Leveraging AI in Oil and Gas: The East African Crude Oil Pipeline (EACOP)
Introduction
Artificial intelligence (AI) continues to transcend society and industry as novel developments and applications are created and implemented. Amid ongoing debate on how technology may disrupt industries, this article aims to assess how AI can increase efficiency across functions of the East African Crude Oil Pipeline (EACOP); An intergovernmental collaboration between the Republic of Uganda and the United Republic of Tanzania.
The landscape of the EACOP project undergoes a transformative evolution with the strategic infusion of AI. Professionals across various sectors must convene to mitigate the externalities and ensure perpetual maintenance of the pipeline with a focus on safety, risk management, refinery processes optimisation and carbon emission reduction. According to a survey conducted by Ernst & Young, 58% of oil & and gas leaders believe the pandemic has increased the need for investment in digital technology. With a focus on the 1,400 kilometres pipeline from Kabale in Hoima Uganda to the port of Tanga in Tanzania, this report explores the specific applications of AI across the entire petroleum value chain, while addressing multifaceted challenges and offering strategic insights. This paradigm shift aligns seamlessly with the global energy industry’s trend towards AI integration, positioning the EACOP project at the forefront of innovation.
Applications Across the Petroleum Value Chain
Upstream Optimisation
Lake Albert serves as the oil source for the EACOP Project. By leveraging subsurface data analysis, governments and organisations gain efficient access to and exploration of oil storage.
Drawing inspiration from pioneering collaborations like Total S.A and Google Cloud, incorporating AI technologies, such as computer vision and natural language processing (NLP) emerges as a pivotal strategy for optimising oil exploration and production processes
Midstream Innovations
The adoption of AI solutions for monitoring across the EACOP project can mean various things. By observing large sets of data, developing proxy subsurface reservoir simulations is made simpler through the application of machine learning (ML). The essence of ML simulations lies in training models on historical data to provide accurate approximations. Once these ML models are trained on a specific dataset, they can be utilised for prediction for other related tasks within a similar domain significantly making the entire ecosystem more versatile.
Within the framework of the agile reservoir model, there are plenty of tools on the market for this specific purpose with integrated cross-department collaboration ability to ensure all specialists easily align on decisions after assessing insights from the data with a more centralised source of truth observed across verticals.
The EACOP’s project’s commitment to environmental guidelines is evident in its use of technology, including fibre optic cable. Serving as robust tools for both the Ugandan and Tanzanian governments, these cables play a dual role as instruments for massive data collection and transmission. Their capabilities extend beyond conventional monitoring, encompassing temperature detection and the early indication of potential oil and gas leakages. Notably, these fibre optic cables not only underpin precise predictions but also empower real-time monitoring of the pipeline, aligning seamlessly with the project’s dedication to environmental sustainability and safe protocols.
Downstream Sustainability
According to Joseph Mukasa, the environmental specialist tasked with the EACOP, there’s a stringent focus on protecting the biodiversity of the wetlands along the pipeline by following regulations including the International Finance Corporation’s (IFC) Performance Standard 6 (PS6) on Biodiversity Conservation and Sustainable Management of Living Natural Resources. ML tools, manifested in Convolutional Neural Networks(CNN), may be trained on satellite imagery of the pipeline providing a proficient solution to pinpointing early oil spill detection whilst including data surrounding the terrain. CNN models can analyse imagery during normal conditions in real-time and alert necessary departments once an outlier image, a potential oil spill, is observed. Earth/satellite observation is seen as a more cost-effective source of data due to the multiple sources covering optical, hyperspectral and synthetic aperture radar of varied resolutions. The collected data may be beneficial in ensuring that the AI monitoring systems are comprehensive and meet the specific criteria required by all necessary regulatory bodies.
Large industrial projects worldwide with numerous moving parts and third parties along the supply chain need to ensure they have robust emergency response plans, rigorous operational resilience frameworks, risk assessments and cybersecurity solutions in place. According to the National Policy of Disaster Preparedness and Management (NPDPM), the Albertine Graben is a seismically prone area of Uganda. What tools are in place to ensure that such natural disasters do not disrupt the BAU (Business as usual) operations within the EACOP project? Automated Disaster recovery plans and testing should be made mandatory for business continuity perhaps by using ML tools that utilise Long Short-Term Memory (LSTM) networks or Gated Recurrent Units (GRUs), which are typically used to predict the location and time of earthquakes. LTSM networks and GRUs have proven to establish mapping rules from complex patterns which makes them an essential tool for identifying features of data to identify the time and location of earthquakes using seismic activity as input.
Addressing Challenges for Seamless Integration
Human Rights and Environmental Issues
Since its inception, the EACOP project has met strict opposition from the international community, with concerns raised regarding human rights and environmental issues. Notably, environmental protests under the banner “STOP EACOP” gained momentum, leading to dissent within Uganda. These tensions have reduced as shareholders demonstrated transparency in outlining sustainable practices. An investment towards ML and AI solutions and integration would demonstrate a greater devotion to environmental sustainability. Such an investment would not only showcase a profound commitment to environmental responsibility but also signal an acknowledgement of the project's vital role as critical infrastructure for both the Tanzanian and Ugandan governments.
Pipeline Cybersecurity
This project's significance extends beyond environmental concerns, encompassing substantial socio-economic benefits. The EACOP aims to contribute to job creation, revenue generation, technology transfer and enhancement along the Uganda-Tanzania trade corridor. As key players in the region with controversial international reputations, these countries face the plausible threat of nation-state aggressors seeking to destabilise and disrupt the project’s completion and operation. Henceforth, it is paramount that while looking to increase reliance on digital solutions, the cybersecurity of the EACOP meets the upper tier. Instances of rogue cyber hacktivists targeting critical infrastructure such as the 2021 Colonial Pipeline ransomware attack in the United States, emphasise the critical need for robust cybersecurity measures. The repercussions of such attacks can be severe, as witnessed by the complete shutdown of the Colonial Pipeline, disrupting the flow of essential fuels along the East Coast.
Large Initial Investment
However, beyond security, there are initial bottlenecks to adopting automation and AI-powered solutions. Implementing automated solutions will require a significant initial investment in hardware, software, and infrastructure. With the project currently being backed worth USD 2 billion, another USD 3 billion may be needed in the form of loans, we’re estimating a larger investment may be necessary to adopt the technology. At the start of the final quarter of the year 2023, large commercial banks had opted out of working on the project, fears of the projected 379 million tonnes of CO2 equivalent emissions over the project's lifetime raise financial, legal and reputational risks. As per regulations, commercial banks in Uganda do not have the capacity to facilitate the required capital investment and require international financiers. International financiers are scared of the environmental risks but also the political risks of Uganda. Nevertheless, by adopting these AI tools, financiers create a greater guarantee of safety and environmental conservation and protection which may sway investors to help reduce these initial costs.
Another pitfall the two governments may encounter is a lack of infrastructure and legal framework surrounding AI technology. While AI remains a potent driver for improving the pipeline, the digital landscape in Uganda and Tanzania remains in its infancy. With a lack of infrastructure and guiding policy to help manoeuvre adopting AI solutions, there is likely to be local and international scrutiny questioning the safety and ethical development and application of these tools along the pipeline.
As alluded to from the onset, the debate surrounding AI applications consistently highlights concerns regarding job security and a general reluctance in the workforce among both employers and employees to embrace these technologies. Overcoming this challenge requires a thoughtful onboarding process. Employers and employees will require initial training and clear communication to ensure that AI tools are perceived as complements, not replacements, offering reassurance about job security.
This onboarding process should focus on educating the workforce about the collaborative nature of AI technologies. In the context of the EACOP project, AI tools and ML models are designed to assist petroleum engineers. They serve as aids, coupling patterns and relations discovered in the data by the model, to the industry knowledge provided by engineers. This strategic approach fosters a coordinated integration of AI, transforming apprehensions into a workforce empowered by advanced modern tools.
If not met with scepticism, the issue of blind acceptance remains apparent. It must be made clear that whilst the relationships and predictions the ML models discover are precise, they are only reflective of the quality of data collected. It is paramount that there are necessary systems in place to ensure precise data collection that adheres to the six dimensions of data quality to avoid inaccuracies that could lead to bias within the AI models used.
Regarding AI policy, Uganda not having a clear policy framework can hinder mutual international collaborations due to a lack of regulatory guidelines. It's paramount that Uganda ensures there is public trust and acceptance regarding the use of emerging technologies to foster a holistic approach that aligns with the nation's technological advancement goals.
Conclusion
In embracing AI as a catalyst for innovation and sustainability, the EACOP project has the potential to redefine industry standards. By strategically addressing challenges and prioritising ethical, safety and environmental considerations, the collaboration between Uganda and Tanzania can set a precedent for responsible AI adoption in large-scale infrastructure projects. The EACOP project, by integrating AI technology, will stand testament to AI for socially responsible sustainable development. As the project progresses the careful balance between innovation and ethical responsibility will shape its legacy and contribute to the broader discourse on AI in critical infrastructure particularly in this area of the world.