How Polymathian uses Amazon ECS Anywhere to optimize underground mine production in near real time

The complexity inherent to operating a successful underground mine is easily missed by those who have never visited one—after all, the sprawling maze of tunnels and workings are hidden deep below the Earth’s surface. These mines are often located in geographically remote areas, making it difficult to get personnel and equipment onsite at short notice, a problem that was highlighted during the COVID-19 pandemic.

Orchestrating the movement of equipment and personnel underground while ensuring a safe operating environment that meets production targets is a daunting task. Space is often so limited that, in most mines, there isn’t enough room for equipment to pass other than in specially designed pullouts, and two machines meeting in the wrong place can result in a significant amount of wasted time to clear the congestion.

In an effort to improve their operations, miners have adopted technology across various stages of the mining process, such as mapping ore bodies and coordinating equipment operation or maintenance. Though these systems work well, they give miners a limited understanding of both the current global operating state of the mine and how their decisions affect the future state. As such, optimizing overall operations is a complex task that requires consideration of hundreds of inputs from multiple systems in near real time.

Solving problems at scale with industrial mathematics

Polymathian specializes in the application of mathematics to solve complex problems for industry. Its blend of advanced analytical expertise and deep industry experience is known as industrial mathematics, and the company uses it to solve problems like improving mining efficiency in surface and underground mines.

ORB is part of Polymathian’s suite of advanced optimization tools that applies this approach to both surface and underground mining. It can be used as both an advanced near-real-time short-interval control tool or to simulate and optimize across various mine planning horizons, from the short-term through to life-of-mine.

As a short-interval control tool, ORB performs a mathematical optimization of every dispatch decision made, taking into account the current state of the mine along with desired objectives and goals. Data from systems such as planning, fleet management, geological modeling, and maintenance systems is considered to ensure the optimal decision is made, given the state of the mine at that time.

For planning, ORB is initialized with the current mine state to create a digital twin that’s then used to simulate events such as draw-point hang-ups or breakdowns, generating a number of best-to-worst-case scenarios of predicted performance. As the planning horizon increases, the challenge becomes more complex—a long-term mine plan can result in a schedule that includes tens of thousands of anticipated tasks over decades, all while taking into account hundreds of constraints like ventilation, drilling capacity, hoist performance, mill throughput, and so on.

The mathematical techniques used to solve these problems are compute intensive and benefit greatly from the scalability and elasticity that the cloud offers. Polymathian built ORB to take advantage of the cloud by using a container-based serverless architecture, ensuring performance while minimizing administrative overhead.

The challenge

Accessing ORB through a standard web browser gives Polymathian and its customers flexibility in how the system is deployed. Typically, a small data-collection application is installed in the customer’s environment to securely transfer the required data into the Amazon Web Services (AWS) Cloud for analysis. Though the majority of Polymathian customers prefer a fully-managed cloud-based ORB deployment, some have a need to run on premises, often due to the unreliable or limited nature of connectivity to remote sites.

In either deployment model, ORB components were packaged as Docker containers, with orchestration performed locally using custom scripts written using a third-party tool. Polymathian would rely on virtual private network (VPN) access to its customer’s network for the ongoing management and support of the application stack. This approach led to significant time and effort being wasted due to difficulties with VPN reliability, delays in procurement of virtual machines, incorrect or limited permissions, and lack of expertise with the container-orchestration tool.

Polymathian needed a solution to support customers hosted in the cloud or on premises, all while using common development and orchestration tools. No small ask!

Moving optimization to the edge

Because the majority of ORB deployments were hosted in AWS, Polymathian already had significant expertise with Amazon Elastic Container Service (Amazon ECS), a fully managed container orchestration service. It approached AWS to see if this expertise could also be used to manage on-premises deployments, and Amazon ECS Anywhere, a feature of Amazon ECS that lets you run and manage container workloads on your infrastructure, was quickly identified as a potential solution.

Amazon ECS Anywhere builds upon the simplicity of Amazon ECS to provide a consistent tooling and API experience for running container-based applications that are hosted on customer-managed infrastructure. An Amazon ECS Anywhere deployment involves installing an agent on on-premises servers, which is linked securely back to the in-region Amazon ECS control plane. Amazon ECS Anywhere then takes care of container management and orchestration in both the cloud and on premises.

The containerization strategy for ORB was revised to use more native Docker functionality, allowing Polymathian to use a single set of tooling and common governance to manage both cloud and customer deployments. VPN access to customer networks, and the associated burden of managing credentials, was removed, and standardizing on Docker meant that recruiting developers with the required skills was easier.

“Before Amazon ECS Anywhere, we would have to navigate the complexities of customer log-ons and virtual environments every time we wanted to deploy a new version” explains Polymathian’s head of sales and marketing Louis Okada. “Now, this needs to be done only once during the Amazon ECS Anywhere setup, allowing developers who are not as familiar with client on-prem servers and operating systems to run deployments and updates, which improves our responsiveness to client change requests”.

Polymathian can now meet demand from clients, whether they’re hosted on premises or in the cloud, while minimizing administrative overhead and improving the customer experience.

Where to next for ORB?

ORB brings an innovative approach to the mine planning and optimization space, one that will only become more critical as automation and autonomy are more widely adopted throughout the industry. By using the power of the cloud while responding in near real time at the edge, ORB will ensure that the right decision is made at the right time—every time.