates overall performance, says the report. These can be shown in relation to the total volume transported (ppm) to provide an overview of the high reliability of pipelines as a bulk-oil mode of transportation (Fig. 2 [84,891 bytes]).

The few very large and random spills somewhat obscure the trend, making the improvement in gross spill volume trend appear less strong than the improvement in spill frequency. Not much progress seems to have been made in reducing the average size of spills; the average spill gross volume is 179 cu m/spill.

The Concawe report says that it might be thought that the trend in the differences between annual gross-volume spill and net-volume spill (that is, recovered spill) should indicate the success in improving clean-up performance. In practice, however, this is not valid, says the report.

For one thing, maximum removal of spilled oil, which is biodegradable, is not necessarily the correct response to minimize environmental damage; this is now better understood.

Another confounding factor is that growth in the pipeline system has been largely for products. To achieve a given visual standard of clean up, less invasive recovery techniques would seem justified for white-oil products than for black crude oil.

Fig. 3 [72,540 bytes] presents the comparison of gross and net average spill sizes. The average spill net volume is 81 cu m/spill: that is, the average recovery of the spilled oil is 55%.

On a per-1,000-km basis, which the report says is a much more meaningful measure on which to assess the trend over time than the absolute volumes, the pattern is as shown in Fig. 4 [52,323 bytes]

This also illustrates that in volume terms, the improvement trend for volumes spilled is rather shallow, says the report, and much less than the improvement in the number of spills. This is partly due to the mix of spill causes changing over the years: the proportion of corrosion spills, small on average, has decreased relative to third-party spills that are larger (Fig. 5 [59,812 bytes]).

Also, for risks posed by pipelines to the environment and to society, it should be recognized that a majority of the spills recorded in the Concawe database are so small that they have been significant only as local nuisances.

Less than 5% (16 spills) of the spills was responsible for 50% of the gross-volume spilt. Whereas some 60% (207 spills), the largest of which was 60 cu m, caused less than 5% of the total gross-volume spilled.

When spills of less than 15 cu m (U.S. basis) are excluded, the comparable Western Europe number of spills over 25 years would decrease to 199 spills from 341.

Concawe categorizes spill causes into five major categories: mechanical failure, operational, corrosion, natural hazard, and third party.

Mechanical failure

There have been 88 mechanical failures out of the total of 341 spills (3.5 spills/year), 26% of the total. Average sizes per spill have been 241 cu m gross and 95 cu m net, 35% of the gross and 30% of the net totals of spill from all causes.

Mechanical failures fall into two categories: construction fault (32) and materials fault (56).

Fig. 6 [50,569 bytes] presents the most common causes of mechanical failures. Of the 88 spills, 34 occurred from the line pipe. The graph also shows the relatively vulnerable spots: pipeline valves, flanges and other fittings, and pump stations. In particular, gaskets and glands in flanges have given trouble.

This indicates, says the report, that adding apparently useful items such as more section block valves risks a slightly negative impact on spill frequency. Thus, minimizing such features is not only an economic factor: the more vulnerable features should be designed out, if possible.

But the trend has been to increase the complexity of control systems (volume and quality-measurement and so forth) that have required additional, potentially more-vulnerable fittings.

Moreover, the rate of improvement in mechanical failures has lagged improvements in other spill causes.

As yet, no evidence suggests an increase in mechanical failures that are potentially age-related; that is, metal fatigue failure of pipelines that are operated close to their elastic limits under cycling-pressure conditions.

If any such pipelines exist, says the report, they are only a very small part of the inventory. The clean record shows that no pipeline has reached an age at which repeat failures are being experienced.


System malfunction and human error caused 25 spills (15 human error, 10 systems failure) at a rate of 1/year or 7% of the total. The result was 1,832 cu m gross volume spilled, 814 cu m net, 3% of both the gross and the net total spill from all causes.

Except for the propensity to cause small spills, the Concawe report says, no general trend appears in these causes. But there is a pattern: only two pipeline operators (of some 70) were responsible for 60% of the human-error spills. Only four other operators incurred 67% of the systems faults.

The underlying causes of this pattern are impossible to determine, Concawe says. Over the 25 years, some pipeline systems have depended highly on manual operations; others on control systems. Also, the complexity of the pipeline systems operated varied widely. The records provide no way of weighting these very different populations.

Moreover, says the report, the number of occurrences is probably insufficient to draw any firm conclusions on the operating standards of individual companies. For many operators, control systems in use today are so radically different from those of the past that past performance is now irrelevant.


There have been 84 occurrences of spills from external corrosion, spilling 7,848 cu m gross volume, 2,460 cu m net. The average spill, 93 cu m gross and 29 cu m net, shows that in general external corrosion results in smaller spills than any of the other causes except operational.

There have been 18 occurrences of internal corrosion resulting in the spill of 3,750 cu m gross, 1,684 cu m net, corresponding to 208 cu m gross and 93 cu m net/spill.

These figures, says the report, are influenced by the occurrence of one particularly large spill in this category: 2,000 cu m gross, 500 cu m net. Otherwise the internal-corrosion, spill-size pattern is very similar to external corrosion.

Overall, corrosion contributed 4.08 of 13.64 spills/year, almost 30%. The spill volume, however, contributed only 19% of the gross and 15% of the net totals from all causes.

The report notes three particularly useful combinations of the pipeline service subdivisions (hot, cold, external, internal) and the corrosion categories (external and internal) that highlight particular aspects: cold pipelines external and internal corrosion, hot pipelines external corrosion, and all pipelines internal corrosion.

Cold pipelines: external, internal corrosion

The main interest in this category concerns whether any evidence indicates that pipelines are starting to reach the end of their lifetimes due to general corrosion.

Concawe calculated the age/spill frequency relationship (spills/year/

1,000 km in the same age bracket) by counting for cold pipelines the number of spills per year due to corrosion in each 5-year bracket from 1 to 5 years through to 61-65 years of age.

This number was then divided by the average length over the 25 years in each age bracket. Fig. 7 [65,469 bytes] shows the results.

The average of such a plot is around 0.1, or 51 spills divided by 25 years = 2 (roughly), divided by 20,800 km of cold pipeline on average in the inventory.

Up to 45 years of age, when there are sufficient data to be meaningful, there is no sign yet of any increasing trend. Results in the 51-55 year age are largely the result of a couple of spills from a single pipeline whose design, construction, and service conditions all bear hallmarks of being atypical. Thus, its performance is not indicative of the prospects for the other pipelines.

An important factor in cold-pipeline corrosion is the much higher incidence of corrosion attacks in features of the pipeline such as road crossings, anchor points, and sleeves.

Only 21 of the 51 incidents occurred from normal underground pipe runs which, given the very great lengths in the inventory, shows that such features are much more vulnerable than normal line pipe.

Industry expects, says the report, that inspections with "intelligent pigs" (instrumented in-line inspection devices) should improve preventive intervention by identifying development of corrosion attack. This should prevent any occurrence of "end of life takeoff" in spill numbers.

Indeed, there is the strong prospect of reducing corrosion spill incidents by catching the corrosion early. It should be noted, however, that intelligent pigging is not possible in all pipelines.

External corrosion: hot pipelines

About six fuel-oil pipelines were responsible for the majority of the 37 hot external-corrosion spills (out of 52) during 1971-80.

These lines evinced particularly severe design and construction problems with heat-insulation coatings that allowed water ingress. They were shutdown or changed service from the combination of their poor performance and the rapid decline in the fuel-oil business in late 1970s.

Those hot pipelines remaining in service along with the few new ones have been better performers but have still been responsible for 13 of the 32 external corrosion spills 1981-95.

On a per-1,000 km basis, even this more recent performance for hot pipelines (average length 600 km) is more than 20 times worse than for cold pipelines (average length 23,700 km).

Internal corrosion: all pipelines

Internal corrosion is much less prevalent than external corrosion: 18 occurrences vs. 84.

One of the pipelines suffering a spill reported that inhibitor was used, one did not report, and the other 16 used no inhibitor. No conclusion can be drawn from this, says Concawe, because the general populations are unknown and the vulnerability to internal corrosion is very complex.

Natural hazard

Natural hazards caused only 14 spills, 10 of which resulted from landslides or subsidence, 2 to flooding, and 2 to other hazards. This category contributes 4% of the total number of spills.

The resulting spill volume was 2,671 cu m gross, 1,004 cu m net, both 4% of the respective totals from all causes. Thus, natural hazard-caused spill sizes are 191 cu m gross, 71 cu m net/spill, very close to the overall average spill sizes.

Third party

Third parties have caused 112 spills: 86 accidentally, 8 maliciously, and 18 incidentally (or prior damaged). The category therefore contributes 4.5 spills/year, 33% of the total number of spills.

The amount spilled as a result was 23,709 cu m gross, 13,316 cu m net, 39% of the gross and 48% of net spills from all causes. The average third-party spill size is 211 cu m gross and 119 cu m net.

The third-party category, therefore, is the largest single cause of spills and is also responsible for the largest proportion of the volume spilled.

Fig. 8 [70,525 bytes] shows the most common causes of accidental third-party spills.

When spills have occurred, says the report, a different pattern emerges between the third-party equipment operators' knowledge of when their activities were near a pipeline and the pipeline-operating companies' knowledge of when work was near their pipelines.

Among the third-party equipment operators, 55% were aware that there was a pipeline near, 35% did not report, and only 10% reported that they were unaware of a pipeline's presence.

Among pipeline operators, however, 45% of the companies knew about the work or did not report (split evenly), and 55% reported that they had no knowledge of the work in progress.

Concawe says that prevention of third-party accidental spills holds the highest priority because of its place in the spill-cause league. It is also the most amenable to improvement through sharing of experiences and comparing operating and work-control practices among pipeline operators from different countries.

The third-party incidental damage category is somewhat of a catchall, says the report, and includes those incidents for which damage was done at some unknown point in a pipeline's life. The damage subsequently deteriorates over time and eventually spills.

In general, they result either from unattributable damage done during original construction or some subsequent construction activity or from damage done in unreported near-miss third-party incidents.

There have been 18 incidental-damage incidents. They all started off from dents, scrapes, and such and share the characteristic that they may well be detectable by intelligent-pig inspections.

Copyright 1998 Oil & Gas Journal. All Rights Reserved.