The study focuses on a risk analysis of the BTC pipeline and integrates state-of-the-art technologies for a comprehensive advanced security analysis (ASA) that includes critical issues such as the geographical and socio-political context along the BTC pipeline. This was addressed in the GIS (Geographical Information System) developed for purpose of integrating satellite imagery together with a number of map layers reflecting both physical and human factors along the BTC pipeline (road networks, topography, vegetation, population density, etc.).
During the course of this analysis, the BTC pipeline was sabotaged by PKK insurgents in August 6th 2008. The geographical and socio-political factors of this sabotage have been weighted and extrapolated to the whole of the pipeline by a geospatial analysis on the GIS layers. As a result, the pipeline has been segmented into several degrees of risks which may prompt additional security actions as proposed in this paper.
Characteristics of the BTC pipeline
The BTC pipeline route passes over 1,100 miles through Azerbaijan (278 miles), Georgia (153 miles), and Turkey (668 miles). Along this route the BTC crosses the Caucasus Mountains and East Anatolia, reaching an altitude of 2,800 m. Its operation requires eight pumping stations (two in Azerbaijan and Georgia each; four in Turkey), four metering stations (one in Azerbaijan and Georgia each; two in Turkey) and more than 100 block valve stations.
Geographical and socio-political conditions along BTC pipeline
For the geospatial analysis of the geographical and socio-political conditions surrounding the pipeline, a GIS has been designed integrating satellite imagery (high, medium and low resolution) with 26 map layers, including among others, ethnic groups, population density, transportation network, towns and villages, land cover, and water bodies. These layers reflect the political, social and geographical context of the pipeline’s route. Fire monitoring along the pipeline route with a Moderate-resolution Imaging Spectroradiometer (MODIS) sensor was also carried out, being of special relevance in the context of the ongoing Russian – Georgian crisis and potential sabotage from different terrorist groups. The most important infrastructures along the pipeline have been also identified on satellite imagery, digitized and integrated in a GIS layer.
Figure 1. Pumping Station along the BTC transect
Analysis of the sabotage at Refahiye Block Valve Station
The sabotage to the BTC pipeline took place on the 6th of August 2008 in the region of Refahiye, Eastern Anatolia. The MODIS sensor in the Terra and Aqua satellites detected hotspots generated by the sabotage providing coordinates, date and time of this event. The sabotage was perpetrated at Block Valve Station number 30 (for additional damage to this infrastructure), along the road which connects the city of Refahiye with Ezrincan. The region lies at the north-eastern border of the Turkish Kurdistan. After the sabotage, Block Valve Stations 31 and 29 were closed. The most important cities around are Refahiye, 14 km to the east (22.000 inhabitants) and Ezrincan, some 50 km to the west (107.000 inhabitants). However, many small villages like Yurtbasi or Alcaatli lie within a radius of 3 kilometers of the sabotage location, providing, therefore potential hiding places for the activists. Other spatial indicators were provided by the GIS layers, characterizing the region where the sabotage took place, as follows:
• Low population density, ranging from 26-50 inhabitants/km² in the exact location of the event to less that 5 inhabitants/km² in the surroundings.
• Low transportation network density. The only main road that crosses the region is the E80 which constitutes the main link between Sivas and Ezrincan. However, some minor roads are present in the region for the purpose of connecting the Sivas – Ezrincan E80 with the surrounding villages. Possible roads taken by the terrorists include: Olgunlar – Sahverdi, Alcaatli – Kirbulak, Ekecik – Kirbulak (in southern direction, from west to east) or, several links from E80 to Akcigdem, Ulukak, Teknecik, Yurtbasi, Ekecik or Kacakkoy (in northern direction, from west to east). The nearest railway runs some 35 km away to the south from the location of the sabotage.
• Low relief values, the sabotage occurred at the bottom of a river valley with slopes under 15°
• The dominant landcover in the region is ”agricultural crops” (LandScan 2005). However, evergreen forest can be observed on the slopes at a distance within 4 kilometers from the location of the sabotage (The forest allows PKK militants to move from one place to another unobserved.
• Last but not least, it is important to mention the proximity to ethnic areas, since the sabotage took place near the Kurdistan border, where the activists would have found logistic support and a safe retreat in the nearby Kurdistan villages mentioned above.
For security reasons, the pipeline runs along its entire length underground (1-10 meters), including the points where the pipeline crosses water courses or roads. As a consequence, the most vulnerable points along the pipeline are related to the infrastructures above the ground such as block valve stations, check valve stations, pumping stations (Fig. 1) and oil terminals.
Analysis indicates that PKK activists have selected a relatively remote area for the location of the sabotage in close proximity to Block Valve Station 30 for maximum damage. The sabotage location is also in the vicinity of their ethnic homeland. The remoteness factor is defined further, in the context of the GIS, by the layers showing relatively low density of population and roads. In these relatively remote areas, activists would have not been observed when digging to set up the explosives
Figure 2. MODIS hotspot detection of the sabotage (in black) overlaid on the Digital Terrain Model of the area
The geospatial analysis in the GIS focuses on the weighting and extrapolation of the geographical and socio-political conditions occurring at the sabotage location to other sections of the pipeline. The objective is to determine the occurrence of similar conditions of risk as possible locations that could attract future sabotages. Relevant information on physical and socio-political characteristics of the BTC influence area has been produced and imported as data for processing tasks. A graphical model has been built inside GIS software, where raster matrix calculations were carried out in a geo-processing environment. (A The Raster dataset is a spatial data model made of rows and columns of cells. Each cell contains attribute values and location coordinates. Groups of cells that share the same value represent geographic features. It is useful for storing data that varies continuously such as in an aerial photograph, satellite image, or elevation surface.) By assigning a weight value to each raster cell and defining the calculation algorithm the model pointed out the most vulnerable locations along the pipeline based on our input data. The benefit of this analysis is that, provided that the relevant layers and suitable resolution are available, it allows a quick and effective way to determine pipeline segments under risks by overlaying critical layers.
The critical layers considered in the geospatial modeling of the vulnerable sections along the BTC pipeline were:
Proximity to present conflict areas (or possible problematic areas in the future)
This layer has been obtained from a combination of political, ethnic and conflict maps which have been digitized and the resulting vectors have been analyzed. Buffer rings have been generated around the conflict areas and have been transformed in different raster classes. A weight has been assigned to each class in relation to its proximity to the conflict areas. Because of the constant threat that PKK is posing to the security of the Turkish BTC segment, and considering that threats have been already put into practice by this terrorist organization (Attack of 6th of August 2008), the highest weight was assigned to the proximity to Kurdistan influence areas. Lower weights were assigned to the Georgian territories under Ossetian threat and Azeri regions close to Nagorno Karabakh.
Transportation network density
The density of the transportation network has been generated in the first instance as a vector dataset containing roads (all types) and railways layers. The vector dataset has been interpolated to raster and four classes have been generated according to thresholds in the transportation density values. Weights have been assigned to each class accordingly. Priority has been given to the low network density.
The population density dataset has been imported from the Landscan product (Landscan, 2005) directly as raster. Different classes have been generated and suitable weights have been assigned to each class. After considering the regional characteristics and analyzing the conditions at the place of the attack, the highest weight was assigned to areas with a population density under 50/km².
The landcover dataset (generated from NOAA’s AVHRR and Landsat 7) which originally contained 24 different classes has been organized in a raster dataset with 10 general landcover classes. Weights have been assigned to each landcover class according to its possibility to provide coverage for sabotages. Priority has been given to different kind of forests (mixed, deciduous, evergreen) and to agricultural fields.
Presence (or proximity) of BTC pipeline installations
The installations, digitized along the pipeline in the original GIS, have been exported in a single raster dataset. Buffer rings have been generated around the installations (valve stations, pump stations, metering stations) at certain radius values. They were subsequently transformed into new classes to which weights have been assigned according to their proximity to the installations.
Figure 3. Spatial characteristics of sabtotage site
These five layers of the GIS were later overlaid and reclassified, giving to each layer a suitable weight related to its importance in the risk analysis. The digital terrain model (SRTM) has also been considered as its analysis provides good results of low visibility areas. The result shows a map with a graduated scale indicating the probability of future attacks based on the modeling of the 6th of August sabotage, BTC pipeline specific characteristics and overall political, social and ethnic indicators of the territory.
Figure 4. Algorithm used for model generation
The most vulnerable pipeline sectors are indicated in the map (Fig. 2 and Fig. 3) as “red hotspots”. These are concentrated in northern Turkey, in the province of Ardahan, in the districts of Ardahan, Hanak and Posof. Of special importance are the two locations in the districts of Ardahan and Posof, respectively, due to local relief characteristics. The first location is situated at an altitude of 1,830 m, on the left hand side of the river Kuruchay, tributary to Kura River. The closest village is Kartalpinar (<1000 inhabitants) and the nearest transport route is the link: Ardahan – Hanak. Immediately south of Kartalpinar the pipeline crosses underground the river of Kuruchay. This fact increases the overall vulnerability indicator of the region, as possible sabotage in the river crossing would make the reconstruction work more complex. The valve station situated here is rather isolated in the sense that, apart from the small village of Kartalpinar and the proximity to Ardahan (~4 miles, 17,000 inhabitants) the surroundings also have a low are population density. To the west, not even small villages appear in the vicinity as the relief is getting more abrupt until reaching heights of more than 2,400 meters. This mountainous region, western of Kartalpinar, can easily provide good hiding places in case of future terrorist attacks. The second location is situated at about 1,450 meters of altitude in the district of Posof on the left hand side of the Çoruh River which flows into Georgia before reaching the Black Sea. The relief is sharp and the slope values are high. The Valve station lies on the bottom of the V-shaped Çoruh river valley and therefore, the visibility in the region is reduced. In combination with the presence of extended patches of pine forest that cover the valley slopes, this offers ideal conditions for hiding in the eventuality of a terrorist attack. The nearby river crossing and the presence of small dirt roads that can serve as good runaway routes are added points that increase the vulnerability of this area.
Figure 5. Overlay of relevant layers in the geospatial analysis for pipeline vulnerability
In Georgia, the most vulnerable areas are located in the central part of the Georgian pipeline sector, in the region of Kvemo Kartli. Additional weak points that increase the vulnerability in the region are the violent ethnic division that the district of Tsalka has experienced in the recent decade and the mountainous relief that makes the access routes difficult. In Azerbaijan, the most vulnerable area has been located in the region of Goranboy in the central-western part of the country.
Figure 6. BTC pipeline vulnerability map with insets showing critical points in northeastern Turkey
3D Spatial analysis
In parallel to the geospatial analysis described above, a 3D infrastructure analysis has been carried out as well. Although it does not pretend to be a complete security analysis it can be regarded as a possible starting point.
The first step was to identify strategic infrastructure along the pipeline course by analyzing high resolution satellite imagery. For the purpose of the exercise, a pumping station located in northern Turkey was identified and chosen for analysis. The next step included the interpretation of the industrial facilities featured in the satellite imagery and the digitizing of the later in GIS software. The resulting map was later imported in a 3D editing software where, according to collateral data, height values have been assigned to the different facilities of the pumping station. These included: oil tanks, water treatment facilities, pressure reduction installations, coolers, antennas, watch towers, fences, etc. The resulting 3D model was later imported in a 3D visualization environment that featured GIS capabilities, i.e. line of sight, visibility areas, possibility of information overlay, etc. The final products included 3D animation and videos of the pumping station.
Figure 7. 3D security analysis
Security threats for BTC pipeline system
The information and communication system (ICS) used at the control station and both terminals is of vital importance for the success of the BTC operation. For economic and competitive reasons all of these operations occur under high time pressure, i.e., the potential for lax security or errors in implementing security procedures is significant. In addition, these valuable data are subject to the following security risks:
1. Tampering with technical data;
2. Theft of strategic data;
3. Intrusion into a valuable company database;
4. Undetected hacking into the system;
5. Data snooping;
6. Introducing malware into the system;
7. Simultaneous destruction of several communication towers.
The additional possibility of insider support for attacks on the ICS would enhance the damage potential significantly.
Figure 8. Detail of the Pumping Station 3D model
One of the most significant security risks for the BTC is represented by the insider threat. The operation of the BTC pipeline system involves the handling, production, storage, and transport of products, which can be toxic, flammable, combustible, or explosive. Malicious insiders have the capability to pose a major security threat for vital sectors of the BTC in view of their specific knowledge and access to various sectors and components.
In view of the design characteristics of the BTC pipeline (speed of oil through pipeline: 2 meters per second (m/s); designed to accommodate throughput of 1 million barrels per day (50 million t/year)) any major simultaneous attack on pipeline and valve stations could result in a significant environmental impact due to an uncontrolled oil spill, as well as a reduced supply to the tankers waiting to be loaded at the Marine Terminal in Ceyhan.
Threats to the marine terminal Ceyhan (Turkey)
The terminal houses seven crude oil storage tanks, a metering station, wastewater treatment, vapor incineration facility, and a jetty; the latter is capable of loading two 300.000 dwt tankers simultaneously (Fig. 4). Based on the experience gained from the global analysis of terror attacks, the following security threats need to be considered:
1. Covert night attack on the maritime terminal
2. Convoy of suicide truck/car bomb attack on the maritime terminal
3. Explosive-laden unmanned aerial vehicle (UAV) attack on the marine terminal
4. Suicide boat attack on jetty and/or tankers with several explosives laden high-speed boats
Figure 9. Detail of the Ceyhan oil terminal by satellite imagery (BTC trace is shown in black)
The next figure shows the resulting damage due to different attack modes. In the simulated scenario a car bomb would be used to blast through the security perimeter, followed closely by the truck bomb attacking the bank of onsite crude oil storage tanks. Oil storage tanks would suffer structural damage initially from the explosives, followed by a detonation due to flammable gases.
The next figure shows the resulting damage due to different attack modes. In the simulated scenario a car bomb would be used to blast through the security perimeter, followed closely by the truck bomb attacking the bank of onsite crude oil storage tanks. Oil storage tanks. Oil storage tanks would suffer structural damage initially from the explosives, followed by a detonation due to flammable gases.
Figure 10. Structural damage (overpressure iso-curves) resulting from the detonation of a convoy of suicide truck bomb/car bomb attacking the Maritime Terminal Ceyhan (Turkey)
1. Whilst an attack on the BTC pipeline system can never be excluded, it is possible to reduce the probability of such an event by the precautionary actions outlined below.
2. Countermeasures are comprised of a technical security system and a system of guards.
3. The technical system relies on a CCTV system and a computerized leak detection system, analyzing pressure drop, detecting acoustic waves transmitted in case of a leak.
4. The guard system foresees military-supported security surveillance, consisting of patrols, CCTV and specially trained soldiers.
5. The pipeline sections outlined in the geospatial analysis of the GIS under similar risk conditions as the section where the PKK sabotage took place should receive consideration for effective countermeasures.
Countermeasures should prioritize areas of higher risks as determined by the geospatial analysis in the GIS (as listed in section 3D Spatial analysis above). In order to identify a possible threat or hostile action against the BTC pipeline or the related infrastructure a combination of state of the art technology, combined with a strategic and tactical concept, is mandatory. A threat assessment and a risk analyses shall govern the security concept and plan of action. The security concept shall cover all events or incidents either man made or accidental. Countermeasures, damage and disaster management are to be developed based on the results of the ADA.
The Onion skin principle (OSP)
The OSP has been developed to protect critical infrastructure by the combination of all kinds of security elements, technologies and strategies as an integral overall concept. The level of reconnaissance, identification, prevention and action shall increase towards the distance to the critical infrastructure (BTC pipeline). The OSP is also considering advanced monitoring and sufficient warning and reaction time. A combination of all available support including the communities and the law abiding public as an integral security approach shall result in an adequate and cost effective protection of the critical infrastructure, the operator, the surrounding environment and population.
Security hardware and technology
• Real time satellite imagery as an overall umbrella with software assisted evaluation module combined with experienced tactical evaluators.
• Local CCTV cameras, sensors, perimeter lines, trenches and protective structures.
• BTC pipelines to be protected by location
• Hardening of the BTC pipeline and inside protection of the pipeline
• Fine-tune of GIS with the development of relevant risk datasets should receive consideration as the integrating system of the technology above.
• Reconnaissance by security service, police, military and civil authorities.
• Patrolling and use of UAV’s, aircrafts, ground patrols, animals and public.
• Reaction and intervention force on various levels.
• First responders, communities pipeline operator and supplier.
• Back ground checks and preventive action against insider attacks and black mailing.
Protection of pipelines and its infrastructure is a rather complex task. The cross country nature of a pipeline over an extreme long distance combined with hot spots (pumping stations, terminals, etc) and the transnational interest in pipeline integrity requires an integrated approach at all levels. Key factors for the secure operation are:
• Political stability, international screening and reconnaissance, national and international intervention in time combined with long term protection if necessary;
• Application of the Onion Skin Principle of various interconnected layers (screening - reconnaissance - intervention), combined with a cost effective, integrated threat and incident management shall govern the overall security policy and strategy of the BTC pipeline
• The segmentation of the pipeline into several risk sectors, as a result of the geospatial analysis in the GIS, provides a good indication as to where the prioritization of countermeasures should be enforced
• The most critical link in the defense strategies are the IFs alarm and reaction time due to the large remote areas covered by the BTC. The protection of the related infrastructure can be achieved with rather conventional security solutions
• Only real-time 24 hours satellite imagery integrated in a custom-designed GIS application together with state of the art sensor technology and software combined with the various layers of fast Intervention Forces (IF) shall protect the BTC adequately and efficiently. IF need to be effective, highly mobile (sea, air, land capability), supported by local first responders and engineering capability provided by BTC management.
This article’s contributors are Friedrich Steinhäusler and Peter Furthner of the University of Salzburg, Division of Physics and Biophysics, Salzburg, Austria and Antonio de la Cruz, Bogdan Palade and Pedro Soares of the European Union Satellite Centre (EUSC), Madrid, Spain