PREDICTIVE ANALYTICS AND MONITORING
A key to an operating plant’s success is the ‘dependability’ and ‘predictability’ of its machines. An enormous amount of time and resources are spent on maintaining plants and yet a significant number of shutdowns are attributable to unforeseen trips and failures. With increased machine or process complexity, failure diagnosis has also become more complex.
Our state-of-the-art, cloud-based predictive analytics and monitoring platform – Turbomechanica®, has been conceptualized with decades of turbomachinery and facilities expertise. The Smart Monitoring and Anomaly detection software platform, developed by engineering and machine learning experts, can easily be plugged into existing data historians or condition monitoring systems. Historical operating data is utilized to build asset-level AI models which continue to learn and evolve with real-time data. Our multi-variate analysis approach provides improved accuracy of predictions, longer lead times to failure, and scalability.
Our platform utilizes data from multiple (tens or even hundreds of) sources to identify anomalies and flag sub-optimal operations based on the underlying relationships among variables.
The analytical sophistication of the platform is further enhanced with the integration of physics-based performance models that augment the AI models providing richer insights into machine performance and health.As part of our domain-driven approach towards the analysis of complex industrial machinery, we have pioneered the use of simulation-based transfer learning methodologies in the field of machine learning to dramatically reduce the volume and variety of historical data required. We have also built-in algorithms that can target specific use cases such as performance degradation, machine vibrations, dry gas seal health, motor insulation health, gas contamination, and several others. This targeted approach has allowed us to detect machine issues that had previously gone undetected by other remote monitoring and diagnostic providers on the same machines.
Our models accurately predict/detect:
- Decrease in machine performance
- Decrease in machine efficiency
- Increase in machine emissions
- Machine faults well ahead of time (several months of lead time in most cases)
Examples of Model Accuracy:
- Equipment performance models such as head, efficiency, power, work input (typical accuracy +/- 2-3%)
- Vibration models (typical accuracy +/- 10%)
- Dry Gas Seal models (typical accuracy +/- 5%)
- Motor stator winding models (typical accuracy +/- 1-2%)
REMOTE MONITORING AND TROUBLESHOOTING
With Mechademy’s remote monitoring and diagnostics services, operating companies can leverage the technical expertise and experience of our SMEs to accelerate fault diagnosis. Over the span of their careers, our SMEs have successfully identified and resolved numerous complex technical issues and companies tap into this rich knowledge base when they contract our services.
We offer a great amount of flexibility in the way we structure our remote monitoring services. From small retainers for monthly consulting and occasional plant visits to comprehensive early fault detection, remote diagnosis, and technical recommendations service, we help customers develop a plan that is best suited to their requirements.
Mechademy’s remote monitoring services are strongly recommended as a supplement to customers who sign up for our predictive analytics platform. Each anomalous event is further evaluated to determine the cause and technical recommendations are provided to prevent reoccurrence and future failure.
BENEFITS:
- Prescriptive actionable insights and alerts
- Deployment on any facility equipment regardless of manufacturer
- No additional hardware requirements
- Easily supplements existing monitoring systems
- Increased equipment uptime
- Increased energy efficiency
- Increased asset lifetime and capital value
- Reduced downtime
- Enhanced operations and maintenance planning
- Reduced maintenance cost
- Reduced operational cost
- Overall enhanced revenues, reliability, competitive advantage, and reputation.
Even before an equipment goes live on the Turbomechanica® platform, our analysis of historical operating data has successfully identified issues such as equipment faults, poor controls, and sub-optimal operations.
Some pain points that we help our customers solve:
Centrifugal Compressors:
- Performance degradation such as internal leaks due to seal wear, erosion, corrosion, and fouling.
- Gas contamination and liquid carry-over.
- Control issues such as improper load sharing, poor lube oil pressure/temperature control.
- Instrument failure.
- Vibration issues such as unbalance, misalignment, rubs, issues related to sub-synchronous or super-synchronous vibrations (oil whip/whirl, fluid flow-based excitation).
- Dry gas seal issues such as seal face degradation, seal ‘hang-up’, dry gas seal contamination, control issues with primary, secondary, or tertiary seal gas.
Gas Turbines:
- Gas turbine real-time power utilization.
- Gas turbine degradation (fouling).
- Fuel gas system changes such as changes in Modified Wobbe Index or heating value.
- Control issues such as poor cooling water temperature control, poor lube oil pressure/temperature control.
- Vibration issues such as internal rubs, oil varnishing issues, misalignment.
- Combustion system issues such as hot spots, liquid carry-over in fuel gas.
Steam Turbines:
- Performance degradation related to corrosion, stress corrosion cracking, erosion, and increased internal leakages.
- Gland steam sealing health.
- Steam governor performance monitoring
- Vibration issues such as unbalance, misalignment, rubs, looseness, cracks.
Centrifugal Pumps:
- Performance degradation such as internal losses due to greater wear ring clearances, losses across side and center bushings, corrosion.
- Process fluid contamination.
- Instrument failure.
- Vibration issues such as unbalance, misalignment, rubs, fluid flow excitation, cavitation.
- Head rise for variable fluid density applications wherein the fluid density is unknown.
Centrifugal Expanders (Gas and Liquid):
- Performance degradation such as internal losses, wear, recirculation, two-phase flow.
- Performance optimization for maximum power, enthalpy recovery.
- Vibration issues such as improper thrust equalization, bearing and wear ring rubs, looseness.
- Control issues such as poor speed control or output power control.
Reciprocating Compressors:
- Performance degradation due to leaks, erosion, corrosion, fouling, and valve failures.
- Detection of mechanical failures such as valve failures, cylinder leakage, damaged piston rings, damaged cylinder packing and overloading.
- Solids, & gas contamination.
- Liquid carry-over (process, water).
- Degradation of inter or after coolers.
- Vibration issues related to poorly designed pulsation suppression systems, unbalanced crankshaft, loose parts, degraded valves, improper lubrication.
- Poor controls and instrument failure.
Screw Compressors:
- Performance degradation due to leaks, erosion, corrosion, and fouling.
- Overloading due to excess compression ratio.
- Solids & gas contamination.
- Liquid carry-over (process, lube oil, water).
- Degradation of inter or after coolers.
- Vibration issues related to lube oil contamination, loose parts, degraded valves, improperly lubricated bearings.
- Poor controls and instrument failure.
Induction and Synchronous Motors & Generators:
- Vibration issues such as unbalance, soft foot, asymmetric air gaps, rotor skew, rubs, looseness, bearing faults.
- Motor stator winding insulation health (hot spots, winding degradation)
Control Valves:
- Mechanical degradation such as blockages, valve stem issues, excess hysteresis, poor operating characteristics.
- Performance issues such as flashing, cavitation, choke flow performance.
- Valve thrust and position monitoring with early fault identification
PROOF OF CONCEPTS:
We encourage our customers to do a Proof of Concept where our platform can be deployed on one or two pieces of facility equipment at very minimal cost so that the benefits of the platform can be realized and appreciated without having to do a full-scale facility-wide deployment.
