Discover how spiderAI™, your assistive copilot for workflows related to condition monitoring, fault diagnosis, reliability engineering, maintenance, and repairs, can bring groundbreaking improvements in decision-making accuracy, productivity, and efficiency.
We are currently releasing it for vibration analysis, allowing maintenance and condition monitoring engineers to independently generate diagnostic reports with world-class expert-level accuracy in just one-tenth of the time previously needed.
Precisely diagnosing the fault and identifying the root cause within 5 minutes of conversing with spiderAI™
In Chemical Processing Plants, particularly in the catalytic cracking process where gas recycling is vital for maintaining reaction conditions, the failure of the gas recycle fan can halt the entire chemical production process.
During a routine inspection, the fan exhibited an unexpected increase in vibration levels in the 3-4x RPM band on the motor’s NDE horizontal reading. If left unaddressed, this could have led to a complete plant shutdown. Given the initial observations, the fault could be anything.
This conversation with spiderAI™ demonstrates how it precisely diagnosed the fault and identified the root cause using its subject matter expertise and mathematical reasoning, thereby preventing a potentially catastrophic failure.
Workflow
Routine Inspection and Initial Observation
During a routine site visit, the engineer observed abnormalities in the 3-4x RPM band on the motor’s NDE horizontal reading. Alarmed by the potential issues and the ticking clock, he realized the urgency to diagnose the root cause of the problem accurately before it could cause significant disruption.
Starting Conversation with spiderAI™
With a preliminary suspicion of a mechanical fault, the engineer initiated a conversation with spiderAI™, entering his initial observations. His curiosity was aroused as he aimed to pinpoint the specific type of mechanical fault.
spiderAI™ Requests for More Observations
With the limited observations, spiderAI™ also showed suspicions of few mechanical faults just like an expert would do. Instead of giving an immediate diagnosis, spiderAI™ asked for more specific observations, similar to how a doctor inquires further from a client.Recognizing that detailed information would lead to a more accurate diagnosis, the engineer quickly gathered and shared the additional data, also noting the observations he was unable to collect.
Diagnosis by spiderAI™
In a concise and precise manner, spiderAI™ delivered the diagnosis, revealing the root cause behind the observations. The fault it identified was unexpected, leaving the engineer stunned. The revelation was so surprising that it prompted the engineer to delve deeper into the matter.
Response from Engineer
Surprised, slightly fearful, and intensely curious, the engineer reasoned why this particular fault was identified over other possibilities. He wanted to understand the reasoning behind spiderAI™’s diagnostic conclusion.
Reasoning by spiderAI™
spiderAI™ provided a thorough explanation, detailing the science and mathematical validation with formulae and calculations. Through this communication and reasoning-driven workflow, the engineer was convinced of the root cause. Confident in the accuracy of the diagnosis, he made the correct decision on the necessary maintenance.
Greeting Each Other
The conversation concluded with the engineer and spiderAI™ exchanging greetings, having successfully resolved the issue.
spiderAI™ Thread
Fault Detected: Broken Rotor Bar
Failure Mode Cracked rotor bars
Effects:
- Increased electrical resistance
- Uneven magnetic field distribution
- Reduced motor efficiency
- Elevated temperatures
- Excessive vibration
- Potential secondary damage to bearings and windings
Comparing spiderAI™ with GPT-4o
The evaluation of spiderAI™ was conducted to compare its performance with the state-of-the-art model GPT-4 as of June 2024. A vignette, containing a detailed fault profile with various attributes of the fault, was used as the ground truth for evaluation. Conversations were masked to ensure unbiased assessment. Three key metrics—Precision, Plausibility, and Helpfulness—were used to evaluate the responses. Each conversation was carefully reviewed and scored on a scale of 0 to 5 for these metrics. Scores were compiled into a comparison table, and the findings highlighted strengths and areas for improvement for each model. This approach ensured a fair, comprehensive, and objective evaluation of the models' performance.
| Metric | spiderAI™ (Score) | GPT-4o (Score) | Reasoning |
|---|---|---|---|
| Precision (Identifying Fault) | High (5) | Medium (3) | spiderAI™ directly identifies rotor bar fault based on symptoms such as harmonics from 1X to 8X RPM and modulation sidebands at 40 CPM. GPT-4o lists rotor bar fault as a possibility but explores other options. |
| Plausibility (Explanation) | High (5) | Medium (3) | spiderAI™ explains the fault with clear technical reasoning, including harmonics, modulation sidebands at PPF, and the regular beat in the time waveform. GPT-4o provides explanations for various possibilities but lacks depth on specific fault mechanisms. |
| Helpfulness (Best Course of Suggested Action) | High (5) | Medium (4) | spiderAI™ is assistive in nature, asking for the right observations such as exact RPM, additional frequency peaks, amplitude data, phase data, and historical data. It also offers a detailed and practical course of action, including inspecting rotor bars, conducting MCSA, and monitoring vibration levels. GPT-4o provides helpful suggestions but lacks specificity. |
Overall Rating:
spiderAI™: High (15/15)
GPT 4o: Moderate (10/15)
GPT 4o Thread
For more updates and information, please contact us at spiderai@minto.ai. We're here to assist you with any inquiries and keep you informed about our latest developments.
Accurately Diagnosing Electrical Discharge Damage in a Hammer Mill Bearing within Minutes of Consulting spiderAI™
In Bio-Ethanol Plants, particularly in the mashing area where wheat is milled for the bio-ethanol process, any malfunction in the hammer mill can lead to significant production delays and financial losses.
During a monthly inspection, the hammer mill’s motor drive end (DE) bearing exhibited an alarming increase in vibration levels, reaching 69 G’s. Spectral data revealed increased activity at both DE and NDE bearings, with harmonics at 3.05 orders of motor turning speed frequency. Without prompt intervention, this issue could escalate, causing severe operational disruptions.
This conversation with spiderAI™ highlights how it swiftly diagnosed the bearing fault and identified the root cause using its expert knowledge on machines and failure modes, preventing a potential machinery failure.
usecase schema/ workflow
Initial Observation and Routine Inspection
During a routine inspection in the mashing area of a Bio-Ethanol Plant, the engineer observed that the peak-to-peak G levels at the motor drive end (DE) bearing of the hammer mill had increased to 69 G’s. Spectral data showed increased activity at both DE and NDE bearings with harmonics at 3.05 orders of motor turning speed frequency in radial and axial directions.
Starting Conversation with spiderAI™
The engineer initiated a conversation with spiderAI™, providing initial observations, including motor RPM, motor type (inverter-driven, variable speed), machine type (hammer mill), drive train, and bearing information.
Diagnosis by spiderAI™
spiderAI™ diagnosed the fault as likely being with the outer race of the SKF 6322 bearing, suggesting potential electrical issues. spiderAI™ recommended inspecting the bearings and electrical components to confirm.
Further Inquiry and Clarification
Upon inspection, the engineer discovered periodic grooves on the bearing's outer race. He queried spiderAI™ for clarification.
Explanation by spiderAI™
spiderAI™ explained that the grooves indicated fluting caused by electrical discharge due to poor grounding or VFD issues. It recommended installing insulated bearings, verifying grounding, and considering shaft grounding devices.
Implementation and Conclusion
The engineer installed insulated bearings and verified proper grounding, following spiderAI™'s recommendations. spiderAI™ also advised implementing a monitoring program for early detection of future issues.
Final Exchange
The engineer thanks spiderAI™ for the precise advice, ensuring the reliable operation of the hammer mill. spiderAI™ offered further support for any future inquiries.
spiderAI™ Thread
Fault Detected: Bearing outer race
Failure Mode: Electrical discharge machining (EDM) of bearing surfaces.
Effects: Bearing outer race damage leading to increased vibration and noise
Cause: High-frequency currents generated by the VFD.
Comparing spiderAI™ with GPT-4o
The evaluation of spiderAI™ was conducted to compare its performance with the state-of-the-art model GPT-4 as of June 2024. A vignette, containing a detailed fault profile with various attributes of the fault, was used as the ground truth for evaluation. Conversations were masked to ensure unbiased assessment. Three key metrics Precision, Plausibility, and Helpfulness were used to evaluate the responses. Each conversation was carefully reviewed and scored on a scale of 0 to 5 for these metrics. Scores were compiled into a comparison table, and the findings highlighted strengths and areas for improvement for each model. This approach ensured a fair, comprehensive, and objective evaluation of the models' performance.
| Metric | spiderAI™ (Score) | GPT-4o (Score) | Reasoning |
|---|---|---|---|
| Precision (Identifying Fault) | 5 | 3 | spiderAI™ directly identifies the outer race defect of the SKF 6322 bearing, matching the root cause described in the vignette. GPT-4o mentions multiple possible causes, reducing precision. |
| Plausibility (Explanation) | 5 | 4 | spiderAI™ provides a detailed explanation linking observed harmonics and electrical erosion to the defect. GPT-4o gives a plausible explanation but considers various issues. |
| Helpfulness (Best Course of Suggested Action) | 5 | 4 | spiderAI™ gives specific recommendations for inspecting the bearing and addressing electrical erosion, including installing insulated bearings. GPT-4o provides useful suggestions but lacks specificity. |
Summary:
Overall, spiderAI™ scores higher across all metrics due to its precise identification of the fault, detailed and plausible explanation, and specific, actionable recommendations. GPT-4o, while offering a comprehensive analysis, includes multiple potential issues and lacks the same level of precision and specificity, resulting in slightly lower scores.
This revised evaluation aligns more closely with the emphasis on precision in identifying the specific fault.
Overall Rating:
spiderAI™: High (15/15)
GPT 4o: Moderate (11/15)
GPT 4o Thread
For more updates and information, please contact us at spiderai@minto.ai. We're here to assist you with any inquiries and keep you informed about our latest developments.
Use Cases Coming Soon!
We are currently building exciting new use cases to enhance your experience. Stay tuned for updates!
For more updates and information, please contact us at spiderai@minto.ai. We're here to assist you with any inquiries and keep you informed about our latest developments.
Use Cases Coming Soon!
We are currently building exciting new use cases to enhance your experience. Stay tuned for updates!
For more updates and information, please contact us at spiderai@minto.ai. We're here to assist you with any inquiries and keep you informed about our latest developments.
Use Cases Coming Soon!
We are currently building exciting new use cases to enhance your experience. Stay tuned for updates!
For more updates and information, please contact us at spiderai@minto.ai. We're here to assist you with any inquiries and keep you informed about our latest developments.