Navigating to net zero: How Cargill unlocks carbon-neutral transport on AWS

Cargill Ocean Transportation (OT), one of the world’s largest transporters of dry and bulk cargo, manages a fleet of approximately 650 chartered vessels, completes over 4,500 voyages annually, and makes thousands of port calls to more than 700 ports. Their cargo primarily consists of iron ore, coal, grain, fertilizer, and sugar, though they have the capability to transport over 100 different dry bulk commodities.

In 2018, Cargill launched its global CO2 Challenge to help decarbonize the shipping sector. Since then, Cargill has raised the bar on reducing carbon emissions and made a commitment to achieve net-zero emissions by 2050.

When new regulatory requirements for emissions reporting were beyond the capabilities of their existing manual processes, Cargill decided to build a solution to satisfy both present and future requirements. Traditional SaaS offerings were not fit-to-purpose or able to address Cargill’s data, which was siloed in many different repositories. Available ERP systems could not match Cargill’s needs, because business and regulatory requirements changed more quickly than the ERP systems could adapt. Cargill needed a flexible and automated means for connecting, integrating, governing and utilizing data from disparate sources, for complying with the new regulations, and for optimizing business and supply chain operations.

As Amazon Web Services (AWS) and Cargill had a long-standing relationship with several, successful large projects deployed and AWS’s AI offerings were well suited to solve these problems, Cargill chose to build with AWS.

In this post we describe the journey from idea to business and environmental results. We will explore how Cargill and AWS were able to build Cargill’s OT DataCloud and business applications to drive regulatory, financial and sustainability results.

Working Backwards from Cargill’s requirements.

AWS proposed analyzing business problems and available data streams and to work backwards from Cargill OT’s challenges toward solutions in a Data-Driven Everything (D2E) Workshop. Cargill and AWS began preparation by collecting data on seven emissions use cases, ranging from reporting to decision support. Cargill decided to focus first on the Carbon Intensity Indicator (CII) and Energy Efficiency Operating Indicator (EEOI) use cases because they were best suited to build the initial data foundation which would be needed for the other use cases. EEOI is a metric used to measure how efficient a vessel is at transporting freight. It is measured in tons of CO2 equivalent emitted per mile. CII is a measure of how efficient a ship is without respect to cargo. CII divides ships into five categories from most efficient (A) to least efficient (E). EEOI is a market-driven requirement and CII is mandated by the MARPOL Annex VI and the Act To Prevent Pollution From Ships. Cargill’s long-term vision is to create meaningful environmental and economic value for their customers. Cargill will do this by expanding access to more sustainable shipping solutions, harnessing AWS’s powerful AI and machine learning capabilities to streamline document processing and boost operational efficiency.

Solution

Overview

During the Data-Driven Everything Workshop, Cargill and AWS scoped out a scalable data foundation that could support numerous use cases and new regulatory requirements. Cargill and AWS built the use cases to maximize reuse of the underlying data. Today, Cargill calls this data foundation the “OT (for Ocean Transport) DataCloud.” The diagram below illustrates Cargill’s conceptual approach.

Carbon Intensity Indicator for vessel efficiency

After the Data-Driven Everything (D2E) Workshop, Cargill began work on the first solution (CII). Within four months Cargill had established the OT DataCloud and built the CII application that classifies Cargill’s ships into five categories based upon their efficiency in order to abide by applicable regulations and allow customers to use the efficiency of the carrying ship as guide to achieving their own sustainability goals. The CII application simplifies regulatory compliance and empowers Cargill and Cargill’s customers to make more sustainable shipping decisions. It also enhances operational visibility and positions Cargill as a trusted, sustainability-focused logistics partner. The diagram below shows the various data sources needed for the CII application on the left, the business logic layer in the center and the data consumption layer on the right. Cargill utilized Amazon S3 to hold the raw data, AWS Glue for ETL, and AWS Lambda to run the CII application. Decision support applications ran against Snowflake on AWS. This initial deployment produced regulatory compliance reports and helped Cargill to make business decisions.

Energy Efficiency Operating Indicator Use Case

Before Cargill deployed the EEOI use case, it took their staff two weeks to produce one EEOI report. Today the EEOI application runs daily with no manual intervention and replaces a manual, spreadsheet-based process of determining a ship’s energy efficiency. Cargill currently saves 1000+ person-hours per year with this application. EEOI also established the data foundation for data-driven business decisions such as making trade-offs between the carbon footprint of a shipping route versus the monetary costs of the route. All officially published data on the emissions footprint of Cargill Ocean Transport’s fleet is generated from the EEOI application. Every year in March/April the process and data are audited by an external firm.

As you can see from the next diagram, Cargill built upon the foundation of the CII use case to add Machine Learning, EEOI and other use cases to the architecture. The applications ran on AWS Lambda and AWS Glue provided ETL functionality. Amazon Textract provided new AI/ML document processing capability for Charterparty processing.

Data driven decisions and Customer-Facing Data Use Cases

Cargill’s next goal was to enable their internal and external customers to use this data to drive sustainability and business decisions. Vessel operators and traders use the analytics services to analyze cost and carbon footprint trajectories, identify trends and anomalies, and perform comparisons/ benchmarks to manage decarbonization mitigations. These results are essential to provide a clear, data-driven path toward Cargill’s goal of achieving zero carbon shipping by 2050.

By putting greenhouse gas (GHG) data into broader context, Cargill empowers their customers to make informed decisions that support their sustainability goals and creates transparency around shipping-related footprints to drive down overall carbon emissions.

Cargill built upon the previous architecture and added new data sources, new AWS AI/ML models and opened the system for customer access. In this iteration Cargill added Amazon SageMaker AI for bespoke AI/ML models Amazon Simple Queue Service and Amazon Simple Notification Service for notifications and to enable microservices architectures with AWS Lambda.

Using AI/ML for Intelligent Document Processing

Cargill then built multiple use cases with AI/ML and Intelligent Document Processing to optimize a vessel’s time in port (Laytime) by electronically analyzing data from shipping documents (like a vessel’s Statement of Facts document and relevant Counterparty Contracts). Performing this analysis manually was not practical due to the large number of documents required and the large number of ships Cargill manages. Cargill’s primary reasons for this use case are to save the ship operators money and to improve the utilization of their vessels. For the Laytime use case Cargill built upon the previous architecture and added a new application for Laytime processing that helps their customers determine the optimal time in port for vessels so that they can plan turnaround time appropriately. The core services and most of the data sources were in place before, but additional use cases and data sources for Laytime calculation are new to this architecture.

If a vessel is in port longer than planned, significant demurrage charges can accrue, and the resultant delays may cause the operator to make time up enroute and therefore increase the carbon footprint of the journey.