Physics-driven Tignis software is helping companies around the world improve the reliability and efficiency of their connected mechanical systems. Our use of digital twins, modeling, machine learning (ML), and a physics engine provides earlier warning of degradation, higher quality alerts, fewer false positives, and faster issue identification and resolution.
But in an era of Big Data and streaming sensor analytics, it takes more than just well-designed software to achieve the performance, scalability, and usability necessitated by industrial plants. To explain how our solution is architected, Tignis released a webinar titled “Augmented Intelligence for Connected Systems.” We believe that sharing our rationale for the underlying architecture will help explain the power, performance, and scalability behind the Tignis condition monitoring and analytics solution.
What follows is a summary of the presentation, previewing the details provided the webinar embedded below.
Webinar title: Augmented Intelligence for Connected Systems
Speaker: Jon Herlocker, President and CEO – Tignis
Functional by Design
The costs of unplanned downtime at large industrial plants can run into the millions of dollars every day. While condition monitoring sensors and analytics provide a rich opportunity to improve plant reliability and efficiency, common obstacles make it challenging to implement and optimize the technologies. For instance, sensor installation can be expensive and a safety risk. Once installed, someone must create and fine-tune rules for each condition being monitored and update the rules as systems change.
Tignis’ cloud-based analytics platform, designed to augment human intelligence concerning asset and system conditions, avoids obstacles by:
- Automatically cleansing and normalizing data
- Automatically generating and testing models
- Automatically monitoring and learning mechanical systems
- Continuously detecting anomalies and trends, even in diverse and complex systems
- Precisely identifying and predicting operational impacts and generating root cause evidence
Being an all-inclusive software-as-a-service (SaaS) solution, the information technology, data science and management, and domain expertise are all handled by Tignis – not the plants. Tignis works with sensors already installed, often saving the plants from having to purchase and install more.
Only two sets of existing data are required for deployment: the customer’s piping and instrumentation diagram (P&ID) library and their data historian with collected sensor data. From the P&ID schematics, 2D digital twins of the systems are created to visualize the asset and system connectivity and flows. Components are dragged and dropped from our proprietary components library and corresponding sensor data is overlaid on the digital replica.
Once live, Tignis continuously monitors huge quantities of condition data from the sensors; cross-checks sensors around and inside problem components to minimize false positives; uses ML to learn characteristics unique to the plant; and identifies system performance issues and conditions requiring predictive maintenance. Tignis users benefit from automatic highlighting of affected areas in the digital twin, display of the most relevant sensor data, mouseover capabilities for added detail, and automatic chart rendering just by scrolling.
As a result, conditions such as degrading system efficiency, faulting or mis-calibrated sensors, errors in control programming, leaking valves, residue in pipes, clogged filters, worn bearings, or worse, can be proactively resolved.
At the platform’s core is the InfluxDB time-series toolkit from InfluxData. Tignis chose this open-source database for its high ingest and storage of time-series data from industrial sensor readings as well as its ability to easily send data for predictive analytics.
InfluxDB is mature technology with a strong user community that came recommended by trusted advisors. It has an intuitive query API, strong documentation, a solid supporting software ecosystem, and requires only occasional RAM adjustments.
Tignis is deployed on a multi-tenant Kubernetes architecture to ensure strong, secure separation between each of Tignis’ customers. The open-source Kubernetes system automates deployment, scaling, and management of containerized applications.
Each new Tignis customer/tenant gets a target Kubernetes cluster with its own separate set of containers and persistent disks. The application resides on the containers and the persistent data resides on Azure managed disks. InfluxDB runs from the container and is configured via Kubernetes.
A graph database stores the system’s objects (chiller, pump, etc.) along with their connections, metadata, and nodes with defining properties (object name, brand, model, etc.).
The Tignis infrastructure as designed has crucial advantages:
Deployment ease: New tenants can be deployed within minutes on the Tignis platform.
Scalability: Scaling to keep up with new and growing customers is straightforward: create a Kubernetes cluster, add containers, nodes, and disks, and increase the amount of RAM if necessary.
Intelligence speed: InfluxDB and Influx Query Language (InfluxQL) enable high-speed interactive analytics and fault diagnostics. For example, an ad hoc user query about a sensor can trigger downsampling of years of historical data and rendering of a thousand points on a chart – all with sub-second response time.
System optimization: The interactive, responsive user interface means less system downtime because anomalies are found and resolved more quickly. It also encourages users to dig deeper into the data and make more-informed decisions.
Flexibility: Tignis can deploy on any Kubernetes cluster with persistent volumes – even the customer’s own Kubernetes cluster should that approach make sense. InfluxDB prevents cloud platform lock-in, meaning it can be moved from the current Azure Cloud to another cloud platform with little effort. Additionally, Tignis can deploy its entire application suite on a developer laptop or a CI/CD machine, improving productivity.
We invite you to view the webinar above for further details as well as challenges, lessons learned, and new Tignis projects. Or, give us a call. We would be happy to demonstrate how Tignis can work for you.