Sondre NilsenØyvind Engan
Our exposition discovered that road bridges in Norway are not inspected properly; damages and vulnerabilities are not followed up properly, and these insufficiencies have contributed to the loss of human lives.More than 17,000 bridges connect the road network throughout Norway. To ensure maintenance and safety of the bridges, the Norwegian Public Roads Administration (NPRA) has a management system for recording of annual inspections, damages and other matters.Using public information laws, VG gained access to 90,000 PDF pages of inspection data, and made the first publicly available national database of bridge safety. Analyzing the records, we found that the regulatory rules of inspections were broken for more than half of the bridges. After big data analysis and travelling across the country visiting 104 bridges in 15 counties, we discovered that many of the most severely damaged bridges had been neglected for years. We exposed 1745 bridges having damages that pose a risk to traffic safety. The past decade insufficient bridge railings have been a decisive factor in 37 fatal accidents.
What makes this project innovative?
Six years ago VG’s newsroom designer Sondre Nilsen finished his degree in civil engineering. For a year he worked as a bridge engineer. Some of his work time was used on bridge inspections. He discovered what appeared to be deviations, inaccuracies and mistakes in the management system. Four years later Sondre worked in VG when a motorway bridge collapsed. His background gave us unique insight into the workings of the Norwegian road administration and their systems.The journalists asked for access to the bridge database to understand how many bridges have severe damages on their capacity. NRPA provided a number, but after a week they admitted the initial number was entirely wrong. This initialized several hypotheses: Are the bridges safe? Are damages being discovered? Are the annual inspections being carried out according to regulatory laws? Do damages cause fatal accidents? Do the maintenance database provide the overview which the NRPA claims it does?The journalists have reconstructed a governmental database and made it available to the public. For the first time, the Norwegian public can check the maintenance status of their bridges of interest. More than 400.000 unique users have visited the map in the first month - and findings from the map have become stories in national, regional and local news outlets across Norway.By color-coding layers in a map we could discover new and unexpected regional differences in how the counties inspected and maintained their bridges.For “The Brobot” (BRO means BRIDGE in Norwegian) we used the techiniques of robotjournalism to offer local and unique articles for all municipalities in Norway:https://www.vg.no/spesial/2017/de-forsomte-broene/kommuneartikkel/index-eng.phpBy visiting and checking bridges themselves, the team have been able to update the database and pressure NPRA to make adjust their way of working. We believe our stories represent a high level of both investigative journalism and innovative presentation.
What was the impact of your project? How did you measure it?
The publication led to a national debate and an improvement project within the NPRA. VG’s exposé was debated in parliament. The minister of transportation demanded major improvements, and has given the NPRA a deadline to completely update the bridge database by February 2018.Several bridges were urgently inspected and repaired due to VG’s publications.For twelve years the department for traffic security in NPRA have requested a national examination of dangerous railings. The minister of transportation now promises that this will be finished in the winter of 2018.Families who lost their loved ones in car accidents learned for the first time that damages on the bridges and railings were a major cause to the fatal outcome.Reader engagement has exceeded our expectations. The interactive bridge map has had more than 400,000 unique visitors. The three main articles have all turned 300,000 which is great numbers in a country of 5 mill. people.
Source and methodology
Our main source for this project is data from the NPRA, mainly from NPRA’s internal bridge database Brutus. We gained access to these data with reference to the Norwegian freedom of information act. However, NPRA did not release the data as database files, which was our primary request. We received the data as PDF files. This forced us to write several internal tools in order to automatically read and sort the data into our own database.We also collected data from NPRA’s open service Norsk vegdatabank (“Norwegian road data bank”). All data sources had one field in common, namely the bridge number unique for each bridge. This made it possible to combine data from different sources.As research progressed, it became clear that important data were missing from the PDFs we had received. Each registered damage has a history file, which was missing. It also turned out that some types of shortages – mainly bad railings – were logged as “vulnerabilities”, not “damages”. These vulnerabilities were not a part of the first data dump. We eventually requested and received both damage history and vulnerabilities.We discovered early that NPRA’s data were not totally reliable. In some cases critical damages had been repaired without the database being updated. In other cases we discovered damages which had not been logged. This made us realise that we had to physically inspect the most important bridges in question. All in all we visited and photographed 104 bridges in 15 counties.A month before we published our project, the NPRA published a report on bridges and fatal accidents. The news story was given to the national broadcaster NRK. We later discovered the report was 19 months old and had been forgotten until we started our investigations. In 2014 VG published an investigative project on NPRA and “The Secret accident reports”. Since our investigations the reports have become public information, and this secured us access to the details of each fatal accident connected to bridges and lack of maintenance.
A MySQL database is the core of both research and presentation in this case.As we received the data from NPRA, we utilized various scripts in order to import the data into the database. First, we used the tool Pdftotext to translate the PDF files into plain text. Second, we had to make our own scripts – partly Python and partly PHP – to recognize patterns in the text files, extract the relevant data and write them to the database.Custom PHP scripts were also used to import data from the open road databank and the spreadsheets we received from NPRA.The database was immediately put into use as a research tool. By displaying bridges on a map and color coding them after different criteria, we discovered angles and connections which would have been very hard to read from the tables.The very same database, and a somewhat streamlined version of the map, is also an important part of our presentation to the readers.The “Brobot” also receives its data from the base.The three main stories were published as what we call “specials”, which means that they are presented outside our normal CMS. A designer has manually styled, coded and illustrated the stories in order to make the relatively advanced topic understandable for all our readers.