Autonomous Control Apps
with effortless data streaming with seamless ML integration with scalable edge deployment with built-in safety guardrails
Build Intelligent Control applications using simple, open tools.
A powerful easy-to-use platform
Uncover the core elements of our platform: fast app development, large-scale edge deployment, and safe autonomous control.
Contextualized Data
Stay in sync with your existing internal systems with a flexible approach for modeling assets and data streams. Import using CSV or programatically through the API.
Development Speed
Simple to use Python SDK to develop and test control apps that can seamless integrate with your machine learning models, rules, or physics-based logic.
Safe Control
Enable autonomous control actions on your edge equipment with real-time guardrails, full audit trail and integration with existing infrastructure.
Seamless Data Streaming
Easily connect and stream data to apps from a wide range of industrial protocols, including OPC UA, ROC, ModBus, MQTT, SQL, and OSI PI.
Offline Operation
Deploy your apps close to your edge equipment for continuous 24/7 operation, free from connectivity concerns. We securely store data and sync it once it's back online.
Flexible Deployment
Kelvin runs on any commercial off-the-shelf hardware, any cloud provider, or on-premises. With Kelvin, you have complete control over your data, ensuring privacy and security.
SmartApps Delivering
Production Uplift
-
Kelvin platform helped a partner ISV deploy and scale their AI model for PCP optimization
-
Kelvin platform helped scale SME best practices and AI model across 1000s of Wells
-
SmartApps enabled contextual decisions by analysing 100s of variables per Well, in real time with, 98%+ engineer approvals
Annual Production Uplift per Well (+1,100 wells)
SmartApp Driven Actions Over 12 Months
ROI in less than 12 months
Oilfield optimization applications
delivering multi-million dollar return
-
Kelvin enabled development, deployment and orchestration of ‘Smart Apps’ across global distributed fleet - 60+ sites, growing to 100+
-
Provisioned real-time contextualized operational data to the SmartApps from heterogenous equipments at Well-Site
-
Enabled Remote and ‘Offline’ operations through secured edge-cloud orchestration and on-demand application mgmt. tools
Annual Reduction in Opex
Reduction in Tech Team Field Visits
Reduction in Operational Downtime
Greater productivity from
existing assets and personnel
-
Plunger velocity targeting (PVT) and non-arrival response (NAR) applications enabled operators to ‘manage by exception’
-
Automated production optimization of wells in steady state operation
-
Automated recovery of wells from non-arrival events
Increased Production
Reduction in Field Costs
ROI on an Annual Basis
Kelvin control driving safer, more productive
and more efficient operations
-
Implemented model orchestration to prioritize optimization strategies
-
Monitoring and management to prevent concurrent plunger arrivals on large multi-well pads
-
Individual well optimization subordinated with a combined effect of maximizing production and minimizing downtime
Reduction in Methane Emitting Events
Increase in Production Volume
Reduction in Field Costs
Supervised Control
Trusted control through app recommendations
Apps provide customized control recommendations that seamlessly integrate with current operational workflows, enabling transparent and secure decision-making for equipment control.
Feedback Loop
Improve app performance with user annotations
How it works
Develop your SmartApp in minutes
Pure Python simplicity — effortlessly stream data into your model and generate control recommendations using simple primitives.
kelvin app create
import asyncio
from kelvin.application import KelvinApp
from kelvin.krn import KRNAsset
from kelvin.message import Recommendation
async def main() -> None:
app = KelvinApp()
await app.connect()
# Receive streaming data from your assets
async for asset_name, df in app.rolling_window(count_size=100).stream():
# Replace with your model predictions
result = model.predict(df)
# And generate Control Recommendations
await app.publish(
Recommendation(
type=result.recommendation,
description=result.description,
resource=KRNAsset(asset_name),
control_changes=[],
)
)
if __name__ == "__main__":
asyncio.run(main())
Deploy at scale with Kelvin Cloud
We handle everything for you — from integrating with your existing infrastructure and streaming contextualized data to apps, to making secure control actions on your edge equipment.
- Full Platform Access
- Tailored Support
- No Credit Card
GitHub
Speed up development with App Templates
Develop in minutes by leveraging the pre-built code samples for common control use cases.
Machine Learning
A Scikit-learn SmartApp Template for showing how to do Machine Learning inference.
Multi-Objective Optimization
A template for applying multi-objective optimization techniques to improve efficiency in the paper-making process, balancing multiple objectives simultaneously.
Event Detection
A template for seamless integration of MLFlow with Databricks, enabling efficient deployment of machine learning models.